(PDF) Dutch disease-led de-industrialization in the Azerbaijan’s economy: analysis of the chemicals industry

i

PHD DISSERTATION

Ibrahim Niftiyev

Szeged, 2022

ii

UNIVERSITY OF SZEGED FACULTY OF ECONOMICS AND BUSINESS

ADMINISTRATION DOCTORAL SCHOOL IN ECONOMICS

DUTCH DISEASE-LED DE-INDUSTRIALIZATION IN THE AZERBAIJAN’S

ECONOMY: ANALYSIS OF THE CHEMICALS INDUSTRY

Phd Dissertation

Supervisor:

Author:

Prof. Dr. Miklós Szanyi

Ibrahim Niftiyev

Institute of Finance and International

Economic Relations

Division of World Economics and

European Economic Integration

University of Szeged

Faculty of Economics and Business

Administration

Doctoral School in Economics

Szeged 2022

iii

DECLARATION

Candidate’s Declaration

I (Ibrahim Niftiyev) hereby declare that the dissertation of this PhD studies is my

original research work and was written by myself under the guidance of my supervisor.

This dissertation has not been presented elsewhere for any other degree, award or

professional qualification. There is no collaborative or jointly owned work in this

thesis, whether published or not. I duly cited all the external sources, published works

(i.e., books, research papers, reports) and data sources in the given place in the text.

Candidate Signature: … .............. Date: 2022.08.26

Name: Ibrahim Niftiyev

Supervisor’s Declaration

I, Miklós Szanyi, hereby declare that the dissertation written by Ibrahim Niftiyev,

entitled (title of the thesis) is his own written work prepared under my supervision.

Based on its professional merits, I support its submission.

Principal Supervisor’s Signature:

Name: Prof. Dr. Miklós Szanyi Date: 2022.08.16

iv

LIST OF PUBLICATIONS

The dissertation is based on the following papers of the author:

Niftiyev, I. (2020). Dutch Disease Symptoms in Azerbaijan Economy. Journal of

Economic Cooperation and Development, 41(3), 33–66.

Niftiyev, I. (2020). The De-industrialization Process In Azerbaijan: Dutch Disease

Syndrome Revisited. In Proceedings of the 4th Central European PhD Workshop on

Technological Change and Development (p. 357–396). Szeged: Faculty of Economics

and Business Administration, Doctoral School in Economics, University of Szeged.

Niftiyev, I. (2021). Dutch Disease Effects in the Azerbaijan Economy: Results of

Multivariate Linear Ordinary Least Squares (OLS) Estimations. HSE Economic

Journal, 25(2), 309–346. DOI: https://doi.org/10.17323/1813-8691-2021-25-2-309-346

Niftiyev, I. (2021). How to Conceptualize the Resource Curse and Dutch Disease

Theories in the Case of Azerbaijan? Initial and Negative Findings. SSRN Electronic

Journal. Available at: https://ssrn.com/abstract=3857936. DOI:

https://doi.org/10.2139/ssrn.3857936

Niftiyev, I. (2022). A comparison of institutional quality in the South Caucasus: Focus

on Azerbaijan. In Proceedings of the European Union’s Contention in the Reshaping

Global Economy (pp. 146–175). Szeged: University of Szeged, Faculty of Economics

and Business Administration, Doctoral School in Economics. DOI:

https://doi.org/10.14232/eucrge.2022.9

Ibadoghlu G., & Niftiyev I. (2022). A Retrospective Analysis of the Azerbaijani

Economy During 30 Years of Independence. Post-Soviet Issues, 9(1), 58–76. DOI:

https://doi.org/10.24975/2313-8920-2022-9-1-58-76

v

Ibadoghlu G., & Niftiyev I. (2022). An assessment of the thirty year post-Soviet

transition quality in Azerbaijan from an economic and social liberalization perspective.

Journal of Life Economics, 9(3), 129–146. DOI: https://doi.org/10.15637/jlecon.9.3.02

Niftiyev, I. (2022). Principal component and regression analysis of the natural resource

curse doctrine in the Azerbaijani economy. Journal of Life Economics, 9(4), 225–239.

DOI: https://doi.org/10.15637/jlecon.9.4.02

vi

ACKNOWLEDGEMENTS

Doctoral studies have been an important part of my life since 2017. This

dissertation and my other completed research would not have been possible without

certain people and institutions that have accompanied, supported, and guided me on this

long but life-changing journey.

First and foremost, I would like to thank Prof. Dr. Miklós Szanyi, from whom I

learned what it means to be a good social scientist. I am very grateful to him because

after consultations and discussions with him, I was able to shape my research and

achieve scientific results. Second, Prof. Dr. Éva Voszka (Director of the Doctoral

School of Economics), who has always guided young PhD students in their research

work through invaluable research workshops at the Faculty of Economics and Business

Administration. Third, the overall academic environment at the University of Szeged,

where research culture meets real challenges, will forever remain in my memory given

the numerous professors, lecturers and academic staff. Thanks to this special research

and academic culture, which encourages the ambition of budding economists and social

scientists, I was able to mature as a social scientist and economist. Finally, I would like

to thank Mr. David P. Curley (University of Szeged, member of MTA-SZTE Research

Group on Artificial Intelligence) and Mr. Ben McKechnie for proofreading this

dissertation and their useful suggestions.

Next, I would like to thank Prof. Dr. Beáta Farkas, who was always deeply

interested in my progress, and Dr. Gábor Dávid Kiss, who often and without hesitation

helped me in both professional and personal matters. The Dean of the Faculty of

Economics and Business Administration, Dr. Péter Kovács, and all the members of the

Institute of Finance and International Economic Relations have been extremely helpful

and supportive throughout my PhD studies, for which I cannot thank them enough.

Finally, I greatly appreciate the evaluations and feedback from Dr. Gábor Dávid Kiss

and Dr. Judit Ricz on my dissertation.

My special appreciation goes to my parents, who have always supported me in

my life choices, both morally and materially, whenever I needed them. Without the

tireless moral support of Mr. Muslum Niftiyev and Mrs. Kamala Niftiyeva (Aghayeva),

it would have been almost impossible to complete this dissertation and my other

published scientific works. I would like to thank my aunt, Tarana Bakirli, for the

countless conversations she had with me to motivate me throughout the dissertation.

Thank you for believing in me and always being by my side.

vii

I cannot thank enough my close friends Kamran Javadov, Huseyn Huseynli,

Agil Yusifli, Vusal Ahmadov and Ulvi Shakhikhanli who supported me and believed in

me. One of my best friends, Eldar Aliyev, who gave his life for his country in 2020,

never hesitated to check in with me about my progress and achievements.

I would like to thank the Hungarian government for granting me the Stipendium

Hungaricum and the Dissertation Scholarship. I would also like to thank the Ministry of

Education of the Republic of Azerbaijan for recommending me for the scholarship

program. My sincere thanks also go to the Center on the European Economy (Mr.

Elkhan Richard Sadik-Zada), Azerbaijan State University of Economics for the help and

support during my PhD studies. Without the support of these institutions, I would never

have been able to start and finish my studies. Thank you very much for everything! I

will always have the best memories of this part of my life.

viii

ABSTRACT

While the oil boom played a positive role in Azerbaijan’s economy by reducing poverty

and increasing overall prosperity, the country’s industrial structure became lopsided and

favored oil and gas production. Typically, the contraction of non-oil tradeable sectors

(e.g., manufacturing) in oil-rich countries places an economic and social burden on

society. The manufacturing sector can stimulate innovative actions triggered by policy

decisions and improve the political and institutional environment. To determine the

reasons for the one-sided industrial production, it is crucial to explain Azerbaijan’s

economy with sound theories and elucidate the largest economic challenges. Thus, the

Dutch disease (DD)-related de-industrialization of non-oil manufacturing is the main

focus of this dissertation. DD is a form of de-industrialization when natural resources

are discovered or commodity prices skyrocket, resulting in unexpectedly high revenues.

This situation shifts the attention of resource-rich governments from non-commodity

production to commodity exports, motivating them to spend natural resource revenues

within a short period of time. However, not all de-industrialization of a particular non-

oil sector is likely to be due to an oil boom. The general institutional environment and

human capital also play critical roles. DD is a useful theory for explaining why in a

small, open, and oil-rich country like Azerbaijan, non-oil manufacturing de-

industrialization has occurred in parallel with the oil boom since 2005/06.

For the process of de-industrialization to occur, several conditions must be met.

In countries rich in natural resources (e.g., oil and natural gas), the natural resource

curse (NRC) and DD may be among the reasons. In advanced countries, this is usually

due to productivity growth or globalization. It is therefore logical to analyze the

economy of Azerbaijan, which is an oil-rich developing country, within the framework

of the NRC and DD. For this reason, a novel, step-by-step explanation of non-oil

manufacturing de-industrialization was developed. For a targeted approach, the

chemicals industry was selected as a case study.

The first objective of this study was to identify whether the non-economic side

of the NRC, namely the institutional part, is indeed applicable to Azerbaijan’s economy.

This was analyzed using principal component analysis, ordinary least squares (OLS),

and dynamic OLS. The statistical methods seemed to confirm that the growth of the oil

industry has had a negative impact on the institutional quality and human capital of

Azerbaijan’s economy.

ix

The second objective was to examine DD effects. This was achieved by

applying unrestricted vector autoregression, linear multivariate OLS, and Bayesian

vector autoregression models. To capture the resource movement and spending effects

of DD in the national economy separately, aggregate models were first estimated.

The third objective was to determine whether the de-industrialization of specific

subsectors of the chemicals industry was indeed related to the oil boom. To achieve this,

quantitative and qualitative research methods were used. Thus, under a quantitative

approach, both short- and long-run linear models were used for estimation (since the

Johansen test revealed long-run cointegration between the variables of interest).

Stepwise, OLS, fully modified OLS, canonical cointegration regression, and robust least

squares were applied. The quantitative results supported the claim that oil- or DD-

induced de-industrialization has had a negative impact on chemical industry output in

Azerbaijan.

In addition, expert interviews helped to gather qualitative data from industry

experts and economists specializing in the industrial side of Azerbaijan’s economy. The

interview results not only supported the quantitative findings but also provided

additional factors and reasons for the de-industrialization of the chemical subsectors and

their ongoing re-industrialization. Examples include rent-seeking behavior that began

with the oil boom in 2005; human capital issues associated with labor supply; outdated

technology; and the inefficiency of current state-owned enterprises.

The final objective was to provide brief policy suggestions to counteract the DD-

related de-industrialization of the non-oil manufacturing sector in Azerbaijan’s

economy. This was achieved by bringing together the empirical results of this

dissertation and reviewing the literature on exchange rates as well as institutional and

industrial policies in different periods. The main policy proposal is to focus on

institutional and monetary aspects of the economy to overcome opportunistic behavior

and monetary pressures that threaten the competitiveness of non-oil production. The oil

boom is negatively affecting the non-oil sectors because of the low quality of

institutions and the inability of human capital to cope with the associated challenges.

The government's imprudent spending policies and the unrestrained real effective

exchange rate require prudent fiscal and monetary policies and developed financial

markets to properly manage the absorptive capacity of the Azerbaijani economy. In

addition, selective industrial policies combined with tariffs and subsidies can encourage

the private sector, and thus the non-oil sectors, to participate in the overall industrial

x

diversification process. These findings are of critical importance to government

agencies and official policy makers, who should base their decisions on well-designed

economic relationships among key sectors and economic indicators. Future research can

focus on other non-oil production sectors besides the chemicals industry and collect

qualitative data, as official statistics do not cover the necessary aspects to analyze oil- or

DD -induced de-industrialization since 2005 and 2006.

 
xi 
TABLE OF CONTENTS 
 
DECLARATION ............................................................................................................. iii 
Supervisor’s Declaration .................................................................................................. iii 
LIST OF PUBLICATIONS ............................................................................................. iv 
ACKNOWLEDGEMENTS ............................................................................................. vi 
ABSTRACT ................................................................................................................... viii 
CHAPTER 1 ..................................................................................................................... 1 
RESEARCH BACKGROUND ........................................................................................ 1 
1.1. Research problem and research gaps ..................................................................... 1 
1.2. Problem statement and study purpose ................................................................... 5 
1.3. Justification of the study ........................................................................................ 6 
1.4. Research objectives ................................................................................................ 9 
1.5. Research questions ................................................................................................. 9 
1.6. Conceptual framework and hypotheses ............................................................... 10 
1.7. Dissertation outline .............................................................................................. 13 
CHAPTER 2 ................................................................................................................... 15 
UNDERSTANDING AZERBAIJAN’S ECONOMY .................................................... 15 
2.1. Azerbaijan’s economy since independence ......................................................... 15 
2.1.1. Macroeconomic stability and stimulation policies ....................................... 17 
2.1.2. Institution Building ....................................................................................... 24 
2.1.3. Liberalization of the economy ...................................................................... 29 
2.1.4. Foreign economic relations ........................................................................... 33 
2.1.5. Creation of a private economy ...................................................................... 36 
2.2. Macroeconomic overview of Azerbaijan’s economy .......................................... 42 
2.3. Summary of the chapter ....................................................................................... 45 
CHAPTER 3 ................................................................................................................... 47 
LITERATURE REVIEW ............................................................................................... 47 
3.1. Natural Resource Curse phenomenon .................................................................. 47 
3.1.1. Main concepts, key terms, and NRC measurement approaches ................... 49 

 
xii 
1.2. Main transmission channels and mechanisms in the NRC doctrine ............. 55 
1.2.1. Role of resource rents and their political and institutional dimensions . 55 
1.2.2. Public investments and the capacity to manage oil revenue .................. 61 
1.2.3. Human capital and education as pathways to the NRC ......................... 64 
Source: The author’s own construction based on the literature review. ................. 69 
1.3. The NRC in Azerbaijan ................................................................................ 69 
1.4. Criticism and limitations of NRC theory ...................................................... 76 
2. Dutch disease ....................................................................................................... 78 
2.1. Dutch disease in Azerbaijan ......................................................................... 84 
2.1.1. Early studies ........................................................................................... 84 
2.1.2. Direct investigations .............................................................................. 87 
2.1.3. Indirect investigations ............................................................................ 90 
3. De-industrialization ............................................................................................. 92 
3.1. Concepts of industrialization and de-industrialization ................................. 92 
3.2. De-industrialization in the case of Azerbaijan .............................................. 98 
4. The chemical industry at a glance ........................................................................ 99 
5. The chemical industry and Azerbaijan’s economy ............................................ 101 
6. Theoretical framework of this dissertation ........................................................ 106 
6. The topicality of NRC and DD for the time being ............................................ 109 
CHAPTER 4 ................................................................................................................. 112 
DATA AND METHODOLOGY .................................................................................. 112 
1. Data and methodology of the analysis of NRC in the Azerbaijan’s economy .. 112 
1.1. Dependent variables .................................................................................... 112 
1.2. Independent variables ................................................................................. 114 
1.3. Empirical strategy ....................................................................................... 115 
2. Data and methodology of the analysis of DD in the Azerbaijan’s economy ..... 117 
2.1. Data and variables of interest ...................................................................... 117 
2.2. Empirical strategy ....................................................................................... 121 
2.2.1. Empirical strategy for REER appreciation .......................................... 121 
2.2.2. Empirical strategy for sectoral implications of REER, NEER, oil prices, 
and oil rents ....................................................................................................... 121 
2.2.3. Empirical strategy for the resource movement effect .......................... 122 
2.2.4. Empirical strategy for spending effect ................................................. 123 

 
xiii 
3. Data and methodology of the analysis of de-industrialization in the Azerbaijan’s 
economy .................................................................................................................... 125 
3.1. Quantitative data ......................................................................................... 125 
3.2. Qualitative data ........................................................................................... 127 
3.4. Methodology of the quantitative analysis ................................................... 129 
3.5. Methodology of the qualitative analysis ..................................................... 131 
CHAPTER 5 ................................................................................................................. 133 
RESULTS ..................................................................................................................... 133 
1. The Analysis of Natural Resource Curse Theory in the Azerbaijan’s Economy
 .................................................................................................................................. 133 
1.1. Figure analysis ............................................................................................ 133 
1.2. Results of one-sample t-test and international comparison ........................ 135 
1.3. Results of PCA ............................................................................................ 135 
1.4. Results of the DOLS and OLS analyses ..................................................... 139 
1.5. Results of the WVS analysis ....................................................................... 141 
1.6 Summary of the section ............................................................................... 149 
2. Analysis of DD .................................................................................................. 151 
2.1. Appreciation of REER ................................................................................ 152 
2.2. Sectoral effects of EDI, REER, NEER, oil prices, and oil rents ................ 155 
2.3. The resource movement effect .................................................................... 161 
2.4. The spending effect ..................................................................................... 165 
2.5. Summary of the section .............................................................................. 166 
3. Oil-Led De-Industrialization of Non-Oil Manufacturing in Azerbaijan’s 
Economy: An Analysis of the Chemical Industry .................................................... 168 
3.1. Descriptive analysis .................................................................................... 169 
3.2. Econometric estimations ............................................................................. 174 
3.3. Results of the qualitative data analysis ....................................................... 179 
3.3.1. Early years and integration of Soviet technology into the chemical 
industry ............................................................................................................. 179 
3.3.2. Recent developments in the chemical industry .................................... 180 
3.3.3. Exchange rate ....................................................................................... 180 
3.3.4. Competitiveness ................................................................................... 182 
3.3.5. Labor .................................................................................................... 183 

 
xiv 
3.3.6. Oil industry .......................................................................................... 184 
3.3.7. Collapsed subsectors ............................................................................ 185 
3.3.8. Developed subsectors .......................................................................... 187 
3.3.9. Investments .......................................................................................... 188 
3.3.10. Production process ............................................................................. 190 
3.3.11. Problems and challenges .................................................................... 192 
3.3.12. Political and institutional environment .............................................. 194 
3.3.13. Re-industrialization of the chemical sector ....................................... 195 
3.4 Summary of the section ............................................................................... 196 
CHAPTER 6 ................................................................................................................. 200 
CONCLUSIONS AND POLICY IMPLICATIONS .................................................... 200 
1. Summary of the findings .................................................................................... 200 
2. Policy recommendations .................................................................................... 204 
2.1. Policy implications: Exchange rate and institutions ................................... 204 
2.2. Industrial policies ........................................................................................ 209 
3. A brief political economy perspective ............................................................... 216 
4. Limitations and recommendations for future studies ........................................ 219 
REFERENCES ............................................................................................................. 222 
APPENDIX ................................................................................................................... 273

 
xv 
LIST OF TABLES 
   
Table 1.1: Research hypotheses related to the natural resource curse (NRC) doctrine in 
Azerbaijan’s economy. ................................................................................................... 11 
Table 1.2: Research hypotheses related to Dutch disease (DD) syndrome in Azerbaijan’s 
economy. ......................................................................................................................... 12 
Table  1.3:  Research  hypotheses  related  to  the  de-industrialization  process  in 
Azerbaijan’s economy. ................................................................................................... 13 
Table 2.1: Selected economic indicators for Azerbaijan (1991–2020). ......................... 43 
Table  2.2:  Structure  of  Azerbaijan’s  economy  in  term  of  investments,  employment, 
industrial output, and trade (in %) .................................................................................. 45 
Table 3.1: A quick overview of the key literature examples. ....................................... 105 
Table 4.1: Dependent variables used in the study ........................................................ 113 
Table 4.2: Explanatory/independent variables used in the study. ................................. 114 
Table 4.3: Independent (explanatory) variables used in the analysis. .......................... 118 
Table 4.4: Dependent variables used in the analysis. ................................................... 119 
Table 4.5: Dependent and explanatory variables of the quantitative analysis. ............. 126 
Table 4.6: Descriptive statistics of the variables of interest for 1995–2020. ................ 127 
Table  4.7:  Technical  details  about  the  interview  process  and  interviewees:  Industry 
experts (Nos. 1–10) and economists (Nos. 11–16). ...................................................... 128 
Table 5.1: Kaiser–Meyer–Olkin (KMO) values and Bartlett’s test results. ................. 136 
Table 5.2:  Communalities  of the variables  related to  institutional quality and  the oil 
sector in Azerbaijan’s economy. ................................................................................... 137 
Table 5.3:  Component matrices of the principal  component analysis (PCA) related to 
institutional quality and the oil sector in Azerbaijan’s economy. ................................. 138 
Table  5.4:  Dynamic  ordinary  least  squares  (DOLS)  results  of  the  oil  factor  and 
institutional quality in Azerbaijan’s economy. ............................................................. 139 
Table  5.5:  OLS  results  of  individual  NRC-related  indicators  against  oil-related 
variables. ....................................................................................................................... 140 
Table 5.6: Importance of politics among Azerbaijani citizens (in %) .......................... 142 
Table 5.7: Interest in politics among Azerbaijani citizens (in %) ................................ 143 
Table 5.8: Intensity of political action in Azerbaijan, as measured by willingness to sign 
a petition (in %). ........................................................................................................... 143 

 
xvi 
Table  5.9:  Intensity of  political  action in  Azerbaijan,  as measured by  willingness to 
participate in a boycott (in %). ...................................................................................... 144 
Table 5.10: Confidence in political parties in Azerbaijan (in %) ................................. 145 
Table 5.11: Voting participation of Azerbaijani citizens in local and national elections 
(in %) ............................................................................................................................ 146 
Table 5.12: Level of confidence in the government among Azerbaijani citizens (in %).
 ...................................................................................................................................... 147 
Table 5.13: Confidence in labor unions among Azerbaijani citizens (in %) ................ 149 
Table 5.14: Confidence in the justice system and courts among Azerbaijani Citizens (in 
%) .................................................................................................................................. 150 
Table 5.15: The variance decomposition of DREER. .................................................. 154 
Table 5.16: VAR Granger causality/block exogeneity Wald tests and pairwise Granger 
causality tests between REER and oil prices, 1995M01–2020M12. ............................ 154 
Table  5.17:  FMOLS  and  CCR  results  of  the  analysis  of  oil  prices  on  the  REER, 
1995M01–2020M12. .................................................................................................... 155 
Table 5.18: Sectoral implications of REER, NEER, oil prices, and oil rents for 1990–
2019. ............................................................................................................................. 156 
Table 5.19: Sectoral implications of REER, NEER, oil prices, and oil rents, for 1990–
2019 (continued). .......................................................................................................... 157 
Table 5.20: Sectoral implications of REER, NEER, oil prices, and oil rents for 1990–
2019 (continued). .......................................................................................................... 158 
Table 5.21: Sectoral implications of EDI, REER,  NEER,  oil  prices,  and  oil  rents  for 
1990–2019. ................................................................................................................... 160 
Table 5.22: Resource movement effect  of Dutch  Disease in Azerbaijan’s  economy for 
2000–2018. ................................................................................................................... 162 
Table 5.23: Results of VAR Granger causality tests. ................................................... 164 
Table 5.24: Spending effect of Dutch Disease in Azerbaijan’s Economy for 2000–2018.
 ...................................................................................................................................... 165 
Table 5.25:  Trade value of subsectors of the  chemical industry (in current USD)  and 
share in total exports (in %) for 1997–2020. ................................................................ 174 
Table 5.26: Regression results for the production of caustic soda (solid form) for 1995–
2020. ............................................................................................................................. 175 
Table 5.27: Regression results for the production of chlorine for 1995–2020. ............ 176 

 
xvii 
Table 5.28: Regression results for the production of hydrochloric acid for 1995–2020.
 ...................................................................................................................................... 176 
Table 5.29: Regression results for the production of isopropyl alcohol for 1995–2020.
 ...................................................................................................................................... 177 
Table 5.30: Regression results for the production of liquid soda for 1995–2020. ....... 178 
Table 5.31: Regression results for sulfuric acid for 1995–2020. .................................. 178 
Table A4.1: Descriptive statistics of the variables of interest used in NRC studies. .... 273 
Table A4.2: Normality test, outlier and missing values of the variables of interest used 
in NRC studies. ............................................................................................................. 273 
Table A4.3: Descriptive statistics of the variables of interest used in DD studies. ...... 273 
Table A4.4: Normality test, outlier and missing values of the variables of interest used 
in DD study. .................................................................................................................. 274 
Table A5.1: Total variance explained of the variables related to institutional quality and 
the oil sector in Azerbaijan’s economy. ........................................................................ 275 
Table  A5.2:  Unit  root  test  results  (augmented  Dickey–Fuller)  of  the  principal 
components used in Chapter 3. ..................................................................................... 275 
Table  A5.3: Unit  root test  results (ADF)  of  REER  and  oil  prices  variables used  in 
section 5.2.1. ................................................................................................................. 276 
Table A5.4: The VAR optimum lag order selection criteria, 1995M01–2020M12. .... 277 
Table A5.5: VAR residual serial correlation LM tests. ................................................ 278 
Table A5.6: Unrestricted co-integration among the variables of interest related to the 
sectoral implication  of  REER, NEER, oil  prices,  and oil  rents, for the  period  1990–
2019. ............................................................................................................................. 278 
Table A5.7: Augmented Dickey–Fuller (ADF) unit root test of the variables of interest 
used in resource movement (employment) VAR model. ............................................. 279 
Table A5.8: VAR optimum lag length criteria. ............................................................ 280 
Table A5.9: VAR residual serial correlation LM tests. ................................................ 281 
Table A5.10: Variance decomposition of manufacturing employment (DMAN). ....... 281 
Table A5.11: Variance decomposition of CPI_ANNUAL_AVER, spending effect. .. 283 
Table A5.12: Unit root tests of the variables of interest. .............................................. 284 
Table A5.13: Correlation matrix of the variables of interest. ....................................... 284 
Table A5.14:  Johansen cointegration analysis of the subsectors of chemicals industry.
 ...................................................................................................................................... 284 
Table A5.15: Interview questions. ................................................................................ 286 

xviii

LIST OF FIGURES

Figure 1.1: Oil rents and institutional quality among the post-Soviet countries. ............. 7

Figure 1.2: Conceptual framework for the study. ........................................................... 11

Figure 3.1: The main transmission channels and results of NRC. .................................. 69

Figure 3.2: Effects of Dutch disease according to the original theory by Corden and

Neary (1982) and Corden (1984). ................................................................................... 81

Figure 3.3: Theoretical framework of the dissertation based on NRC and DD theories.

...................................................................................................................................... 107

Figure 5.1: Worldwide governance indicators for Azerbaijan, 1996–2020. ................ 134

Figure 5.2: Distribution of year-over-year average growth rates for institutional quality,

based on the development phases of Azerbaijan’s economy (index values). ............... 134

Figure 5.3: Comparison of institutional quality of Azerbaijan with non-resource post-

Soviet countries during the oil boom period (2005–2014). .......................................... 135

Figure 5.4: Scree plot of the variables related to institutional quality and the oil sector in

Azerbaijan’s economy. ................................................................................................. 137

Figure. 5.5: Component plot in rotated space of institutional quality and the oil sector in

Azerbaijan’s economy. ................................................................................................. 138

Figure 5.6: Democracy and government responsibility in Azerbaijan, 2017–2020. .... 148

Figure 5.7: Impulse response functions of the VAR model, 1995M01–2020M12. ..... 153

Figure 5.8: Impulse response functions of manufacturing employment in VAR models.

...................................................................................................................................... 163

Figure 5.9: Impulse response functions of the CPI used in the Bayesian VAR model. 166

Figure 5.10: Subsectors of the chemical industry that experienced output de-

industrialization for 1995–2020. ................................................................................... 170

Figure 5.11: Subsectors of the chemical industry that did not experience output de-

industrialization for 1995–2020. ................................................................................... 172

Figure 5.12: Exports of the chemical industry (1996–2020). ....................................... 173

Figure A5.1: Auto-Regressive (AR) characteristic polynomial inverse roots o the VAR

model. ........................................................................................................................... 277

Figure A5.2: Autocorrelations with approximate 2 standard error bounds for REER

appreciation ................................................................................................................... 278

Figure A5.3: Inverse roots of auto-regressive (AR) characteristic polynomial. ........... 280

Figure A5.4: Autocorrelations with approximate 2 standard error bounds. ................. 281

xix

Figure A5.5: All impulse response functions used in the resource movement VAR

analysis. ......................................................................................................................... 282

Figure A5.6: Auto-regressive (AR) characteristic polynomial inverse roots of the VAR

model for spending effect. ............................................................................................ 282

Figure A5.7: Autocorrelations with approximate 2 standard error bounds for spending

effect. ............................................................................................................................ 283

Figure A5.8: Codes and groupings of the qualitative analysis of the industry experts 285

Figure A5.9: Codes and groupings of the qualitative analysis of the economists ........ 285

Figure A5.10: Exports of chemical subsectors in the Azerbaijan economy, 1994–2020.

...................................................................................................................................... 287

Figure A5.11: Exports of chemical subsectors in the Azerbaijan economy, in thousand

AZN 1994–2020. .......................................................................................................... 288

xx

LIST OF ABBREVIATIONS

2SLS

Two-Stage Least Squares

ACG

Azeri, Chirag, and Guneshli

ADF

Augmented Dickey-Fuller

AIC

Akaike Information Criterion

AIH

Azerbaijan Investment Holding

ANB

Azerbaijan National Bank

AR

Auto-Regressive

ARDL

Autoregressive Distributed Lags

AZAL

Azerbaijan Airlines

AZN

Azerbaijan Manat

BICEX

Baku Interbank Currency Exchange

BR

British Petroleum

BTC

Baku–Tbilisi–Ceyhan

BTC

Baku-Tbilisi-Jeyhan

BVAR

Bayesian Vector Autoregressive

CBAR

Central Bank of the Republic of Azerbaijan

CEE

Central and Eastern Europe

CIE

Cancellation of the Inspections in Entrepreneurship

CIS

Commonwealth of Independent States

CJSC

Closed Joint-stock Company

CPI

Consumer Price Index

CPS

Cyber-Physical Systems

CUSUM

Cumulative Sum

CUSUMSQ

Cumulative Sum of Squares

DD

Dutch disease

DOLS

Dynamic Ordinary Least Squares

EBRD

European Bank of Reconstruction and Development

EDI

Extractives Dependency Index

EGDI

E-Government Development Index

EITI

Extractive Industries Transparency Initiative

EU

European Union

FDI

Foreign Direct Investments

xxi

FMOLS

Fully Modified Ordinary Least Squares

FSU

Former Soviet Union

GCI

Global Competitiveness Index

GDP

Gross Domestic Product

GNP

Gross National Product

GVC

Global Value Chains

HDI

Human Development Index

HDPE

High-Density Polyethylene

HQ

Hannan–Quinn

IBA

International Bank of Azerbaijan

ICTs

Information Communication Technologies

IDA

International Development Association

IFC

International Finance Organization

IFRS

International Financial Reporting Standards

IMF

International monetary Fund

IRF

Impulse Response Function

JBN

Jarque–Bera test

KMO

Kaiser-Meyer-Olkin

KPMG

Klynveld Peat Marwick Goerdeler

LLC

Limited Liability Company

LM

Lagrange Multiplier

MENA

Middle East and North Africa

MNCs

Multinational Companies

MPC

Marginal Propensity to Consume

MSMEs

Micro and Medium-Sized Enterprises

NGOs

Non-Government Organizations

NRC

Natural Resource Curse

OEC

Observatory of Economic Complexity

OECD

Organization of Economic Cooperation and

Development

OJSC

Open Joint-stock Company

OLS

Ordinary Least Squares

PCA

Principal Component Analysis

xxii

PP

Polypropylene

PU

Production Union

R & D

Research and Development

RCVI

Resource Curse Vulnerability Index

REER

Real Effective Exchange Rate

REER

Real Effective Exchange Rate

RLS

Robust Least Squares

SAC

Structural Adjustment Credit

SB

Booming sectors

SC

Schwarz Criterion

SCIP

Sumgait Chemical Industrial Park

SDGs

Sustainable Development Goals

SE

Forecast Error

SL

Lagging sectors

SME

Small and medium-sized enterprises

SNT

Non-tradeable sectors

SOCAR

State Oil Company of Azerbaijan Republic

SOE

State-Owned Enterprises

SOFAZ

State Oil Fund of the Republic of Azerbaijan

SSCRA

State Statistical Committee of the Republic of

Azerbaijan

SWFs

Sovereign Wealth Funds

TANAP

Trans-Anatolian Gas Pipeline

TAP

Trans Adriatic Pipeline

TFP

Total Factor Productivity

TOT

Terms of Trade

UK

The United Kingdom

UNEP

United Nations Environment Programme

USD

United States Dollars

USSR

Union of Soviet Socialist Republics

VAR

Vector Autoregressive

VECM

Vector Error Correction Method

VIF

variance inflation factors

xxiii

WGI

Worldwide Governance Indicators

WTO

World Trade Organization

WUA

Water User Associations

WVS

World Value Survey

1

CHAPTER 1

RESEARCH BACKGROUND

In this chapter, the research problem and gaps are introduced to conceptualize the

purpose and objectives of the current study. Then, the general conceptual framework is

presented. This includes the research questions and hypotheses that were established to

analyze the collected data as well as the theoretical frameworks used to analyze

Azerbaijan’s economy. Finally, an outline of the dissertation is presented.

1.1. Research problem and research gaps

Industrialization—particularly manufacturing—has helped developed countries to

achieve higher income levels and living standards, especially in the early years of their

development (Szirmai–Verspagen 2015). In addition, approximately 80% of developing

countries’ exports have shifted from commodity extraction to industrial production-

based goods since 1960 (Gelb 2010). This shift has created opportunities for low-

income countries to participate in global value chains (GVCs)

1

across a variety of

sectors. Manufacturing-centered economies increase the complexity of exports and

ensure economic prosperity for their citizens. A recent study also demonstrated that the

great importance of manufacturing as an engine of economic development has not

changed significantly since 1990 (Haraguchi et al. 2017). In other words, manufacturing

continues to contribute significantly to the wealth of nations. Yet, the nature of

manufacturing itself is changing rapidly in our fast-growing world. For example,

increasing digitization has affected the structure of manufacturing and business models,

as consumer markets have become smaller and more fragmented but also more

interconnected (Paulus–Rohmer 2016). For instance, Brettel et al. (2017) discussed the

term “Industry 4.0,” proposed by Schwab (2015), as a new term for conceptualizing

changes in the manufacturing landscape. In other words, the growth engine of the

modern economic world comprises big data-driven cyber-physical

2

systems that enable

1

“GVC” means value added in the process of design, production, marketing, distribution, and support for

the consumer of a product or service where firms and workers perform in inter-firm networks on a global

scale (Gereffi–Fernandez–Stark 2011).

2

Cyber-Physical Systems (CPS) basically mean the management of the physical processes via computer

algorithms and the concept of big data basically means the enhanced data collection techniques used in

day-to-day activities, storing the data in large and complex data sets that will be used in predictions and

forecasting.

2

modularization through knowledge-based service platforms to identify and satisfy

newly emerging consumer demands (Paulus–Rohmer 2016).

In parallel with the development of manufacturing in the world economy, which

creates high value-added, the role of the service sector has also increased in many

countries. However, without a well-developed industry, the additional benefit of

services may remain low. For this reason, since Kaldor’s (1960) initial theorization of

manufacturing as an engine of growth, much research has focused on industrialization

as an engine of growth and prosperity.

3

However, it is unlikely that every country will benefit from manufacturing

(Cantore et al. 2017). With the increasing importance of the globalization process and

the reintegration of former centrally planned economies into the world economy, the

emphasis since the late 1980s has been on the use of specialization and trade with

minimal adjustment costs (Pomfret 2001). The rise of GVCs for weakly industrialized

countries could be seen as a push to industrialize production. However, without

significant adjustments in domestic productivity, it is likely that less productive

countries will remain confined to low value-generating activities for an extended period

(Pahl–Timmer 2020).

While some mineral-rich developing countries have achieved economic

diversification, other developing countries and mineral- or primary sector-based

economies have been less successful; thus, they have received less attention in the

literature. Moreover, several economists have argued that long-term sustainable

development cannot be achieved in mineral-rich and weakly industrialized countries

because they deprive themselves of the positive externalities that manufacturing

generates (e.g., learning by doing; Magud–Sosa 2013; Matsuyama 1992; Hausmann et

al. 2007). This is why various researchers have consistently emphasized the need to

strengthen manufacturing and resource-independent industrialization to sustain

economic prosperity (McMillan–Rodrik–VerduzcoGallo 2014; de Vries–Timmer–de

Vries 2015).

3

Kaldor’s (1957; 1966) work stressed the importance of manufacturing in economic growth due to the

increasing returns that it provides, and it also underlies the importance of demand and the external factors

that influence it (see Pata and Zengin (2020), for an analysis of post-World War II England). Also,

Kaldor (1957) introduced the technical progress function, which characterizes how new industrial plants

increase labor productivity, offer intermediate and capital goods, and improve production techniques.

Lastly, Kaldor (1957) mentioned that a country’s industrial development boosts domestic demand and

external demand, and this also reinvigorates economic growth and development. With these ideas, the

economics scholars saw why industry has always been a priority for developed countries.

3

A country’s economic performance and well-being depend on its production and

exports (Hausmann et al. 2007). Likewise, Rodrik (2008) argued that rich countries

became rich because they produced innovative products and not just traditional goods in

abundance. The economic diversification of production activities enables improved

growth performance in the long run because it increases the economy’s sources of

income, contributes to job creation, and leads to balanced development between urban

and rural areas (Albassam 2015). Furthermore, the pursuit of mineral extraction and

export-oriented economic development creates dependency, which is associated with

negative impacts. Examples include real effective exchange rate (REER) appreciation

and rent-seeking behavior.

4

However, it is not easy for low-income and developing

countries to shift from traditional production to more diverse production habits to

increase value-added exports (Hidalgo et. al 2007). Typically, a rapid shift in

production methods requires complex policy implementation that leads to structural

changes and higher living standards (Hidalgo et. al 2007).

Azerbaijan is a small and oil-rich post-Soviet country. It has rich oil and gas

resources and an unbalanced economy; it underwent a painful transition process from a

command economy to a market economy in the early 1990s; and it has pursued

extractive industry-driven economic development since 1994. As a result, Azerbaijan

has a lopsided economic structure that is vulnerable to international commodity shocks,

as was the case from 2014 to 2015. The main driver of gross domestic product (GDP) is

the oil industry, and the country’s main exports are based on crude oil and petroleum

products.

Because of the country’s overdependence on oil exports and revenues,

Azerbaijan’s economy has been studied by some using the natural resource curse

(NRC)

5

doctrine and Dutch disease

6

(DD). NRC and DD are similar concepts, but they

also differ because NRC is a broad term for the theoretical branch of natural resource

management. However, the concept of NRC focuses precisely on the economic

4

The term rent-seeking was introduced by Ann Krueger. Rent-seeking is an economic term that describes

a situation in which an entity seeks to acquire wealth without providing anything in return. The term

"rent-seeking" is derived from the economic meaning of "rent," which is defined as economic gain

obtained through the skillful or possibly manipulative use of resources.

5

NRC refers to the side effects of economic performance of oil-, mineral-, and other resource-rich

countries compared to non-resource-rich countries. However, oil-, mineral-, and other resource-rich

countries are considered to have better opportunities to grow rapidly and develop their societies because

of available short-term revenue opportunities.

6

DD is an expression of the contradiction that arises when good news, such as the discovery of enormous

oil reserves, harms a country's overall economy. It may have its origin in a substantial influx of foreign

capital to exploit a newly discovered resource.

4

crowding out mechanisms and domestic inflationary pressures of booming sectors over

non-booming sectors. In other words, while DD reflects only the economic aspects of

NRC, NRC itself is not limited to economic issues of resource dependence, as it also

encompasses political, institutional, and governance dimensions. Furthermore, the other

aspects of the NRC and DD, such as deindustrialization and de-agriculturalization, have

not been studied in depth.

7

The literature still lacks unequivocal arguments on the

negative consequences of extractive industry-dominated economic structures in

Azerbaijan. This has resulted in both theoretical and practical gaps in guidance for

decision makers, policy makers, and academics in how to conceptualize Azerbaijan’s

economy within a widely accepted and tested theoretical framework. The strong and

positive relationship between Azerbaijan’s GDP and oil prices, or simply the

appreciation of the REER—as claimed in previous studies—is not sufficient to explain

why the NRC or DD is solely due to oil wealth. Undoubtedly, the NRC and DD theories

should be considered, but the production and export of difficulties in certain sectors of

the economy, such as non-oil tradeable sectors (e.g., industrial producers), should also

be analyzed. To develop effective and targeted industrial policies that reduce the

potentially harmful effects of an oil-based economy, new research should focus on oil-

related de-industrialization and industrial diversification in Azerbaijan. If the results

provide a clearer picture of the negative consequences of the oil industry’s dominance

in Azerbaijan’s economy, both scholars and policymakers would be able to more

effectively shape their policies.

When examining the economy of Azerbaijan, the mere assumption of oil

dependence and the need for diversification lead researchers to naively expect signs of

the NRC and effects of DD. However, the supposed final theoretical outcome of the

NRC or DD is the de-industrialization of nonbooming sectors in terms of output and

employment. Therefore, after testing the NRC doctrine and the DD model, this study

analyzed Azerbaijan’s economy in terms of the de-industrialization of the chemicals

industry. The aim was to more accurately capture the negative consequences of oil-

based economic growth and development.

7

This dissertation will mainly focus on de-industrialization; however, the agricultural sector is also an

economic priority, as it employs the highest share of the labor force in Azerbaijan. Therefore, some parts

of the quantitative analysis will include some data from the agricultural sector.

5

To date, only a handful of studies (mainly journal articles) have addressed the

de-industrialization process.

8

Troubling, yet-to-be-resolved issues are the stage at which

oil revenues were misdirected and why Azerbaijan’s economy has been unable to

develop non-oil manufacturing. Without solutions, research on Azerbaijan’s economy

will not be specific or relevant enough to provide comprehensive solutions for a more

diversified economic structure. In times of low commodity prices, the current industrial

structure certainly threatens national income, employment, and the monetary side of the

economy. Finally, Szirmai and Verspagen (2015) argued that former centrally planned

economies are underrepresented in studies that treat manufacturing as an engine of

growth. The present study therefore sought to fill this research gap by extending NRC

and DD studies on Azerbaijan.

1.2. Problem statement and study purpose

Lower manufacturing value-added and exports have been demonstrated to have adverse

effects on long-term sustainable economic development in mineral-rich countries, such

as Nigeria (Schubert 2006), Russia (Bogetic et al. 2006), and Ghana (Acquah-Andoh et

al. 2018). If institutional, political, and governance aspects of the economy fail in

addition to the economic crowding-out mechanisms of DD, a country is likely to

become dependent on a single commodity. This, in turn, will lead to a constrained

growth environment for noncommodity tradeable sectors, as one of the main factors

determining the competitiveness of a given economy is macroeconomic stability

(Khyareh–Rostami 2022). However, monetary pressures and procyclical fiscal policies

usually hinder the competitiveness of commodity-rich countries.

Industrialization is often argued to be an engine of balanced economic growth

and development, whereas the opposite (i.e., de-industrialization) is allegedly harmful

to a country. In any study on the impact of Azerbaijan’s oil industry, the country’s

postcommunist legacy, small size, and de facto oil wealth must be considered as part of

the background. Simply assuming that de-industrialization has been a negative

development since the collapse of the USSR would not be helpful for furthering

knowledge about Azerbaijan’s economy. In other words, resource-poor countries have

no other option than industrialization through their manufacturing sector. Meanwhile,

mineral-rich countries tend to use their available natural resources in the short term to

8

See Hasanov (2013) for a brief consideration of “relative de-industrialization” and Niftiyev (2020a) for

a more focused but descriptive argument against de-industrialization.

6

avoid painful reforms and changes. In Azerbaijan, the decreasing role of the

manufacturing sector and the related structural changes can be attributed to oil.

In this dissertation, the research design and main theses were based on the idea

that the de-industrialization process of Azerbaijan’s economy since 1995 has been an

extension of the NRC doctrine and its economic explanation (i.e., DD). This is not the

first study to examine the NRC and DD in the case of Azerbaijan; rather, it was

motivated by recent developments in Azerbaijan’s economy, such as decreased GDP,

devaluation of the national currency, and increased domestic price levels. These

developments followed the volatility in international commodity markets in 2014 and

2015. The threats posed by volatile oil prices appear to be caused by the low

diversification of Azerbaijan’s economy and poor oil revenue management. Therefore,

the relevance of oil-related adverse effects in booming sectors was studied using both

quantitative and qualitative research methods.

1.3. Justification of the study

The case of Azerbaijan provides a unique opportunity to deepen the studies of NRC and

DD to get a complete picture of the negative impact of the lopsided industrial structure

on the rest of the economy. The end of the oil boom (as measured by oil prices) in 2014

had a devastating impact on Azerbaijan. GDP per capita fell from $7,891 in 2014 to

$5,500 and $3,880 in 2015 and 2016, respectively, but all of this was due to inefficient

management of oil revenues, which various international economists and experts had

warned about since 1995, when the economy entered the recovery phase. Some other

post-Soviet countries (e.g., Russia and Kazakhstan) share the same fate, but their GDP

per capita and its recovery in the post-boom period (between 2015 and 2020) were

higher. Moreover, oil prices have been rising rapidly from the end of 2021 (the end of

the major pandemic waves of COVID-19) and February 2022 (the beginning of the

Russo-Ukrainian war), leading to higher oil revenue expectations. These facts point to

the contextual limitations of the previous studies, in which changing external factors

(such as the oil price decline and GDP collapse) were not adequately considered and

discussed, leading to inconclusiveness. After the macroeconomic events of 2014 and

2015, there is a need to develop informed decision-making in policy circles and

government agencies, which is also supported by academic research. For this reason,

this study fills this substantial contextual gap.

7

Azerbaijan was the most dependent on oil rents among the 15 FSU countries

(even more so than other oil-rich countries such as Russia, Kazakhstan, and

Turkmenistan), while indicators of institutional quality were low (see Figure 1.1.,

panels a and b). This suggests that Azerbaijan's success or failure in achieving stable,

long-term economic development depends on a proper understanding of the oil industry.

For this reason, Azerbaijan is likely to have more statistically relevant trends than some

other FSU countries (e.g., Russia, Kazakhstan) where the non-oil manufacturing

industry is still active, which can be methodologically analyzed in the NRC and DD

frameworks.

Figure 1.1: Oil rents and institutional quality among the post-Soviet countries.

a. Relationship between oil rents as % of

GDP (X-axis) and the voice and

accountability index (Y-axis) between

1996 and 2020 among post-Soviet

countries.

a. Relationship between oil rents as % of

GDP (X-axis) and control of corruption

(Y-axis) between 1996 and 2020 among

post-Soviet countries.

Source: World Bank.

In the NRC and DD studies, it is usually argued that a booming sector and

dependence on natural resources lead to a rentier state that is not interested in

diversifying domestic production and exports, but the question "how exactly does this

lopsidedness occur?" is not answered. For this reason, the case study of a particular

chemical subsectors (as an individual representative of the system concept of "non-oil

production") allows us to address this conceptual gap and also to address the

Armenia

Azerbaijan

Belarus

Estonia

Georgia

Kazakhstan

Kyrgyzystan

Lithuania

Latvia

Russia

Tajikistan

Turkmenistan

Ukraine

Uzbekistan

-2,5

-2,0

-1,5

-1,0

-0,5

0,0

0,5

1,0

1,5

0,0 10,0 20,0 30,0

Armenia

Azerbaijan

Belarus

Estonia

Georgia

Kazakhstan

Kyrgyzystan

Lituania

Latvia

Moldova Russia

Tajikistan

Turkmen

istan

Ukraine

Uzbekistan

-2,0

-1,5

-1,0

-0,5

0,0

0,5

1,0

1,5

0,0 10,0 20,0 30,0

8

methodological limitations of previous studies (e.g., insufficient samples, biased

research designs, especially when only quantitative methods are used).

Focusing this work on a single country and sector allowed to avoid the problems

associated with case selection, levels, and scope of comparative studies and allowed for

a more focused and in-depth examination of the specific NRC and DD signs in

Azerbaijan. Although Azerbaijan can be compared to at least some post-Soviet oil-rich

countries, it was initially challenging to focus precisely on establishing equivalence

between the countries and their non-oil-producing industries. Although the very goal of

comparative studies is to look for commonalities and divergences, it remains difficult to

identify universal patterns in social science research (Mills–van de Bunt–de Brujin

2006). Moreover, the causality required to capture the effects of NRC and DD and their

impact on subsectors appeared to be more pronounced in the chemical sector in

Azerbaijan than in other non-oil manufacturing sectors such as textiles, machinery,

food, and clothing. In other words, only certain subsectors of the chemical industry

(e.g., chlorine, soda, sulphuric acid) promised a causal relationship between the rise of

the oil industry and the key variables of DD when observed. In contrast, other non-oil

manufacturing sectors and subsectors of the chemical industry (e.g., polyethylene,

oxygen, and nitrogen production) did not convince descriptively that they were de-

industrialized compared to the oil industry, according to the initial visual inspection of

the time series. In fact, Azerbaijan was one of the most important centers of chemical

production during the years of the Soviet Union, and it was logical to expect its de-

industrialization due to a lack of competitiveness. To have a de-industrialized sector,

there must first be a well-established industry. Since Azerbaijan was and is a country

rich in oil and natural gas, the central government of the USSR made several decisions

to establish production facilities in the chemical sector. Finally, the current industrial

policy of the government to diversify domestic production is mainly based on two

sectors: agriculture and chemicals. Therefore, the analysis of the chemical subsectors

that have been de-industrialized and are currently being re-industrialized under the

assumption of possible negative effects of the oil industry promises fruitful theoretical

and empirical insights.

The author's previous education and ability to directly observe the main

macroeconomic events in Azerbaijan related to specific subsectors allowed obtaining

valuable and detailed information (e.g., expert interviews, calculation of the Extractive

Dependency Index, or EDI) that could only be analyzed in the context of a country- and

9

sector-specific approach instead of a comparative and cross-sector scenario. Thanks to

the country-specific research design, the necessary data sources and variables of interest

could be operationalized, and their contextual interpretation could be carried out in a

more certain way. The author's educational background and pre-existing knowledge of

the Azerbaijani economy played a critical role in the selection of the Azerbaijani

economy and its non-oil manufacturing sectors to develop a deeper understanding of

NRC and DD phenomena for the future phases of this line of research. Finally, the

application of the original theory of DD allows a generalization of the observed results,

even though the present study is an investigation of the phenomenon in a single country.

1.4. Research objectives

The main objective of this study was to identify the adverse effects of the booming oil

sector on lagging sectors in Azerbaijan. The specific objectives of this study were as

follows:

1. To examine whether the dominance of the oil sector has negatively affected

Azerbaijan’s economy at the level of politics, institutions, governance,

education, and human capital;

2. To determine whether the economic explanation provided by the NRC doctrine

(i.e., DD) is applicable to Azerbaijan’s economy and whether it hinders non-oil

manufacturing growth and development;

3. To demonstrate the relevance of the de-industrialization process that results

from resource dependence and how it decreases the opportunity for sustainable

and long-term prosperity;

4. To analyze available policy alternatives in the short, medium, and long term

using the proposed explanatory framework.

1.5. Research questions

To achieve the aforementioned research objectives, the following key research

questions and side questions were formulated:

1. What are the main stages of development of Azerbaijan’s economy reflected in

the main macroeconomic indicators? Side questions: Are there subtle differences

between the stages of development that were influenced by the oil sector? What

were the main policy measures taken by the government to make the transition

10

from a command economy to an economic system based on a free market

economy?

2. What impact has the oil sector had on institutional quality in Azerbaijan? Side

questions: Has the oil boom led to lower spending on education and health care

(corresponding to human capital in the case of the NRC)? Did public opinion

reflect a societal desire to more effectively manage oil revenues during the oil

boom period in Azerbaijan’s economy?

3. Has the oil boom in Azerbaijan’s economy created signs or effects of DD since

the completion of major oil and natural gas projects (one possible consequence

being the de-industrialization of non-oil manufacturing sectors)?

4. Has the chemicals industry experienced de-industrialization of its production

since 1995 due to the oil sector? Side question: What are the reasons for the

differences between the deindustrialized and reindustrialized subsectors of the

chemicals industry?

1.6. Conceptual framework and hypotheses

This section presents the conceptual framework and research hypotheses. Figure 1.1

depicts the research phases and interconnected parts required to explain the DD-induced

de-industrialization of Azerbaijan’s economy. The impact of the oil sector on

Azerbaijan’s non-oil manufacturing sectors is explained using both the noneconomic

and economic components of the NRC doctrine. The noneconomic side emphasizes the

role of institutions, education, and health care, as well as social cohesion and political

will. This allowed the ability of the Azerbaijani government to manage oil revenues to

be demonstrated. Furthermore, DD as an economic explanation was tested by

examining how changes in resource movement and spending can lead to direct or

indirect de-industrialization.

This study adopted a step-by-step deductive approach and was based on several

hypotheses. These were derived from the NRC and DD theories, in which de-

industrialization is assumed to be a negative effect of the oil boom. Based on the

conceptual framework proposed in Figure 1.1, the process of de-industrialization was

expected to be observed at the fiscal and monetary levels due to the negative impact of

oil revenue mismanagement. The hypotheses are described below.

11

Set 1: Presence of the NRC

Based on previous studies, relying only on economic explanations (which mainly

employ the DD hypothesis) or simply assuming that oil-rich countries suffer the same

fate is not sufficient for determining whether the oil sector in Azerbaijan has created

backward non-oil sectors. Noneconomic explanations of the NRC doctrine emphasize

institutional, political, and governance indicators (Gelb 1988; Deacon 2011; Heller

2006; de Medeiros Costa–dos Santos 2013; Abdulahi et al. 2019). The data set provided

by the World Bank’s Worldwide Governance Indicators allows the statistical correlation

and association between oil industry dominance and institutional quality to be analyzed.

The NRC-related hypotheses used in this study are presented in Table 1.1:

Figure 1.2: Conceptual framework for the study.

Source: Author’s illustration.

Note: Here, H1–H5 denote the four key hypotheses and the expected sign of the relationship between the

oil sector and national economy of Azerbaijan.

Table 1.1: Research hypotheses related to the natural resource curse (NRC) doctrine in

Azerbaijan’s economy.

1

H01: The oil sector does not have any significant influence on institutional

quality: 1=2=...==0;

Ha1: Oil-related variables have a negative influence on political and

12

institutional quality:  < 0 for at least one .

2

H02: There is no statistically significant association between the oil-related

variables and human capital indicators, such as education and health care:

1=2=...==0;

Ha2: Oil-related variables (e.g., oil rents, oil dependency, and oil abundance)

have a negative relationship with human capital indicators, such as education

and health care:  < 0 for at least one .

Set 2: Diagnosing DD

Studies related to DD should be well planned and empirically based. The main problem

with previous studies on DD in Azerbaijan (Gahramanov–Fan 2002; Yıldırım Mızrak–

Gurbanov 2013; Zulfigarov–Neuenkirch 2019; Şanlisoy–Ekinci 2019) is their one-sided

investigation of exchange rate issues and GDP without proper consideration of the

original theory of Corden and Neary (1982) and Corden (1984). In other words, these

studies have only analyzed the relationship among Azerbaijan’s oil prices, GDP, and

REER. However, research on DD should also consider the sectoral movement of

resources and government expenditure. Accordingly, the second set of hypotheses is

presented in Table 1.2:

Table 1.2: Research hypotheses related to Dutch disease (DD) syndrome in

Azerbaijan’s economy.

1

H03: Oil prices do not have positive relationship with the Azerbaijani REER:

1=2=...==0;

Ha3: Oil prices appreciate the Azerbaijan’s REER:  > 0 for at least one .

2

H04: The nominal or real effective exchange rate and oil-related variables do

not have a negative relationship (growth-reducing) with non-oil

manufacturing: 1=2=...==0;

Ha4: The nominal or real effective exchange rate and oil-related variables

have a negative relationship (growth-reducing) with non-oil manufacturing:

 < 0 for at least one .

3

H05: Higher oil prices and the appreciation of the REER do not have a

statistically significant and theoretically meaningful impact on non-oil

manufacturing output and employment in Azerbaijan: 1=2=...==0;

13

Ha5: Higher oil prices and the appreciation of the REER had either a direct

or indirect impact on sectoral output and employment in the non-oil

manufacturing sector in Azerbaijan:  < 0 for at least one .

4

H06: Oil revenue does not create any inflationary effects through government

revenue or spending and population income9: 1=2=...==0;

Ha6: Oil revenue creates inflationary effects through government revenue or

spending and population income:  < 0 for at least one .

Set 3: De-industrialization

De-industrialization has not been studied in much depth for Azerbaijan. The theories of

DD and NRC allow the occurrence of de-industrialization to be conceptualized. The

main hypothesis examined in Chapter 5 is presented in Table 1.3:

Table 1.3: Research hypotheses related to the de-industrialization process in

Azerbaijan’s economy.

1

H07: Dutch disease in Azerbaijan has not led to the de-industrialization of

non-oil tradeable industrial sectors since 1995, especially in the chemicals

industry: 1=2=...==0;

Ha7: Dutch disease in Azerbaijan has led to the de-industrialization of non-

oil tradeable industrial sectors since 1995, especially in the chemicals

industry:  < 0 for at least one .

1.7. Dissertation outline

This dissertation consists of six chapters, the remainder of which are organized as

follows: Chapter 2 reviews the main macroeconomic indicators and policy decisions in

Azerbaijan since 1991 and presents the necessary facts about the country's economy.

Chapter 3 reviews the literature on the level of NRC, DD, and de-industrialization.

Chapter 4 presents the data collection process and methods used. Chapter 5 reports on

the results. Chapters 3 and 4 present the methods and results at three levels: NRC, DD,

and de-industrialization of the chemical industry. Finally, Chapter 6 reviews the

economic policy responses, policy implications, and suggestions for future research

9

Total income is the sum of primary incomes including salaries of employees, incomes from

entrepreneurial activities, incomes from property, and current and capital transfers.

14

based on a literature review of other oil-rich countries, where the NRC and DD have

been common obstacles to overcoming excessive oil dependence.

15

CHAPTER 2

UNDERSTANDING AZERBAIJAN’S ECONOMY

This chapter presents an overview of the developmental stages of Azerbaijan’s

economy. First, Section 2.1 examines macroeconomic stability and stimulation policies,

the liberalization of the economy, institution building, foreign economic relations, and

the development of the private economy. Then, Section 2.2 describes selected

macroeconomic indicators. Finally, Section 2.3 provides a summary of the chapter.

2.1. Azerbaijan’s economy since independence

To demonstrate the importance of national economies’ developmental stages, the

institutionalist school of economic literature emphasizes the historical origins of current

institutions and the causes of long-term economic consequences (Acemoglu et al. 2001,

2002). Azerbaijan moved away from a centrally planned command economy in 1991

when the Union of Soviet Socialist Republics (USSR) collapsed. The Soviet era

comprehensively and ideologically shaped the economic traditions, activities, and

behaviors of economic actors in Azerbaijan (read: the superiority of the communist

economy). Independence required a shift in thinking – not only in rejecting a Marxist-

Leninist political economy but also in learning how markets could reform the

economies of the newly independent post-Soviet countries (Zaostrovtsev 2016). Thus,

the collapse of the USSR created various conditions for building a national economy in

Azerbaijan.

Between 1920 and 1991, the command economy and socialism eradicated

market principles and weakened the entrepreneurial abilities of Azerbaijan’s citizens

(Cornell 2015). In the early years of independence, Azerbaijan was a semideveloped

and socially secular post-Soviet Muslim country, not an economically liberal polity

(Khalilzada 2019). Then, in the late 1990s and early 2000s, preparations for and a rapid

transition to oil-driven economic development occurred. As a newly established state,

Azerbaijan displayed complex patterns of political and economic development.

Nevertheless, its economy can be divided into modern stages of development based on

certain political and economic events.

For instance, Güneş (2020) offered a two-stage chronology: a period of chaos

and instability (1991–1995) and a period of macroeconomic stability and dynamic

economic development (1996 to the present). This approach encapsulates the economic

16

realities of Azerbaijan’s economy; however, by focusing only on the main political

events, it fails to represent or oversimplifies key developmental events and decisions.

As cited in Amirbekov (2015), Rustamov (2010) took a different approach by dividing

Azerbaijan’s economic development into the following stages: transformational

recession (1992–1995), stabilization and recovery (1996–2000), investment-led

growth

10

(2001–2004), and the oil boom (2005–2008). However, this approach ignores

critical aspects of the transition from a command to a market economy (e.g., waves of

privatization and GDP recovery) and shortens the duration of the oil boom.

11

To

effectively capture the recent phases of Azerbaijan’s economic development and rapid

macroeconomic changes, a more comprehensive approach is required.

Aliyev and Suleymanov (2015) analyzed the macroeconomy of Azerbaijan in

three stages, namely the recession (1991–1994), the restructuring or recovery period

(1995–2004), and the oil boom period (2005 onwards).

12

This approach is more in line

with the time frame of crucial economic and political events in Azerbaijan. However, it

is necessary to consider a fourth stage from 2015 to 2020, which corresponds to the

post-boom period. Thus, it captures the consequences of the price decline in

international commodity markets. Therefore, a step-by-step analysis of Azerbaijan’s

post-Soviet economy allows factors to be identified that have led to successful

outcomes and hindered processes of transition and diversification. This modified

framework helps to explain how the extractive industry led to a growing gap between

oil production and non-oil tradeable sectors within Azerbaijan’s economy.

The four-step assessment is done by focusing on different aspects of

Azerbaijan's macroeconomy between 1991 and 2020. Subsection 2.1.1 provides an

overview of macroeconomic stability and stimulation policies. Subsection 2.1.2 focuses

on institution building. Then. Subsections 2.1.3 and 2.1.4 provide an overview of

liberalization and foreign economic relations, respectively. Finally, Subsection 2.1.5

addresses the creation of a private sector.

10

Mainly because of the positive role of Foreign Direct Investments (FDI).

11

The oil boom period should not be limited to the extraction boom as in Rustamov’s (2010)

consideration. Rather, it should also include the revenue boom that usually occurs during following the

extraction boom. This is why the widely accepted oil boom period in Azerbaijan covers the range

between 2005 and 2014.

12

Soyak and Nesirova (2003) also divided the modern history of the Azerbaijan economy into periods

that correspond to 1991 to 1995 and 1995 and later. However, at the time of writing (2003), the authors

did not have enough data to include other economic events and indicators, nor access to political and

governmental decisions that would have enabled them to present a more up-to-date development timeline.

17

2.1.1. Macroeconomic stability and stimulation policies

From the late 1980s, Azerbaijan experienced severe periods of recession, which

intensified when the USSR collapsed in 1991 (IMF 1995). Azerbaijan’s economy

contracted by an average of 63% between 1989 and 1995, compared with 42% in other

countries of the former Soviet Union (FSU, World Bank 2003). Moreover, Azerbaijan’s

cumulative real GDP declined by 61%, approximately 30% higher than those of other

post-Soviet countries. In the early years of independence, Azerbaijani industry was

unable to deliver marketable products. This was because manufacturing enterprises

produced low-quality products at high prices and depended on the centrally controlled

inefficient division of labor (Cornell 2015).

The main cause of macroeconomic instability during the recession period was

hyperinflation. Compared with 1993, the consumer price index (CPI) increased by

1,664% in 1994 (IMF 1995). The National Bank of Azerbaijan (ANB)

13

printed money

to finance the large fiscal deficits caused by hyperinflation (IMF 1997). Because of

continued hyperinflation and a significant lack of confidence in the manat (Azerbaijan’s

currency), foreign currency holdings surged dramatically and the nominal exchange rate

depreciated in 1994 (IMF 1995, 1997). This led to rapid dollarization and a

deterioration in the terms of trade (TOT), which peaked in 1994 (IMF 1995). The

resulting price increases for imports exacerbated the rise in the CPI, as imported food

and other products were heavily weighted in the consumer basket. In addition, the

Azerbaijani government failed to initiate the developments necessary to stabilize the

macroeconomy. However, in 1995, the state introduced several policy measures to

tackle the most pressing problems related to macroeconomic stability. The recovery

stage is discussed later.

At the beginning of the recession period, the nationwide banking crisis

exacerbated macroeconomic instability at the level of the national economy. The state

responded by passing the National Bank Law in February 1992 and the Law on Banks

and Banking Operations in August 1992. As a result, the EBRD (1995) reported the

emergence of 200 small commercial banks, which accounted for 10% of total loans to

enterprises and individuals. New minimum capital requirements for commercial banks

led to their consolidation by the end of 1995, reducing their number (EBRD 1995). In

addition, the “National Bank of Azerbaijan was granted broad powers of prudential

13

The ANB is now known as the Central Bank of the Republic of Azerbaijan (CBAR).

18

regulation” to effectively regulate the banking sector against further banking crises

(EBRD 1995: 35); however, enforcement was weak. Underdeveloped financial

institutions and a lack of will to reform slowed the country’s transition to a market

economy (Schroeder 1996).

During the recession, Azerbaijan lacked financial institutions outside of the

banking sector, such as active investment funds, which could have stimulated the

financial system through private capital (EBRD 1995). Moreover, the lack of a stock

exchange drastically limited the government’s ability to implement macroeconomic

stability and stimulus measures to facilitate the transition to a market economy (EBRD

1995).

14

To stimulate foreign exchange trading, the ANB established a facility in

August 1994 called the Baku Interbank Currency Exchange (BICEX) for commercial

banks, promoting trading in non-cash foreign exchange (IMF 1995).

At the end of the recession period, the government applied to “Rehabilitation

Credit” from the World Bank to stimulate the national economy and overcome obstacles

on the reform agenda (World Bank 1995). The rehabilitation loan was intended to

support structural reforms to stabilize a country’s macroeconomy and accelerate its

transition to a market economy (World Bank 1996). It provided financial support for

Azerbaijan’s efforts to implement key transformation measures. Another economic

stimulus package was the state program “Entrepreneurship Development in Azerbaijan”

(1993–1995). This sought to increase the number of small and medium-sized enterprises

(SMEs) to rebuild foreign economic relations and boost production. The program

provided the necessary infrastructure and a favorable environment for local

entrepreneurs to do business. For instance, in addition to providing technical assistance

to new businesses, the state reduced unnecessary audits by state agencies as well as

corruption rates (Mustafayev 2016a). Moreover, the number of licensed economic

activities was decreased to encourage the emergence of new entrepreneurs in the

national economy (Mustafayev 2016a). However, the following factors still threatened

macroeconomic stability: high inflation, erosion of real wages, a weakened exchange

rate, depleted international reserves, the six-year military conflict with Armenia over the

Nagorno-Karabakh region,

15

and high political uncertainty (IMF 1997).

14

However, it should be mentioned that the Law on Securities and Stock Exchange was passed in

November 1992.

15

War and political turmoil led to many uncertainties in business prospects within the national economy

of Azerbaijan. The country experienced continuous shortages in food and raw materials for production

and high inflation rates, which decreased well-being (Cornell 2015). Furthermore, the war with Armenia

19

The “Contract of the Century”

16

and the 1994 ceasefire with Armenia led to

greater macroeconomic stability in Azerbaijan throughout the recovery period, which

began in 1995. Structural reforms supervised by the World Bank and IMF facilitated

this stability. Surprisingly, even before the oil boom (2005–2014), Azerbaijan became

the fastest growing economy in the post-Soviet landscape; its GDP grew by 10% year-

on-year in 1998 and by 7.2% in 1999 (Aras et al. 2016; Cornell 2015). Moreover, the

government implemented radical economic reforms to overcome economic imbalances,

hyperinflation, and declining living standards during the recovery phase (Kaynak–

Nasirova 2005).

In contrast to the recession period, the recovery stage began with strict and

efficient tax collection by the state and sharp expenditure compression actions

(contractionary fiscal policy), ensuring fiscal consolidation (IMF 1995). “As a result,

central bank financing of the government budget deficit declined from 11% of GDP in

1994 to 1.5% of GDP in the first nine months of 1995” (IMF 1995). Similarly, growth

in the domestic broad money

17

supply declined from 18% to 8% at the end of 1995

compared with the end of 1994 (IMF 1995). Moreover, the availability of large external

financing due to new oil contracts led to an increase in the government budget and an

improvement in monetary stability; during the recovery period, these factors led to a

sharp decline in budget financing by the central bank (IMF 1995).

18

In addition, the

nominal exchange rate stabilized, while the average monthly inflation rate fell to 1.5%

meant that two key sectors in Azerbaijan—agriculture and industry—could not retain enough investment

for restructuring and revitalizing national economic development (Luong–Weinthal 2001). According to

Starr et al. (1998), the great powers’ geopolitical preference for Armenia delayed the transition process in

Azerbaijan.

16

The Contract of the Century was an agreement between 12 large oil extraction companies to utilize

cheap, high-quality, and abundant oil fields such as Azeri, Chirag, and Guneshli (ACG) located in the

Caspian Sea, near the Absheron Peninsula. The resources of the Caspian basin were a viable option for

Western countries to choose over those in Persian Gulf countries. In addition, during the early years of

Azerbaijan’s independence, the collapse of industry meant that the country was unable to generate

competitive products for world markets. During the Soviet years, Azerbaijan’s economy had already been

based on natural resources. Later, the Contract of the Century initiated a new stage of resource-led

development in Azerbaijan.

17

“Broad money” refers to the amount of money in circulation in a given economy. It is viewed as the

most comprehensive approach to assess a country’s money supply, taking into account narrow money and

other assets that can be quickly changed into cash to purchase goods and services. Source: Investopedia

(2021).

18

Azerbaijan earned a USD 91 million signing bonus for finalizing an oil deal with foreign oil

corporations in 1995, which was channeled into the budget. Although the monetary impacts of these

foreign exchange inflows were not completely sterilized (a sudden injection of a huge amount of foreign

currency creates inflationary pressures that a government must address), they offered a significant source

of budget financing without jeopardizing monetary stability.

20

in 1995 (IMF 1995). Thus, the expansion of credit and money was brought under

control.

In 1997, Azerbaijani production recovered, mainly due to new investments in

the country’s oil fields (IMF 1998). The unemployment rate remained high (approx.

19%) during the first half of the recovery stage; nevertheless, the IMF (1998) reported a

decline in industrial employment and an increase in agricultural employment due to the

growing number of private farms created by new land reforms.

In the early years of recovery, stimulation programs included the International

Development Association (IDA)’s comprehensive restructuring programs, the World

Bank’s Third Structural Adjustment Credit (SAC III), and the Azerbaijani government’s

State Program for Socioeconomic Development of the Regions of the Republic of

Azerbaijan for 2004–2008. The critical component of SAC III was the privatization of

the utilities sector; here, private sector development was supported by business

environment studies and consulting services by specialists for the privatization of

medium and large industrial enterprises (also referred to as conditional credits; World

Bank 1999a). Furthermore, the aforementioned State Program promised new jobs and

economic diversification by reducing disparities in socioeconomic development

between the capital and Azerbaijan’s regions.

However, according to various experts, the macroeconomic and stimulation

measures were not sufficient for overcoming the major challenges in the recovery

phase. So-called “light oil money” hindered balanced and sustainable economic

development, as state officials became increasingly lax in fighting for a share of oil

revenues (Ibadoglu 2008). This strengthened corrupt ties between state officials and

businesses and made the construction sector the main priority for state budget spending.

Only expensive infrastructure projects allowed for a quick transfer of oil money from

the state budget into the hands of the corrupt (Ibadoglu 2008).

Meanwhile, the expected diversification of the economy through greater private

sector participation remained sluggish. According to the World Bank’s “Doing Business

2004” report, an average of 106 days were required to start a business in Azerbaijan,

compared with only 25 and 30 days in Armenia and Georgia, respectively (Doing

Business 2004).

19

New jobs and economic diversification leading to macroeconomic

19

In addition, opening a business in Azerbaijan required 14 procedures and cost USD 119. By

comparison, Armenia required 10 procedures and USD 68 to open a business, and Georgia required nine

procedures and USD 171 to open a business (Doing Business 2004).

21

stability and stimulation are only possible if the private sector is well developed and

supported. However, at the end of the recovery period, structural changes accelerated at

the expense of non-oil manufacturing and agriculture, despite loans from international

organizations and government programs favoring oil tradeable sectors (Ibadoglu 2008).

During the oil boom period, major macroeconomic events and stimulation

packages were implemented primarily because the government was quick to spend

accumulated oil revenues. The government promoted expensive infrastructure projects

and increased public wages, pensions, and social benefits to stimulate the consumer—or

demand—side of the economy in the short term. Moreover, the overspending of oil

revenues led to fiscal deficits in the state budget. This led the government to increase its

use of funds from the State Oil Fund of the Republic of Azerbaijan (SOFAZ) to finance

fiscal deficits (Suleymanov–Aliyev 2015). As a result, the state had a considerable

propensity to spend immediately instead of pursuing a selective policy.

Nevertheless, the IMF (2010) reported “spectacular growth rates” in Azerbaijan

during the oil boom. The country’s GDP grew by an average of 18.8% between 2005

and 2010, the poverty rate fell from 45% in 2003 to 11% in 2009, and the

unemployment rate reached its lowest level at 11% in 2009 (compared with 13% in

2008; IMF 2010). The exchange rate also stabilized: “The de jure exchange rate regime

has been pegged to a euro-dollar basket since March 2008, but the de facto regime has

been stabilized against the U.S. dollar since June 2008” (IMF 2010: 3). The receipt of

substantial oil revenues was accompanied by a strengthening of the manat. The

exchange rate was USD 1 to AZN 1 in 2006, which changed to USD 1 to AZN 0.80 in

2008 (Mehtiyev 2017). The national currency continued to appreciate until the end of

the oil boom. At the end of 2014, the exchange rate was USD 1 to AZN 0.7844, thus

remaining stable (Mehtiyev 2017).

However, the global financial crisis and the increasing importance of the oil

sector undermined the sustainability of Azerbaijan’s economy. Oil revenues fell by

35%, non-oil revenues were 17% below projections, and total exports fell by 30% in

2009 (IMF 2010).

20

As a result of the crisis, several state-owned enterprises (SOEs) and

banks had difficulty extending short-term foreign liabilities, leading to a liquidity

shortage in the banking system and a dramatic drop in credit growth (IMF 2010).

20

With declining production in the construction and non-oil manufacturing sectors, non-oil GDP growth

fell from 15.7% in 2008 to 3% in 2009, further reducing demand for credit and weakening bank loan

portfolios.

22

Moreover, Mehtiyev (2017) reported that foreign direct investment (FDI) in the non-oil

manufacturing sector declined significantly between 2010 and 2015. By contrast, most

FDI went to the oil sector. During the oil boom, domestic capital owners viewed the

non-oil manufacturing and agriculture sectors as risky and not worthy of investment

(Mehtiyev 2017); thus, domestic investment contributed little to the non-oil

development of Azerbaijan’s non-oil economy. In addition, 70% of domestic

investment (mostly from the state) went to infrastructure development and construction

(Mehtiyev 2017).

During the oil boom, President Ilham Aliyev adopted the State Program for

Socioeconomic Development of the Regions of the Azerbaijan Republic for 2009–2013

as well as that for 2014–2018. They were aimed at increasing industrial production in

the regions, creating a business-friendly environment, and encouraging domestic and

foreign investment. The State Statistical Committee of the Republic of Azerbaijan

(SSCRA; 2014) reported successful results of the abovementioned state programs.

These included increased overall GDP in rural regions, increased competitiveness

(measured by the global competitiveness index), higher FDI in non-oil sectors, and

reduced inflation and poverty. However, Ibadoglu (2008) reported an increasing

monopolization of Azerbaijan’s economy and erroneous calculations by the state

regarding the minimum wage. In 2006, the minimum living wage was AZN 58 per

month, while in 2007 it was only AZN 64; according to alternative calculations, the

actual minimum living wage was AZN 90. In addition, Ibadoglu (2008) found increased

signs of DD, expressed in higher REER and CPI as well as in the resource movement

effect.

21

From 2015, a decrease in oil prices was felt through negative and lower GDP

growth rates, higher budget deficits, and a decline in imports (Mukhtarov 2018). In

addition, the Central Bank’s strategic reserves fell by USD 8.74 billion and bottomed

out at around USD 5 billion, the same level as in 2009 (Bayramov 2016). The decrease

in oil revenues jeopardized the medium- and long-term fiscal balance and plans, while

SOFAZ recorded a larger budget deficit (AZN 4.4 billion) in 2017 than in 2016

(Ahmadov 2016; Mukhtarov 2018). In addition, the banking sector fell into crisis, and

CBAR drastically raised interest rates (Mehtiyev 2017). These developments forced the

government to painfully rebalance Azerbaijan’s macroeconomic dynamics. The

21

In fact, FDI exclusively flowed into the oil sector, while the non-oil manufacturing sector was starved

of investments needed to gain momentum.

23

commodity price crisis also entailed distressing new realities for citizens in terms of

income and welfare. As a result, the first years of the post-boom period were chaotic

and uncertain, as the government was unable to design a short-term adjustment plan

(Ahmadov 2016).

Notably, the devaluation of the manat was the most remarkable macroeconomic

event in the post-boom period. It and other difficulties related to the decline in oil prices

first led to higher public debt as a share of GDP, which undid Azerbaijan’s leading

status as a relatively low-debt country among the post-Soviet countries (Ahmadov

2016). Although the government asserted that devaluation and a floating exchange rate

should increase “the government’s maneuverability to adjust the value of the national

currency to oil prices” (Niftiyev 2020a), in practice a floating exchange rate did not

occur. The exchange rate has been fixed at approximately AZN 1 to USD 1.70 and EUR

2 since 2015. Second, devaluation led to more expensive imports, rising inflation,

public panic (leading to the conversion of deposits from manats to U.S. dollars), and an

unofficial dollarization process. Third, the social impact of the crisis took the form of

economic insecurity as food prices rose while the labor force did not experience any

real increase in wages (Guliyev 2016). In addition, the devaluation hurt citizens who

had mortgages in foreign currencies as their loans became more expensive. Another

consequence of the devaluation was job cuts. The energy, telecommunications, and

banking sectors reacted the fastest, laying off approximately 250–300 employees very

shortly after the devaluation (Guliyev 2016). Finally, throughout the post-boom period,

local and uncoordinated protests among the population occurred in response to

increasing poverty and unemployment in rural areas (Guliyev 2016).

To mitigate the negative impact on the manat’s value, the president signed a

decree to increase the salaries and pensions of state employees (Guliyev 2016). In

addition, the government promoted the devaluation as an opportunity to increase the

competitiveness of non-oil exports and private sector participation in non-oil GDP

(World Bank 2019).

22

According to Azerbaijani media reports, the devaluation boosted

production in both the oil and non-oil tradeable sectors, supported domestic and foreign

investments, and increased non-oil tax revenues (Azerbaijan Gazette 2019; Mustafayev

2016b). However, Bayramov (2020) argued that the reforms and stimulation programs

did not bring about the desired regeneration of the national economy. In fact, the

22

The devaluation of the national currency made exports cheaper, therefore provided a competitive edge

for the producers and exporters.

24

increase in the share of non-oil manufacturing in total output and total exports was due

to declines in oil production and exports. In other words, non-oil sectors grew not

because of increased competitiveness of non-oil tradeable sectors but because of

decreased oil production.

In February 2021, President Ilham Aliyev signed a new decree entitled

“Azerbaijan 2030: National Priorities for Socioeconomic Development” (Aliyev 2021).

The decree set out five priorities for an effective macroeconomic policy framework.

They aim to ensure sustainable macroeconomic stability and strengthen the medium-

and long-term “drivers” of economic development. Examples of such drivers include

the modernization of human capital, expansion of the digital economy, and full

economic sovereignty. The program emphasizes development principles such as the

need for a steadily growing competitive economy; a society based on dynamic and

inclusive social justice; competitive human capital and space for modern innovation; a

great return of the liberated territories

23

; and a clean environment and “a pro-green”

country.

24

Thus, the government has continued its attempts to stimulate and regulate the

economy after low commodity prices. However, after the second quarter of 2016, oil

prices began to rise. Again, this caused the government to lose motivation to implement

reforms and invest in human capital, which are critical for long-term sustainable

economic growth and development. It is also important to have an overview of

institutional changes based on the stages of development of the Azerbaijan economy to

understand why certain policies have been successful while others have failed to

achieve macroeconomic stability.

2.1.2. Institution Building

Many problems and challenges emerged during the recession period; however, the

Azerbaijani government made various decisions and decrees to begin the process of

institution building. In fact, institution building became a critical priority for ensuring

23

In 2020 Azerbaijan and Armenia had the Second Karabakh War that ended with Azerbaijan’s return of

the previously occupied Nagorno Karabakh and surrounding districts. This enlarged the de facto

territories of the Republic of Azerbaijan. Liberated areas were then included in the economic

development programs.

24

In December of 2016, the development program entitled Strategic Roadmap created a priority for

increasing the share of the renewable energy sources in the overall energy production in Azerbaijan. The

government became more interested in green growth and renewable energy sources because exporting oil

and natural gas is more profitable than domestic consumption (Vidadili et al. 2017). Therefore,

“Azerbaijan 2030: National Priorities for Socio-Economic Development” offered a different perspective

of the state’s strategy for renewable energy production during the post-boom period.

25

the country’s smooth transition to a market economy after its socialist legacy. For

instance, the State Committee on Property Issues, Anti-Monopoly Committee,

Entrepreneurship Support Fund, and Committee on Foreign Investment were all

established between 1992 and 1994 (Yunusov 2012). Moreover, Azerbaijan adopted the

Law on Protection and Promotion of Foreign Investment in 1992 and has signed

bilateral investment and tax treaties with many countries (Pashayev 2013). The manat

(shortened to AZM, but AZN after the denomination in 2005) became the country’s

official currency in 1992, but its actual circulation only began in 1994 (Aras et al.

2016).

25

During the recession period, legal regulations in newly independent Azerbaijan

initially included the Law of the Republic of Azerbaijan, titled “Property in the

Republic of Azerbaijan” (November 9, 1991); the Decree of the President of the

Republic of Azerbaijan, titled “On Commercialization of the Activity of Trade

Enterprise” (August 1, 1992); and the Law of the Republic of Azerbaijan on

Destatization and Privatization in 1993 (Im et al. 1993). Structural reforms and

important legislative regulations, such as the Tax Code, Land Code, Customs Code, and

Labor Code, were developed during the early years of independence (Pashayev 2013).

Value-added tax and the National Bank Law were introduced in 1992, and the

Execution Bankruptcy Law was adopted in June 1994 (Aras et al. 2016).

Despite a limited capacity, considerable success was achieved in institution

building during the recovery period (World Bank 2003). However, in the first half of

the period, there was a failure to address “deep structural reforms such as large-scale

enterprise privatization, banking reform, regulation and operation of public utilities, and

delivery of social services were slower than expected” (World Bank 1999a: 5). The

second half delivered mixed but improved results with regard to institutional reforms

intended to address the legacy of inefficient production, monopolies, market failures,

outdated technology, and other issues (World Bank 2003).

The establishment of SOFAZ in 1999 is considered one of the key

socioeconomic developments of the recovery period. Primarily, it improved

transparency and accountability (Wakeman Lin et al. 2003) in oil revenue management.

In addition, Boyarchuk (2012) identified SOFAZ as the main factor that ensured short-

25

It was assumed that Azerbaijan citizens needed time to gain trust in the manat over the Russian ruble,

which had circulated within the country for a long time. However, as a national currency was an integral

part of Azerbaijan’s sovereignty, the government spent huge resources to build up the manat’s value.

26

and medium-term stability in oil-sponsored government spending, especially with

regard to energy projects.

Furthermore, a law called “On Land Reform” was introduced in 1996 during the

recovery period. It provided legislative and regulatory mechanisms for dissolving

collective and state farms. Until 1996, farmers followed the old Soviet-style

organization of labor (i.e., collective unions) by using available resources in agricultural

production (World Bank 1996). This created fuzzy property rights and several issues in

the distribution and marketing of final products during the late recession and early

recovery periods. Later, institutional support for the agriculture sector was provided

through the Adaptable Credit Plan and complemented by International Finance

Organization (IFC) activities to increase Azerbaijan’s competitiveness (World Bank

1999a).

Although institution building became more systematic during the recovery

period, Azerbaijan’s anticipated large oil revenue troubled international experts. The

World Bank (1999b) recommended implementing vigorous programs to combat

corruption and public sector failures, which might in turn strengthen market

mechanisms, provide property rights, and regulate the financial sector. A transparent

and nondiscriminatory business environment was expected to increase efficiency and

quality of life in poverty-stricken regions as well as improve social services. However,

during this period, Azerbaijan’s institutional quality failed to reach a level sufficient for

managing oil revenue and preventing corruption and poor governance (Ahmadov et al.

2013; World Bank 1999a).

Furthermore, during the oil boom, SOFAZ’s spending of oil revenue on

immediate needs and long-term development goals was insufficient for achieving

sustainability in the economy. However, future generations must benefit from oil

wealth; thus, “further institutional building [was] needed for transparent project

identification, planning, prioritization and execution” (World Bank Group 2015: 9). The

IMF (2012) suggested broadening the tax base, reducing compliance costs, and

enhancing transparency through tax reforms to boost non-oil tradeable sectors.

However, underdevelopment and underinvestment in the judicial system and public

courts for resolving business conflicts reflected a weak rule of law in Azerbaijan (World

Bank 2006b). The World Bank Group (2015) reported weak institutional capacity and

coordination, limited budget transparency, and low citizen influence. These factors led

27

to institutional gaps in the monitoring and implementation of public policies at the end

of the oil boom period.

Nevertheless, the government introduced two crucial institutional innovations

during said period. The first was measures that accompanied agricultural privatization

in rural areas, such as “land reform and rural credit commissions, town hall credit

meetings, Water User Associations (WUA), and Community Associations” (Thurman

2004: 16). The second was the establishment of the State Agency for Public Service and

Social Innovations under the President of the Republic of Azerbaijan (ASAN), which

aims to decrease corruption and dysfunctionality in the public sector (Chantzi 2013).

26

ASAN seeks to reduce additional expenses and lost time as well as increase the level of

professionalism in public service delivery, strengthen confidence in state bureaucracy,

ensure access to electronic public services, increase transparency, and strengthen the

fight against corruption.

The macroeconomic events of the post-boom period (e.g., devaluation of the

national currency and low growth rates) strongly indicated a lack of institutional

capacity for effectively combating the adverse effects of sharp commodity price slumps

(Bayramov 2016). However, the state adopted several institutional measures to manage

the negative outcomes of oil dependency, which were harshly revealed in the post-boom

period. Ahmadov (2016) grouped the state’s initiatives after 2014–2015 into three

categories: strengthening financial security and predictability, liberalizing the economy

and improving the entrepreneurship space, and establishing institutional reforms to

support the development of non-oil sectors. First, strengthening financial security and

predictability enabled the government to stabilize the exchange rate and decrease

speculation with foreign currencies on the black market. To this end, CBAR reserves

and a new entity called the Financial Market Control Chamber were established.

Second, to liberalize the economy and improve the entrepreneurship space, institutional

reforms were implemented to support the development of non-oil sectors, addressing

failures in the privatization process during the oil supercycle (i.e., the oil boom).

Nevertheless, the efficiency of these responses remains uncertain; moreover, the

informal sector’s large share in Azerbaijan’s GDP complicates the use of traditional

policy tools such as money supply and bank credits.

26

ASAN gathered 100 public services from 25 government entities, which were meant to increase in the

quantity after the establishment of e-governance (up to 477 services).

28

Third, in terms of institutional reforms, the institutionalization of non-oil

development became the government’s main focus. The Strategic Roadmap for the

National Economic Outlook of the Republic of Azerbaijan was adopted in 2016, which

sought to better target underdeveloped economic sectors through appropriate policies.

Niftiyev (2020a) summarized the following additional institutional developments during

the post-boom period: the establishment of the Center for Economic Reforms Analysis

and Communication

27

and the Financial Stability Committee, decrees and laws on

Additional Actions for Investment Promotions, Additional Measures for Promotion of

the Non-oil Products, and Cancellation of the Inspections in Entrepreneurship (CIE).

Among these developments, the CIE law is particularly notable. It entered into

force in November 2015 and contained the following first and fundamental item:

“[I]nspections in the area of entrepreneurship in the territory of the Republic of

Azerbaijan will be suspended until January 1, 2022” (Aliyev 2015). The law was

adopted to prevent state agents from engaging in corruption. It was not the first attempt

to decrease the arbitrariness of state interventions in business.

28

Poor enforcement of the

rule of law and widespread corruption had previously led to an extremely high number

of inspections of businesses, the main aim of which was to collect bribes. Although the

government’s efforts to significantly reduce the arbitrariness of state inspectors and

civil servants mostly failed, CIE’s passing ensured that only the most crucial

inspections were conducted.

29

Thus, entrepreneurs were able to continue and even

increase their participation in the creation of value-added products during the post-

boom period.

27

The Center for Economic Reforms Analysis and Communication seeks to provide policy suggestions

to guide government reforms based on microeconomic and macroeconomic research. The Financial

Stability Committee recommends policy actions to reform macroeconomic and financial stability in the

national economy. Additional Actions to Investment Promotions, Additional Measures on Promotion of

the Non-oil Products, and the Cancellation of the Inspections in Entrepreneurship cover governmental

support to increase non-oil output by a variety of mechanisms, which range from financial to governance-

related policies.

28

In the official 2002 decree, President Haydar Aliyev stated, “Today, the work of a number of central,

city, and district executive authorities, which implement state regulation of entrepreneurship, cannot be

considered satisfactory. Thus, various central and local executive authorities, including tax, customs, law

enforcement and sanitary-epidemiological services, interfere in the activities of entrepreneurs, obstructing

their work by conducting various inspections. Bureaucratic pressures and biased decisions by courts in

resolving economic disputes are common. All this causes legitimate dissatisfaction of people engaged in

entrepreneurial activities, it does not allow them to realize their entrepreneurial initiatives and make full

use of internal potential” (Aliyev 2002).

29

The law states, “the provisions of this Law shall not apply to inspections carried out by the Prosecutor

General’s Office of the Republic of Azerbaijan in connection with the investigation of corruption

crimes.”

29

In addition, the Azerbaijan Investment Holding (AIH) was another development

in institution building during this period. AIH is a public legal body established through

a decree by the President of the Republic of Azerbaijan dated August 7, 2020. Its goal is

to improve the management and operation of state-owned companies and enterprises as

well as businesses with a share of state capital. It represents the first stage in the full

privatization of problematic SOEs, as it attempts to improve the economic efficiency

and transparency of their investment programs, ensure their competitiveness, and

improve their financial health and sustainability (AIH 2021). However, AIH is a

relatively new development, and thus, real results have yet to be seen.

2.1.3. Liberalization of the economy

During the recession period, price liberalization was the government’s main concern.

The government intended to focus exclusively on the price ragualtions of the natural

monopolies (World Bank 1995). “In 1992, 70–80% of the producer and consumer

prices were liberalized, with further rounds of liberalization later in 1992 and 1993,

leaving bread and energy as the main goods under the price controls” (EBRD 1995). In

addition, the prices of utilities and petroleum products remained under state control,

while those of housing, rent, transportation fares, electricity, and other public utilities

were administratively set, remaining far below cost recovery levels (IMF 1995).

30

Overall, price liberalization during this period was rapid but inefficient. The

government tailored its structural adjustment reforms according to increasing energy

prices.

During the recession period, the most critical steps in liberalizing Azerbaijan’s

financial system were to utilize oil resources and eliminate barriers to investing in

extractive industries (Rakov 2020; Agaev 2012). At the end of the period, the state

eliminated all quotas and licensing restrictions for both imports and exports, except for

certain “strategic goods” such as oil and cotton (EBRD 1995). Another aspect of the

large-scale economic liberalization was wage liberalization; until May 1994, wage

ceilings had been imposed on enterprises (EBRD 1995). Lastly, the government’s

measures throughout this period to address monetary policy resulted in the liberalization

of interest rates for most of 1995 (EBRD 1995).

30

According to the World Bank’s (1995) Institution Building Technical Assistance Project, the

government liberalized other prices (including bread prices) and energy prices near the end of the

recession period.

30

During the recovery period, the government focused on land privatization and

farm restructuring as liberalization policies to help overcome agricultural production

crises (Thurman 2004; World Bank 2006b). In fact, overall agricultural output rose by

53% in 2003 compared with 1995 levels, and household and other small farms

increased their participation in agricultural production. Moreover, government imports

of grain were eliminated, and quota and licensing requirements were removed at the end

of March 1995 to reduce the state’s dominant role in foreign trade (World Bank 1995).

In January 1995, the government fully liberalized domestic prices by abolishing state

orders for enterprises and ending bread subsidies (IMF 1995). In other words, the state

was still making production plans for domestic enterprises until 1995, but it ceded full

control to them over how much to produce during the recovery. Furthermore, domestic

“prices of oil and oil products were raised to about half world levels at the beginning of

1995, with the intention to reach parity with world market prices within the next few

months” (EBRD 1995: 35).

The recovery period also saw the large-scale liberalization of financial

transactions and exchange rate regimes. According to Rakov (2020), Azerbaijan

gradually pursued a policy of liberalizing in the capital account. In 1995, it established a

regime of soft pegging the national currency’s exchange rate, which continued post-

boom. In terms of financial openness (the ratio of total capital flows as a proportion of

GDP), Rakov (2020) ranked Azerbaijan second among the Caspian basin countries in

2018, as the inflow and outflow of FDI to and from the country were free and

constantly changing. In addition, markets for exchange auctions and Central Bank

refinanced credits were established to more efficiently allocate available financial

resources, which slightly increased transparency. Furthermore, the application of the

modern tax code, more transparent budget execution and accounting, and a tighter

internal and external audit capacity supported the rapid adaptation of Azerbaijan’s

economy in favor of liberalization.

Moreover, the anticipated large-scale privatization of the telecommunications,

aviation, chemical, and industrial sectors made modest progress during the recovery

period (EBRD 2003; 2004).

31

Attempts to privatize gas and water utility SOEs

continued; in September 2002, four newly constructed regional electric supply

infrastructures were entrusted with 25-year private management contracts to increase

31

Despite this, the government announced its plan to sell its stakes in Azercell and Bakcell, two cellular

phone companies, by the end of 2003 (EBRD 2003).

31

their efficiency and attract investment (EBRD 2003). However, these liberalization

steps were not continued after oil revenues began to enter the country, and the

agreements between the state and private sector were annulled in 2006. The government

also gained full control over the electricity supply infrastructure (Mehtiyev 2017). In

other words, large oil revenues prevented improvements in Azerbaijan’s economic

efficiency (Mehtiyev 2017), jeopardizing large-scale privatization and complicating

further developments in liberalization (EBRD 2003). At the end of the recovery period,

the EBRD (2004: 98) reported that “the main railway and shipping companies will also

be transformed into joint-stock companies in advance of planned privatisation.”

However, the SOEs Azerbaijan Railways CJSC and Azerbaijan Caspian Shipping CJSC

have still not undergone liberalization.

32

During the oil boom period, only a handful of state liberalization measures were

taken. This was partly due to the main privatization targets being completed

33

and the

large oil revenue allowing the government to relax. Thus, rapidly generated oil rents

provided new opportunities for the state to avoid relying on Azerbaijan’s non-oil

productive capacity, and it turned their attention to expenditures on patronage and

security forces (Guliyev 2013). Nevertheless, the IMF (2005) reported substantial

progress in exchange rates and trade liberalization toward the end of the recovery and at

the beginning of the oil boom. Moreover, during the latter period, Azerbaijan

maintained a trade system that was free from non-tariff restrictions; in fact, the average

tariff was 6% (IMF 2005a). Furthermore, “the liberalization of the telecommunications

market had opened up opportunities for the private sector,” leading tariffs for unlimited

broadband internet to plummet (EBRD 2015: p. 97).

In early 2005, the IMF recommended privatizing the state-owned Kapital Bank

and the International Bank of Azerbaijan (IBA; IMF 2005a). However, only Kapital

Bank was privatized. Since IBA was the largest bank in the South Caucasus, the state

continued to play a prominent role in the banking sector, which entailed its own set of

complications. Hasanov (2017) concluded that a lack of transparency, large-scale

corruption, poor management, weak supervision, and false audits caused IBA’s assets to

32

In the railway and shipping sectors, one major development was the transfer of Azerbaijan Railways

CJSC and Azerbaijan Caspian Shipping CJSC to Investment Holding for cooperative management.

However, the supervisory boards created by Investment Holding to increase efficiency, transparency, and

governance were not composed of independent managers but rather the usual state officials from various

parts of the government.

33

At the beginning of the recovery period (1996–1998) a total of 6,303 small enterprises were privatized.

In 2004, this number dropped to 1,934 (Ibadoghlu 2021).

32

become problematic (e.g., fewer liquid assets, problems meeting capital adequacy

requirements, and increased foreign liabilities). Moreover, “failure to take adequate

measures, despite the continuous violation of requirements of banking legislation and

the norms of the Central Bank of Azerbaijan,” seriously undermined the bank’s

management practices (Hasanov 2017: 39).

Although small-scale privatization continued during the post-boom period, most

of the liberalization process was almost complete, and hence, it was no longer the

government’s main focus. The liberalization of the economy and the improvement of

entrepreneurship simply included presidential decrees to produce non-oil exportable

goods, reform customs, and simplify licenses and permits for businesses during the

post-boom period. The EBRD (2020) reported that the authorities partly liberalized the

foreign currency operations regime. Specifically, imports with a total cost of less than

USD 10,000 were excluded from reporting to currency control authorities under new

guidelines to speed up imports. This amendment led to a reduction in delays caused by

bureaucratic operations. However, the post-boom period drew attention to SOEs, as the

Azerbaijani government needed to improve governance to reduce costs after sharp

commodity price downturns in 2014. The Supervisory Board of the State Oil Company

of the Azerbaijan Republic (SOCAR) was established by order of President Ilham

Aliyev on January 23, 2021 to exercise general management and control over the

activities of SOCAR (Ibadoghlu 2021). Similarly, supervisory boards at Azerbaijan

Airlines CJSC, BakuBus LLC, Azerbaijan Railways CJSC, and Baku Metro CJSC were

approved by presidential decree on March 30, 2021 (Ibadoghlu 2021).

Furthermore, the Strategic Roadmap from 2016 (introduced in Section 2.1.2)

established many priorities and goals for liberalizing various sectors of Azerbaijan’s

economy. For instance, it strongly recommended accelerating economic liberalization to

optimize the costs and revenues of utilities and unused industrial facilities. The roadmap

also underlined the necessity of liberalizing airlines and railways, since SOEs such as

AZAL and Azerbaijan Railways CJSC currently provide expensive and below-par

services. In addition, a new priority for the government is to develop the tourism sector;

however, without liberalization, prices and services cannot be optimized to lure tourists.

Since oil prices have been rising since 2017, none of these reforms have been

33

implemented to the desired extent. Nevertheless, the monitoring of the roadmap

promises modest progress in the near future.

34

2.1.4. Foreign economic relations

Following the collapse of the USSR, the devastating consequences of breaking ties with

other FSU countries and a decline in production were intensified by the collapse of

inter-republic trade arrangements and payment systems (Taymas 1993, as quoted in

Aliyev–Suleymanov 2015). According to Cornell (2015), the Soviet era brought

industrialization, urbanization, infrastructure development, and high levels of education;

however, the command economy and central government ensured that the integration of

member countries into GVCs would be harsh if it became necessary (mainly if the

Soviet Union collapsed). Soviet-style economic management ensured specialization in

only a handful of sectors and production according to the specific and planned demands

of other countries (Cornell 2015). Consequently, the collapse of the Soviet Union led to

a collapse in supply and demand for the main goods and services in nearly all FSU

countries.

Like other Commonwealth of Independent States (CIS) countries, Azerbaijan

experienced a significant shift in external trade flows during the recession period.

Following the disintegration of the Soviet Union, the country sought to retain centrally

planned commercial links with CIS nations through a series of bilateral trade

agreements. Under these agreements, trade volumes were typically determined at prices

substantially lower than world market levels. During the Soviet era, production and

trade among member countries were not based on principles such as economic

efficiency and competitiveness; therefore, the collapse of the USSR left the FSU

countries in a position in which supply linkages could not rebuild foreign trade.

Bilateral agreements began to crumble in 1992 as suppliers pursued better rates in

international markets. As a result, the share of non-CIS nations in trade steadily

increased, and previous supply linkages became disadvantageous in a more competitive

economic environment (IMF 1995).

During the recovery period, the role of energy projects was also more clearly

observable in Azerbaijan’s foreign economic relations compared with the recession

period. This small post-Soviet country in the South Caucasus began to wield its

34

See “Monitoring and Evaluation 2016–2021” by the Center for Analysis of Economic Reforms and

Communication for more details

34

geopolitical power in pipeline negotiations with the West, presenting itself as an

alternative to the European Union (EU)’s energy security diversification plans

(Abdullayev 2017). After the transit crisis between Russia and Ukraine, the EU

Parliament and the Council of the EU classified the Southern Gas Corridor as “NG3”

gas pipelines from the Caspian Basin and Middle East to Europe (Abdullayev 2017).

This allowed Azerbaijan to access new markets and build political relations with EU

countries, especially Bulgaria, Romania, Hungary, and Austria, which supported the

emergence of new energy streams (Abdullayev 2017).

During the recovery period, several economic projects, a massive inflow of FDI,

new agreements, and infrastructure projects to transfer oil and gas were completed in a

short period of time. For instance, the Contract of the Century, also known as the ACG

Fields Agreement, was a mega-project valued at USD 11 billion, with an estimated

producible oil volume of 730 million tons and natural gas reserves of 96 billion m3

(Aras et al. 2016).

35

Other significant projects in which Azerbaijan’s oil and gas

industry played a key role included the Nabucco gas pipeline (later renamed Western

Nabucco), the Trans-Anatolian Gas Pipeline (TANAP), the Trans-Adriatic Pipeline

(TAP), the Interconnector Turkey–Greece–Italy, and the Azerbaijan–Georgia–Romania

Interconnector (Abdullayev 2017).

The structure of foreign trade in Azerbaijan also underwent rapid change during

the recovery period (Mogulveskij and Točickaja 2005). At the end of June 2004, high

oil prices led to a 34% year-on-year increase in the value of exports. Imports also grew,

primarily in capital items connected to the expansion of the oil and gas sector. As a

result, the trade balance dramatically deteriorated, rising from 1.9% of GDP in 2003 to

2.7% in 2004. In addition, the share of machinery, equipment, and transportation in

imports rose from 19.2% to 40.3% between 1995 and 2003. This indicated the use of

imports to reequip and modernize Azerbaijan’s weakened economy.

The oil boom period began with decreased imports of capital goods due to the

completion of large oil and natural gas projects (EBRD 2006). As a result, a huge trade

surplus of nearly 4% of projected annual GDP was recorded in 2006, compared with a

deficit of 6% in 2005 (EBRD 2006). In addition, the ratio of gross public external debt

35

To deliver oil extracted from the ACG fields, Azerbaijan initiated the Baku–Tbilisi–Ceyhan (BTC)

pipeline project, which was signed by the presidents of four countries (i.e., Azerbaijan, Georgia, Turkey,

and Kazakhstan) in 1999; in 2006, the pipeline went into operation (Bayramov 2019).

35

to GDP declined from 18.6% in 2004 to 13.1% at the end of 2005 as the result of the

government’s policy to reduce external debt (EBRD 2006).

Kocharli (2011) discussed the rapid decrease in the manufacturing sector’s share

of exports from 1995 to 2008. In addition, Ibrahimova (2017) highlighted that food

products accounted for 41.5% of total imports in 1995, compared with 19% in 2000 and

14.9% in 2015. Pylin (2015) stressed the decreased share of agricultural and non-oil

manufacturing exports, which began during the oil boom period.

36

Azerbaijan’s main trade partners also exhibited changing patterns during the oil

boom. At the time, the country’s main export partners were Italy, Indonesia, Thailand,

Germany, and Israel (Pylin 2015). Italy was the main importer of Azerbaijani crude oil,

which Pylin (2015) attributed to its optimal geographic position in the Mediterranean

Sea. This meant that it could easily trade with Azerbaijan with mediation from Turkey.

However, established political and cultural relations also played a decisive role in

determining the direction and intensity of economic relations between them (Bernardini

2017).

37

In addition, Russia was a leading trade partner from 2000 to 2013; it mainly

imported agricultural products from and exported manufactured consumer goods to

Azerbaijan. However, oil-led consumption in Azerbaijan shifted import patterns toward

more competitive EU countries, such as the United Kingdom, Germany, and France

(Pylin 2015).

The oil boom period coincided with the global financial crisis of 2008–2009,

when Azerbaijan’s oil exports to the EU significantly decreased due to general

diminished consumption. Meanwhile, increasing economic activities in Asian countries

such as Thailand, India, and Indonesia allowed Azerbaijan to dominate new markets

(Pylin 2015). This was also made possible by its newly established marine

infrastructure, which included 345 tankers. This played a crucial role in the country’s

foreign economic relations during the oil boom period (Pylin 2015).

Tahirova et al. (2021) highlighted Azerbaijan’s increased propensity to trade

with post-Soviet countries during the post-boom period. Thus, the potential to expand

the structure and volume of foreign trade turnover between Azerbaijan and other post-

36

Ibrahimova (2017) attributed this to the extensive development of agriculture, a lack of state support

for some food products, and the development of agriculture that was in practice based only on domestic

needs. Therefore, the production of agricultural products that were previously exported from Azerbaijian

sharply decreased, and the full export potential of agricultural products was not achieved. These problems

remain ongoing, and the crowding-out effects of the oil boom became more visible during the oil boom

period.

37

Bernardini (2017) also emphasized the role of the increased student exchanges between Azerbaijan and

Italy. And there was an increase in Italy’s cultural and artistic products to Azerbaijan.

36

Soviet nations has not been realized. To meet modern challenges, it is therefore

necessary to develop and implement more adequate mechanisms and forms of

partnership. Tahirova et al. (2021) cited bilateral meetings, business forums, and other

events that are currently being conducted to increase the productivity of foreign

economic relations and cooperation.

38

During the post-boom period, Azerbaijan’s foreign economic relations have

reflected two major changes, namely governmental support to boost non-oil trade and

national currency devaluation. Together, these changes may work in Azerbaijan’s favor

if appropriate industrial and other economic policies succeed. However, there is no

guarantee that such reforms and changes will result in the development of non-oil

tradeable sectors. This represents a missed opportunity to decrease the country’s

dependence on oil.

2.1.5. Creation of a private economy

In 1991, all former Soviet property in Azerbaijan was nationalized, but no individual

restitution law was enacted (EBRD 1995). The main phase of privatization started in

1993 and continued for 11 years. More than 29,000 small and 1,500 medium-sized

enterprises were privatized, whereas large state-owned companies were privatized

slowly (Baranick–Salayeva 2005).

39

A delay in presidential approval for a new

company law (passed by the parliament in spring 1993) significantly stalled the

development of the private economy during the recession period (EBRD 1995). The

government eliminated subsidized credits for SOEs from two major state banks, namely

Agroprom and Prominvest, to help restructure and privatize SOEs (EBRD 1995).

Im et al. (1993) described the early features of privatization in Azerbaijan in a

case-by-case approach that consisted of an annual multi-track path and considered

various sizes of SOEs. This particular approach proved to be remarkably slow. The

initial strategy was not mass privatization; rather, the main focus was on medium-sized

and large enterprises, relegating small enterprises to future stages of privatization. The

38

For example, on December 17, 2019, a bilateral business forum was held in Baku, where the most

pressing problems and prospects to enhance the business environment between Azerbaijan and the

Ukraine were discussed with the participation of the presidents of both countries. This event was attended

by 100 companies from Azerbaijan and 90 companies from the Ukraine. Tahirova et al. (2021) also

outlined Azerbaijan’s increased cooperation with China, Kazakhstan, Belarus, Georgia, and Turkey via

trade houses established in capital cities to promote mutual trade.

39

The sluggish pace of privatization was also seen in the land and housing sectors, as the relevant

legislation remained under development until 1993. Moreover, according to the EBRD (1995), most

registered private enterprises were inactive.

37

government identified five main approaches for privatizing state assets: leasing, sales to

labor unions, transfer of assets to a joint company, sales through tenders, and auctions

(Im et al. 1993).

40

However, mass privatization was quicker as large and diverse groups

of buyers could receive a substantial portion of public assets either for free or at a

minimal charge through a voucher system (Lieberman et al. 1995). Indeed, the voucher

system – introduced in Azerbaijan in 1996 – intensified the privatization process in

Eastern European and FSU countries (Alexandrowicz 1994).

Despite abundant foreign assistance and the government’s willingness to

expedite the creation of a private economy, the recession period merely provided an

introduction to the privatization process. Due to short-lived administrations, the

government lacked the domestic capacity to systematically implement reforms. In

addition, political and economic difficulties slowed the pace of privatization in

Azerbaijan between 1991 and 1994, according to a World Bank (1995) report.

Moreover, additional factors were required to ensure successful privatization:

“[E]conomic reform, including creating conditions for a stable macroeconomic

environment, trade liberalization, price liberalization, financial sector reform,

elimination of subsidies, a pro-competition policy, and regulatory reform may be

important elements in a successful privatization program” (Lieberman 1993: 15).

However, more successful privatization outcomes were observed during the

recovery period. In 1995, then-president Heydar Aliyev signed the State Program of

Privatization of State Property in the Republic of Azerbaijan in 1995–1998. The

program institutionalized the privatization process, attempting to apply up-to-date

lessons learned from other countries that had undergone a similar transition to a market

economy. The voucher system was introduced to involve most of the population, while

privileged sales to labor unions began to be regulated more systematically. From

January 1, 1997, all Azerbaijani citizens received four vouchers that corresponded to

one privatization share of state property. Overall, 32 million privatization vouchers

were produced, accounting for 65% of state property.

41

The second stage of privatization began in August 2000 when a law called “On

Privatization of State Property” and the Second State Program of Privatization entered

40

Other types of privatization included deep discounts, liberal arrangements for employee and

management buyouts, transfers of the state property to entrepreneurs at book value, management

contracts, leases, and contracting out state property.

41

This paragraph is based on material entitled “Ten years of privatization in Azerbaijan,” which the

author is unable to cite due to numerous limitations (e.g., lack of author name, publication year,

publishing house, etc.).

38

into force. The main difference between the first and second stages of privatization was

the increased involvement of labor unions. In addition, the pace of privatization

increased among small state enterprises and foreigners were allowed to participate.

42

Moreover, the second State Program not only focused on the transfer of state property

to the population but also considered the development of businesses after privatization.

Thus, tax and other benefits, loans on concessional terms, energy and gas supply, the

writing off of debts to the state, and other benefits were offered to privatized

enterprises.

During the early years of the recovery period, the private sector accounted for

approximately 30% of GDP and 25% of employment (EBRD 1997).

43

The creation of a

private economy was stimulated by cutting direct subsidies from the state budget in

spring 1995, including subsidies for bread (EBRD 1995). Although the government’s

full attention was directed to the privatization of SOEs, its preparation for a market

economy was slow due to a lack of corporatization, legal and regulatory frameworks,

and low buyer interest arising from excessive privatization prices (World Bank 1995;

1996).

44

However, after the introduction of the privatization program in 1997, “1,065

medium-sized and large enterprises were transformed into joint-stock companies and

privatized by the end of 1999” (EBRD 2000: 134). Moreover, in September 1998, the

state implemented competitive tenders to increase case-by-case privatization. However,

by the end of the first quarter of 2000, “only five companies had been privatised

through this mechanism” (EBRD 2000: 134). Moreover, the establishment of the State

Property Ministry in February 2000 and a new privatization law adopted in May 2000

failed to develop concrete and transparent processes to assist privatization (EBRD

2000). In addition, the validity of previously provided vouchers and options (the bulk of

which were owned by foreigners) was extended beyond August 2000 (EBRD 2000).

Nevertheless, the recovery period delivered the most critical change in

Azerbaijan’s economy – namely the Land Reform Law of 1996, which turned land into

42

During the second stage of privatization, foreign investors used options with vouchers to buy state

property. According to the law (On Privatization of State Property), an option is a non-cash registered

security that gives foreigners and investors the right to participate in privatization through vouchers.

Accordingly, foreign investors participated in specialized check auctions by offering options. In addition,

they were able to participate in privatization without providing options (at the expense of Azerbaijan’s

profits) when they reinvested their net profits into the country.

43

This excludes enterprises with less than 50% private ownership. In addition, the EBRD (1997) reported

that 40% of the GDP was privately produced based on the unofficial economy.

44

In 1995 and 1996, the state’s role in the economy was still too prominent, and the early years of the

recovery period required an environment that was conducive to sustaining private sector development in

Azerbaijan (World Bank 1996).

39

private property (Hasanov 2001). Under this law, 90% of arable land (22% of total land)

was to be privately owned, 45% of total land was reserved for state ownership, and 33%

was reserved for municipal ownership (EBRD 1997). Around 2000, the privatization of

agricultural land and the issuance of land titles were nearly complete (EBRD 2000).

During the second half of the recovery period, the greatest challenge for the state

was to improve the business environment. The EBRD (2005) reported that corruption

was the main obstacle to the development of the private economy in Azerbaijan. An

anticorruption law was adopted in 2005, but the EBRD (2005: 102) mentioned that its

“implementation [was] still uncertain.” The state’s SME development initiative, which

began in August 2002, aimed to eliminate unnecessary licensing and regulations as well

as corruption and other barriers to growth in the private sector (EBRD 2003).

45

In

addition, it was not only the private sector that struggled but also SOEs, which remained

highly inefficient and opaque. Indeed, poor governance among SOEs was especially

notable during the second half of the recovery period (EBRD 2005). In June 2005, the

government declared that it would improve the supervision and financial discipline of

crucial SOEs, including SOCAR. At the end of this period, the state adopted certain

measures to address the slowdown in the development of the private economy. For

instance, new antimonopoly legislation was introduced in July 2005. In addition, the

World Bank assisted in the development of investment legislation and competition laws

(EBRD 2005).

During the oil boom period, Azerbaijan’s business environment significantly

improved; however, entrepreneurial opportunities in the private share of non-oil sectors

remained scarce due to the strong positions of monopolies, restricted entry to various

economic sectors, corruption, and bureaucratic delays (EBRD 2008). The World Bank’s

“Doing Business 2009” report ranked Azerbaijan as one of the top countries for

reforming its business environment among 181 economies. In January 2008, “one-stop

shops”

46

were established to simplify business registration and licensing procedures; by

then, the number of registered firms had already dramatically increased (EBRD 2008).

During the first half of the oil boom period, electronic tax filing was introduced and the

process of registering real estate was simplified (EBRD 2008). In addition, starting from

45

It was anticipated that the newly formed Business Council would provide suggestions to further boost

the investment climate during the second half of recovery period (EBRD 2003). In the end, the oil boom

made privatization efforts less important because officials were more interested in the oil rents.

46

In previous years, entrepreneurs had to visit several governmental units to register and start their

businesses. This situation led to numerous corruption cases in Azerbaijan.

40

2008, the use of international financial reporting standards (IFRS) was made obligatory

for all large companies (EBRD 2008). All of these measures improved Azerbaijan’s

business environment and supported the creation of a private economy.

However, the World Bank’s (2009) Enterprise Surveys Country Note Series

reported several persistent systemic flaws that hindered the development of the private

economy. Specifically, domestic firms still faced greater oppression and discrimination

from government officials (e.g., bribery and unnecessary tax controls) than foreign

companies during the oil boom period, in addition to gender discrimination (e.g.,

female-run firms received fewer visits from officials than male-run firms). However, the

same report also indicated a significant decrease in corruption and improvement in

access to finance in 2009 compared with 2005. In addition, the main impediment for

Azerbaijani entrepreneurs was difficulties in securing initial investors, partners, and

funding to start a business (Kuriokose 2013). Failing was also problematic, as changing

taxation regulations and bankruptcy legislation were among the most critical legal and

regulatory constraints that discouraged entrepreneurs from pursuing their ideas in the

Azerbaijani market. Based on other findings and recommendations from Kuriokose

(2013), it could be argued that the Azerbaijani government did not really engage in

reform to boost high-growth entrepreneurialism, which could have also assisted non-oil

diversification. Instead, the oil boom was the government’s main focus since it was the

major contributor to the economy. Meanwhile, a lack of competition, risk capital, and

industry-relevant skills led to the stagnation of entrepreneurship.

The second half of the oil boom period was notable due to rapid developments

in the information and communication technologies (ICT) sector, which became the

second-largest industry to receive foreign investment after the oil industry. This

stimulated the development of a private economy, especially in non-oil manufacturing.

The EBRD (2014) reported an average increase of 25–30% between 2003 and 2013. “In

July 2014 the State Fund for the Development of IT, which was established in 2012,

awarded grants to 31 start-up projects in areas such as high-technology, e-payment

software applications, air navigation systems and e-government” (EBRD 2014: p. 97).

In 2014, the drop in oil prices led government officials to redress the imbalance

by pursuing diversification, investing in physical and human capital with greater

participation from the private economy, and raising awareness about the fragile status of

Azerbaijan’s economic structure (World Bank 2019). “Article 14 of the Law of the

Republic of Azerbaijan on the State Budget of the Republic of Azerbaijan for 2021

41

states that the financing of the state budget deficit should be carried out through

privatization and revenues from other sources” (Ibadoghlu 2021). In fact, during the

early years of the post-boom period, the government rapidly remapped new policies to

stimulate privatization. On May 19, 2016, President Ilham Aliyev signed a decree to

improve the privatization of state property. It established new privatization methods,

such as individual project-based privatization

47

and investment competition

48

(Ibadoghlu 2021). Furthermore, the Cabinet of Ministers was instructed to develop and

submit the new Draft Law of the Republic of Azerbaijan on the Privatization of State

Property. After five years, the Cabinet still had not provided a reworked legal

framework for regulating the new phase of privatization (Ibadoghlu 2021). In addition,

new attempts to privatize SOEs continued to fail due to their economic unattractiveness

(Ibadoghlu 2021). For example, Dashkasan Filizsaflashdirma OJSC had more than 270

million tons of iron ore reserves and was considered one of the most strategic SOEs in

the non-oil extractive industry. The Azerbaijani government invited potential investors

from the USA, the EU, the United Arab Emirates, and India and assigned the private

organization KPMG

49

as a private advisor (emlak.gov.az 2019). The deadline for

applications and offers was May 31, 2018; however, as of today, Dashkasan

Filizsaflashdirma OJSC has not been privatized. Its case highlights the challenges of the

new privatization phase, which is strongly interlinked with the institutional and

governance failures in Azerbaijan’s economy that began during the oil boom period

(Ibadoghlu 2021).

In October 2016, the Cabinet of Ministers decided to provide subsidies for

several export products in non-oil sectors from March 1, 2017 to December 31, 2020

(Ibrahimova 2017).

50

In addition, tax reforms entered into force in January 2019. They

were intended to improve transparency and accountability as well as to encourage

47

Individual project-based privatization is a process that tries to identify appropriate investors via

proposed projects that take into consideration the features of the enterprise and it addresses the

government’s planned goals.

48

Investment competitions combine a limited version of some forms of demand, forcing the state to

stimulate private capital investment in exchange for low profits. Thus, when property is sold through

investment competitions, the investor undertakes a three-, five-, or sometimes even 10-year investment

commitment to develop the enterprise. By applying an effective control mechanism to the fulfillment of

investment obligations, it may require the modernization of an owned enterprise based on financial

investments and the effective organization of its activities.

49

Klynveld Peat Marwick Goerdeler (KPMG) is one of the big four multinational accounting

organizations.

50

This law covered 25 commodity items, of which six (except for naphthalene ointment) were light

industry products, while the others were food products. The latter included natural honey, dried fruits,

national sweets, spices, fruit and vegetable juices, canned food, fresh fruit, mineral waters, grapevines,

and other alcoholic and non-alcoholic drinks (Ibrahimova 2017).

42

businesses to formalize (i.e., emerge from the shadow economy) and participate in

economic diversification.

51

Thus, the tax rate for micro-, small, and medium-sized

enterprises (MSMEs) was reduced from 4% to 2%.

Moreover, the regional support infrastructure for SMEs was strengthened

(EBRD 2020: p. 3) during the post-boom period. In early 2020, Azerbaijan’s newly

founded Agency for the Development of SMEs launched the inaugural House of SMEs.

This platform, based in Khachmaz in the northern part of Azerbaijan, gathers

government and private sector services in a single location. These services include

assistance with the development of business and marketing plans; business knowledge

enhancement; and the acquisition of registration and various licenses to access financial

resources, supply chain and infrastructure networks, domestic and foreign markets, and

trade facilitation. The post-boom period has also been memorable due to an extended

development program called the Regional State Development Program of the

Azerbaijan Republic on Economic and Social Development for 2019–2023. The

program includes several measures designed to stimulate export-oriented

entrepreneurship and increase local employment and production among others (Niftiyev

2020a).

2.2. Macroeconomic overview of Azerbaijan’s economy

During the recession period, Azerbaijan’s economic growth was −16.4% and its per

capita GDP was USD 142.3 per annum (see Table 2.1). The first half of the recovery

period saw a GDP growth rate of 2.5%. Between 2000 and 2004, Azerbaijan’s GDP

rose by 10% and continued to exhibit strong growth during the first half of the oil boom

period, reaching 18.8% between 2005 and 2010. However, the second half of said

period saw a growth rate of 2.3%. The international commodity price downturn severely

affected Azerbaijan’s economy during the post-boom period, which led to a growth rate

of −0.4%. Similarly, as Table 2.1 indicates, the per capita GDP recovered during the

recovery period, achieving a 17.1% growth rate between 2005 and 2010 compared with

−17.8% during the recession period. Therefore, the greatest jump in per capita GDP

occurred during the first half of the oil boom period, while the second half brought a

51

In January 2020, President Ilham Aliyev said that, owing to the “fight against the shadow economy,”

tax authorities collected AZN 120 million in addition to the forecasted amount, and customs authorities

collected AZN 40 million.

43

sharp decline (albeit with 1% growth). However, the post-boom period was

characterized by a significant slowdown in per capita GDP (−1.3%).

Table 2.1: Selected economic indicators for Azerbaijan (1991–2020).

Recession

period

Recovery

period

Oil boom

period

Post-boom

period

1991–1994

1995–

1999

2000–

2004

2005–

2010

2011–

2014

2015–2020

Growth

GDP (current prices, billions of USD)

0,802

3,717

6,635

35,555

71,261

44,950

GDP growth (% per year)

−16.5

2.5

10.0

18.8

2.3

−0.4

GDP (USD, constant 2015)

14,536

9,521

14,715

37,689

49,422

52,206

Per capita GDP (USD, current prices)

142.3

473.0

810.1

4,045.2

7,259.3

4,548.0

Per capita GDP (USD, constant 2015)

1,964.2

1,213.8

1,797.8

4,306.5

5,315.2

5,280.6

Per capita GDP growth (% per year)

−17.8

1.5

9.1

17.1

1.0

−1.3

Economic openness

Exports (% of GDP)

53.5

27.4

43.0

61.5

50.3

45.3

FDI (% of GDP)

N/A

19.1

31.7

15.0

6.0

6.5

Terms of trade1

N/A

97.5

157.0

200.3

126.3

REER2

22.6

87.9

79.8

106.5

133.8

98.9

Structural change (value-added, % of

GDP)

Agriculture, forestry, and fishing

29.0

21.4

13.7

6.7

5.2

5.9

Manufacturing

17.3

8.9

7.2

5.4

4.3

5.0

Industry (including construction)

31.5

35.4

46.3

61.9

58.0

47.4

Services

34.3

37.4

33.3

25.6

30.7

38.8

Oil rents3 (% of GDP)

Oil rents in Azerbaijan

10.6

13.1

25.1

33.9

26.0

18.1

Average oil rents in Azerbaijan

between 1991 and 2019

21.7

Average oil rents in other oil-rich

post-Soviet countries4

5.8

7.5

18.1

13.7

8.9

Source: World Bank, World Development Indicators.

Notes: “N/A” means “not applicable”; GDP = gross domestic product; FDI = foreign direct investment;

all numbers are rounded up to the first decimal place.

1 In percent. Base year is 2000 or 2000 = 100%.

2 In percent. Base year is 2007 or 2007 = 100%.

3 Oil rents are the difference between the value of crude oil production in regional prices and total costs of

production.

4 The oil-rich post-Soviet countries are Azerbaijan, Kazakhstan, Russia, Turkmenistan, and Uzbekistan.

As a percentage of GDP, exports grew from 27.4% to 43% during the second

half of the recovery period (between 2000 and 2004), peaking at 61.5% between 2005

and 2010. Exports also accounted for a high percentage of GDP (50.3%) during the

second half of the oil boom period due to increased oil exports. However, they

accounted for 45.3% of GDP during the post-boom period, which meant a decrease in

export growth. Moreover, the positive trend of FDI in Azerbaijan’s economy began to

decline during the oil boom period, largely due to the completion of upstream and

downstream projects in the oil industry; it bottomed out at 6% of GDP between 2011

44

and 2014. In addition, a similar level of FDI flows was observed during the post-boom

period (6.5% of GDP). TOT and the REER exhibited significant upward trends during

the recovery period, but both indicators rapidly decreased during the post-boom period;

TOT was 200.3% from 2011 to 2014 and dropped to 126.3% from 2015 to 2020, while

the REER was 133.8% from 2011 to 2014 and declined to 98.9% from 2015 to 2020.

In terms of structural changes in Azerbaijan’s economy, the share of total value-

added in the agriculture, forestry and fishery, and manufacturing sectors decreased from

the recession period to the post-boom period. However, due to both the extractive

sectors and the construction boom, the industrial sector’s share of total value-added rose

to 61.9% from 46.3% during the oil boom period. Moreover, the service sector’s share

of total value-added ranged from 25.6% to 38.8% between 1991 and 2020.

Lastly, because of the high shares of oil rents in the economy, Azerbaijan can be

considered an oil-dependent country (see Table 2.1). In each developmental stage of the

economy, oil rents have occupied a higher percentage of GDP compared with other oil-

rich countries in the post-Soviet space. Between 1991 and 2020, oil rents accounted for

21.7% of GDP per year on average; this threshold was surpassed during the second half

of the recovery period and during the oil boom period.

Table 2.2 presents the polarized structure of Azerbaijan’s economy in terms of

overall investments, employment, output, and trade. More specifically, the oil boom

(63.90%) and post-boom periods (61.20%) were marked by the mining industry’s

dominance in both the national economy and FDI. However, investments in the

manufacturing industry sector nearly doubled during the post-boom period compared

with the oil boom period.

Furthermore, the employment section in Table 2.2 indicates the mining

industry’s declining share in total employment. This highlights its capital-intensive

nature and the low levels of benefit that can be obtained from this industry in terms of

salaries and wages. Nevertheless, the agriculture and service sectors are the two largest

employers in Azerbaijan. The service sector has continued to occupy the highest share

of total employment since the beginning of the recovery period.

In terms of the structure of industrial output, the mining industry and the

manufacture of refined petroleum products reached 84.27% during the oil boom period,

61.12 and 19.16 percentage points higher than during the recession and recovery

periods, respectively. However, non-oil manufacturing accounted for 66.10% of total

industrial output during the recession period, falling to 19.59% during the recovery

45

period. During the oil boom period, this figure further decreased by 9.52 percentage

points. Only the post-boom period saw an increase in non-oil manufacturing, which is

mainly attributable to the mining industry’s lower output due to the end of the

commodities boom.

Table 2.2: Structure of Azerbaijan’s economy in term of investments, employment,

industrial output, and trade (in %).

Recession

period

Recovery

period

Oil boom

period

Post-

boom

period

1991–1994

1995–1999

2000–2004

2005–2010

Overall investments

Mining

–

63.90

61.20

Manufacturing

–

14.20

26.70

Employment

Mining industry

–

1.04

0.99

0.81

Manufacturing industry

11.20

4.87

4.92

5.14

Agriculture

32.60

35.86

37.96

36.26

Services

43.30

51.26

56.13

57.78

Industrial output

Mining industry

and manufacture of refined petroleum products

23.15

65.11

84.27

75.36

Manufacturing industry

66.10

19.59

10.07

18.66

Other

10.75

15.30

5.66

5.98

Trade (exports)

Manufacturing industry

–

10.50

4.70

4.00

Agriculture

–

4.80

4.20

4.30

Other

–

0.00

0.20

0.80

Source: State Statistical Committee of the Republic of Azerbaijan.

Notes: (i) The category “Other” under “Industrial output” includes electricity, gas, and steam production;

the distribution of supply; water supply and waste treatment; and disposal; (ii) total values may not equal

100% in all cases because some calculations excluded categories such as “Other”; (iii) the recovery

period only encompasses the period 1999–2004 due to data limitations; (iv) under “Trade (exports),” the

recovery period begins in 1996; (v) under “Trade (exports),” the category “Mining industry” includes

inedible crude materials; (vi) under “Trade (exports),” the category “Other” includes commodities that are

not included in the structure of exports according to the Standard International Trade Classification

(SITC).

Lastly, the trade structure of Azerbaijan’s economy indicates that the oil-based

mining industry had the largest share in exports while manufacturing bottomed out at

around 4% during the oil boom and post-boom periods. In addition, the agriculture

sector has not exhibited any increase in exports since the oil boom.

2.3. Summary of the chapter

This chapter summarized the main stages of the development of Azerbaijan's economy

in terms of macroeconomic stability and stimulation, institution building, liberalization

46

of the economy, foreign economic relations, and creation of a private sector. The

modern economic development of Azerbaijan can be divided into four distinct

developmental stages, namely the recession period (1991–1994), the recovery or

restructuring period (1995–2004), the oil boom period (2005–2014), and the post-boom

period (2015–2020). Despite limited information on the recession and post-boom

periods, it can be said that Azerbaijan survived the severe economic downturn of the

recession period and began to rapidly recover from the devasted economy during the

restructuring period. The impact of the oil boom on Azerbaijan’s economy was more

clear cut compared with the previous two stages, leading to the mining industry’s

increased share in investments (both domestic and FDI), employment, output, and

exports. This structural imbalance has created a set of challenges that are currently

being investigated in diversification studies on Azerbaijan’s economy (Bayramov–

Abbas 2017; Guliyev 2020; Ahmadova et al. 2021; Hamidova 2021).

Oil-based economic development fueled economic indicators such as GDP,

GDP per capita, FDI, and exports. However, long-term sustainable economic growth

and development, which could reduce a country’s dependence on volatile commodity

markets, have not been achieved. Oil prices determine the trajectory of the main

macroeconomic indicators and the structural distribution of Azerbaijan’s economy,

which in turn makes balanced development uncertain. Issues related to crucial policy

reforms and institutional governance either stop or fall to the wayside when oil prices

rise. Shortsightedly, the Azerbaijani government increased its efforts to reform

institutions and governance only when oil prices were low. However, when oil revenues

have risen, this has led to a disinclination among government officials and politicians,

creating a gap in the reform agenda. In such circumstances, macroeconomic instability,

institutional deficiency, a lack of political will and plans for liberalization, and

underdevelopment of the private economy in Azerbaijan have allowed DD and the NRC

to grow. Therefore, the next chapter is an empirical test of the NRC phenomenon as a

prerequisite for an economic explanation, which is DD. Therefore, the next chapter is an

empirical analysis of the NRC phenomenon in Azerbaijan. This needs to be done before

the theory of DD can be used to provide an economic explanation for the negative

effects of the oil boom.

47

CHAPTER 3

LITERATURE REVIEW

3.1. Natural Resource Curse phenomenon

The term NRC refers to the slower economic growth of resource-rich countries relative

to resource-poor countries (Auty 1993). Numerous studies have provided a solid

foundation for resource curse-related studies, enabling an enhanced understanding of

the economic structure of resource-rich countries relative to resource-poor countries.

NRC theory has been discussed since 1970; pioneering papers were by Sachs and

Warner (1997; 1998; 1999; 2001), who found an inverse relationship between natural

resource abundance and resource dependence,

52

and GDP performance in cross-country

studies. They also highlighted the fact that mineral-rich countries tend to be expensive

countries, which hinders export-led industrialization in the long term. Furthermore,

Auty (2001) revealed that income per capita growth was higher in resource-poor

countries than resource-rich countries between 1960 and 1990. In fact, among the

largest mineral exporters, the annual GDP per capita growth rate decreased from 1980

to 1993 following the boom period of 1970 to 1980 (Mikesell 1997).

53

Mikesell (1997)

also noted that the average annual GDP growth rates of mineral exporters declined after

commodity prices collapsed from 1980 to 1993. In a more recent study, Sharma and Pal

(2020) found evidence for the existence of the resource curse phenomenon in the short

and long term based on a panel of 111 countries from 1996 to 2015. They observed a

negative impact of resource dependence on economic growth.

If a downward trend occurs in main commodity prices in the long term, then the

NRC may pose a serious threat to mineral-rich countries (Arezki et al. 2014). This could

lead to trade deterioration or simply the contraction of mineral revenue.

54

A growing

body of literature related to the NRC and DD has cited other risks. For instance, through

52

A country is considered to be abundant in natural resources if more than 40% of its national income is

generated by extractive industries, according to Auty–Warhurts (1993). Sachs and Warner (1997)

calculated resource dependence as the ratio of primary exports over GDP.

53

The exceptions are Chile, Jamaica, Papua New Guinea, and Oman.

54

Prebisch (1950) and Singer (1950) formulated the first notable hypothesis regarding the negative long-

term trend in raw material prices. Since then, numerous debates have taken place on this topic. For

example, Frankel (2012) argued that the resource curse is not well suited to the Prebisch-Singer

hypothesis because a consistent negative trend in commodity prices is not observed. However, in contrast

to Frankel (2012), Harvey et al. (2010) found a negative trend in major commodity prices in the long run

(which supports the Prebisch-Singer hypothesis), but not every commodity displayed a negative trend.

Some commodities did not show any trend at all in Harvey et al.’s (2010) study.

48

the effects of DD,

55

REER appreciation significantly reduces the productive capacity of

non-resource tradeable sectors (Krugman 1987), encourages corruption, and decreases

bureaucratic quality (Busse–Gröning 2013). The resource curse also hinders knowledge

accumulation and capital formation (Welsch 2000), which harms education levels as the

need to invest in education to provide specialized human capital to crowded-out

manufacturing sectors is reduced (Wadho 2014). Moreover, a study found that

“knowledge accumulation and capital formation are inversely related to the natural-

resource intensity”

56

(Welch 2000: 62).

Some countries, such as Norway, Botswana, Indonesia (Gurbanov–Merkel

2009), Chile (Havro–Santiso 2017), and Iceland (Gylfason–Zoega 2006), have managed

to increase their economic growth and distribute their natural resource revenues more or

less well by minimizing the negative impacts of resource abundance through

institutional arrangements. Nevertheless, their structural problems and social challenges

remain to some extent. Thus, general claims of the existence of the NRC in a country or

region should be handled very carefully. If institutions function well and the state

distributes income equally and efficiently, natural resources may be a blessing rather

than a curse, boosting economic growth (Acemoglu et al. 2005). However, if a country

becomes dependent on a single commodity as a source of revenue during its

developmental stages and has weak institutions, macroeconomic destabilization may be

inevitable due to volatile commodity prices and political challenges (Venables 2016).

Furthermore, the effects of the NRC may vary over time. For instance, Collier–

Goderis (2007) demonstrated that the short-term effects of commodity booms may be

positive, but their long-term influence may be very harmful. Hence, the resource curse

thesis may be more visible over a long time span than over a short one. Thus,

policymakers should focus on long-term policy solutions to lessen the NRC’s effects.

Although a popular field of research, NRC theory must be investigated against a

variety of theoretical considerations to more accurately establish the interplay between

resource wealth, economics, and politics (Torvik 2009). Azerbaijan is no exception—its

economy heavily depends on the oil and gas sector, while non-oil manufacturing has

55

According to Corden and Neary (1982), Corden (1984), and Brahmbhatt et al.’s (2010) original theory,

the DD phenomenon results from structural change caused by the discovery of large mineral resources or

extractive industry-based development. DD means a loss of competitiveness due to an appreciated real

exchange rate, a productivity decrease in non-mineral sectors, and heavy dependency on state

expenditures.

56

Here, natural resource intensity measures the efficiency of the resource use spent to produce one unit of

GDP (Lorentzen 2008). Resource-rich countries are considered to be highly resource-intensive, as GDP

heavily depends on resource extraction and exports.

49

decreased and become reliant on government subsidies (Niftiyev 2020c). Moreover, as

the main objective of the present study was to examine economic policy solutions to

deindustrialization in Azerbaijan, an investigation of the resource curse thesis could

answer frequently asked research questions about Azerbaijan’s economy. Theoretical,

descriptive, and empirical studies are not conclusive on this topic. Each successful year

offers a casual glance through the lenses of various determinants of the NRC, which

must be analyzed to obtain an understanding of the topic.

3.1.1. Main concepts, key terms, and NRC measurement approaches

With regard to NRC syndrome, the primary interest lies in understanding the impact of

natural resources—especially sub-soil hydrocarbon resources such as oil—on economic

growth. Natural resources affect economic growth through both macroeconomic

indicators (directly) and social institutions (indirectly; Gylfason–Zoega 2006).

Favorable commodity prices boost the flow of mineral revenue into the exporter

country. However, as soon as prices noticeably decrease, the country will experience

macroeconomic challenges that can leave lasting scars. Moreover, if the country suffers

from corruption, inefficient government spending, low human capital (assets), and rent-

seeking behavior from politicians, slower growth will be compounded.

To test the relevance of NRC theory to Azerbaijan’s economy, appropriate

variables first needed to be chosen. Concepts such as “natural resource-rich” and

“natural resource-poor” countries must be clarified, as the appropriate use of

terminology is critical in any theoretical analysis. To understand the NRC doctrine, it is

crucial to define the term “natural resource capital,” which encompasses not only

natural resource stocks (renewable and non-renewable) but also land and ecosystems

(OECD 2011a). In general, NRC studies cover a mineral resource that enters non-

renewable natural resource stocks (OECD 2011b). The concept of natural resources has

a broad meaning, namely the natural and environmental resource wealth available to an

economy (Barbier 2002: 488); however, Ross (2013) claimed that the resources curse

only applies to oil resources that have fostered authoritarian governments and civil wars

in developing or underdeveloped countries since the 1970s. According to Ross (2013),

four aspects of oil lead to the resource curse: large rents, unusual sources of government

revenue, secrecy, and volatile prices.

57

Sala-i-Martin and Subramanian (2003), Murshed

57

Extractive industries generate enormous amounts of income within a short time period, which in turn

increases the scale of the government revenue and expenditure. Ross (2013) evaluated extractive

50

(2004), and Isham et al. (2005) have also made similar arguments—that the central role

of point-source natural resources in the economic structure is the main reason for NRC

syndrome.

Point-source natural resources refer to dense concentrations of resources (e.g.,

oil and minerals) on a geographically narrow scale, as opposed to diffuse resources that

encompass wider areas such as forests (World Trade Organization 2010). Statistical

evidence suggests that point-source natural resources indirectly hinder economic

development through institutions as they are easier to capture and control—especially in

non-democratic regimes (Isham et al. 2005; Mehlum et al. 2006; Wick–Bulte 2009).

Therefore, NRC theory is heavily skewed toward mineral-rich countries. However, it is

not limited to energy resources.

58

If more than 10% of a country’s GDP is based on the mining sector and more

than 40% of foreign earnings are derived from mineral revenue, then that country can be

viewed as a mineral-rich, resource-rich, or resource-abundant country (Auty–Warhurts

1993). However, the NRC doctrine is not applicable to every single resource-abundant

country. Rather, chronic resource dependency (economic channel), sociopolitical

conditions, and institutional failures are the factors that can hinder a country from

overcoming the effects of the NRC, thus establishing long-term sustainable economic

development.

Sachs and Warner (1997) defined “resource dependence” as the ratio of primary

products to the gross national product (GNP) of a country.

59

If a country is resource-

abundant, then a high probability exists that the government will side with increasing

extraction and exports to benefit from boom periods in international commodity cycles.

However, Ding and Field (2004) presented an argument regarding the objective

preconditions of natural resources for influencing the economic structure. In other

words, a country’s economic structure simply responds to the technological capacities

to transform natural resource stocks into economic growth. Hence, key terms,

definitions, and naturally the resource types are vital for concluding economically

industry-generated rents as non-tax revenue (but his approach is non-specific). Oil prices are extremely

volatile, which creates considerable uncertainty for national economies; Ross (2013) argued that

“secrecy” referred to hiding oil revenue from the public in offshore accounts.

58

As cited in Vahabi (2018), scholars have suggested including forest resources (Price 2003; Harewell,

Farah and Blundell 2011), non-fuel natural resources (Sorens 2011), and overall commodities (Besley and

Persson 2011; Bazzi and Blattman 2014) to capture the broader impact of natural resources.

59

By “primary products,” Sachs and Warner (1997) mainly meant produced, extracted, or cultivated

goods and services. Moreover, Ding and Field (2004: 3) viewed this index as misleading, because “[i]t

registers primarily the sectoral importance of primary industries, in the economy and in terms of exports.”

51

meaningful results. Thus, distinctions among the conceptual dimensions of NRC theory

and spillover effects among natural resource types should be made to capture relevant

effects on the economy (Blanco–Grier 2012). If the definitions and economic conditions

of a particular resource-rich country mean that it is likely to fall under the NRC

doctrine, then policy failures that originate from political and institutional channels can

carry negative connotations of its resource-abundance.

Negative perceptions of natural resource abundance are relatively novel. Rosser

(2006a) chronologically outlined the evolution of how natural resources are perceived; a

positive, growth-enhancing perception was observed between 1950 and 1980, while the

NRC doctrine gained more credence after 1980. Economists such as Lewis (1955),

Rostow (1960), Rostow (1961), Watkins (1963), Innis (1963), Kruger (1980), and

Balassa (1980) defended the soundness of extractive-led industrialization by citing

historical development patterns and the natural evolution of advanced national

economies. In addition, accumulated mineral revenue boosts economic growth and

reduces credit constraints, eventually leading to take-off among low-income countries.

Notably, countries rich in natural resources can experience a “big push” to alleviate

poverty and prepare themselves to produce high value-added products (Rosenstein–

Rodan 1943). The relevance of extractive-led industrialization is indicated by the

examples of the United States (De Long–Williamson 1999), the United Kingdom (Van

Neuss 2015; Stevens 2015), Australia (Lowe 2012), and Norway (Fagerberg–Mowery–

Verspagen 2009; Ville–Wicken 2013). However, the following questions arise: Is this

approach to the role of natural resources in economic growth and development since the

1980s still pertinent?

60

Furthermore, are natural resources conducive to industrialization

in developing, resource-rich, low-income, and transitional countries? According to Ali

et al. (2018), natural resources, especially oil endowments, appear to fail to make

citizens happy in oil-rich countries; in other words, there is an inverse relationship

between oil rents

61

and happiness.

Ali et al.’s (2018) observation enables the most critical intellectual premise of

NRC theory to be examined, namely the relationship between quality of life (in

economic terms) and oil wealth. It is not sufficient to examine the associations between

60

After the 1980s, an overall negative trend in commodity prices was limited to be as flexible as the

period of pre-80s provided. Moreover, since the 1980s there is more data to decide whether mineral-

richness had led to civil wars, the crowding-out process of the non-mineral productive sectors, or other

effects that are traditionally assigned to the resource curse doctrine.

61

According to the World Bank (2020), “oil rents are the difference between the value of crude oil

production at world prices and total costs of production.”

52

economic growth variables, such as GDP, GDP per capita, GDP growth, income levels,

and resource abundance (or dependence). Instead, specific target variables should be

studied, such as political and institutional attributes, human capital development,

bureaucratic quality, and the drive to integrate the extractive industry into the rest of the

economy. To this end, resource curse-related studies must address the direct and

indirect impact channels (i.e., transmission channels) of natural resources. Thus, the

following sections outline the most relevant channels in the context of NRC theory and

the transmission channels of the NRC before empirically testing the theory’s relevance

in the case of Azerbaijan, a small, oil-rich, and post-Soviet country.

The definition of resource abundance itself presents a challenge when selecting a

measurement level to capture its effects. As cited by Stevens (2015), the most popular

measures of natural resource abundance are dependence on primary products (Sachs–

Warner 2001), per capita land area (Wood–Berge 1997), employment in the primary

sector (Gylfason et al. 1999), export concentration, and population (Syrquin–Chenery

1989).

The type of natural resource is also relevant, as cross-country studies have

demonstrated. When measuring diamond production, for example, resource abundance

is positively and significantly related to income growth, whereas oil abundance (as

measured by oil production) is not significantly related to growth (Daniele 2011). Ores

and mineral resources do not appear to significantly impede or harm growth as

measured by economic indicators; however, negative influences on fuel resources and

standards of living have been found, as measured by GDP per capita and the human

development index (HDI; Pendergast et al. 2011). Resource dependence measured by

the share of metals and ores in total exports has a strong and significant relationship

with average growth (Daniele 2011). Furthermore, Stiglitz (2005) suggested using green

GDP to measure sustainable growth against resource depletion, because understanding

how a country grows richer or poorer in terms of how exhaustible resources affect the

environment is crucial.

62

In the NRC literature, the other most critical factor in terms of impact channels

is oil prices (or commodity prices). The role of oil prices in export revenue and fiscal

spending is critical in oil-rich countries and leads to procyclicality (Bova et al. 2018).

Positive oil price shocks or oil price upturns increase GDP growth, which Brückner et

62

Stiglitz (2005) argued that green GDP is a key indicator for measuring sustainable growth dynamics

and that it should be included in theoretical and empirical research on resource curse topics.

53

al. (2012) argued also supports democratic transition. However, negative oil price

shocks undermine the financial development of resource-rich countries, which in turn

increases their dependence on resource exports (Mlachila–Ouedraogo 2020). A strong

dependence on oil prices places oil-exporting countries in a vulnerable position in the

face of oil price fluctuations; for example, between 1990 and 2014, such dependence in

regions such as Sub-Saharan Africa led to lower GDP growth rates, unstable political

indicators, and a deterioration in human capital assets (Vandyck et al. 2011).

Consequently, a general tendency seems to exist that oil reserves and increased oil

production and mineral exports are followed by signs of the NRC.

Sachs and Warner (1995) pioneered research on the NRC by documenting an

inverse, significant, and robust association between resource-based exports

63

as a share

of GDP and economic growth. As the body of literature grew, various frameworks and

characteristic features emerged to explain NRC theory in resource-abundant economies.

According to Torvik (2019), the following indicators should be followed to understand

the NRC: DD, rent-seeking, the political economy, civil conflicts, large public sectors,

and huge, economically inefficient, but politically beneficial domestic investments.

Moreover, Gylfason–Zoega (2006) found an inverse relationship between natural

resource dependence and economic growth.

Although pioneering studies have argued that an adverse relationship exists

between natural resource abundance and economic growth, the findings are mixed

regarding the impact of mineral resource wealth on economic growth and development.

Some studies have found negative effects, while others have found the opposite—how

can this be the case? According to Havranek et al. (2016), the reasons depend on the

heterogeneity of the data and methods applied. Furthermore, NRC syndrome strongly

depends on how it is measured and how economic growth is modeled (Rambaldi et al.

2020). Moreover, resource dependence and the distinction between resource types,

investment levels, and institutional quality influence the relationship between resource

abundance and economic growth (Havranek et. al 2016). In a recent study, Shahbaz et

al. (2019) re-examined NRC theory retrospectively in terms of resource abundance and

dependence for the period 1980–2015 in 35 countries. They concluded that resource

abundance leads to economic growth, while resource dependence negatively affects the

real GDP of mineral-rich countries. Nevertheless, although research related to slower

63

Resource-based exports include agriculture, minerals, and fuels.

54

growth resulting from mineral richness has rapidly evolved and provided empirically

sound evidence for the NRC, the methodologies applied and the cause–effect

relationships remain ambiguous (Davis–Tilton 2005). However, new case studies may

accelerate our understanding of the NRC.

In addition to the types of data used to analyze the NRC doctrine, the type of

resource is another integral part of the analysis. Natural, physical, and human capital

contribute to human welfare through the production function of the national economy

(Barbier 2002); however, different types of natural resources may affect standards of

living, economic growth, and welfare differently. If certain regulatory and institutional

preconditions do not exist, adverse effects or suboptimal performance may be

inevitable. However, the resource curse does not only concern the negative association

between natural resource abundance/dependence and income levels. Wick and Bulte

(2009) noted that the NRC includes several aspects. For instance, developmental

indicators such as the HDI, life expectancy, hunger, and undernourishment may be

suitable as alternative measures of resource abundance. Similarly, Gylfason–Zoega

(2003: 288) demonstrated that natural resources impede growth and increase

inequality: “[I]f the distribution of ownership of natural resources is more unequal than

the distribution of other forms of wealth, the inequality of the distribution of income,

education or land is directly related to the share of natural resources in national

income.” Moreover, Pendergast et al. (2011) reported that higher corruption and rent-

seeking are associated with fuel resources, while forest resources attract fewer rent-

seeking behaviors and corruption. According to Williams (2011), countries rich in

point-source resources are less transparent, as measured by the Release of Information

Index. In addition, transparency issues are the direct result of resource revenue as they

tend to endanger sustainable economic growth.

Another key concept in natural resources originates from the work of Boschini

et al. (2007: 2), who introduced a new classification system of resource abundance

called “appropriability.” Appropriability means “how easy it is to realize large

economic gains, within relatively short period of time, from having control over it.” In

other words, resource abundance in itself is not harmful to economic growth; rather,

other factors such as technical and institutional appropriability determine the fate of

resource utilization. Technical appropriability means that certain resources (e.g.,

diamonds) tend to produce rent-seeking behaviors because they are easy to store and

transport. Institutional appropriability means the ability to regulate and control all

55

relevant aspects of resource production and revenue for positive economic growth,

rather than for negative economic growth.

In the next section, a brief cross-country survey clarifies how an Azerbaijan-

specific analysis can be developed using the available literature and statistical data. It

also outlines the main transmission mechanisms in NRC theory described in the

literature.

3.1.2. Main transmission channels and mechanisms in the NRC doctrine

While some resource-rich countries have managed to diversify their economies (e.g.,

Norway and Chile), many still struggle to do so. When countries fail to address

institutional, governance, and human capital challenges, they tend to experience the

NRC (Gelb 2010). This is part of a larger problem known as the “rent curse” (Auty–

Furlonge 2019). To assist in the diagnosis of NRC syndrome, it is crucial to trace its

exact pathways, transmission channels, and mechanisms.

The literature provides a wide range of examples that assist in analyzing

alternative channels and diagnosing the NRC phenomenon. According to Karabegović

(2009), there are three main transmission channels for the resource curse, namely

economic, political, and institutional. The economic channel mainly includes the effects

of DD. By contrast, the political and institutional channels are characterized by a low

propensity to tax non-resource sectors, which creates social groups uninterested in

political accountability. It also creates a wealthy elite who resist reforms and

transformation and seek to establish long-term sustainable development (i.e., rent-

seeking). In addition, as a transmission channel, public investments and the capacity to

manage oil revenue represent the overall situations in resource-rich countries, enabling

one to understand whether windfall gains

64

explain why they fail to achieve long-term

sustainable growth and development.

3.1.2.1. Role of resource rents and their political and institutional dimensions

The NRC literature features a plethora of economic explanations of resource rents.

Based on Gelb’s (1988) work, Deacon (2011) argued that conventional economic

explanations of lagging growth within the NRC doctrine fail to capture the big picture.

Thus, a rentier state model or the rent-seeking behavior of states helps to explain the

64

Windfall profits are unexpected increases in income for a commodity-exporting economy that may be

due to unforeseen factors such as price increases in international commodity markets. Windfall profits are

temporary in nature; as long-term observations have shown.

56

institutional and policymaking aspects of slower growth (Di John 2011). According to

Heller (2006), an analysis of resource-rich countries should focus on political

institutions to understand the success or failure of managing commodity revenue.

Political and institutional explanations of the NRC attempt to account for the possible

sources of slower or lagging growth in non-resource sectors. Political and institutional

factors are considered the most significant transmission channel because resource-rich

countries are likely to experience boom after-effects.

Therefore, the question arises of whether resource abundance leads to

substandard institutions or, conversely, whether substandard institutions result in the

NRC. Are corrupt politicians the cause of a boom in the extractive industry or is it

already decided beforehand? Here, Engerman–Sokoloff’s (1997) assertion that natural

factor abundance or technology shape the institutional evolution of a society can serve

as a starting point (the Staples thesis). This statement was supported by Luong–

Weinthal (2006), who mentioned that resource-rich countries deliberately avoid

efficient institution-building processes. This is because increasing transparency, rule of

law, and political accountability may endanger their control over policymaking and the

distribution of export rents. A lack of transparency primarily leads to rent-seeking

behavior by politicians in the form of direct payments, subsidies, or inefficient and

suboptimal infrastructure expenditures in return for electoral support (Robinson–Torvik

2005). Mehlum et al. (2006) claimed that the NRC occurs due to low-quality

institutions, which is similar to Collier–Goderis’s (2007) argument that bad governance

leads to the NRC. However, if economic policies can mitigate the adverse effects of

natural resource revenue, then the NRC is preventable (Acemoglu et al. 2003).

To track the NRC, many studies have focused on the relationship between

extractive industries and non-economic indicators, such as the rule of law, government

efficiency, democracy, and corruption in the face of private and state institutions. Ross

(1999) classified NRC studies into three main categories: those that provide cognitive

explanations, focusing on the failures of policymakers; those that provide societal

explanations, describing interest groups, elites, or social classes that gain power and in

turn create growth-decreasing policies to protect their own interests and power; and

those that provide state-centered explanations, covering how state institutions weaken

as a result of resource booms.

Bulte et al. (2005) stated that to understand how the NRC works, linkages

between institutional structure and resource endowments should be prioritized.

57

Therefore, institutional explanations of NRC syndrome encompass failures or success

stories about how necessary regulations and decisions eliminate the shortfalls of

revenue mismanagement (Torvik 2002; Bulte et al. 2005; Baggio–Papryakis 2010;

Deacon–Rode 2015). Revenue mismanagement occurs when acquired resource revenue

is inefficiently spent as a result of political lobbying to increase the wealth of very

limited circles (Gilberthorpe–Papyrakis 2015).

Moreover, if special interest groups capture resource rents and crowd out the

majority of citizens, long-term social and infrastructure investments will be delayed

(Pendergast et al. 2011). Karabegović (2009) noted that inefficient and unproductive

economic activities result from the government’s lack of attention to and care for the

non-resource sector. In such cases, rent-seeking and corruption will negatively affect

living standards (Pendergast et al. 2011) due to the immobility of natural resources.

“Immobility” means that, unlike other forms of capital (e.g., machinery, equipment, and

labor), natural resources cannot be freely and easily moved from one location to

another; this forces authorities to heavily regulate and tax them, thereby creating

unproductive economic activities (Karabegović 2009). Furthermore, politicians can

intervene in resource allocation processes, influence institutions, and distort policies to

benefit from mineral rents (Ross 2001; Stevens–Dietsche 2008; Orogun 2010). Tsui’s

(2010) model demonstrated that natural resource wealth can a blessing, not a problem

that should be overcome to successfully create economic growth. However, without

high-quality institutions to manage the effects of mineral wealth, a country is less likely

to benefit from its resources.

Furthermore, Hodler (2006) argued that oil windfalls extensively impact

property rights, decrease non-resource production, and subsequently lower income.

Institutions and human capital are also critical in the development process (Acemoglu et

al. 2014). In a cross-country regression, Cabrales–Hauk (2011) indicated that in terms

of human capital accumulation and capital formation, institutions play a significant role

in determining whether natural resources are actually a blessing or a curse in the long

term.

Although the institutional and political dimensions of the NRC have been

extensively investigated, no universal consensus exists regarding the relationship of

natural resource abundance or dependence with institutional quality. Objective reasons

exist to explain this deficiency in the overall position regarding the elements and

conceptual parts of the NRC doctrine. For instance, democracy and institutional quality

58

are difficult to measure, and empirical and econometric research usually fails to

accurately capture multidimensional connections (Paolo 2008). Notably, a study argued

that the NRC is more widespread in presidential and authoritarian regimes than in

democratic regimes (Kim–Lee 2018).

In developing countries, democratic institutions are critical for helping to catch

up with frontier countries (Butkiewicz–Yanikkaya 2006). NRC literature acknowledges

the role of education as an engine of growth; it is also linked to political regimes,

elections, taxation, and revolutions. Cabrales and Hauk (2007) argued that natural

resources can harm democracy and institutional and human capital accumulation. Their

model indicated that if the government’s main source of income is natural resources,

politicians are not interested in involving well-educated people in productive activities

as it would weaken their position. Well-educated people could gain a higher share of the

income, become more deliberate in their political choices, and potentially mount a

revolution to replace failing politicians. In addition, according to Wantchekon (2002),

the likelihood of an authoritarian regime existing increases if a country is resource-rich.

This is because oil rents are thought to increase personalism in political regimes, which

in turn nourishes a divergence from democracy (Fails 2019). Therefore, the link

between institutions and democracy seems to be very strong.

However, a certain strand of literature opposes the aforementioned approach.

After analyzing global data from 1960–2009, Brooks and Kurts (2016) argued that oil

reserves and production do not necessarily hinder political regimes from transforming

from a democracy into an autocracy, since both oil and democracy are innate to a

country’s early industrialization. Nevertheless, developed countries tend to be more

democratic than developing countries because they experienced and made progress in

industrialization. Usually, these aspects are not accounted for in statistical research.

Similarly, Herb (2005) argued that any negative effects of resources on democracy are

negligible rather than a universal truth. If a country is oil-rich and non-democratic but

there is a tendency toward democracy in neighboring countries, then that country is

highly likely to experience similar social and political trends. This phenomenon is

known as the interdependency of democracy (Brooks–Kurts 2016).

Dunning (2008) argued for an opposite approach to the associations between

democracy and resource wealth. Citing the example of Latin America, he argued that

resources have been a blessing for the region due to the redistribution policies

implemented by states due to high income inequality. Under high income inequality,

59

authorities tend to apply redistribution policies; if income is not relatively equally

distributed among citizens, political rulers are encouraged to use mineral revenue to buy

off their political opponents, which occurs in non-democratic regimes.

To obtain an enhanced understanding of the NRC thesis, some scholars have

focused on the rule of law rather than democracy. Butkiewicz and Yanikkaya (2006:

660) claimed that “the rule of law, but not democracy, enhances economic growth.”

Similarly, Norman (2009) and Kolstad (2008) have reported that mineral abundance is

negatively associated with the development of the rule of law among resource-rich

countries. In a recent study, Adani and Ricciuti (2014) reported negative effects of

resource rents on institutions in a large sample of countries. However, they also

mentioned the importance of time lags in analyses of the association between the NRC

and institutions. Usually, policymakers require time to reshape institutions; sometimes,

an indirect effect rather than a direct association exists between institutional quality and

resource rents.

In addition, the effects of oil on democracy depend on its regional distribution.

Ahmadov (2014) found that oil wealth in Latin America was positively associated with

democracy but negatively so in the Middle East and North Africa (MENA) region. A

negative link was also found in Sub-Saharan Africa. In countries with weak institutions

and high autonomy over social processes on the part of politicians, resource booms

allow governments to use resource rents to maintain power. Even if resource booms

create efficiency in the extractive industry, the rest of the economy suffers from

inefficiencies due to clientelism and resource allocations by politicians who wish to

maintain their power (Robinson et al. 2006). Thus, according to Ahmadov (2014), oil

does not negatively affect democracy directly, but rather indirectly through particular

channels.

Moreover, Konte (2013) found that democracy plays a crucial role in

determining whether natural resources are a blessing or a curse for growth, which was

not true for education and institutions. Whether the country is democratic or autocratic

also does not play a role. However, as measured through democracy and the rule of law,

institutions play a significant role in overcoming rent-seeking behavior, which leads to

growth collapse in resource-rich countries, especially when point-source resources are

abundant (Mavrotas et al. 2011).

In addition, Butkiewicz and Yanikkaya (2010) emphasized the importance of

trade openness, which determines growth dynamics. If trade openness is high, resource

60

owners in mineral-rich countries can benefit from higher returns on resource exports

and increased non-resource imports. Azerki and Van der Ploeg (2007) found that the

NRC was less prevalent in countries where trade openness was high. In addition, in

reference to the ideas of Falkinger and Grossmann (2005), Butkiewicz and Yanikkaya

(2010) noted that failure to improve non-resource sectors, in turn creating an

uneducated labor force, usually represents the misuse of political power by resource

owners, who create cheap factors of production for domestic markets. However, Kurtz

and Brooks (2011) contradicted this by claiming that trade plays a neutral role in

determining the NRC.

The links between growth, democracy, and energy resources ought to be

continually investigated until the predictability of such research notably increases

(Paolo 2008). From all relevant discussions and debates, institutional, political, and

democracy-related variables should be used to help enhance the understanding of the

effect of natural resource wealth on countries’ economic growth and development.

Yang and Lam (2008) analyzed 17 oil-rich countries and found that oil booms

did not lead to lower GDP or decreased investment. The latter was only observed in

Venezuela and Columbia, but they are considered institutionally well-endowed. The

authors also found that oil booms do not change institutional quality, whereas Daniele

(2011) argued that if high resource wealth is accompanied by poor governance, then

institutional quality decreases.

In the short term, huge mineral revenue may decrease the domestic tax burden

and stimulate the private sector to increase its economic activities. However, once

natural resources are depleted, the new equilibrium will entail higher costs than the pre-

equilibrium situation (Bornhorst et al. 2009). Besides royalties or tax revenues, central

governments derive little benefit from mining activities in the form of value-added

because most extracted resources are exported abroad to be manufactured (Davis–Tilton

2005). Empirically speaking, Bornhort et al. (2009) found a negative association

between oil and gas revenue and domestic non-hydrocarbon revenues, as measured by

non-oil tax efforts. A 1% increase in hydrocarbon revenue lowered non-oil revenue by

0.2%, which could in turn decrease governance quality and public control over the

government. Furthermore, if the financial sectors of a national economy are

underdeveloped, oil revenue does not contribute to economic growth (Erdogan 2020).

Lastly, in mineral-rich economies, government officials are usually optimistic

about windfall revenue. During price slumps, they borrow from abroad and prepare the

61

economy for a second oil boom. Thus, if the subsequent boom fails to generate the

necessary revenue in the medium term, the economy finds itself with a foreign debt

burden (Auty–Warhurst 1993).

3.1.2.2. Public investments and the capacity to manage oil revenue

Government spending increases as resource extraction rises.

65

This is usually related to

attempts by governments of mineral-rich countries to smooth consumption over time

due to the uncertain nature of mineral revenue (Stijns 2006). On the one hand, optimal

government expenditure can solve many economic problems in a country through fiscal

policy; on the other hand, natural capital intensity was argued to lessen growth both

directly and indirectly by “reducing equality, secondary-school enrollment rates and

investments rates” (Gylfason–Zoega 2003: 289).

66

However, spending should be

smoothed within a manageable range rather than relying on or attempting to accurately

predict oil prices to manage revenue volatility (Collier 2011). Similarly, Lane and

Tornell (1997) emphasized that windfall gains in resource-rich countries lead to

competition among interest groups; they exhaust public goods, leading to inefficient

public investments. Interest groups’ ability to affect government spending to serve their

own interests (Meltzer–Richard 1981) may lead to significant delays in structural

reforms or other policies for enhancing long-term economic growth (Leite–Weidman

1999). In turn, Baumol (1990) mentioned issues related to rates of return in non-

resource sectors, which indicate the magnitude of society’s economic incentives to

engage with the rest of the national economy (Karabegović 2009). In other words, a

decrease in rates of return in non-resource sectors will harm sustainable economic

growth due to fewer incentives to invest in non-resource sectors, even if a significant

influx of mineral revenue and foreign investments occurs. In such cases, based on

institutional and political factors, public investments emerge as a distinct channel for

the NRC.

67

65

It is wrong to think that resource exporter countries do not invest the right proportion of their windfall

revenue. However, due to increased opportunities for investments from the windfall revenue, this does

not mean that resource richness encourages the accumulation of human capital to the desired level.

66

For example, the nature of social institutions, which was calculated via variables that measure civil

liberties, has an indirect effect on investments, education, and growth (Gylfason–Zoega 2006).

67

An important note should be added to clarify the role of institutions in terms of direct or indirect

impact. Pessoa (2008) drew attention to the social cohesion aspect of measuring institutional quality,

which has rarely been addressed. The resource curse should not be solely linked to the quality of

institutions that guarantee the work of market mechanisms; institutional quality in the dimension of the

social cohesion is crucial to understanding this phenomenon. Manca (2014) defines social cohesion as

“the extent of connectedness and solidarity among groups in society.” In other words, society and

62

In resource-rich countries, three pillars of investment management exist to

prevent adverse effects of booming sectors. These are capacity building to manage

public investments, creation of the necessary environment for boosting private

investments, and reduction of the unit cost of public and private investments (Collier

2011).

68

However, in developing countries with inefficient government institutions,

public investment reduces growth because it crowds out private investments and

discourages savings (Butkiewicz–Yanikkaya 2011). Thus, rates of investment and

saving, which ensure long-term sustainability and non-mineral development in small

oil-exporter countries, are not at the optimal level (Farzin 1999). Extractive industry-

based economies tend to have a weaker production linkage (in terms of the non-mineral

manufacturing of tradeables) but a higher government revenue linkage (increased

government spending and expenditures during commodity price booms) to the rest of

the economy. However, despite higher rates of investments, oil-rich countries tend to

have slower economic growth, which strengthens the inefficiency of investments

(Auty–Warhurst 1993). Meanwhile, the savings rate should be high if the expected

depletion of resources is near. Hence, investment and saving policies should be tailored

to the domestic realities of resource-rich, low-income countries

69

(Collier 2011).

Bhattacharyya–Collier (2014) highlighted that public investments are a crucial

channel for tracking the effects of the NRC because they are clearer and more direct. In

essence, resource rents embody government revenue that is under the control of the

government (Bhattacharyya–Collier 2014). If resource revenue does not increase the

capital stock, then signs of the NRC may well reveal themselves through the avenues

mentioned here. The authors investigated 45 countries from 1970 to 2005 and

concluded that the log per capita public capital stock decreased by 7% if a 1% increase

in resource rents occurred. In addition, the adverse effects of point-source resources

communities also play an essential role in determining whether natural resources serve the overall well-

being or not. Therefore, the measurement techniques and methods of institutional quality across the

countries cannot allow us to find the ultimate measure for institutional quality that could reflect the true

relationship between the resource abundance or dependence and institutional quality.

68

The essential purpose of these pillars highlights the role of institutions, which should manage mineral

revenue in such a way that prevents the signs of the resource curse.

69

Notably, zero or negative rates of the genuine saving (which includes three types of investment:

produced, natural, and human and calculated based on the following formula: GS = investment in

produced capital − net foreign borrowing + net official transfers − a depreciation of produced capital −

net depreciation of natural capital + current education expenditures) or adjusted new saving is a common

picture for resource-rich countries which show the unsustainable consumption of the resource rents, as

well as policy, fails to address the future growth issue.

63

were greater than those of diffuse resources, which favors the NRC doctrine

(Bhattacharyya–Collier 2014).

NRC syndrome occurs through government investments when the public fails to

monitor increased public spending originating from mineral revenue. Wiens (2014)

argued that if institutional preparedness fails to address citizens’ interests, then the lack

of capable politicians or specialists for adequately managing oil revenue will lead to

resource misallocation and inefficient investments. This can manifest as the subnational

resource curse, which refers to the regional distribution of fiscal resources being marked

by inefficiencies and corruption. Hoyos (2019) demonstrated this in Peru.

70

Another critical element of the capacity to manage revenue in mineral-rich

countries is the role of sovereign wealth funds (SWFs). According to the IMF, SWFs

are stabilization funds that usually function as a buffer against commodity price

volatilities. These savings funds allocate resource revenue for future generations and

help investment corporations to grasp the benefits of better investment opportunities,

usually overseas (Allen–Caruana 2008). SWFs enable countries to decouple their

national economy from the volatility associated with international commodity markets.

Shabsigh and Ilahi (2007) argued that by establishing an oil fund in an oil-exporter

country, SWFs can also play a role in industrial development and the pension system.

This extends their role by conferring additional status, such as development funds or

contingent pension reserve funds.

Oil funds or SWFs in oil-exporter countries have been popular since the 1970s.

Yo-Chong (2010) stated that the key characteristic of SWFs is that they are state-

owned. In oil-rich countries, state ownership and strict government control raise

multiple concerns in the case of non-transparent public governance. This is because

SWFs by default play a crucial role in managing a country’s absorption capacity after it

receives large amounts of revenue. Bahgat (2011) stated that in fast-growing economies

such as China and Singapore, oil-rich countries have started to experience current

account surpluses following increases in oil prices. This has initiated new attempts to

establish oil funds. In turn, this has led to enhanced macroeconomic balance

management, as the excess revenue is invested abroad, maintaining inflation levels

within the acceptable range. However, the process is not always smooth.

70

The author investigated Huari and Espinar regions and identified resource curse signs, such as

inefficient investments, the dependence of the employment on the mining sector value chains, corruption

and fiscal overspending.

64

Moreover, evidence suggests that SWFs have failed to ensure macroeconomic

stability in some oil-rich countries, as they have been unable to decrease the high rates

of interdependence between resource earnings and government spending (Davis et al.

2001).

71

In addition, Koh (2017) claimed that the effectiveness of oil funds depends on

the country’s existing institutional quality; thus, a country with low institutional quality

cannot solve macroeconomic stability issues arising from the boom and bust cycles of

the economy, even if an SWF is established. This sobering fact demonstrates that even

if oil funds are benevolent, their role and association with the rest of the economy must

be examined very carefully.

In addition, as oil funds have limited fiscal benefits (Shabsigh–Ilahi 2007), their

efficacy is in doubt. The fiscal benefits usually expected from SWFs can be achieved

through improved fiscal policies and administration (Shabsigh–Ilahi 2007). Moreover,

although SWFs have become highly popular in oil-rich countries, their efficiency

depends on the quality of the government and the role of institutions.

In sum, the management of oil revenue in resource-rich countries has failed to

achieve high levels of government effectiveness, decreasing the marginal utility of

additional revenue (Ossowski et al. 2008). Political and institutional factors play a

significant role in determining whether corruption and rent-seeking behavior will

decrease the economic benefits of oil revenue in the long run. Thus, public investments,

as measured by government expenditure, are a significant transmission channel for the

NRC in oil-rich countries.

3.1.2.3. Human capital and education as pathways to the NRC

Human capital is “the skills the labor force possesses and it is regarded as a resource or

asset” (Golding 2016). Economic growth cannot be sustained without the necessary

dedication to education and human capital development in economic reforms (Ranis et

al. 2000). Therefore, human development is an integral part of economic growth in the

modern global economy (Bloom et al. 2004; Hartwig 2010). However, the term is

relatively new in the jargon of economists (Golding 2016). Still, the relationship

between human capital and economic development may not always be so clear cut—it

may depend on the country. According to Schultz (1961) and Becker (1964), human

71

The case of certain countries deserve secial attention (see Davis et al. 2001). For instance, while

Norway’s experience with SWFs has been successful, Papua New Guinea closed off its Mineral

Stabilizations Fund due to failures in smoothing budgetary expenditures. Similarly, Venezuela’s

Macroeconomic Stabilization Fund failed to smooth its expansionary fiscal policy.

65

capital characterizes the set of knowledge, experience, expertise, skills, and abilities that

an individual gains over time through education, healthcare, and training.

Actually, human capital is as important as corruption as a transmission channel

for the NRC in mineral-rich countries (Papyrakis–Gerlagh 2004). This can be traced

back to political and institutional factors. Politicians in oil-rich countries do not

prioritize human capital development, as their position does not depend on high levels

of education or human capital (Wigley 2017). In other words, politicians do not allocate

the necessary resources to society because institutional failures ensure their power

without the input of the electorate, which displays its will through elections. However,

Szirmai and Verspagen (2015) demonstrated that the manufacturing sector can have a

positive impact on GDP if a highly educated workforce is employed. Therefore, the

education or human capital-related resource curse relies on effects that decrease the

time and financial resources invested in the education system (Wadho 2014).

In manufacturing, human capital is one of the most critical production factors.

During a resource boom in a given country, institutional failures will lead to decreased

human capital accumulation. According to Lucas (1988) and Romer’s (1990) popular

endogenous growth models, human capital, technology, and research and development

are viewed as products of the education system that significantly contribute to the

welfare of society. In other words, if human development is neglected, then sustainable

economic growth and development will be impossible in the long run. In fact, economic

success in advanced and emerging economies is rooted in the education system, human

capital development, and investments in information technologies and equipment,

which are key drivers of competitiveness in the face of new globalized challenges

(Castelló–Doménech 2002; Jorgenson–Khuong 2003; Wilson–Briscoe 2004; Ashton et

al. 2005; Kefela 2010). Another reason for human capital’s importance to economic

growth is that, since 1990, increasing amounts of human capital have been required to

achieve higher value-added and succeed among the competition.

Developing the education system and human capital eases technological

adoption and provides skilled labor, which are required to manage innovative activities

and accelerate economic growth (Awan–Khan 2015). Some even argue that human

capital is more important than institutions for economic growth (see Glaeser et al.

2004). Similarly, as a crucial part of human capital development, government health

expenditures may boost economic growth. Bloom et al. (2004) concluded that health

has a positive and significant effect on economic growth, as measured by the growth of

66

factor inputs, technological innovation, and technological diffusion. More precisely, the

authors calculated a 4% increase in output if life expectancy increased by one year. In

addition, GDP per capita would increase by 0.5% with a 1% rise in health spending; the

same increase in health spending would also boost the survival rate of children aged

under 5 years by 0.6% (Beldacci et al. 2008). However, mineral-rich countries were

found to underperform with regard to health indicators when child mortality and HIV

infection rates were used as proxies (de Soysa–Gizelis 2013).

Countries may ensure effective economic development through rural

development, industrialization, the attraction of new firms, and subsequently job

creation if high-quality infrastructure is present (Doeksen et al. 1998). The healthcare

system and a strong economy are closely interdependent, often in ways that establish

sustainable economic development (Ivinson 2002). An increase in health spending

guarantees not only an increase in people’s quality of life but also direct economic

benefits through employment and income (Doeksen et al. 1998). However, in resource-

rich countries, spending on education and health is low and human capital accumulation

is lower. Moreover, the literature seems to favor human capital as a transmission

channel for the NRC among resource-rich countries and regions.

The situation in these countries, however, tells a different story. According to

statistical analyses and current literature, skilled human capital grows more slowly in

resource-rich economies (Suslova–Volchkova 2012). Studies have frequently observed

that education and human capital accumulation do not increase in parallel with resource

wealth in such countries (Cabrales–Hauk 2007). Resource abundance and the HDI are

negatively correlated; thus, some resource-rich countries fail to provide proportional

access to safe water and have an undernourished population (Bulte et al. 2005).

Educated workers are the mediators of technological diffusion, which fuels growth.

Growth, in turn, is positively and significantly associated with average years of school

attainment for men at the secondary and higher levels. Thus, the diminishing utility of

education in resource-rich countries endangers non-resource sectors (Barro 2001).

The alarm over commodity-exporter countries arises from a lack of long-term

sustainable economic growth and a diversified export basket. In the short term, although

natural resource exports increase a country’s income level, economic growth will

decline without appropriate human capital development (Bravo Ortega–De Gregorio

2005). Gylfason (2001) linked this feature to the fact that resource-rich countries tend to

have a capital-intensive rather than a human capital-centered economic structure. In

67

fact, Cockx and Francken (2016) found an adverse relationship between resource

dependence and point-source natural resources and public spending on education. This

indicates that governments of natural resource-rich countries do not prioritize public

education as the share of natural resources of GDP increases over time. Resource-rich

countries must invest in education and human capital if they wish to diversify their

national economies and sustain long-term economic development and prosperity

(Papyrakis–Gerlagh 2004). Barro (2011) also found a nonsignificant association

between female school attainment and growth, which suggests the underperformance of

female labor utilization in resource-rich countries. Moreover, promoting women’s

education would indirectly enhance growth by lowering fertility rates.

For the period 1970–2004, Behbudi et al. (2010) demonstrated that human

capital is in fact a transmission channel for the NRC if human capital levels are low and

petroleum-exporter countries neglect to develop it.

72

In a similar study of 35 resource-

rich countries from 1980 to 2015, Shahbaz et al. (2019) found that in countries such as

Saudi Arabia, Cameroon, Botswana, and Nigeria, where human capital endowments are

low, natural resources were a curse rather a blessing. This was because non-resource

productive economic activities were negatively affected by resource dependence.

The wise use of windfall revenue may boost human capital in high-income,

resource-rich countries and help prevent a sole specialization in resource production and

exports. A key finding of Aldave and García-Peñalosa (2009) was that natural resources

create corruption and reduce education, in turn shrinking growth. Thus, resource

abundance affects human capital in resource-rich countries.

Moreover, natural resource producer and exporter countries have lower levels of

education (as measured by school enrollment level). This is because real exchange rate

appreciation decreases the competitiveness of non-resource sectors, thereby lowering

the need for human capital to produce manufactured goods (Gylfason et al. 1999).

Gylfason et al. (1999) concluded that a low-quality education system raises training

costs, which disincentivizes secondary sectors

73

from enhancing worker skills and

building innovation-based production processes. Therefore, secondary sectors may

suffer from the high cost of hiring, learning, and growing because lucrative primary

72

Human capital may not only compensate for negative effects arising from the economic dominance of

the extractive industry as the Scandinavian example shows. If human capital is abundant, the joint

development of high-technology and extractive industrial sectors can occur in parallel if there are strong

linkage between resource and non-resource manufacturing (Bravo–Ortega–De Gregorio 2005).

73

By “secondary sectors” Gylfason et al. (1999) mean non-resource tradeable sectors.

68

sectors pay higher wages. Secondary sectors may also increase payments to prevent

employees from leaving.

Measuring the quantity and quality of education poses a serious challenge due to

the variety of ways to conduct surveys and use analytical methods in the literature. For

example, Hanushek and Wößmann (2007) measured the quality of education in terms of

international test scores and level of education by years of schooling as the main

determinants of economic growth.

However, some express contradictions and a degree of skepticism toward the

role of education in economic growth. Compared with studies by Mankiw et al. (1992),

Mankiw (1997), Benhabib and Spiegel (1994), and Barro and Sala-i-Martin (1995), the

cross-country regressions of Caselli et al. (1996) and Pritchett (1996) have

demonstrated weak or even negative associations between economic growth and

education.

The literature also contains many conflicting results concerning the link between

human capital and resource wealth. Pineda and Rodríguez (2010) claimed that the

interaction between human development and natural resource abundance is positive and

significant. In particular, natural resources significantly affected non-income

components of the HDI, such as literacy and life expectancy, from 1970 to 2005. The

impact of resources on human development has also been uneven. For instance, the

authors found that Latin America has benefited from natural resources less than the

average of other resource-rich regions. Moreover, Wigley (2017) found a negative

association between oil and gas abundance and child mortality in 167 countries between

1961 and 2011.

In addition, Shao and Yang (2014) advocated for the efficient allocation of

production factors. They claimed that resource abundance and the development of

resource-based industries can be a blessing if mineral revenue is properly and efficiently

invested into improving education and increasing educational opportunities and

quality.

74

Thus, whether natural resource wealth has the effect of a curse or blessing is

conditional. If investments in human capital development continue, oil-rich countries

74

In spite of this, there are opposing ideas. Greasley–Madsen (2010) argued that a country’s geological

and geographic properties matter more than its institutional and educational standards. Wright and

Czelusta (2004) cited the United States, Sweden, and Norway as examples of countries where natural

resources were used to initiate the industrialization process and where education and institutional

properties were developed only after this process was started. Greasley–Madsen (2010) said that

knowledge creation that results from mineral extraction positively affects growth, while land abundance

has the opposite effect.

69

can benefit from resource wealth, eventually reducing the negative effects of oil (Kurtz–

Brooks 2011). Nevertheless, it is vital to combat the detrimental effects of extractive

industry development, such as real exchange rate appreciation, de-industrialization, and

macroeconomic volatility. These are usually considered the curse of natural resources

(Shao–Yang 2014) and directly and indirectly affect education and human capital.

Figure 3. 1 summarizes the overall NRC phenomenon in a graphical way.

Figure 3.1: The main transmission channels and results of NRC.

Source: The author’s own construction based on the literature review.

3.1.3. The NRC in Azerbaijan

This section examines the relevance of the NRC doctrine in the case of Azerbaijan.

Since the mid-1990s, Azerbaijan has been repeatedly mentioned when economists have

expressed concerns about upcoming signs of the NRC and DD. This is due to the

country’s agreements signed with multinational companies in the extractive industries

and expectations of huge mineral revenue; specifically, concerned economists have

argued that the domestic absorption capacity is lacking for such an influx of capital and

revenue (Tsalik 2003). Biresselioglu et al. (2019) classified Azerbaijan as a country

highly vulnerable to the NRC, ranking it among the top 10 countries labeled “high,” as

NRC

Political

Rent-seeking Corruption

Low propensity to tax non-

resource sectors

The availability of excess

revenues increases the cost

of taxation

Institutional Unregulated gov. spendings Lack of fiscal mechanisms

to regulate spendings of

resource revenues

Social

Low human capital quality Low education and

healtcare spendings

Low social cohesion Low political will of the

society

Economic Dutch disease Booming sector crowds out

non-booming sectors and

makes the overall economy

dependent on itself

70

measured by the Resource Curse Vulnerability Index (RCVI).

75

This indicated a lack of

economic diversification, economic planning, and industrial development policies.

Consequently, the literature reviewed in this section demonstrates that, starting from the

mid-to-late 1990s, a growing body of studies sounded the alarm about the presence of

the NRC in Azerbaijan’s economy.

Following an analysis of large oil and gas projects, Hoffman (1999) argued that

the chances of converting oil revenue into widespread economic growth were low if an

appropriate tax-collecting apparatus and independent statistics remained absent in

Azerbaijan. Some have argued for the possibility of the NRC or DD while relying on

the experiences of other resource-rich countries. The rationale is that, since the NRC has

occurred in other resource-rich countries, it could also happen in Azerbaijan. For

example, Esanov et al. (2001) argued that domestic energy sectors and fiscal

management are two key challenges for Azerbaijan due to a lack of necessary

experience in managing oil revenue; similar challenges were observed in other case

studies.

76

In early articles on the NRC in Azerbaijan, the government’s decisions and new

spending habits were sources of great concern. Some authors claimed that oil revenue

was spent in a nontransparent manner that did not support non-oil development, which

could ensure long-term sustainable development; in addition, the distribution of benefits

was problematic at the national level (Gulbrandsen–Moe 2007). Khanna’s (2011)

descriptions of Azerbaijan’s oil boom period highlighted the government’s low

willingness to redistribute oil revenue, market-distorting interventions by the state, and

the influential position of oligarchs. Achieving independence from the Soviet Union did

not appear to inspire Azerbaijan to manage its oil revenue in a desirable way.

Consequently, if the management of oil revenue fails, the reasons behind the fiasco

point to the relevance of the NRC.

As an economic explanation for the NRC, concerns about DD have been

frequently voiced.

77

Singh and Laurila (1999) asserted that DD syndrome was present

(due to exchange rate overshooting) in Azerbaijan from the early 2000s. In addition, a

low level of democracy and weak rule of law seriously diminished the government’s

75

For more details, see Biresselioglu et al. (2019).

76

Similarly, Bandiera et al. (2008) documented that if Azerbaijan overspends, then it could be a net

debtor if the oil prices fall (see also Budina–Wijnbergen 2008). Also, the study from Esanov et al. (2001)

incorporated the case studies of Kazakhstan, Turkmenistan, and Uzbekistan.

77

The results of a literature review on DD in Azerbaijan are elaborated on in Chapter 4 of this

dissertation.

71

capacity to design policies that could successfully manage revenue (Tsalik 2003).

Wynne Russel et al. (2004) reported DD and signs of the NRC as economic risk factors

for sustainable economic development and the catching-up process after the collapse of

Soviet rule in Azerbaijan.

Amineh (2006) claimed that resource-rich post-Soviet countries, such as

Azerbaijan, Turkmenistan, and Kazakhstan, would not be able to successfully

industrialize due to issues arising from the NRC. Similarly, Esanov et al. (2005) argued

that political reforms in resource-rich transition countries do not favor a deterministic

model of policy formation.

In fact, according to Kronenberg (2004), substantial differences exist between

resource-rich and resource-poor transitional countries. He argued that, while resource-

poor Central and Eastern European (CEE) countries performed well in catching up with

developed economies, resource-rich countries seemed to lag. This was mainly due to

corruption inherited from the Soviet era, which entailed a high level of state capture.

Franke et al. (2009) argued in favor of the existence of the NRC in Azerbaijan because

of the lack of an alternative political elite as well as a substandard democracy;

moreover, they argued that a lopsided economic structure was established after high

mineral revenue flowed into the country. Early concerns about the NRC in Azerbaijan

were based on a lack of budgetary processes, long-term policy planning, public sector

constraints in front of private sector development, and dependence on oil prices (Tsalik

2003). If oil prices crashed, poverty and unemployment would increase (Franke et al.

2009).

Observations and analyses of political and institutional variables in Azerbaijan

have supported the relevance of NRC syndrome. Bhatty (2002) considered corruption,

weak state capacity, and impediments to trade the main signals of the political and

institutional channel for the NRC. Bayulgen (2005) argued that oil rents encouraged an

authoritarian regime, resulting in the accumulation of power in the hands of the

president. Later, O’Lear (2007) provided evidence of the NRC based on survey data

from Azerbaijani citizens. According to his findings, an oil-dominated economy, a high

concentration of wealth in the hands of a ruling elite, political legitimacy problems, and

centralized political control were clear signs of the NRC. Other indications of the

NRC’s political and institutional channel include internal and external patronage

networks, clientelism (Bayulgen 2005; Guliyev 2009), autocracy (Schubert 2006;

Pomfret 2011; Kendall–Taylor 2012; Radnitz 2012), problems with political freedom

72

and democracy (Altstadt 2017), transparency and accountability issues in revenue

spending (Wakeman-Linn et al. 2003; Franke et al. 2009), and strong resistance to the

development of the private sector (Kalyuzhnova–Kaser 2005). A further indication is

neopatrimonialism, which refers to informal personalized rule combined with pyramidal

power structures (Franke–Gawrich 2010; Heinrich 2010).

In Azerbaijan’s economy, institutional failures have hindered effective

macroeconomic management, especially poor revenue management, which originates

from SOFAZ (Mirzeyev 2007). Oil revenue created confidence among political circles,

leading to inefficiency and poor performance in the taxation of the population (Shaw

2013). In turn, low taxation led to a shallow aspiration for democracy (Almaz 2015).

Then, there are no less benign conditions for signs of the NRC to develop in a rentier

state, leading to a lack of willingness to diversify the economy. Guliyev (2013) argued

that the government and political elites are uninterested in the diversification of

Azerbaijan’s economy because regime opponents could accumulate power and become

autonomous power centers. The voices of civil society and nongovernment

organizations (NGOs) being limited was another source of concern regarding the

country’s deteriorating political and institutional development in 2014 (Sovacool–

Andrews 2014). This creates a chain reaction in Azerbaijan’s economy, where the

effects of the NRC are embodied in the form of weak monitoring of the oil sector and

revenue by civil society; notably, the latter could ensure transparency and accountability

(Aslanli 2018).

Because of the country’s political and institutional failures and oil-dominated

economic structure, certain studies have focused on rent-seeking and bribery in

Azerbaijan. According to Sadigov’s (2014) survey and analysis, bribery in Azerbaijan is

strongly correlated with individual respect for the rule of law and the overall

jurisprudential system. Higher levels of corruption undermine the power of said system,

which leads to unsatisfactory cases in the legal system.

The other possible channel of the NRC in Azerbaijan’s economy is FDI. The

flow of FDI mainly into extractive industries harms non-resource economic growth

because these industries are highly capital-intensive; furthermore, multinational

companies (MNCs) cannot generate poverty-reducing employment and wage-boosting

economic activities at the national level (Asiedu 2013). Vanderhill et al. (2019) argued

that Western FDI in Azerbaijan has not led to positive changes in the institutional

environment, which could boost FDI into non-resource sectors. However, Mammadova

73

and Coskun (2015) argued that FDI and technology transfers have had a significant and

positive impact on non-oil sectors. According to them, any issues associated with

lagging non-oil sectors should be addressed within the fact that Azerbaijan is a

developing country and such countries usually face difficulties in catching up with

advanced economies. However, in line with early concerns about the NRC in

Azerbaijan, Lee (2005) mentioned that FDI only favors the ruling elites, leaving many

economically disadvantaged citizens behind.

Along with institutional and revenue management failures, the education system

in Azerbaijan has shown signs that point to the existence of the NRC. However, studies

that directly assess the impact of natural resources on education and human capital are

difficult to find. Based on input–output analyses of 2006, 2008, and 2009, Sadik-Zada

et al. (2020) argued that although human capital in Azerbaijan is abundant, the

extractive industry’s low job creation and production linkages are obstacles to

reinvesting mineral revenue for policymakers. This argument supports the belief that

manufacturing sectors in Azerbaijan are shrinking, which demonstrates the crowding-

out effect of the extractive industry.

Moreover, a shrinking non-oil manufacturing sector reduces the demand for

skilled workers and specialists. Thus, the education system in any given country

operates within a complex ecosystem of political and institutional subsystems, which

are integral parts of a more complex social structure (Ursul–Ursul 2013). One disaster

leads to another, creating a vicious circle. For instance, Sadigov (2014) argued that the

education system in Azerbaijan fails to teach marketable skills to students; furthermore,

a large market exists for acquiring diplomas by dishonest means, as having one is a

precondition for obtaining a job or increasing one’s social status. This has led to a dual

perception of bribery and the education system among citizens. Recently, Shahbazov

and Afandiyev (2020) found that Azerbaijani students consider white-collar

78

crimes to

be more serious than usual crimes. Although education expenditure (along with gross

capital formation and population growth) in Azerbaijan increased with economic

growth from 1995 to 2018 (Mukhtarov et al. 2020), whether non-oil tradeable sectors

have benefited is unclear.

78

Based on Sutherland’s (1983) definition, white-collar crime means “a crime committed by a person of

respectability and high social status in the course of his occupation”

74

Another integral part of human capital is the healthcare system. Although the

government’s perceived capability substantially increased just before the oil revenue

boom, the quality of medical service delivery remained below par (Ibrahimov et al.

2010; Aliyev et al. 2011). Some have argued that this is due to unofficial out-of-pocket

payments and corruption (Habibov et al. 2018); others have mentioned factors such as

decision-making, the auditing of budgets, and standardization issues (Ünal–Tagiyev

2016); while others mentioned the inefficient use of healthcare resources (Bonilla-

Chacin et al. 2018). The occasional failure of healthcare legislation and a lack of

incentives to build an efficient healthcare system—despite the availability of enormous

oil revenues—could be viewed as a sign of the NRC (Ibrahim et al. 2010). If the

population cannot afford the high cost of medical services because of unofficial

payments requested by doctors, then what is so good about being an oil-rich country is

unclear (Aliyev et al. 2011). Because of the close nexus between political and

institutional incompetence and the healthcare system, Rzayeva (2013) argued that the

state has failed to restructure and deliver high-quality medical services to the majority

of the population. The state’s healthcare expenditure has also remained low (Aliyev et

al. 2011). A common perception among Azerbaijani citizens is that the public

healthcare system has not changed much since Soviet times and that public policy has

failed to eradicate the marginalization of a large part of society (Hohmann 2014). Thus,

without the systemic transformation of the healthcare system, human capital cannot be

fully developed in Azerbaijan. Although this could be gradually accomplished with the

help of oil revenue, some authors are highly skeptical about any radical transformation

that could result in a sustainable, efficient, and socially inclusive healthcare system.

Notwithstanding, some authors evaluated the early 2000s in Azerbaijan as

economically successful; however, there is still a long way to go. Mehmet (2006)

considered Azerbaijan being authoritarian or democratic as equally likely, mentioning

successful decisions to manage oil wealth (e.g., the establishment of SOFAZ). An

example of a successful attempt to manage oil revenue is the integration of civil society

participation in the management process (Bayramov 2009). Clemens (2007) offered

several optimistic predictions about the future of Azerbaijan’s economy, such as

effective anti-corruption drives, the high potential of non-oil sectors, and an upward

trend in the development of financial markets. Îlhan (2007) also addressed institutional

challenges in Azerbaijan, stressing the need to strengthen domestic and global civil

actors to successfully combat the NRC, as Norway did. Other articles have emphasized

75

the successful avoidance of debt accumulation and oil-fueled spending (e.g., Malik

2010). However, the aforementioned articles are not so recent and do not account for

the rapidly changing domestic realities of oil revenue management, the development of

non-oil sectors, and political transformations.

In addition, Aridemir (2018) outlined some serious steps taken by Azerbaijan

with regard to non-oil sectors, especially in cooperation with Turkey to attract FDI. If

FDI can be increased in non-oil sectors, then even if domestic institutions fail to redirect

oil revenue into them, an opportunity to develop non-oil sectors exists.

79

Weeks (2008)

argued that capital controls and the fixed exchange rate of the AZN against USD has

helped Azerbaijan to minimize the effects of the NRC. However, tangible progress in

non-oil development through FDI is relatively rare.

Hübner (2011) listed several obstacles to non-oil FDI that can impede serious

developments. These include sector monopolies, informal barriers to market entry,

bribery, and unfavorable monetary conditions. Together, they represent the regime’s

failure to develop non-oil sectors while FDI flowed exclusively into the energy sector,

creating the illusion of good governance (Frayne 2012). However, Hübner (2011) also

cited positive developments that have empowered non-oil sectors, such as the rapid

development of the banking sector, the proliferation of businesses through the one-stop

shop,

80

and the establishment of a state investment company to attract FDI. Overall, FDI

provides incomplete insights for understanding the NRC in Azerbaijan.

Another approach to tackling the NRC lies in the view that even if its effects are

deeply rooted in the economy, they have not transformed Azerbaijan into an

authoritarian regime, since transparency and accountability issues were already

prominent in the pre-oil boom period (Walker–House 2008). However, based on British

Petroleum (BP) documentation, Flegel (2012) argued that Azerbaijan is part of a tiny

group of countries where the main oil and gas projects are publicly available. Moreover,

the EITI has supported Azerbaijan’s institution-building efforts to manage oil revenue

(Öge 2014). However, the country stopped participating in this process in 2017 (EITI

2020).

81

Economists have voiced their concerns about this development as it decreased

79

However, this is a vicious cycle. Still, due to the economic channels of the resource curse, non-oil

sectors may not seem attractive, and this will hinder the growth of non-oil sectors.

80

By one-stop shop, Hübner (2011) meant “the Azerbaijan Service and Assessment Network, known as

ASAN, which means “easy” in Azerbaijani, is a multi-purpose service network that provides easy access

to public services and streamlines government-to-citizen communication” (Mehdiyev 2020).

81

Which joined in 2007.

76

the country’s economic attractiveness for developing non-oil tradeable sectors (Alili–

Bittner 2017).

Despite the apparent signs of the NRC, economists have expressed a high degree

of optimism about the possibility of sustainable economic development in Azerbaijan;

said optimism is conditional on oil revenue being effectively managed, trade reforms

being implemented, and corruption and rent-seeking behavior being brought under

control (Ismayil 2015). A particular policy preference should be given to the strategic

use of accumulated revenue to achieve productivity growth that can ensure an increase

in competitiveness, which is achievable for the government (Musayev 2016).

Nevertheless, a critical approach to successful oil revenue management should not be

ignored, as recently Gurbanov et al. (2017) demonstrated that the government’s

expenditure on capital has not translated into industrial growth; in theory, this should be

realizable because the 2015–2020 State Program on Industrial Development was

adopted as an economic reform tool for realizing growth in non-oil tradeable

manufacturing.

82

In the case of Azerbaijan, some insights obtained through the literature review

largely follow theoretical expectations, but others require further investigation. The

inclusion of new variables that measure relevant aspects of the NRC and multiple

methodologies would allow for an improved conceptualization of the NRC as well as

significantly extend knowledge on the topic. Despite statements by various authors

about the government’s successful fight against NRC syndrome, they do not have

enough explanatory power to allow conclusions to be drawn either way regarding its

existence in Azerbaijan. However, the limitations of the NRC should also be noted,

which are detailed in the final subsection of the literature review as follows.

3.1.4. Criticism and limitations of NRC theory

Despite the popularity of NRC theory in academic research since the early 1990s, it has

also been heavily criticized. Simply assuming an unconditional negative association

between resource exports, abundance, and production may lead to misleading results.

Mineral economies are not necessarily doomed to experience slower growth because

resource revenue provides foreign exchange, which is a source of government spending

and resource-based industrialization (Auty–Warhurst 1993). Instead, the so-called NRC

has direct and indirect effects, and many things may help to alleviate them. According

82

See also Hamidova (2018) for the policy recommendations for diversifying the Azerbaijan economy.

77

to Mapon and Tsasa (2019), the problem of the NRC does not exist; instead, they

asserted that it is the failure of government leadership to develop prodevelopmental

institutions to suppress the rent-seeking behavior of interest groups that channels

mineral revenue toward private benefit. Moreover, several scholars have found that

natural resources can have a positive impact on the labor force (Lederman–Maloney

2003), real GDP, and institutional quality (Brunnschweiler 2008). Thus, what the NRC

doctrine asserts is questionable.

Certain relationships between economic indicators seem to meet the critical

approach to NRC syndrome. In terms of direct and indirect association, natural resource

abundance and income per capita do not exhibit a direct negative association (Canuto

and Cavallari, 2012). A finding of Ding and Field (2005) was that resource assets are

positively associated with growth, but that resource dependency is negatively associated

with growth, as predicted by Sachs and Warner (2001).

Lederman and Maloney (2003) based their statement “there is no resource

curse” on an analysis of the connection between growth, natural resource exports, and

intra-industry trade. The authors found a positive impact of the share of primary exports

in total GDP and the labor force on growth and criticized Sachs and Warner’s (1993)

NRC syndrome.

NRC theory was also challenged by the cross-country survey of Brunnschweiler

(2008), which employed OLS and 2SLS estimation methods. In the study, an

examination of natural capital, as measured in USD per capita and selected institutional

indicators (i.e., the rule of law), revealed that mineral resources did not have a negative

impact on real GDP growth and institutional quality from 1970 to 2000.

Moreover, Brunnschweiler and Bulte (2008) found a negative correlation

between resource dependence and income per capita from 1970 to 2000, indicating the

validity of the NRC. However, a positive correlation appeared after the explanatory

variable was changed from resource dependency to resource abundance. The authors

emphasized the endogeneity problem of resource dependence (as an explanatory

variable) in the correlation analysis on growth and conflict. They also found just the

opposite: a causal relationship among conflicts and weak institutions and the NRC,

which would mean that the NRC is a result rather than a cause of underdevelopment.

As mentioned earlier, natural resources are not necessarily an evil and

sometimes even lead to improvements in a country’s economic indicators. For instance,

evidence from 31 OECD countries (including resource-rich ones) suggested that

78

globalization and natural resources improve financial development (Zaidi et al., 2019).

If institutional quality in a resource-rich country is high, the gross savings rate should

also be high; however, if corruption hinders institutional quality, genuine savings

should substantially decrease (Dietz et al. 2007).

Despite these critical approaches to NRC syndrome, the doctrine was assumed

to be valid in the present study to analyze the impact of oil revenue and associated

economic activities on economic growth and development. This assumption was also

made to ascertain whether observed non-oil economic deterioration can be attributed to

the oil sector in Azerbaijan.

3.2. Dutch disease

The history of the DD phenomenon goes back to the economy of the Netherlands,

where manufacturing sharply declined after the discovery and exports of North Sea

natural gas (Humphreys et al. 2007). This economic phenomenon, which went against

the expectation of improved economic and social outcomes in the long run, drew the

attention of both academia and popular media. The Economist (1977) coined the term

“Dutch disease” and, since then, DD has become a popular model for conceptualizing

sectoral changes in resource-rich economies.

The Dutch case had several notable characteristics, which served to ground the

DD hypothesis. Examples include an overvalued national currency, increased

government spending, and decreased non-resource exports. However, why should a

country experience adverse effects due to a booming sector? Eerd (2010) argued that

declining manufacturing in the Netherlands resulted from the negative macroeconomic

effects of large-scale projects and increased government spending occasioned by natural

gas revenue. These factors created unsustainable financial mechanisms for financing

these large investments in the long run. In particular, previously competitive domestic

tradeable sectors lost competitiveness due to higher local prices and real exchange

appreciation (Barder 2006). Imports became cheaper, removing an economic incentive

for local producers to create value-added. Moreover, Eerd (2012) stressed the role of

Dutch politicians, who failed to resist the temptation to spend the revenue that flowed

into the country within a short period, which also required little accountability to inject

cash into the national economy. Then, increased energy prices multiplied the utility of

the Slochteren natural gas field for the Netherlands economy, which in turn impacted

79

growth, inflation, the labor market, the external balance, fiscal accounts, the industrial

structure, and social policy (Ciuriak 2014).

Based on the Dutch case, Eerd (2012) asserted that DD could happen in both

developing and developed countries. However, Friedman (2006) insisted that only oil-

rich countries tend to experience the effects of DD, as they are vulnerable to the impact

of oil booms through the oil price channel. In addition, resource abundance and oil

prices jointly identify the economic behavior of the state through politicians and

political regimes. Unlike Friedman (2006) and Eerd (2010), Wacziarg (2012) failed to

find any statistically significant correlation between oil prices and political regimes.

Although mineral resources (especially in oil- and gas-rich countries) serve as

well-known case studies for DD, the Australian gold booms of the 19th century, the

Colombian coffee boom of the 1970s, and gold and silver extraction by Spanish and

Portuguese colonial powers are also considered events that gave rise to DD (Humphreys

et al. 2007). In other words, all of the abovementioned economic events changed the

economic structure as well as fiscal and monetary policy of countries challenged by

sustainable long-term economic development. This brings us to a major point—whether

DD could provide an economic explanation for the NRC doctrine.

Today, DD is the most common approach for representing the macroeconomic

challenges of resource-rich countries (Henstridge–Roe 2020). It is usually defined as a

chronic exchange rate appreciation resulting from foreign currency inflow and

inefficient management, which increases price levels in the economy (Van Wijnbergen

1986; Bresser–Pereira 2013; Barder 2006). Sachs and Larrain (1993) argued that the

mismanagement of windfall gains harms the productivity and competitiveness of non-

mineral tradeable sectors and abruptly decreases the country’s supply of these goods

and services. Therefore, the most distinct symptoms of DD are an appreciated exchange

rate, domestic inflation, and reduced international competitiveness of domestic

exporters as a result of an overvalued national currency (Stiglitz 2005). The main reason

for these symptoms are an increase in the price of the main export mineral or

agricultural commodities, which in turn triggers a resource misallocation process

through relative prices in economic sectors (Badeeb et al. 2016; Murshed 2004). Then, a

deindustrialization process occurs (Corden–Neary 1982). De-industrialization refers to a

shrinking non-resource manufacturing and industrial capacity (Koritz 1991; Gregory et

80

al. 2009), whereas de-agriculturalization refers to a declining role of agrarian sectors in

the economy

83

(Johnston 1970; Gollin 2010).

The first model for the DD hypothesis was presented by Corden and Neary

(1982). The model assumes a three-sector economy: a resource tradeable sector

(booming) or SB; a non-resource tradeable sector (lagging) or SL, which mainly consists

of agriculture and manufacturing; and a non-tradeable sector (tertiary sectors, usually

services) or SNT. According to Corden (1984),

84

world markets determine prices in both

resource tradeable and non-resource tradeable sectors, while domestic markets dictate

prices in non-tradeable sectors. Another critical concept in DD theory is the real

exchange rate, which is represented as the ratio of non-tradeable sector prices to

tradeable sector prices. Other assumptions in the theory are as follows: two production

factors exist in the economy (labor and capital); different sectors have various levels of

labor-to-capital ratios; capital is fixed but labor is mobile in the short run; the labor

market is flexible and full employment exists in the economy; and internal demand in

the national economy is determined by household consumption (as summarized by

Mironov–Petronevich 2015).

DD occurs through two effects (either alone or jointly): the spending effect and

the resource movement effect. According to Corden (1984), sources of the spending

effect can direct spending by resource owners (directly through profit-making

companies if the government does not own a high share of booming sectors) or

government expenditures (indirectly through collected taxes and subsequent state

spending). In both cases, there is a higher demand for non-tradeable goods, as this

demand drives and leads to an increase in the price levels of non-tradeable sectors in

terms of manufacturing goods, which Corden referred to as “real appreciation” (Nülle–

Davis 2018). Government spending triggers consumption sourced by imports at the

expense of domestic manufacturers, as explained by the DD hypothesis, which

discourages necessary non-resource production activities (Humphreys et al. 2007). In

addition, the spending effect may be triggered if an expectation exists of high capital

83

However, one important distinction should be mentioned here. Usually, de-industrialization and de-

agriculturalization in developed countries are understood as the economy’s natural structural change

process against the background of increasing per capita income and productivity. Thus, Corden and

Neary’s (1982) Dutch disease-related de-industrialization and de-agriculturalization should be understood

differently. In other words, de-industrialization or de-agriculturalization are attributable to booming

sectors that give rise to the effects of DD (Palma 2008; Üngör 2013).

84

A non-mathematical summary of the DD model.

81

inflows into the country due to new agreements for extractive exploration and extraction

(Rodriguez–Sachs 1999).

The second effect is the resource movement effect. As the name suggests,

because of an increase in marginal labor productivity, labor resources move from

lagging sectors toward booming sectors to meet increased need there. Resources

moving away from lagging sectors will lead to direct de-industrialization; thus, the

market for non-tradeable sectors is not involved and the real exchange rate will not

appreciate. However, if labor resources move from non-booming sectors to non-

tradeable sectors because of the spending effect (in addition to the resource movement

effect), then they will also move from non-tradeable sectors into lagging sectors (a

combination of the resource movement and spending effects). Figure 3. 2 (panel a and

b) depicts these effects in more detail:

Figure 3.2: Effects of Dutch disease according to the original theory by Corden and

Neary (1982) and Corden (1984).

a. Resource movement effect

b. Spending effect

Source: The author’s own construction based on Corden (1984).

Notes: SB = booming sectors; SL = lagging sectors; and SNT = non-tradeable sectors. W and P denote

wages and prices, respectively.

Technical improvement; resource

discovery or rise of factor prices

Increased demand for labor in SB

Shift of labor resources

From SLto SB (direct deindustrialization)

From SNT to SB

From SLto SNT

Fall in the output of SL(direct

deindustrialization) or/and fall in the

output of SNT (indirect deindustrialization)

The rent of specific factors of SLdeclines;

W increases in SBand SNT in terms of SL

Becasue of increased demand for labor in

SBand SNT, W* increases (only in the

resource movement effect)

The spending effect can increase or

decrease W*. The prices of SNT rise,

causing W/P to fall and W* to rise

From SLto SB (direct deindustrialization)

From SNT to SB

From SLto SNT

Fall in the output of SL(direct

deindustrialization) or/and fall in the

output of SNT (indirect deindustrialization)

82

The literature has provided theoretical frameworks that assist in conceptualizing

the effects of DD on an economy; however, measuring these effects entails several

challenges due to the difficulty of determining the counterfactual size of tradeable

sectors in the economy (Brahmbhatt et al. 2010).

85

Sachs and Warner (1999) argued that

the growth-generation process in resource-rich countries depends on increasing returns

to scale in economic sectors. If increasing returns to scale occur in non-tradeable

sectors, mineral revenue generated from resource booms can be channeled to non-

tradeable sectors and increase economic growth. However, if they occur in the tradeable

manufacturing sector, resource booms would hinder economic growth due to the effects

of DD. In other words, the precise identification of productivity and return to capital

indicators is fundamental in DD studies.

Furthermore, resource-based industries are not labor-intensive, resulting in them

making a weak contribution to a country’s employment rate (Roemer 1979).

86

Oil

booms lead to higher wages in booming sectors, overly rapid consumption, and higher

government spending, which increase domestic price levels and hence decrease the

competitiveness of agriculture and manufacturing. As indicated by the original case of

the Netherlands, a declining ratio of non-petroleum products to GDP translates to a loss

of competitiveness (Gylfason 1984). Thus, resource booms cannot permanently raise

GDP and per capita GDP if the country’s exports heavily depend on primary goods. In

addition to real exchange rate appreciation, this also leads to high-level protectionism in

manufacturing (Auty–Warhurst 1993), which is usually accompanied by rent-seeking

behavior by firms.

Overcoming DD is a prerequisite for sustainable long-term growth (Auty–

Warhurts 1993). If a country is mineral-rich, it may experience receding economic

development due to increasing technological gaps and divergence in productivity vis-à-

vis trading partners who are not mineral-rich (Cherif 2013). Therefore, any

displacement process of the manufacturing industry in a developing country should be

minimized. Otherwise, the strong positive externalities resulting from diversification

and adaptation to the current economic environment cannot be provided by the

85

According to Chenery and Syrquin (1975) and Brahmbhatt et al. (2010), the counterfactual refers to

measuring what would have been the size of tradeable sectors in the absence of natural resources.

86

Of course, exceptions do exist. In the south-central United States, there have been no signs of the

resource curse despite the prominent role of the natural gas sector. According to Weber (2014), this is

because every gas-related job created at least one non-mining job, and the resources did not have a

crowding-out effect on non-resource sectors. However, Weber (2014) also highlighted the volatility of

resource revenue, which creates risks for local and state government revenue and spending. And,

environmental and health effects deserve a mention here.

83

economic structure. However, the growth-enhancing potential of lagging economic

activities should also be studied; otherwise, support for non-competitive and

economically unviable sectors may fuel rent-seeking behavior (Ciuriak 2014).

In addition, Auty (2007) mentioned three institutions that can limit DD: a capital

fund that smoothes the revenue or spending stream and ensures a competitive exchange

rate; cooperation with the Extractive Industries Transparency Initiative (EITI) to

decrease rent-seeking possibilities and establish organizational units to evaluate public

investments; and the establishment of a public investment evaluation unit to regulate

mineral rent deployment and increase the government’s propensity to invest in

productive economic activities. Stiglitz (2005) also mentioned stabilization funds as an

institutional tool and integral part of macroeconomic policies for countering the NRC.

Stabilization funds can play a significant role by diminishing the rent-seeking activities

of governments of resource-rich countries, establishing transparent spending

frameworks, and decreasing the natural propensity of central governments to spend

mineral revenue. However, depending on the political regime, sovereign wealth funds

may fail in their mission if resource wealth becomes easily manipulated.

Nevertheless, DD may not pose a serious risk to economies compared with other

transmission channels for the NRC. Sala-i-Martin and Subramanian (2013) argued that

waste and corruption from oil resources, rather than the DD hypothesis, are the true

sources of long-term economic underperformance.

87

Moreover, theoreticians and

scholars have asserted that DD is not a “disease” but rather an adjustment process of the

economy toward comparative advantage (Sachs–Warner 1955); furthermore, it can be

viewed as a form of structural disequilibrium rooted in booming sectors and

macroeconomic failures (Davis 1995; Kojo 2014).

Thus, even the conceptualization of developing and developed countries within

DD syndrome is a scientific puzzle. Still, DD presents a valuable opportunity worth

examining in resource-rich countries as well as a useful analytical framework for

explaining their economies (Ebrahim-Zadeh 2003). A careful examination of a national

economy according to booming, lagging, and non-tradeable sectors is a prerequisite for

diagnosing the DD phenomenon or simply tracking its signs. Notably, for mineral-rich

87

The authors gave an example taken from Nigeria arguing that all Nigerian citizens should have a right

to benefit from oil revenue directly. They address the popular policy solutions for the economic

mismanagement sourced by huge mineral revenue such as prudent macroeconomic policies and fiscal

policy. Privatization and trade liberalization are irrelevant. This should allow for the Nigerian government

to address the key issue related to the oil abundance, namely oil revenue management. Still, it is hard to

argue that DD is directly related to the rule of law (Mehlum et al. 2006).

84

countries, DD as a theoretical explanation for economic underperformance or a

transmission channel for the NRC is not a strict rule; rather, it should be treated as an

expectation (Davis 1995).

3.2.1. Dutch disease in Azerbaijan

In this section, a brief literature review of the DD phenomenon is provided only in the

case of Azerbaijan’s economy. In Azerbaijan, the main theory used by economists to

model the economy has been the DD theory since the beginning of the recovery period

(1995–2004). While some researchers have merely expressed concern about the

expected effects of DD, which typically occur in oil-rich countries, others have

provided empirical evidence of their existence. In other words, some economists have

not been able to point to any meaningful patterns of DD in Azerbaijan, while others

have vehemently defended them. However, since the commodity supercycle ended

(2014–2015), the popularity of studies using DD has increased, with Azerbaijan serving

as an example due to the devastating external shocks observed in 2015. Therefore, the

obvious signs of DD in Azerbaijan’s economy paint a complicated picture for scholars.

Further studies are required to clarify the underlying mechanisms responsible for the

DD syndrome in Azerbaijan’s economy. Such a clarification would assist in

understanding the process of the de-industrialization of certain non-oil manufacturing

sectors.

This literature review section is divided into three subsections, namely early

studies (4.1.1), direct studies (4.1.2), and indirect studies on DD (4.1.3). This

classification allows for the critical evaluation of the results obtained through empirical

methods as well as an investigation of the possible presence of DD in Azerbaijan’s

economy.

3.2.1.1. Early studies

In Azerbaijan, economic growth has been driven by extractive industries since 1994,

when the Contract of the Century was signed.

88

Since then, the non-oil manufacturing

sectors have been neglected by the incumbent government. Within 4 years of the

Contract of the Century, the government’s exclusive focus on the oil sector was

criticized by scholars and economists. They stated that this one-sided policy could lead

to the NRC and DD in the long run if policy makers do not make the necessary changes.

88

See Chapter 2 for more details.

85

Early studies on the relevance of DD to Azerbaijan’s economy mostly lacked

precise research based on country-specific data and careful reasoning. They addressed

concerns about the resource-based development of the economy compared with that of

other mineral-rich countries. By contrast, later studies adopted a more focused approach

(e.g., Hasanov 2013, Majidli 2022). Several authors have believed that DD could

manifest due to the Azerbaijani government’s lax attitude toward reforms in the non-oil

sector, low incentives for private sector development, and weak institutional and

governmental capacities. For example, Singh and Laurila (1999) noted the euphoria of

Azerbaijani officials after they signed oil and gas contracts with Western MNCs due to

easy money in the form of FDI. The authors also argued that the upcoming boom in oil

revenues could overshadow non-oil industrial development. Governmental authorities

could always race for oil rents and neglect the work required to reform non-oil

manufacturing sectors. Moreover, Singh and Laurila (1999) highlighted the possibility

of exchange rate overshooting, which could result from rising oil revenues in the

medium and long term. This tends to occur when governments continue reckless public

spending policies fueled by high oil revenues. These claims were well founded and

based on the experience of other mineral-rich countries examined for the NRC or DD.

Reports by the World Bank (1996; 1999; 2003), the EBRD (1995; 1997; 2000; 2003),

and the IMF (1995; 1998) also warned Azerbaijani officials of the possible DD effects

of economic development focused unilaterally on the oil industry. Nevertheless, from a

methodological perspective, the simple assumption that DD is caused by the role of the

extractive industry in Azerbaijan’s economy is unsatisfactory. A more targeted and

specific approach should be adopted to track the signs and symptoms of DD.

After empirically comparing FSU and CEE countries, De Broeck and Sløk

(2001) claimed that exchange rate appreciation posed a real threat to post-Soviet

countries. The authors found that in the 1990s, exchange rate appreciation was caused

by productivity growth in CEE. However, resource-rich post-Soviet countries failed to

cope with undesirable exchange rate appreciation in parallel with low productivity

growth rates. However, as Laurens (2002) noted, prior to the implementation of the

largest pipeline project (i.e., the BTC), there was only concern about the signs of DD,

not the necessary preconditions for it in Azerbaijan.

89

Among these concerns was the

lack of incentives for the Azerbaijani government to promote the non-oil sectors. Before

89

However, O’Lear (2001) emphasized the increasing role of oil prices in the Azerbaijan economy could

be seen even 4-5 years before the BTC became operational.

86

the construction of the BTC, a bonanza of oil revenues did not occur and the signs of

DD were not clear enough in official statistics. After the oil boom, some experts

expected significant growth in non-oil production through domestic investment and

diversification policies. Azerbaijan actually lacked the financial resources as well as the

will to sponsor any significant non-oil manufacturing projects until 2005. As official

statistics and published articles demonstrate, this situation did not change after the oil

boom began in 2005 and peaked in 2011 (Ibadoghlu 2021).

Although Rosenberg and Saavalainen (1998) were optimistic about the

management of Azerbaijan’s oil revenues, they still suggested taking action against the

expected signs of DD. They observed high inflation, low public consumption, rising

output of the oil and non-tradeable sectors, and a shrinking non-oil manufacturing

sector. Following Laurila and Singh (2001), one can conclude that the lack of

preparation for the huge mineral revenues was due to the high level of corruption in the

government. Simultaneously, a weak financial sector, an underdeveloped private sector,

and a lack of political democracy exacerbated expectations of negative effects from oil-

driven economic growth. Moreover, concerns about DD were usually accompanied by

reasons related to the political economy and economic diversification. Therefore, when

Kaser (2003) and Mahnovski (2003) stated that the low levels of political and economic

diversification of Azerbaijan’s economy posed a risk and fell under the NRC or DD,

this can be understood as a logical conclusion to observations of Azerbaijan at that time.

Their main argument was based on similar economic expectations of the resource-based

development strategies in the Caspian Basin countries. These were authoritarian and

exhibited rent-seeking behavior, which was a common political and economic reality.

Furthermore, Kamrava (2001) was one of the first to predict that DD in

Azerbaijan’s economy was likely. The author focused on the uneven outcome of the

privatization campaign, inefficient bureaucracy, and transparency problems. These were

the main institutional challenges to building sustainable economic development. A

heavy dependence on oil, corruption, and a reluctance to privatize the banking sector

exacerbated the problems in Azerbaijan’s economy (Kryukov 2005). Political,

governance, and institutional failures following a systemic change usually lead to

failure in efficiently managing windfall revenues and building a diversified economic

structure (O’Lear 2001). These developments resulted in the inability of domestic

economic sectors to absorb the colossal oil revenues and productivity.

87

Studies that have directly analyzed the signs and symptoms of DD in

Azerbaijan’s economy were mostly conducted in 2009 and 2010, but the process is still

ongoing. The authors cited in the next subsection took a more focused approach, and the

available statistical data allowed them to be more methodologically conclusive.

3.2.1.2. Direct investigations

Gahramanov and Fan (2002) conducted the first empirical and direct analysis of DD in

Azerbaijan’s economy. They used an adapted version of the Balassa–Samuelson model,

which explains how a country’s national currency appreciates in value due to efficiency

growth in tradeable sectors. Indeed, the oil boom in Azerbaijan occurred in a tradeable

sector. In other words, domestic inflation and REER appreciation are the main signs of

DD in the oil sector, and they were also included in the Balassa–Samuelson model.

Thus, this approach could explain why REER appreciation by the oil industry is

possible. However, the authors excluded the possibility of DD in Azerbaijan’s economy

because the expected monetary signs of DD, such as oil-related inflation and exchange

rate appreciation, were absent. Notably, however, in 2001/02 when the paper was

written/published, the full extent of the oil boom was not yet apparent. Moreover, the

statistical data were poor for such a study. When the BTC became operational in 2005

and 2006, more statistical data, systematic studies, and empirical research on DD in

Azerbaijan began to appear. In fact, Hwang et al. (2010) argued that there was no well-

established long-term relationship between oil prices and the Azerbaijani exchange rate

before the construction of the BTC. Therefore, the work of Gahramanov and Fan (2002)

lacked the data necessary to fully investigate the DD phenomenon in Azerbaijan.

Hwang et al. (2010) concluded that due to the construction of the BTC,

increased oil exports and oil prices affected the exchange rate more than domestic

productivity growth. This strengthened the local currency and weakened the

competitiveness of non-oil manufacturing sectors. Hasanov and Samadova (2010)

documented a negative impact of REER appreciation on non-oil GDP

90

between 2002

Q3 and 2009 Q3 by applying the vector error correction method (VECM). Similarly,

Hasanov (2010) demonstrated that Azerbaijani currency appreciates by 0.7% when oil

prices increase by 1%. Moreover, the author discussed the signs of DD in light of the

cointegration relationship between oil prices and real exchange rates. Studies by

Huseynov (2009), Ağazade (2018), and Dikkaya and Doyar (2017) have also found

90

Not only non-oil manufacturing, but also non-tradeable sectors.

88

similar relationships between oil price shocks and price levels in Azerbaijan. The

aforementioned studies have emphasized the patterns of chronic exchange rate and

inflation increases coinciding with DD syndrome.

As available statistical data accumulated with developments in Azerbaijan’s

economy, more thorough studies also appeared on DD. Hasanov (2013) conducted one

of the most crucial. According to his results, a positive long-run relationship exists

between public spending and non-oil GDP. They also indicated that the spending effect

of DD is more pronounced compared with the resource movement effect, as oil

revenues are the main source of government spending. In other words, the transfers

from SOFAZ to the state budget financed public spending rather than tax revenues. In

Azerbaijan, the weak presence of the resource movement effect is mainly related to the

capital intensity of the oil sector. Uçan and Ünal (2018) also found evidence of the

spending effect of DD by applying methods such as FMOLS regression and the Granger

causality test between 1996 and 2016 in Azerbaijan.

Yun (2018) argued that Azerbaijan has undergone significant de-

industrialization—a crucial element of DD theory—because the oil industry has

dominated GDP output and exports and crowded out non-oil tradeable sectors.

Furthermore, chronic appreciation of the REER exacerbated the decline in non-oil

competitiveness, even more so than in other oil-rich FSU countries (e.g., Russia and

Kazakhstan). This provides support for claims of oil-related de-industrialization in

Azerbaijan (Niftiyev 2020a; Azizov 2021). Moreover, Yun (2018) recommended that

policymakers implement structural reforms to overcome vulnerabilities to international

commodity markets; otherwise, oil prices will continue to determine the main directions

of the ongoing de-industrialization of Azerbaijan’s economy.

Zulfugarov and Neuenkirch (2019) analyzed the decline in quarterly oil and

non-oil GDP after negative oil price shocks in Azerbaijan between 2002Q1 and

2018Q1. Applying linear vector autoregressive models, they found that GDP per capita

and total trade turnover had been positively affected by oil price increases (Zulfugarov–

Neuenkirch 2019; Mukhtarov et al. 2021). In addition, oil price slumps have

significantly affected Azerbaijan’s economy through reduced government subsidies for

non-oil sectors and the less profitable oil sector (Zulfugarov–Neuenkirch 2019). Recent

studies by Czech and Niftiyev (2021) and Shahin et al. (2021) have also demonstrated

that oil prices strongly determine the value of the manat. During periods of high oil

prices, the national currency appreciates, resulting in low competitiveness of non-oil

89

manufacturing exports. By contrast, during periods of low oil prices, such exports

become more competitive.

Since 2014–15 (i.e., the end of the commodity boom), the number of studies

examining DD syndrome in Azerbaijan’s economy has increased. For instance, Niftiyev

and Czech (2020) documented the impact of DD on vegetable export markets.

According to their findings, vegetable exports were lower during periods of high oil

dependence, as measured by the extractives dependence index (EDI). Moreover,

vegetable exports were higher when the EDI decreased. The EDI considers many DD-

related economic indicators, such as the share of booming and non-booming sectors in

value-added, the revenues they generate, and the fiscal balance of the economy. The

study of Niftiyev and Czech (2020) suggested that DD’s impact can occur not only at

the aggregate level, such as oil, non-oil, or non-tradeable, but also at the subsector level.

In Azerbaijan’s economy after the recession period (1991–94), the dependence

on oil, appreciation of the national currency, increase of real wages in the mining sector,

slowdown of the non-oil industry, and growth of the state-sponsored non-tradeable

sector coincided. Niftiyev (2020a) claimed that these signs of DD were present in the

economy. Niftiyev (2020b) applied the same logic to labor markets to determine

whether there were signs against REER appreciation. He concluded that the resource

movement effect was rather weak compared with the spending effect of DD, which was

also reflected in employment dynamics.

Moreover, Yıldırım Mızrak and Gurbanov (2013) stated that the spending effect

was more visible due to high government spending on infrastructure projects. A certain

strand of literature also focused on the sloppy application of techniques that led some to

conclude that signs of DD existed in Azerbaijan. They either predicted the wrong trends

in economic indicators or failed to examine the necessary and crucial parts of DD

syndrome based on the original theory of Corden and Neary (1982) and Corden (1984).

Bayraç and Çemrek (2019), for example, erroneously concluded that Azerbaijan has

signs of DD. This was based on a causal relationship between oil consumption and

economic growth. The finding was not supported by any DD-based theoretical

framework, and the authors overlooked critical aspects of DD, such as the movement of

resources from lagging to booming sectors and government spending. The same fallacy

was committed by Şanlisoy and Ekinci (2019) when they analyzed the relationship

between GDP and crude oil prices. The authors ignored the fact that DD cannot be

identified solely by examining the relationship between GDP and crude oil prices. They,

90

like Bayraç and Çemrek (2019), also attempted to examine DD in Azerbaijan’s

economy, but they failed to examine the crucial elements of the DD framework, such as

the REER, distribution of sectoral output, and employment. This led the authors to

categorically deny the existence of DD syndrome in Azerbaijan’s economy.

Furthermore, Majidli (2022) found that non-oil exports are negatively affected

by oil production, mainly through the resource movement effect of DD. He also found

that an increase in oil prices positively affects non-oil exports in the long run. This

positive effect is due to the subsidies that the Azerbaijan government usually provides

after oil revenues are transferred from SOFAZ to the state budget.

3.2.1.3. Indirect investigations

Indirect investigations of DD in Azerbaijan’s economy mainly include studies that

analyze fiscal policy, government spending behavior, or domestic inflation dynamics,

which are the key aspects of DD. Such studies are all mainly concerned with the

spending effect of DD.

The main features of the spending effect include high government spending

during periods of increased commodity prices (procyclical fiscal policy) and high shares

of windfall transfers to the state budget. Non-tradeable sectors also increase their share

of total value-added due to increased state-sponsored infrastructure projects. Usui

(2007) noted that Azerbaijan was careless with its oil revenues and spent immediately

during oil price spikes. No institutional regulations were implemented, and the

government saved less for future generations. This has led to procyclical fiscal policies

and made the country vulnerable to unpredictable oil price fluctuations. After applying

OLS, autoregressive distributed lags (ARDLs), and other techniques, Aliyev et al.

(2016) also found a positive long-run relationship between public spending and non-oil

GDP between 2000Q1 and 2015Q2. Their results were consistent with those of

Hasanov (2013), who found that oil revenues are the main source of public expenditure.

They also drive up domestic prices.

An oil boom drives up expectations and contributes to authorities’

overestimation of further revenues. Hayat et al. (2013) noted an appreciation of the

REER in Azerbaijan due to the government overestimating oil reserves and relying

more on mineral revenues. This in turn leads to a loss of control over government

expenditure, especially when the government embarks on costly public infrastructure

projects. Such projects are also often accompanied by foreign borrowing. The overall

91

result is a loss of fiscal discipline, which fuels inflation, foreign debt accumulation, and

the spread of corruption. Therefore, in the case of Azerbaijan, an expected nexus may

exist between economic growth and government spending, which is actually a

significant part of the spending effect of DD. For instance, Sabiroglu and Bashirov

(2013) tested Wagner’s law, which examines the relationship between economic growth

and the public sector’s share of the economy. Although their study did not address DD

syndrome, the authors examined the role of oil prices in government spending in

Azerbaijan. They found that public sector spending increased sharply after favorable oil

price peaks, beginning in 2005. Their findings support the results of an earlier study by

Koeda and Kramarenko (2008), who found that non-oil sector growth and total factor

productivity growth in Azerbaijan were caused by high government spending.

Moreover, Guliyeva et al. (2021) claimed that oil revenues and government spending in

Azerbaijan’s state budget are strongly linked to oil prices, resulting in inefficient and

procyclical fiscal policies.

The appreciation of the national currency may reflect the change in trade

dynamics. For example, Dikkaya et al. (2018) analyzed the relationship between oil

prices and the volume of imports from Turkey to Azerbaijan. The authors reported an

increase in imports during periods of higher oil prices, implying that Azerbaijan has

more purchasing power because of oil revenues. Simply put, imports became cheaper,

leading to a high propensity to consume during the oil boom period. These results

highlight the importance of currency appreciation and depreciation in relation to key

macroeconomic indicators. Similarly, Bahmani-Oskooee and Jamilov (2014) reported

positive and significant responses of non-oil sectors in Azerbaijan to currency

devaluation shocks. When the local currency is devalued, exports to major trading

partners in the EU increase. These authors’ conclusion invites a reexamination of the

impact of DD on trade patterns (i.e., harmful and debilitating effects of an appreciated

exchange rate).

In summary, the aforementioned studies, even if they did not deal directly with

DD, have covered the crucial elements of the theory, such as oil prices, government

spending, exchange rate appreciation, and reckless fiscal policy.

92

3.3. De-industrialization

This section discusses the literature on de-industrialization at both the global and

Azerbaijani levels. It then presents the chemical industry as a whole and its major

trends.

3.3.1. Concepts of industrialization and de-industrialization

The Merriam-Webster English dictionary defines a nation’s industry as “manufacturing

activity as a whole,” which essentially means “the process of producing products

through the use of machinery and factories” (Merriam–Webster 2021). Industrial

production enables the production of material goods that cannot be cultivated on land

(Hewitt et al. 1992). Due to productivity advantages over other sectors and higher

positive externalities resulting from its growth, industrialization—particularly

manufacturing that produces durable goods—is considered an engine of economic

growth throughout the global economy (Szirmai 2012). These definitions assume that

industrialization is a society-wide process of increasing the production of goods and

services, in turn leading to greater wealth (Kiely 2005). Manufacturing has played a

central role in raising the level of economic development and contributed to poverty

reduction (Szirmai 2012). By contrast, de-industrialization can be seen as a threat that

forces policy makers to continue wealth creation through industrialization.

91

A consensus exists that Clark (1957) followed by Kaldor (1966) were the first to

use the term “de-industrialization” in their articles (Kandžija et al. 2017). De-

industrialization has been defined as the decline in the share of industry in total output

and employment (Saeger 1997; Tregenna 2014; Rodrik 2016), which is the opposite of

industrialization. Nevertheless, it differs from terms such as destructuring or

restructuring of industry (Koritz 1991).

92

Similarly, Gregory et al. (2009) described de-

industrialization as a decline in industrial activity and capacity, particularly in

manufacturing. Villanueva and Jiang (2018: 162) defined de-industrialization “as the

91

Industrialization necessitates a greater use of new technological processes and ways of research and

development, as well as a more skilled and productive labor force and entrepreneurial cadre (Cypher–

Dietz 2008). Industrialization increases physical capital investment at the enterprise level and across the

economy, including the creation of physical and social infrastructure (Cypher–Dietz 2008). As a result,

the key message of this literature review is that authorities should pay attention to role and share of

manufacturing in national economies and take action if the de-industrialization phenomenon occurs on a

large scale as manufacturing positively impacts economic growth (Szirmai–Verspagen 2015; Cantore et

al. 2017; Attiah 2019).

92

Sood and Kohli (1985) describe restructuring of the industry as the natural response to the challenges

like declining productivity, economic losses, and diminishing market shares. Therefore, the restructuring

of industries or enterprises allows them to adjust for increased competitiveness. Also, unlike de-

industrialization, restructuring does not lead to massive industrial employment and output collapse.

93

change in the production technique that involves a decrease in labor productivity and an

increase in capital productivity.” In the UK in the 1960s and 1970s, concerns about the

slowdown in economic development caused by industry led economists to begin studies

of de-industrialization. For instance, Kaldor (1966) focused on the British case, and

Singh (1977) characterized de-industrialization as a structural imbalance in the UK.

While productivity and price competitiveness were rising in the UK, manufacturing was

losing its overall competitiveness due to inefficiencies in production. Put differently,

Singh (1977) argued that a decline in manufacturing employment cannot be considered

de-industrialization if productivity and output levels remain the same. In the UK, de-

industrialization was a symptom rather than a cause of structural problems.

The process of economic development and becoming a developed nation is

attributed to the evolution of economies from agriculture to manufacturing, and from

manufacturing to services (Clark 1941; Kuznets 1957; Chenery 1979; Fuchs 1980;

Cypher–Dietz 2008). Economic development and growth lead to structural changes “in

what is produced, in how production is organized, in labor use, and for whom

production is ultimately destined” (e.g., the domestic market versus the export market;

Cypher–Dietz 2008: 268). Economic research took off in the second half of the 20th

century, with a clear emphasis on the importance of manufacturing and

industrialization.

Manufacturing boosts economic growth by shifting from activities with low to

those with high productivity levels (as cited in Pandian, 2017; Lewis 1954; Cornwall

1977). This provides a way out of an unfavorable specialization in primary goods

(Frank 1967). Supply and demand—and hence price levels for primary goods and

natural commodities—tend to be volatile. According to Prebisch (1950, 1959, 1964)

and Singer (1950, 1975), the so-called old orthodoxy argues that as countries’

economies continue to grow, demand for manufactured goods will increase more than

the demand for primary goods. In turn, the terms of trade (TOT) for exporters of

primary goods will deteriorate relative to those of industrialized countries.

93

Countries

that overproduce, exporting primary goods and importing mostly manufactured goods,

do not only experience instability in their TOT due to volatile commodity prices; their

purchasing power for manufactured goods will also deteriorate in the long term

(Cypher–Dietz 2008). Kindleberger (1956, 1958) supported Prebisch and Singer’s

93

The Prebisch–Singer hypothesis.

94

thesis but found no evidence for the difference in TOT between primary and

manufactured goods at that time; rather, he found a significant difference between the

TOT of underdeveloped and industrialized (developed) countries.

94

This was actually a

roundabout way of confirming the Prebisch–Singer hypothesis. Later, Harvey et al.

(2010: 376) provided more direct evidence. They found that “overall, eleven major

commodities

95

provide new and robust evidence of a long-run decline in their relative

prices.” Thus, industrialization creates a productive domestic economic structure for

achieving higher economic growth, which allows developing countries to have similar

export and import patterns (Cypher–Dietz 2008). Ultimately, the message of the old

orthodoxy was that no country can develop without a strong manufacturing sector

because the economic value of primary goods is unstable (Acharya 2007).

Rowthorn and Ramaswamy (1997) argued that manufacturing is a

technologically advanced sector of the economy in which more can be produced over

time with less labor. This is not readily possible in other sectors. In particular,

manufactured products are highly standardized and mass produced, while certain

services (e.g., medical care and catering) cannot be easily standardized. Although the

telecommunications sector is as technologically advanced as manufacturing,

productivity growth in key service subsectors is slower. Therefore, as the role of

services in the economy increases and the share of manufacturing decreases, the average

long-term growth of the economy will depend on productivity growth in the service

sector. In the short and medium term, however, the service sector may not be able to

support the growth of the economy.

In the literature, more recent examples demonstrate the importance of

industrialization for a country to reach a higher level of prosperity. For example,

Fagerberg and Versbagen (1999, 2002) regressed real GDP growth rates and

manufacturing growth rates in 76 countries. They concluded that industrialization led to

higher growth rates in Latin America and the East Asian region. By contrast, the

benefits of industrialization were smaller in advanced economies, with the exception of

the 1950–1973 period. Other recent studies have focused on low- and middle-income

countries, such as South Africa and other African countries (Olamade–Oni 2016;

94

Between 1913 and 1952, “the net barter terms of trade of Western Europe improved by 50% vis-a-vis

the “underdeveloped areas of the world outside of Europe” (Kindleberger, 1955, p. 290).

95

Aluminum, coffee, jute, silver, sugar, tea, wool, zinc, hide, tobacco, and wheat are just a few examples.

95

Moholwa 2017), Turkey (Ozturk–Altinoz 2018), and some Central and Eastern

European countries (Ulbrych 2017). They have attempted to defend the thesis that

manufacturing plays a crucial role in the growth of the economy.

Industrialization has also played a vital role in emerging economies catching up

with high-income countries (Rodrik 2011; Szirmai et al. 2013). Hausman et al. (2007)

and Hidalgo et al. (2007) have demonstrated that manufacturing helps achieve higher

levels of economies of scope “with countries that can produce larger varieties of goods”

(Haraguchi et al. 2019). Thus, developing countries can rapidly grow their economies.

Rodrik (2007) argued that rich countries become rich because of the wide variety of

goods they produce, not just because of their ever-increasing industrial output per se.

Rodrik (2013) also analyzed a large sample of countries and documented unconditional

convergence of productivity only in manufacturing. This gives low-income countries

the opportunity to rapidly exploit their economic potential to catch up with high-income

countries. However, both developed and developing countries face major risks from de-

industrialization.

According to Rowthorn and Wells (1987), de-industrialization can be both

positive and negative. If the service sectors can absorb the excess labor created by

unemployment in manufacturing, then this is positive de-industrialization. However, if

the service sector cannot absorb the excess labor, the economy would experience

negative de-industrialization.

96

Rowthorn and Wells (1987) also proposed a third type

of de-industrialization, namely trade de-industrialization. This occurs when net

manufacturing exports move away from non-manufacturing exports of goods and

services. Trade de-industrialization illustrates the external factors that can lead to de-

industrialization. In other words, the following are usually considered internal factors of

de-industrialization: “result[s] of the interactions among changing preference patterns

between manufactured and services, the faster growth of productivity in manufacturing

96

Felipe et al. (2018) argued that manufacturing employment is more important than manufacturing

output, as in the countries where manufacturing employment takes more than a share of 18% of overall

employment, the economic well-being—measured by the population income—is high. This is not

surprising, as the literature pointed out that manufacturing sectors successfully absorb the excess

agriculture labor force (McMillan–Rodrik 2011; Timmer et al. 2015), and manufacturing share of

employment statistically significantly affect economic growth (Pandian 2017). Also, manufacturing

employs social labor rather than isolated labor, thus benefiting the overall economy when the state must

protect and nourish industries to support nation-building according to Frederick List’s approach (German-

U.S. economist who believed tariffs on imported goods would stimulate domestic development, Cowen–

Shenton 1996). However, agriculture and industry employment has been decreasing across the global

economy since the 1970s to reach 30% in overall employment altogether (Iversen–Cusack 2000).

Notwithstanding, the importance of manufacturing employment for economic growth is still important for

both developed and less developed countries (Pandian 2017).

96

compared to services, and the associated relative decline in the price of manufactured

goods” (Rowthorn–Ramaswamy 1999: 34).

A particular type of de-industrialization is also found in middle-income and

developing countries, which Rodrik (2016) termed “premature de-industrialization.”

Premature de-industrialization means that certain developing countries peak in

manufacturing at earlier stages of economic development than developed industrialized

nations do. Haraguchi et al. (2019) highlighted that the rise of the service sector and the

rapid development of breakthrough technologies—such as automation—have led to a

decline in the role of industrialization in developing countries’ economic growth.

However, in transitional and post-socialist countries, de-industrialization may occur due

to the transition process from a command economy to a market economy. Mickiewicz

and Zalewska (2001, 2002, 2006) referred to this type of de-industrialization as “forced

de-industrialization.” Tomljanović et al. (2018) summarized the aforementioned

authors’ main ideas and concluded the following: If the transition period is efficient,

industry employment is low only during the transition period; once the economy

recovers from severe transition shocks, industry employment increases; however, if the

transition period is inefficient, then only industrial employment in the primary sectors

will prevail.

Su and Yao (2017) argued that the weak performance of the manufacturing

sector and the strong position of the service sector may send bad signals to the

governments of developing countries, as only the manufacturing sector provides

positive externalities on services. Furthermore, according to their findings, the

manufacturing sector not only creates incentives for saving among the population but

also promotes technological accumulation in the country. Nevertheless, between 1990

and 2010, manufacturing employment began to have a less positive effect on growth in

less developed countries worldwide (Pandian 2017). Pandian’s findings invite one to

consider the importance of manufacturing employment in a more complex sense. That

is, the reorganization of manufacturing activities worldwide is changing the ability of

manufacturing to contribute to growth in less developed countries. Nevertheless,

manufacturing has a positive impact on growth and is necessary for development.

Here, the following question arises: Do the most current studies support the

assumption that there is de-industrialization in developing countries? Kruse et al. (2021)

argued that it was too early to state that developing countries have experienced de-

industrialization. The authors documented significant employment industrialization for

97

many countries in sub-Saharan Africa and Asia. They mentioned that a noticeable de-

industrialization process occurred until the early 2000s, but then the trend was reversed

(Kruse et al. 2021). Although Erubman and Vries (2021) supported these findings, they

also reported low and even negative productivity levels for many sub-Saharan African

countries, where poverty cannot be reduced due to sluggish structural transformation. In

other words, even if developing countries somehow defeat de-industrialization, they

will repeat the success stories of the old industrialized countries. This is because

productivity growth in manufacturing, measured by the median level of per capita

income, does not approach that of high-income countries.

Although some countries de-industrialize faster than others, Rowthorn and

Ramaswamy (1999) argued that de-industrialization is not necessarily undesirable. They

argue that the majority of the labor force is employed in manufacturing and services;

therefore, the evolution of employment shares is largely determined by output and

productivity growth in these two sectors. Despite similar growth rates in output, the

authors found that labor productivity grew much faster in manufacturing than in

services. Thus, unemployed workers in manufacturing are absorbed by the service

sector, but the service sector has slow productivity growth (also see Rowthorn–

Ramaswamy 1997).

De-industrialization is the systematic reduction of the productive capacity of the

economy (Bluestone–Harrison 1982, as cited in Kandžija et al. 2017; Bluestone 1984).

Its determinants and causal factors can be internal or external. The main internal

determinants include the following: GDP per capita, economic expansion or recession,

structural changes (Kandžija et al. 2017), labor productivity growth, investment in

manufacturing, and declining demand for manufactured goods—or a shift in consumer

spending to services (Rowthorn–Ramaswamy 1997). The external factors of de-

industrialization include trade patterns (Kucera 2003; Bogliaccini 2013) and

globalization (Saeger 1997; Alderson 1999; Van Neuss 2018). In fact, several studies

once argued that increased North–South trade led to de-industrialization among the

advanced economies—namely the USA, European countries, Japan, and even the Asian

Tigers—due to the rapid growth of labor-intensive manufacturing in low-income

countries (Rowthorn–Ramaswamy 1997).

97

The literature has clearly captured the role

of North–South trade in de-industrialization (Burgstaller 1987; Sachs–Shatz 1994;

97

This is the model that explains how trade between the North or core and the South or periphery leads to

the growth of developing countries (Abdenur 2002).

98

Wood 1994, 1995; Saeger 1997); however, Rowthorn and Ramaswamy (1997) argued

that it has little to do with de-industrialization. De-industrialization began in successful

advanced countries as the result of effective economic development due to productivity

growth and the rise of services (Rowthorn–Ramaswamy 1997). Moreover, real domestic

spending on manufactured goods remained relatively stable; instead, it was spent on

imported goods (Rowthorn–Ramaswamy 1997). Kang and Lee (2011) demonstrated

that de-industrialization is mainly caused by internal and external factors, namely labor

productivity growth, changes in demand, and trade patterns.

Although it is increasingly difficult to follow the classical industrialization-led

economic growth based on productivity growth, recent studies have reported the

growth-enhancing role of FDI in manufacturing. In some economies, FDI in services

fails to boost overall growth and even leads to de-industrialization (Doytch–Uctum

2011). Moreover, advanced economies are responding to growing competitive pressures

from China and India with the Fourth Industrial Revolution (Industry 4.0

98

,

99

), which

provides flexibility and quality in production systems to meet the demands of

innovative business models (Khan–Turowski 2016). Manufacturing that generates high

added value and ensures sustainable production requires a higher degree of

digitalization (Möller 2016) and “the promotion of a tight connection between science

and the economy” (Zhou 2013: 7). In other words, industrial production continues to

evolve but retains its role as growth engine, despite the ongoing de-industrialization in

developed and developing countries.

3.3.2. De-industrialization in the case of Azerbaijan

De-industrialization as a separate issue has received little attention in Azerbaijan’s

economy (for early reflections, see Malik 2010; Gurbanov 2013). Some studies have

mentioned de-industrialization or relative de-industrialization based on the increasing

role of the service sector due to government spending (e.g., Hasanov 2013). Other

studies have been descriptive and unable to clearly explain the de-industrialization

process (e.g., Niftiyev 2018). To study the broader economic outcomes of oil-led

growth and development since 1991 and gain a thorough understanding of their impact

98

Khan and Turowski (2016: 17) defined Industry 4.0 as “a revolution enabled by the application of

advanced technologies (like IT) at the production level to bring new values and services for customers

and the organization itself.”

99

Industry 4.0 and the idea of sustainable growth are examples of an “industrial renaissance,” which

Yoshikawa (1995) said were necessary to keep manufacturing as a source of growth and prosperity.

99

on the diversification status of the national economy, the de-industrialization process in

Azerbaijan must be analyzed.

Since their independence in 1991, post-Soviet resource-rich countries have been

studied for signs of DD- or REER-induced de-industrialization. Usually, empirical

studies use the REER as an indicator of the competitiveness of tradeable sectors. For

example, Egert (2012) reported significant REER and NEER appreciations in

Southwest Asia—including Azerbaijan—when oil prices increased. Yun’s (2018) recent

study on selected post-Soviet oil-exporting countries demonstrated that DD and de-

industrialization pose real threats to post-Soviet countries. His results indicated that the

de-industrialization process is more pronounced in Azerbaijan than in Russia and

Kazakhstan, as measured by the decline in manufacturing employment. However, he

did not find any significant impact of REER appreciation on manufacturing

employment in Azerbaijan or other oil-exporting post-Soviet countries.

According to Mikayilov and Najafov (2019), not only did REER appreciation

negatively impact the share of manufacturing in Azerbaijan’s GDP but also high interest

rates. High interest rates prevented manufacturing firms achieving sustainable

production levels due to limited access to financial capital. Noteworthily, the authors

found a negative correlation between the ratio of credit to value-added and the share of

manufacturing in GDP. High interest rates contributed negatively to manufacturing

firms’ costs, reducing their competitiveness and production capacity (Mikayilov–

Najafov 2019).

The de-industrialization literature on Azerbaijan’s economy is sparse, as this

process has hardly been studied. Therefore, the present study contributes to the

literature by by elucidating a specific non-oil manufacturing subsector, namely the

chemical industry. Thus, it provides a focused perspective on the industrial realities in

Azerbaijan.

3.4. The chemical industry at a glance

Heaton (1993) asserted that although the use of chemicals dates back to ancient

civilizations, the chemical industry in its present form did not emerge until the Second

Industrial Revolution of the 1800s. The chemical industry makes extensive use of

“fossil-resources such as minerals, natural gas, oil, coal, air, and water as raw materials

to produce more than 60,000 different chemical-based products” (Mohan–Katakojwala

2021: 1). In the early years of this industry’s rise, alkali for soapmaking, silica and

100

sodium carbonate for glassmaking, and bleaching powder for cotton were the main

inorganic chemicals that it produced for other manufacturing sectors (Heaton 1993).

Then, in the 1860s, the organic chemical industry began to develop following “William

Henry Perkin’s discovery of the first synthetic dyestuff—mauve” (Heaton 1993: 1).

From 1950 onwards, the organic chemical industry experienced explosive growth,

which accelerated with the development of petrochemicals in the 1960s and 1970s; in

particular, this was because of the increasing demand for synthetic polymers, such as

nylon, polyester, epoxy resins, polypropylene, and polyethylene (Heaton 1993). The

chemical industry started to provide various resources for a wide range of economic

sectors.

The products of the chemical industry are used in “health, agriculture,

transportation, food, energy, environment, consumer goods, etc., [and] include the

product categories of pharmaceuticals, agrochemicals, pesticides, basic and specialty

chemicals, fibers, detergents, functional materials, petrochemicals, polymers, and fuels”

(Mohan–Katakojwala 2021: 1). The chemical and petrochemical industries supply

inputs to technology-related companies in the form of reagents or intermediates.

However, the chemical industry supplies the most critical inputs for certain sectors of

the economy, including the following: pharmaceuticals, rubber and plastic products,

construction, computers and electronic equipment, furniture, cars and trailers,

agricultural product and food packaging, and containers (e.g., bags, trays, and mats;

Abbasov–Aliyev 2018). The chemical industry takes small quantities of raw materials,

such as oil and natural gas, and converts them into high value-added industrial or

consumer goods (Heaton 1993). The relative value of typical petrochemical products is

10 times that of crude oil, while typical consumer goods have a value 50 times that of

the raw material (Heaton 1993). In addition, the chemical industry is a high-technology

industry that benefits from recent advances in electronics, engineering, and computer

technology, which are capital- rather than labor-intensive (Heaton 1993). The chemical

industry has high value per unit as opposed to high-volume products (Heaton 1993).

Following the First Industrial Revolution, the chemical industry experienced

different growth rates in different countries. According to Aftalion (2001), each country

concentrated on specific areas of the European continent. Germany, for example,

established its first chemical laboratories and, by 1914, the German chemical industry

had a 75% share of the world market (Lesch 2000). Italy produced over 900,000 tons of

superphosphates per year at the time (Aftalion 2001). The Nitro Nobel Company,

101

founded by Alfred Nobel in Sweden in 1864, specialized in the production of dynamite

(Aftalion 2001). The production of fertilizers, carbides, and chlorates also took off in

Sweden due to increasing hydroelectric power generation. Aftalion (2001) stated that

the French Revolution isolated France from the rest of the world, which favored the

chemical industry. The decline in imports from countries such as Italy, Spain, the UK,

Peru, and India forced local elites to focus on domestic raw materials to produce various

chemicals for the rest of the economy. As a result, by the end of 1810, French industry

was able to produce “20,000 tons of sulfuric acid, 10,000–15,000 tons of Leblanc soda,

and 600 tons of hydrochloric acid” (Aftalion 2001: 14). By 1835–36, Britain’s chemical

industry was self-sufficient, producing sulfuric acid, Leblanc soda, hydrochloric acid,

and similar chemicals (Aftalion 2001).

In developed countries today, it is safe to assume that the chemical industry has

already matured and even slowed. However, around the early 1990s, oil-rich developing

countries such as Saudi Arabia and Mexico expanded their production of basic alkene

and aromatic petrochemical intermediates, achieving high rates of growth in chemical

manufacturing (Heaton 1993). Developing energy-rich countries focused on the specific

chemical subsectors, where they could obtain stable raw materials and cheap energy

sources (Jonnard et al. 1985). As a result, countries such as Russia, Brazil, India,

Indonesia, China, and South Africa increased their share of annual global chemical sales

from 13% to 28% (UNEP 2012). Moreover, despite its late entry into the chemical

market, China has been the world’s largest chemical producer in terms of revenue since

2011 (Hong et al. 2019, as cited in Chen–Reniers 2021). All of this demonstrates the

importance of considering the chemical industry as a critical component of the

industrialization of developing countries, of which Azerbaijan is one. Azerbaijan is rich

in oil, natural gas, and other minerals. In the years of the USSR, chemicals represented

one of the main directions of industrialization; however, the years since independence

present a complicated picture.

3.5. The chemical industry and Azerbaijan’s economy

The development of Azerbaijan’s chemical industry was closely connected to oil

refining. The first commercial and industrial oil extraction in the world occurred on the

Absheron Peninsula in 1847; later, the Dubin brothers opened an oil refinery in 1859

(Bagirov 1996). Since the mid-18th century, oil production and refining in Azerbaijan

have steadily developed and opened up new opportunities for other industries, such as

102

the chemical industry. In 1913, for example, the Nobel brothers succeeded in producing

sulfuric acid using the Tentelev contact process in Baku (Aftalion 2001). Sulfuric acid

was used primarily in oil refineries to wash impurities out of gasoline and other refined

products (The Columbia Electronic Encyclopedia 2012). Moreover, sulfuric acid

produced in Baku was intended for the production of other chemicals, such as sodium

sulfate, hydrochloric acid, chlorine (through the Weldon process), and bleaching

powder (Aftalion 2001). The production of sulfuric acid continued during the Soviet

Union period. At the beginning of the 20th century, the chemical industry established

various factories in Azerbaijan, which produced, among others, sulfuric acid and caustic

soda. Among these chemical enterprises, Azerbaijani entrepreneurs accounted for 10%

of all owners (Seyidzada 1998).

In 1928, five-year plans (1928/29–1932/33) aimed at developing industry were

adopted (Ibrahimli–Aziz 2011). However, few industries were developed in the early

years of the Soviet Union. The rapid development of the chemical industry in

Azerbaijan actually dates back to the 1950s and 1960s.

In 1938, the construction of synthetic rubber factories and other chemical plants

began in Sumgait, Azerbaijan’s second largest city. With the outbreak of the Second

World War, the construction works in Sumgait were stopped. After the war ended in

1945, however, heavy industrial plants—chemical, pipe-rolling, synthetic-rubber,

aluminum, and superphosphate—were put into operation there. The Sumgait Chemical

Plant, commissioned in 1966, was considered the largest petrochemical plant in Europe

at the time (Azerbaijans.com). In the 1970s and 1980s, these factories were operating at

full capacity, exporting products to almost all parts of the FSU. This economic growth

helped to make Sumgait one of the largest cities in Azerbaijan (Azerbaijans.com).

In 1981, the construction of the EP-300 complex for the production of ethylene

and propylene, a valuable petrochemical product, began in Sumgait, based on modern,

high performance, and waste-free technology with a high production capacity. Today,

based on the EP-300 plant, the Sumgait Ethylene-Polyethylene Plant currently produces

high-quality polyethylene, which is in high demand in world markets. It is also a crucial

source of raw materials for the growing chemicals complex in Azerbaijan (Strategic

Road Map for the National Economic Prospects of the Republic of Azerbaijan 2016).

The recent development of the chemical industry has also benefited significantly

from Azerbaijan's rich hydrocarbon resources. However, compared to the years before

and during the Soviet Union, the post-Soviet years were mainly known for the fact that

103

the chemical industry was supported by SOCAR. In other words, not only the inputs for

the production process, but also the constant state financing and investments were tied

to the oil sector. Nevertheless, in terms of production, the chemical industry is a part of

Azerbaijan's economy that is not part of the oil sector.

Suleymanov and Turkan (2019) summarized the main developments in the

Azerbaijan’s chemical industry as of 2001 as follows: To successfully develop one of

the key sectors of the Republic of Azerbaijan (i.e., the chemical industry), lead it out of

the crisis, and solve related problems, the state-owned enterprise Azerikimya

Production Union (PU) was established on the basis of existing enterprises. Since the

establishment of Azerikimiya PU did not lead to the desired results, the Decree of the

President of the Republic of Azerbaijan dated March 21, 2001, titled “On Privatization

of Enterprises of the Chemical Industry,” privatized most of the enterprises that were

part of Azerikimiya PU. Four enterprises remained as part of Azerikimya PU: Sintez-

Kauchuk, Organic Synthesis, Surfactants, and Ethylene-Polyethylene. Later, the Sintez-

Kauchuk Plant was transferred to the Ethylene-Polyethylene Plant, and by the Decree of

the President of the Republic of Azerbaijan dated April 2, 2010, titled “On

Improvement of Management Mechanisms in Petrochemical Industry,” Azerikimiya PU

was transferred to SOCAR.

100

Kerimli et al. (2021) reported that a new era began with the establishment of

SOCAR Polymer LLC in 2013, the first public–private partnership in Azerbaijan in the

chemical and petrochemical industry: “The company’s polypropylene (PP) and high-

density polyethylene (HDPE) production facilities are located at the Sumgait Chemical

Industrial Park (SCIP) site” (socarpolymer.az). Kerimli et al. (2021) assumed that

Azerbaijan’s Absheron economic region can develop clusters of SMEs to exploit

existing oil and gas resources. However, obstacles exist to achieving stability in the

chemical industry, the largest of which are as follows: bureaucratic barriers, difficulties

in financing business activities, political sanctions that increase the costs and risks of

operations, and a high tax burden encouraging the shadow economy (Kerimli et al.

2021). Nevertheless, recent developments in the chemical and petrochemical industries

in Azerbaijan, as well as in the institutional, economic, and political environment, have

provided new opportunities and challenges for shaping the non-oil manufacturing

sector.

100

The Azerkhimiya PU was made a part of SOCAR by a decree from the President of the Republic of

Azerbaijan on April 22, 2010.

104

After the painful devaluation decision in 2015, the Azerbaijani government

adopted a document titled “Strategic Roadmap for the National Economic Prospects of

the Republic of Azerbaijan,” which prioritized the chemical and petrochemical

industries. In the 21st century, developing Azerbaijan’s chemical industry is crucial.

The post-oil boom era is characterized by increasing global economic influences and

lower oil prices; thus, the country needs to develop industries outside of the oil sector

that are competitive and capable of contributing to the growth of the national economy

(Gamidova 2017).

101

Various infrastructure supports were also considered for the budding chemical

companies.

102

The government claimed the following:

Work continued on the creation of external and internal infrastructures in

the Sumgait Chemical Industrial Park, office, consulting, laboratory,

business start-up, training and vocational education services, and other

necessary infrastructure facilities for the effective implementation of

entrepreneurial activities. The implementation of the “Polymer” project by

the State Oil Company of the Republic of Azerbaijan in the Park is the

largest project of its kind and scale in the petrochemical industry of

Azerbaijan in the last 40 years. In addition, large-diameter corrugated

polyethylene pipes, steel pipes, mechanical and hydraulic equipment, and

glass panels based on float technology are produced at the park (rolling in

a hot bath). The country’s first pesticide production plant has been put into

operation. (New Production Facilities and Prospects, Strategic Roadmap for

the National Economic Prospects of the Republic of Azerbaijan 2016)

Although certain development programs have prioritized the development of the

chemical industry, such as the aforementioned Strategic Roadmap and “Azerbaijan

2020: Vision for the Future,” FDI in these sectors remains low or nonexistent;

101

The Ministry of Agriculture, together with the Ministry of Economy, will consider providing the

necessary support to the private sector in setting up fertilizer (as well as biofertilizer) plants, if necessary.

One of the potential opportunities in the oil and gas sector is the construction of fertilizer plants, and the

measures taken in the agricultural sector will be coordinated according to such opportunities. The work to

be done in this area is taken into account in the 3rd strategic goal of the “Strategic Roadmap for the

Development of the Oil and Gas Industry (including chemical products) of the Republic of Azerbaijan.”

Measure 4.5.5: Promoting local fertilizer production, Strategic Road Map for the Republic of Azerbaijan's

National Economic Prospects 2016).

102

SME entities representing different sectors can make better use of the opportunities created by special

zones. For example, entities operating in the chemical industry may participate in the joint management

of costly sewerage and wastewater infrastructure (7.1.6. Priority 1.6. Special industrial zones and on the

creation of clusters for SMEs, Strategic Road Map for the National Economic Prospects of the Republic

of Azerbaijan 2016).

105

moreover, energy efficiency is low, production capacity is small, and export

participation could be better (Abbasov–Aliyev 2018). The production of household

chemicals (e.g., detergents and drain cleaners) is also weak in Azerbaijan’s economy.

This limits the development of subsectors of light industry as they are deprived of cheap

inputs (Abbasov–Aliyev 2018). The expansion of oil refining is expected to produce

raw materials, which can be used to increase the capacity of large technological units in

the country’s chemical and petrochemical enterprises (Gamidova 2017). However, some

subsectors of the chemical industry, such as chlorine, hydrochloric acid, caustic soda,

liquid soda, sulfuric acid, and isopropyl alcohol, have either declined sharply or simply

ceased operations (SSCRA 2020).

Overall, despite the chemical industry’s enormous economic potential and other

benefits that it offers many countries, it is a controversial manufacturing sector because

of the high risks that it poses (Malich et al. 1998, as referenced in Dakkoune et al.

2018). Environmental pollution and safety problems are typical negative externalities of

this industry (Chen–Reniers 2020). For example, its facilities are usually located in

densely populated areas and operate with hazardous materials (Reneirs et al. 2006, as

referenced in Dakkoune et al. 2018). Thus, severe industrial accidents could seriously

damage the surrounding environment and urban areas, which the government must

consider in industrial planning (Gomez et al. 2008). Even though the chemical industry

has played a key role in the development of manufacturing, countries are now using

new economic models, such as the circular economy and Industry 4.0, to achieve

sustainable levels of growth and development (Cucciniello–Cespi 2018; Cortés Serrano

et al. 2018; Keijer et al. 2019).

Table 3.1. summarizes the key literature examples and their messages in a

tabular way.

Table 3.1: A quick overview of the key literature examples.

Author(s)

Main findings, factors and messages

Natural resource curse (NRC)

Auty (1993; 2001); Sachs and

Warner (1997; 1998; 1999; 2001)

Slower growth of resource-poor countries compared to resource-rich

countries.

Gylfason and Zoega (2006)

The NRC's general framework: macroeconomic and social

institutions.

Ross (2013)

A strong emphasis on oil as a reason for NRC

Sala-i-Martin and Subramanian

(2003), Murshed (2004), and

Isham et al. (2005)

Agreed with Ross (2013) that point-source natural resources lead to

NRC.

Isham et al. (2005), Mehlum et al.

Strong emphasis on non-democracies where natural resources

106

(2006), and Wick and Bulte

(2009)

become a "curse" rather than a "blessing."

Auty and Warhurst (1993);

Mapon and Tsasa (2019);

(Brunnschweiler 2008)

These authors critiqued the NRC phenomenon by simply pointing to

the role of institutions and governance and highlighting some

positive effects of natural resources on labor and institutions.

Dutch disease (DD)

Corden and Neary (1982) and

Corden (1984)

The first original theory of DD.

Gahramanov and Fan (2002)

The first empirical study of DD on the example of Azerbaijani

economy. The authors rejected DD due to lack of monetary signs of

the theory.

Hasanov (2013)

The first comprehensive work demonstrating the presence of DD in

the Azerbaijani economy due to the strong co-integration relationship

between oil prices and GDP and inflationary pressures from

government spending.

Niftiyev (2021) and Majidli

(2022)

Comprehensive study of the DD phenomenon in Azerbaijan,

indicating the presence of resource movement and the spending

effects of the theory.

Bayraç and Çemrek (2019)

These works concluded that there are no DD signs or effects in

Azerbaijan.

De-industrialization

Clark (1957) and Kaldor (1966)

The authors were the first to use the term "de-industrialization" in

their articles.

Kandžija et al. (2017); Rowthorn

and Ramaswamy (1997)

The main internal determinants include the following: GDP per

capita, economic expansion or recession, structural changes, labor

productivity growth, investment in manufacturing, and declining

demand for manufactured goods-or a shift in consumer spending to

services.

Kucera (2003); Bogliaccini

(2013); Saeger (1997); Alderson

(1999); and Van Neuss (2018)

External causes of de-industrialization: international trade and

globalization.

Rodrik (2016)

Referred to as "premature de-industrialization."

Malik (2010); Gurbanov (2013)

and Hasanov (2013)

Initial reflections on the phenomenon of de-industrialization in the

Azerbaijani economy.

Niftiyev (2020a); Azizov (2021)

and Niftiyev (2022)

Targeted study and identification of the visible patterns of de-

industrialization outside the oil sector in Azerbaijan.

Yun (2018)

A panel study identifying Azerbaijan as a prominent country that de-

industrialized while the REER soared.

Source: The author’s own construction based on the literature review.

3.6. Theoretical framework of this dissertation

Based on the literature review, Figure 3.3 represents the theoretical framework of this

dissertation. In order to address the NRC-triggered DD-led de-industrialization process

in the Azerbaijani economy, the chosen theoretical framework should clearly describe

the complex interactions between the key aspects of both theories. Figure 3.3 shows the

key fundamentals and outcomes of resource-based economic growth and development

by integrating the key NRC and DD aspects relevant to economic growth in an oil-rich

country. The discussion in the subsequent paragraphs provides detailed theoretical

explanations of how DD relates to the de-industrialization of lagging sectors and how it

affects economic growth.

107

Figure 3.3: Theoretical framework of the dissertation based on NRC and DD theories.

Source: Author’s own construction based on the literature review.

While natural resources could have a positive impact on growth if there is

capital accumulation and adequate investment in productive activities, the key

observation is that windfall revenues lead to a "rentier state" effect, where easy access

to resource revenues discourages the government from collecting non-resource taxes.

This also results in the government being less responsive to citizens' demands (Mlambo

2022) and failing to manage oil revenues, leading to corruption and a poor institutional

environment (Kurečić–Seba 2016, as cited in Mlambo 2022). Low political and

institutional quality, in turn, has a direct negative impact on economic growth, as recent

studies by Natkhov and Polishchuk (2019), Bosco and Poggi (2020), Olabisi (2020),

Butkiewicz and Yanikkaya (2010), Ullah (2020), and Oyinlola et al. (2020) show. In

other words, according to the above studies, low political, institutional, and governance

quality hinders SMEs' access to financial capital, increases income inequality, fails to

contribute to poverty reduction, and fails to accumulate adequate human capital that

could provide value-added processed products to resource-rich countries. Indeed, the

108

deleterious effects of weak institutions and poor governance are still evident in many

oil-rich countries (Ajida–Soyemi 2022; Aziz 2022; Kansheba–Marobhe 2022). As

indicated in the conceptual framework in Chapter 1, low social trust in politicians could

lead to sufficient leeway for a few rent-seeking groups to benefit from high oil revenues

in Azerbaijan.

The next more complicated and DD-oriented part of the theoretical framework

describes how natural resource-based growth makes certain sectors unprofitable and

leads to crowding-out effects when labor and capital begin to move between sectors of

the economy. In short, if capital and labor are mobile in the three-sector model of the

economy, a sector that experiences a boom will attract labor and capital, making some

sectors unprofitable and unproductive. This leads to a lopsided economic structure in

which only one good is normally produced and exported. In this case, the non-tradeable

sectors become the next best sector for labor and investment. However, governments of

resource-rich countries tend to invest in unproductive activities that do not increase the

productive capacity of the economy, so the absorptive capacity of the economy is low.

Moreover, this leads to inflation and appreciation of the REER, which reduces

competitiveness and thus hampers growth.

Figure 3.3 shows that a lopsided economy has an indirect negative impact on

growth as the economy requires less human capital and macroeconomic instability

increases as commodity price fluctuations lead to lower national incomes due to low

economic complexity. Domestic inflation, usually due to imprudent procyclical fiscal

policies in commodity-rich countries, leads to higher costs of domestic production for

tradeable sectors. As a result, less human capital is demanded by the economy. If this

government and market failure is not corrected, commodity revenues will only be

invested in non-tradable sectors. In this scenario, the price level of non-tradeables will

further raise the general domestic price level, leading to an appreciation of the REER.

Thus, due to the appreciation of the REER and macroeconomic volatility, the lopsided

economic structure has a negative impact on long-term growth.

Although NRC and DD theories explain the main expectations of resource-based

economic growth and the directions of its effects, individual-level aspects (e.g., de-

industrialization of subsectors) have not yet been studied in depth. It is difficult to say

whether system-level events can be generalized to individual-level cases. This gap

suggests taking a case study approach after aggregate-level analysis and

operationalizing new variables to model more accurate reflections of NRC or DD

109

phenomena. Accordingly, this dissertation uses newly constructed variables such as

EDI, MPC, the oil boom, and economic shocks (dummy variables) to answer this

theoretical question. The rationale for using these variables is that traditional variables

such as oil prices and oil industry output may not be sufficient to capture the impact of

the oil sector on the non-oil sectors of the economy.

3.6. The topicality of NRC and DD for the time being

From recent debates and publications, we can see that NRC and DD are still among the

most important challenges facing resource-rich countries and their regions or provinces.

Certainly, much has changed since the early 1980s and 1900s, when scholars first

defined and modeled the key drivers of NRC and DD (Auty 1993). While 30–40 years

ago the focus was on long-term sustainable growth and development without heavy

reliance on international commodity markets, today there are additional factors. For

example, green growth, the Sustainable Development Goals (SDGs), and institutions

are the main motivators for continuing NRC and DD studies because economic growth

and the creation of tangible wealth is a gradual and evolutionary process rather than an

immediate one. For example, previous and current studies on the role of natural

resources have discussed that resource rents should serve the prosperity of society, but

this has been slow or nonexistent. A few publications are cited below to demonstrate the

continuing relevance of these theories to resource-rich economies.

In the case of single-country studies of large economies, Lee–He (2022) showed

that resource-rich Chinese provinces were unable to achieve high green total factor

productivity, while resource-poor provinces were able to achieve higher levels. In

addition, Lee–He (2022) also sheds light on the importance of the role of institutions, a

much-debated topic in the case of resource-rich economies. For example, market-

oriented institutions have helped turn the "resource curse" into a "resource blessing" by

increasing the allocative efficiency of markets and firms in China. Yu et al. (2022)

reached similar conclusions to Lee–He (2022) in the case of the Chinese economy, but

also emphasized the harmful effects of a special situation called resource dependence.

In other words, the resource wealth of certain regions did not hinder the economic

growth or productivity of those regions, but when a region becomes dependent on

resource rents, there is a clear sign of NRC. The approach used by Yu et al. (2022) was

a generalized weighted least squares support vector panel regression model for the data

sample between 2009 and 2018.

110

In panel data studies, NRC continues to dominate the literature when resource-

rich poor countries and developing countries are analyzed in terms of their key

economic indices. For example, Ajide (2022) examined sub-Saharan African countries

between 1995 and 2018 and concluded that natural resource dependence or

overdependence has hindered the economic complexity of African countries. In other

words, the parallel growth of non-commodity-producing sectors vis-à-vis extractive

industries has not been sustained by the government, indicating the practice of rent

mismanagement. The recent study by Inuwa et al. (2022a) reached similar conclusions:

the ten most commodity-rich African countries suffer from the negative effects of

commodity dependence, particularly in terms of institutional quality and financial

development. Other panel data-based studies by Inuwa et al. (2022b) also found a

negative relationship between natural resources and their impact on growth in the GCC

region and emphasized the crucial role of institutions in the whole process. In the

MENA region, the quality of institutions also remains low (e.g., widespread corruption),

while oil revenues are a large part of the economy (Chebab et al. 2022). As a result,

resource-rich regions continue to dominate the literature in the absence of conclusive

attempts to manage resource rents in a more inclusive and sustainable manner.

It seems that the importance of the NRC theory continues to be supported by

studies from a single country. Dell'Anno and Maddah (2022) studied rent-seeking

behavior and resource rents in the case of Iran and claimed that institutions play a

crucial role in transforming resource rents into long-term sustainable development when

a certain threshold is not exceeded in terms of resource dependence. The other example

comes from Kazakhstan, where DD has been a long-debated academic topic.

Kelesbayev et al. (2022) found that oil prices largely determine the main dynamics of

the REER and CPI. There is also clear DD evidence in the Kuwaiti economy, as Azom

(2022) found that the non-oil manufacturing sector declined while the non-tradeable

sector increased due to positive oil productivity shocks based on the dynamic stochastic

general equilibrium model. As a result, key macroeconomic indicators are highly

correlated with economic activity related to the oil sector (vulnerability to commodity

price shocks).

Other empirical studies reached a different conclusion: Gombodorj and Pető

(2022) found no negative impact of mining on agriculture in Mongolia, but commodity

wealth does not contribute to poverty reduction either. Similarly, in the case of Algeria,

Maachi and Benloulou (2022) found no important empirical evidence for the DD signs.

111

Maachi and Benloulou (2022) argued that the exchange rate channel was well

controlled, and oil prices in Algeria did not lead to an appreciation of the national

currency between 1990 and 2016.

We observe more sophisticated methodological tests and up-to-date data sets

that improve our understanding in the case of single-country and panel data, as

presented in some of the publications cited above. In all studies, the institutional and

human capital channels showed statistically significant results in relation to the NRC

doctrine. Considering the current data samples and methodologies, all these studies

show the timeliness of NRC and DD, as there is still no conclusive evidence that

resource-rich countries have been able to transform their resource wealth into economic

prosperity. While some countries, such as Chile, Norway, and Malaysia, have been able

to diversify their economies by redirecting resource rents to productive sectors of their

economies, the vast majority of resource-rich countries remain dependent on resource

rents (Lashitew et al. 2021). This dependence is based on public spending by domestic

actors and less transparent policies (Brollo et al. 2013). The fact that so few countries

have been able to break free of NRC and DD reflects the complex nature of resource-

based economic growth and development and the quality of the choices their economies

have had to make.

112

CHAPTER 4

DATA AND METHODOLOGY

In this section, it is possible to find all the data and methodological explanations of the

empirical analysis of the dissertation. Because the study included three major empirical

phases—the NRC analysis, the DD analysis, and the chemical industry analysis—the

methodological aspects of each phase are discussed in separate subsections.

4.1. Data and methodology of the analysis of NRC in the Azerbaijan’s

economy

To test the validity of the NRC thesis in Azerbaijan, the present study collected

indicators related to institutions, governance, and human capital (i.e., health and

education spending) to examine whether any statistically significant associations could

be found with oil-related variables. In addition, a figure analysis of the main political,

institutional, and governance indicators was conducted in the initial stage of the

empirical analysis to allow comparisons between the oil boom and post-boom periods.

Lastly, a bottom-up perspective was applied through the evaluation of qualitative data

from World Values Survey (WVS). The selected survey questions were related to the

importance of politics, trust in politicians and the government, and Azerbaijani citizens’

propensity to take political action, reflecting the changing societal perceptions of the

political government between 1994 and 2020.

4.1.1. Dependent variables

Table 4.1 lists the dependent variables that were used in the principal component

analysis (PCA) and regression analysis. In the main, these are the index variables that

were used to measure institutional quality in Azerbaijan. The total government

expenditure on education (TGEE) and out-of-pocket expenditure on health care services

(OP_EXP) represent the percentage shares of GDP and current health care expenditure,

respectively.

The political and institutional channel of the resource curse in Azerbaijan was

traced through the following variables: political stability and absence of

violence/terrorism (POL_ST; hereinafter “the political stability index” or “political

stability”), the rule of law (RULE_O_LAW), the voice and accountability index

(VO_AND_ACC), and latent human rights protection scores (H_RIGHTS; hereinafter

113

“human rights scores”). The first four variables were obtained from the Worldwide

Governance Indicators (WGI) provided by the World Bank, while the last variable was

taken from the data set of Schnakenberg and Fariss (2014), referred to by Fariss (2019)

as “Latent Human Rights Protection Scores.”

Table 4.1: Dependent variables used in the study.

Variable name

Abbreviation

Measurement

Source

Resource

curse

channel

Political stability and

absence of violence

POL_ST

Index value: −2.5 (weak) to

2.5 (strong)

World Bank –

Worldwide

Governance

Indicators

Political and

institutional

channel

Rule of law

RULE_O_LAW

Voice and accountability

VO_AND_ACC

Human rights scores

H_RIGHTS

−3.8 (weak) to 5.4 (strong)

(Schnakenberg

& Fariss,

2014; Fariss,

2019)

Government effectiveness

GOV_EFFEC

Index value: −2.5 (weak) to

2.5 (Strong)

World Bank

(Worldwide

Governance

Indicators)

Capacity to

manage oil

revenue

channel

Regulatory quality

REG_QUAL

Government integrity

GOV_INT

Scale from 0 to 100, where

100 indicates very little

corruption

Total government

expenditures on education

TGEE

% of GDP

World Bank

Education

and human

capital

channel

Out-of-pocket expenditures

on health services

OP_EXP

% of current health

expenditure

Furthermore, POL_ST measured perceptions of the likelihood of political

instability or politically motivated violence, including terrorism. The control of

corruption captured perceptions of the extent to which public power is exercised for

private gain, including both petty and grand forms of corruption, and state “capture” by

elites and private interests. RULE_O_LAW represented perceptions of the extent to

which agents have confidence in and abide by the rules of society, particularly the

quality of contract enforcement, property rights, and the police and courts, as well as the

likelihood of crime and violence. VO_AND_ACC captured perceptions of the extent to

which citizens can participate in selecting their government as well as freedom of

expression, association, and the press. Finally, H_RIGHTS measured the quality of the

overall environment for human rights in a country.

114

All of the variables related to the political and institutional channel, excluding

human rights scores, ranged between −2.5 and +2.5 (the higher the better). Human

rights scores ranged from −3.8 (minimum) to 5.4 (maximum). The examined period was

from 1996 to 2019.

4.1.2. Independent variables

Table 4.2 lists the explanatory variables applied in this study’s regression analysis.

Similar to the dependent variables, the time period covered was 1996 to 2019. Of the

independent variables, only the extractives dependency index (EDI) was calculated

according to Hailu and Kipgen’s (2017) methodology. The formula and accompanying

explanations are as follows:

  󰇟 󰇛  󰇜󰇠  󰇟 󰇛  󰇜󰇠  󰇟 󰇛  󰇜 , (1)

where

EDI is the extractives dependence index for a country at time t;

EIX is the revenue from the extractive industry, expressed as a share of total export

revenue;

HTM is the export revenue from high-skill and technology-intensive manufacturing as

a share of global HTM exported in year t;

Rev is the share of revenue from the extractive industry in total fiscal revenue;

NIPC is non-resource income, including tax revenue, profits, and capital gains as a

percentage of GDP;

EVA is the share of the extractives industries’ value-added in GDP; and

MVA is the countrywide non-resource manufacturing potential, as measured by per

capita manufacturing value-added.

Table 4.2: Explanatory/independent variables used in the study.

Variable name

Abbreviation

Levels of measurement and

description

Source

Extractive dependency

index (EDI)

EDI

Index value

Calculation

based on Hailu–

Kipgen (2017)

Oil rents

OIL_RENTS

% of GDP

World Bank

Oil exports/GDP

OIL_ EXP

Share of oil exports in GDP, %

Oil foreign direct

investment (FDI)

OIL_FDI

Annual FDI to oil sector

Mehtiyev

(2019)

SOFAZ’s share of the

state budget

SOFAZ’s_SH

% of total state budget

Annual reports

from SOFAZ

Economic shocks

ECON_SHOCK

Dummy variable where 1 = 2008–

2009 and 1 = 2014–2015

Based on

economic

downturn years

provided by the

115

World Bank

EDI is a multidimensional variable that captured the share of natural resource

and non-natural resource sectors in the national economy. Oil rents (OIL_RENTS)

represented the difference between the value of crude oil production in international

prices and the total cost of production, while the oil exports to GDP ratio (OIL_

EXP/GDP) represented the role of exports in the overall economy in Azerbaijan. FDI in

the oil industry (OIL_FDI) was another potential channel for booming sectors to

influence the variables of interest. SOFAZ’s share of the state budget (SOFAZ’s_SH)

measured the state budget’s performance in relation to the oil revenue transfers from

SOFAZ. Lastly, economic shocks (ECON_SHOCK) was a dummy variable for the

global financial crisis in 2008 and 2009 and sharp commodity price downturns in 2014

and 2015.

All additional information (e.g., descriptive statistics, normality tests) on the

data analyzed in Chapter 5 related to NRC can be found in the Appendix section (see

Table A4.1 and A4.2).

4.1.3. Empirical strategy

Considering the wide range of the collected data set, the main empirical stage started

with PCA. PCA is beneficial when the data set is large and several variables need to be

examined (Bro–Smilde 2014). Jolliffe’s (1990) early study on PCA stressed that if the

correlation between variables is strong, it may be decreased to discover “a true

dimension” of the data set that would deliver the same information with the least

information loss. This reduction yields “components,” which help one to identify

patterns across various data series (Ringnér 2008). Ringnér (2008) also emphasized the

independence of components rather than them being uncorrelated. If the original

variable quantity a can be reduced to b using newly constructed index variables or

components, a large amount of information can be analyzed using a relatively simple

technique. PCA is often used as a pre-analysis of variables of interest and also as an

analytical bridge for further investigation.

Here, PCA provided the main components for analyzing institutional quality and

its relation to the oil sector. Varimax rotation was used in the PCA to maximize the

variance of the factor loadings (Dien 2010). The main components were then saved as

116

individual time series and regressed against each other using the dynamic ordinary least

squares (DOLS) method. The model specification is as follows:

  



  , (2)

where institutional_quality is the first component of PCA at time t; Oil_factor is the

second component of PCA at time t; and  is the error terms. Furthermore, Oil_factor

was added along with lags, allowing to find the best way to build the model and to test

how stable the results were.

This study also used the ordinary least squares (OLS) technique to test the

impact of individual oil-related variables on the selected human capital variables. Three

models related to this are presented as follows:

    

 (3)

    

 (4)

    󰆒 (5)

In the above-listed models, OP_Expenses denotes the out-of-pocket expenses on

health care; TGEE is the total government expenditure on education; Human Rights is

the human rights scores at time t; and  is the intercept in all models. Then, Oil rents,

EDI, Oil Exports, Economic Shocks, Mining Industry, and SOFAZ’s share are the

explanatory variables at time t. Lastly,  is the error terms at time t.

Lastly, large-scale cross-national WVS questions were evaluated to apply a

bottom-up perspective on Azerbaijan’s institutional quality between 1994 and 2020.

More specifically, the survey results were from the WVS waves between 1994 and

1998, 2005 and 2009, 2010 and 2014, and 2017 and 2020 when the respondents from

Azerbaijan participated in the surveys. Individuals from dozens of nations worldwide

self-report their values, attitudes, and beliefs regularly to the WVS (Barrios 2015).

People are asked how they feel about their lives, including how happy they think they

117

are. They are also asked questions about politics and society in general, including what

they think of competition. The surveys also ask each person about their gender, income,

and religious beliefs, among other things (Barros 2015). Thus, the WVS was a valuable

source for assessing the social perceptions and cohesion in Azerbaijan and how they

have changed over economic development stages (e.g., recession, transition, oil boom,

and post-boom).

4.2. Data and methodology of the analysis of DD in the Azerbaijan’s

economy

This section presents the data and methodology of the study. First, Section 4.2.1

provides an overview of the data sources, sample size, variables and their measures, and

sources. Then, Section 4.2.2 discusses the empirical strategies used to test the DD

hypothesis for Azerbaijan.

4.2.1. Data and variables of interest

The main data sources for this study were the State Statistical Committee of the

Republic of Azerbaijan (SSCRA) and the World Bank. The collected data were

aggregated (grouped) into the following three sectors, as in the literature review section:

SB – booming sectors that comprise the extraction of crude oil and natural gas as well as

petroleum manufacturing; SL – lagging sectors that comprise non-oil manufacturing, the

production of utilities, and agriculture; and SNT – non-tradeable/tertiary sectors that are

a combination of service sectors, such as construction, transportation, and trade.

Table 4.3 contains the names and descriptions of the explanatory variables, units

of measurement, sources, and the section in which they are used. Descriptive statistics,

outlier years, missing values, and the Shapiro–Wilk normality test for the explanatory

and dependent variables presented in this chapter can be found in the Appendix (see

Table A4.3 and A4.4).

The first subsection of DD chapter (5.2.1) contains monthly oil prices (i.e., the

average spot price of Brent). A VAR model was constructed to test whether

international oil prices lead to an appreciation of the Azerbaijani REER. The second

subsection (5.2.2) reports how the REER, nominal effective exchange rate (NEER), oil

prices, oil rents, and EDI affected the three-sector model of Azerbaijan’s economy

between 1990 and 2019. Effective exchange rates were included in two forms, namely

nominal (NEER_66) and real (REER_66) terms based on 66 trading partners.

118

Separating the exchange rates into real and nominal allowed the exchange rate to be

tracked with and without inflation effects. The third and fourth subsections (5.2.3 and

5.2.4) present results on the resource movement and spending effects of DD,

respectively. Compared with other studies (e.g., Hasanov 2013), this study analyzed two

effects of DD theory separately, which allowed for an enhanced conceptualization of

DD in Azerbaijan.

Table 4.3: Independent (explanatory) variables used in the analysis.

Section

Variable name

Description

Measurement

unit

Source

5.2.1.

OIL_PRICES

Crude oil, average spot price

of Brent, Dubai and West

Texas Intermediate, equally

weighted

USD per barrel,

monthly

IndexMundi

5.2.2.

REER_66

Real effective exchange rate

based on 66 trading partners

In percent, 2007

= 100%, annual

Bruegel data

sets, Real

effective

exchange

rates for 178

countries: A

new database

NEER_66

Nominal effective exchange

rate based on 66 trading

partners

OIL_PRICES

Europe Brent Spot Price

FOB, Brent trademark

USD per barrel,

annual

U.S. Energy

Information

Administrati

on

OIL_RENTS

Oil rents of Azerbaijan’s

economy

% share of GDP,

annual

The World

Bank

EDI

Extractives dependence

index

Index value,

annual

Calculation

based on

Hailu–

Kipgen

(2017)

OIL_BOOM

Oil boom period

2005–2014=1,

annual

Dummy

variable

5.2.3.

OUTPUT_SB_GR

Output growth rate of

booming sectors

In %, year-over-

year

SSCRA

OUTPUT_SL_GR

Output growth rate of

lagging sectors

OUTPUT_SNT_GR

Output growth rate of non-

tradeable sectors

INC_AZN

Population income in the

Azerbaijani manat

In AZN, annual

INC_USD

Population income in USD

In USD, annual

REER

Real effective exchange rate

Index value,

quarterly

Bruegel data

sets

Oil_P

Oil prices of the BRENT

trademark

In USD, quarterly

International

Energy

Agency

MIN_EMP

Mining employment

In thousands of

persons, quarterly

SSCRA

SERV_EMP

Services employment

5.2.4.

MPC

Marginal propensity to

consume

Calculated values

Own

calculation

based on

SSCRA data

119

OUTPUT_SNT

Output of non-tradeable

sectors

In AZN, annual

SSCRA

INC_POP_AZN

Population income in the

Azerbaijani manat

INC_POP_USD

Population income in USD

In USD, annual

GOV_SPEND_USD

Government spending

In USD, annual

World Bank

GOV_SPEND_SHA

RE_GDP

Government spending as a

share of GDP

In %, monthly

ST_BUD_EXP

Expenditure of the state

budget

In millions AZN,

monthly

CBAR

Table 4.4 lists the dependent variables of the study. In the first subsection

(5.2.1), only one dependent variable is used, namely the REER (REER_66). In the

second subsection (5.2.2), 10 dependent variables are used to outline the sectoral

distribution of the influence of REER, NEER, oil prices, oil rents, EDI, and the oil

boom period. Only two of them, namely exports of booming (SB_EXP_SH) and

lagging sectors (SL_EXP_SH), are experimental in nature, since DD studies do not

usually examine how the sectoral manifestations of DD show up in exports.

Table 4.4: Dependent variables used in the analysis.

Section

Variable name

Description

Measurement

unit

Source

5.2.1.

REER_66

Real effective exchange

rate based on 66 trading

partners

In %, quarterly

Bruegel data sets

5.2.2.

SB_OUT_SH

Share of booming sectors

in industrial production

% of overall

industrial

production, annual

State Statistical

Committee of

the Republic of

Azerbaijan

(SSCRA)

SM_OUT_SH

Share of

manufacturing/lagging

sectors in industrial

production

SM_VA_SH

Manufacturing value-

added

% of GDP, annual

World Bank

SA_VA_SH

Agriculture, forestry, and

fishery value-added

SNT_VA_SH

Services value-added

SB_EMP_SH

Share of booming sectors

in overall employment

% of total

employment,

annual

SSCRA, World

Bank

SL_EMP_SH

Share of lagging sectors in

overall employment

SNT_EMP_SH

Share of non-tradeable

sectors in overall

employment

SB_EXP_SH

Share of booming sectors

in total exports

% of total exports,

annual

SL_EXP_SH

Share of lagging sectors in

total exports

5.2.3

OUT_SB_GR

Output growth rate of

booming sectors

In year-over-year

%, annual

SSCRA

120

OUT_SL_GR

Output growth rate of

lagging sectors

OUT_SNT_GR

Output growth rate of non-

tradeable sectors

EMP_SB_GR

Employment growth rate

of booming sectors

EMP_SL_GR

Employment growth rate

of booming sectors

EMP_SNT_GR

Employment growth rate

of booming sectors

MAN_EMP

Manufacturing

employment

Quarterly,

thousand people

5.2.4

CPI_GR

Consumer price index

growth rate

In %, annual

Central Bank of

the Republic of

Azerbaijan

(CBAR)

REER_GR

Growth of the real

effective exchange rate

NEER_GR

Growth of the nominal

effective exchange rate

CPI_ANAVE

Consumer price index

Annualized

average, in %,

monthly

The export and import data provided by SSCRA are limited to the period 1996–

2019. Missing values for the period 1990–1995 were taken from World Bank statistical

handbooks (1993, 1996) and inserted into the data set. There were also missing values

for the EDI (for 1990–1993) because data for individual components of the index were

not available for these years. Therefore, the missing values were filled using the Trend

function of the online Google Sheets application using the least squares method. Since

OLS and FMOLS are sensitive to outliers, the outlier values were examined using

boxplots and then Winsorized using the “Trimming and Winsorizing” add-in of Eviews

version 11.

Section 5.2.3 analyzes how the growth rate of output and employment of one

sector affects the output or employment of another sector, for which the OLS technique

was used based on annual data. The initial linear models revealed that manufacturing

employment may be negatively affected by the DD-related variables. For this reason,

quarterly manufacturing employment between 2000 and 2020 was analyzed in the

context of REER, oil prices, services, and mining employment (see Subsection 5.2.3 for

more details).

Similar to the estimation of the resource movement effect, the spending effect

was also modeled using the OLS technique. Subsequently, a BVAR model was

constructed for a more targeted approach, which involved analyzing the short-term

relationship between CPI and state budget spending using monthly data.

121

4.2.2. Empirical strategy

This section summarizes the main empirical strategies that were used to obtain the

results presented in each subsection of Section 5.2.

4.2.2.1. Empirical strategy for REER appreciation

The methodology used to obtain the results in Section 5.2.1 was the standard

unrestricted VAR model. This was used to identify the responses of Azerbaijan’s REER

to oil price shocks. The empirical form of the VAR analysis is defined as follows:

        , (6)

where  is the REER at time t;  denotes oil prices at time t; and  is the

error term. The expected relationship from the VAR model was positive responses of

REER to oil price shocks.

4.2.2.2. Empirical strategy for sectoral implications of REER, NEER, oil prices,

and oil rents

According to the original theory by Corden and Neary (1982) and Corden (1984),

testing the sectoral effects of the key variables of DD was crucial. To achieve this goal

in the case of Azerbaijan’s economy, the FMOLS method was applied, which is an

optimal single-equation linear modeling approach (Philips–Loretan 1991). It effectively

addresses problems of serial correlation and endogeneity among independent variables

based on the OLS technique (Philips–Hansen 1990). Serial correlation and endogeneity

usually occur due to the presence of interactions of the cointegrating process among the

explanatory variables in linear models (Rahman et al. 2021). Moreover, FMOLS

effectively addressed the nonstationarity of the data set by adding linear and quadratic

trends as well as deterministic regressors, which allowed the quality of the modeling to

be improved. Therefore, the sectoral distribution of the impacts of REER, NEER, oil

prices, and oil rents in Azerbaijan was tested, and the results are presented in Section

5.2.1. The only control variable used was the exchange rate (NEER and REER). The

distinction between the NEER and the REER allowed the inflation effects to be

observed and whether sectoral output or employment was affected by oil-related

inflationary pressures to be determined. The model used in Section 5.2.2 was as

follows:

    󰇛 󰇜  

    󰇠 , (7)

122

where  is the intercept,  is the given sector (i.e., booming, lagging [agriculture,

manufacturing, or combined], or non-tradeable) at time t,  is the EDI at time t,

 is the REER at time t,  is the NEER at time t, OIL_BOOM is a dummy

variable, OIL_PRICES denotes oil prices at time t, OIL_RENTS denotes oil rents as the

share of GDP at time t, and TREND is the linear trend of the models. REER and NEER

were the control variables used in the analysis.

4.2.2.3. Empirical strategy for the resource movement effect

The empirical strategy employed to obtain the results in Section 5.2.3 was modeled

after Mironov and Petronovich’s (2015) study on the Russian economy. They applied

the OLS estimation technique to analyze the three-sector model of DD in the Russian

case.

OLS estimation is one of the most popular statistical methods in the social

sciences (Hutcheson–Sofroniou 1999); furthermore, it remains one of the most widely

used multivariate techniques for quantifying hypothesized relationships among

variables of interest (Krueger–Lewis–Beck 2008). The empirical model for the resource

movement effect of DD for Azerbaijan is as follows:

    

󰇛 󰇜   , (8)

where Output GR and Employment GR are the output and employment year-over-year

growth rates of a given sector (SB, SL or SNT) at time t;  is the intercept;   , and

 are the coefficients of the explanatory variables; and  , and  are the

year-over-year growth rates of the sectors at time t. Furthermore,

 and  are the population

income in two currencies, namely USD and AZN, which were the control variables, and

 is the error term of the equation at time t. Notably, the abbreviation “GR” means

growth rate among the listed formulas; moreover, “/” does not indicate the division sign

but rather multiple dependent variables in the same formula.

To ensure a more focused approach to the resource movement effect and reveal

direct or indirect de-industrialization processes in manufacturing employment in

Azerbaijan’s economy, the following unrestricted standard VAR model was applied for

the time period 2001Q1–2020Q4:

123

    



   



 , (9)

where  is the vector of endogenous variables,  is a matrix of k autoregressive

coefficients at lag i,  is a vector of q exogenous variables,  is a matrix of q

exogenous variable coefficients, and  is the error term in the macro-econometric

technique. As in Koitsiwe and Adachi’s (2015) paper, the error term was assumed to

have no serial correlation and covariance matrix.

This econometric technique was developed following a thorough analysis of the

work of Koitsiwe and Adachi (2015; a DD study on Australia). Equation (4) can also be

expressed as follows:

         

       

     , (10)

where  is the oil price,  is the real effective exchange rate,  is

mining employment,  denotes the services employment figure, and  is the

error term. The data were transformed to their first difference because the time series in

their leveled form had a unit root.

4.2.2.4. Empirical strategy for spending effect

Similar to the resource movement effect of DD, this study tested the spending effect of

DD through OLS. The model for this is represented by the following formula:

    

󰇛 󰇜   , (11)

where   and  are the year-over-year growth rates of the

CPI, REER, and NEER;  is the intercept;  , and  are the coefficients of the

explanatory variables;  is the marginal propensity to consume at time t;

denotes the output of the non-tradeable sectors; POP_INC_USD is the population

income in USD at time t; GOV_SPEN_USD is the government spending in USD;

GOV_SPEN_SH_GDP is the percentage share of government spending in GDP; and 

is the error term of the equation at time t.

To ensure a more focused analysis on the spending effect, Bayesian vector

autoregression (BVAR) was also applied using the following formula:

    



 , (12)

124

where CPI_ANAVE is the annualized average monthly consumer price index at time t

and lag i,  is the coefficient,  is the coefficient, ST_BUD_EXP is the expenditure of

the state budget, and  is the error term at time t and lag i.

BVAR has become popular for macroeconomic estimation with data of a small

sample size, relatively simple systems, and with few or multiple predictors (as cited in

Petrevski et al. 2015; De Mol et al. 2008; Banbura et al. 2010; Koop–Potter 2003;

Wright 2003; Stock–Watson 2005, 2006). In a BVAR model, coefficients are often

treated as random values centered around a mean (Chandramowli–Lahr 2012). Most

importantly, BVAR requires one to consider several prior types when estimating a

model. Model parameters in BVAR are treated as stochastic variables with prior

probabilities rather than constant values, unlike typical VAR models. A study

demonstrated priors in BVAR models to be highly beneficial for reducing difficulties in

overparameterization (shrinkage), regardless of whether forecasting or estimating

(Koop–Korobilis 2010). These shrinkages imposed by the set of priors can take the

form of constraints on parameters. They can also be equal to zero. According to Koop

and Korobilis (2010), this has led to a significant increase in the use of Bayesian

approaches because prior knowledge provides a rational and formally consistent way to

introduce shrinkage.

The Normal–Wishart prior, selected for this study, is a special version of the

Minnesota prior. The Normal–Wishart prior is a standard approach in macroeconomics

(see Kadiyala–Karlsson 1997; Banbura et al. 2010). Moreover, it relaxes some

assumptions of BVAR, which was a necessary step in estimating the spending effect of

DD in Azerbaijan’s economy, since the model was small and simple, rather than large

and multidimensional. According to Koop and Korobilis (2010: 11), the Normal–

Wishart prior “assumes that each equation has the same explanatory variables and it

restricts the prior covariance of the coefficients in any two equations by making them

proportional to one another.”

Next, hyperparameters had to be chosen for the BVAR modeling. These are

often set to values previously reported in the literature (Amir-Ahmadi et al. 2020).

There are two types of priors in BVAR under the Normal–Wishart prior, namely

coefficient and residual priors, and their values range from 0 to 1 (Chandramowli–Lahr

2012). Here, the AR(1) coefficient prior (Mu1) was set to 1 due to the nonstationarity of

the data (Ouliaris et al. 2016: 57), while the overall tightness residual prior (Lambda1)

was set to 0.5 to manage the variance of the higher-order lags of the endogenous

125

variable (Ouliaris et al. 2016), since the maximum lags are 12 months. The result of

these steps proved fruitful, as the estimate reflected the expected direction of the nexus

between CPI and the state budget spending according to DD theory.

Lastly, lag selection was set to 12 based on the monthly data type, as the AIC

and other lag selection tests did not yield meaningful results. Variables in the BVAR

model were estimated without any transformation and were in their nonstationary form.

4.3. Data and methodology of the analysis of de-industrialization in the

Azerbaijan’s economy

Because both quantitative and qualitative data were used in this study, the following

subsections describe the sources and strategies used to collect data (subsections 4.3.1

and 4.3.2). Subsequently, the quantitative and qualitative methods are explained in

subsections 4.3.3 and 4.3.4.

4.3.1. Quantitative data

In a given country, the employment and output shares of industries in total employment

and value-added are the standard measures for estimating industry dynamics (Mayer

2018). The analysis presented in this chapter was based on the natural volume of

production in the chemical industry. After a descriptive review of official data, it

became clear that output at current or real prices does not capture the elements of non-

oil manufacturing required to understand de-industrialization.

Empirical modeling was applied to the data range of 1995–2020. Table 4.5

describes the variables and their explanation, measurement, and source. An initial figure

analysis of chemical industry subsectors was performed to select the dependent

variables. The selection of explanatory variables was based on DD theory and recent

literature on Azerbaijan’s economy. For example, Majidli and Guliyev (2020) found

that oil prices played a statistically significant role in the growth of Azerbaijan’s non-oil

GDP from 2005 to 2019. The importance of the REER in studies of Azerbaijan’s non-

oil economy has been repeatedly documented (Hasanov 2010; Hasanov et al. 2017;

Zulfigarov–Neuenkirch 2019). Considering the original DD theory of Corden and

Neary (1982) and Corden (1984), variables such as service sector employment

(SERVICES_EMP) and oil boom (OIL_BOOM) were included in the econometric

models. A dummy variable (OIL_BOOM) was also included to determine what impact

the oil industry’s production, revenue, and export boom might have on chemical

126

subsectors. Finally, the real labor productivity variable in the chemical industry

(RLP_CHEMICAL) tested whether the de-industrialization of Azerbaijan’s economy is

due to productivity gains or losses or to the oil industry.

Table 4.5: Dependent and explanatory variables of the quantitative analysis.

Variable name

Explanation of the variable

Measurement

Source

Explanatory variables

REER

Real effective exchange rate

in %; 2007 = 100%;

calculated based on

170 trading partners

Bruegel

OIL_PRICE

Global crude oil prices

USD per barrel

British Petroleum

– Statistical

Review of World

Energy

SERVICES_EMP

Employment of the service

sector

in thousands of

persons

State Statistical

Committee of the

Republic of

Azerbaijan

RLP_CHEMICAL

Real labor productivity in the

chemical industry

Ratio of output over

employment in the

chemical industry

OIL_BOOM

Dummy variable for the oil

boom period between 2005

and 2014

2006–2014 = 1,

other years = 0

Author’s own

construction based

on macroeconomic

data

Dependent variables

CAUS_SODA

Production of caustic soda

in tons

State Statistical

Committee of the

Republic of

Azerbaijan

CHLORINE

Production of chlorine

HYD_ACID

Production of hydrochloric

acid

IZOP_ALC

Production of isopropyl

alcohol

LIQ_SODA

Production of liquid soda

SUL_ACID

Production of sulfuric acid

thousands of tons

No data were missing in the data set. The only missing value—for chlorine

production for 2018—was replaced by linear interpolation based on the data range of

1995–2017. The descriptive statistics for the variables of interest are presented in Table

4.6.

The descriptive statistics indicated that, among the explanatory variables, oil

prices and labor productivity in the chemical industry have high volatility, as measured

by the coefficient of variation. The skewness estimation results indicated that the REER

and labor productivity in the chemical industry are skewed, while oil prices and

employment data in the service sector have a fairly symmetrical distribution.

127

Table 4.6: Descriptive statistics of the variables of interest for 1995–2020.

Min

Max

Mean

St. Dev.

Coef. Of

Variation

Skewness

Kurt.

Explanatory Variables

REER

79.00

139.56

102.59

18.56

0.18

0.66

−0.95

OIL_PRICE

12.72

111.67

54.66

31.78

0.58

0.48

−0.93

SERVICES_EMP

841.00

1418.10

1125.50

157.72

0.14

0.00

−1.00

RLP_CHEMICAL

0.03

0.31

0.09

0.06

0.70

1.83

3.91

Dependent variables

CAUS_SODA

0.00

7647.80

1263.92

2190.42

1.73

1.56

1.45

CHLORINE

0.00

17035.00

4661.31

4879.59

1.05

0.92

−0.03

HYD_ACID

0.00

8.10

3.81

2.91

0.76

−0.10

−1.53

IZOP_ALC

0.00

24159.20

12248.39

6155.20

0.50

0.11

0.04

LIQ_SODA

0.00

36400.00

14575.23

12523.37

0.86

0.03

−1.45

SUL_ACID

0.00

52.50

16.41

14.66

0.89

0.57

−0.17

Notes: 1) Min, max, St. Dev., and Kurt. denote minimum, maximum, standard

deviation, and kurtosis respectively; 2) coefficient of variation (Coef. Of Variation) is

the ratio of standard deviation over the mean of a variable.

Noteworthily, all subsectors studied here had minimum values of 0, implying a

complete collapse of production, high standard deviation, and variance. In particular,

caustic soda and chlorine production varied considerably, as measured by a coefficient

of variation of 1.73 and 1.05, respectively. Furthermore, the skewness and kurtosis

values indicated that the distribution of the dependent variables was symmetrical,

except for caustic soda production (skewness = 1.56).

4.3.2. Qualitative data

The data sources for the qualitative analysis were semistructured and structured expert

interviews. In this study, the experts were people with extensive knowledge or skills in

a particular area based on their research, experience, or professional background. In a

broader sense, an expert is someone who is an authority in their field—in this case, the

field was the chemical industry and its economics. A total of 16 expert interviews were

conducted in February and March 2022. The experts interviewed included 10 industry

specialists and six economists. The interview process was in accordance with the

Charter of Fundamental Rights of the European Union and the European Code of

Conduct for Research Integrity. The industry experts were selected through

recommendations from universities and research centers in Azerbaijan. In addition,

professional social media networks such as LinkedIn were used to select potential

respondents based on their professional experience. Furthermore, the economists were

selected based on their experience, publications, and other research activities related to

industrial development and the chemical industry in Azerbaijan. Moreover, before each

128

interview, the aim and scope of the research were explained to them and they were

assured of the anonymity of their answers. Thus, it was possible to create a genuine

atmosphere of conversation around the research topic. Moreover, all interviewees were

asked for their consent to be recorded. Only one interviewee—an industry expert—

declined this request. Nevertheless, structured notes were taken to record the process of

data generation. These were then reviewed and a written text of the interview was

reproduced for the qualitative analysis software package Quirkos. Table 4.7 presents

key information about the qualitative data sources:

Table 4.7: Technical details about the interview process and interviewees: Industry

experts (Nos. 1–10) and economists (Nos. 11–16).

No.

Code

Name of Institution

Experience

in Years

Language

of the

Interview

Type of Interview

Date

1.

TG1

SOCAR Polymer

3

ENG

Video interview

12/02/2022

2.

NN7

Sangachal Garadagh Oil and

Gas Terminal, Middle East

Petroleum

20

AZE

Video interview

16/02/2022

3.

OA9

Sobsan LLC

7

AZE

Video interview

17/02/2022

4.

MH4

SOCAR Polymer

7.5

ENG

Video interview

19/02/2022

5.

MH1

Synthetic rubber production,

SOCAR, Azerikimya PU

57

AZE

Voice call interview

26/02/2022

6.

AB6

Former soda PF

40

AZE

Voice call interview

01/03/2022

7.

RG6

Heydar Aliyev Oil Refinery

40

AZE

Voice call interview

07/03/2022

8.

IE2

Azerikimya PU

1.5

AZE

Voice call interview

09/03/2022

9.

ISH8

Azerikimya PU

35

AZE

Voice message

interview

17/03/2022

10.

EA1

Former chlorine PF

15

AZE

Reproduced from

the interview notes

16/03/2022

11.

GI7

London School of Economics

AZE

Video interview

28/02/2022

12.

AM3

Azerbaijan State University

of Economics

32

AZE

Video interview

03/03/2022

13.

IA8

Khazar University, Eurasia

Extractive Industries

Knowledge Hub

20

AZE

Video interview

01/03/2022

14.

SB1

Entrepreneurship

Development Foundation

27

AZE

Written interview

14/03/2022

15.

ERS8

Ruhr-Universität Bochum

15

ENG

Written interview

16/03/2022

16.

EB4

Azerbaijan State University

of Economics

16

AZE

Written interview

12/03/2022

Notes: Each code consists of the interviewee’s first and last names combined with a random number. PU

stands for production union; PF means production facility; and ENG and AZE are the English and

Azerbaijani languages, respectively.

The qualitative analysis followed a similar data generation procedure to that of

Montrone et al. (2021). Initially, only a small sample of industry experts was identified

through the reference procedure. Following the initial set of interviews, the initial

sample size was iteratively expanded through the snowballing procedure (Gardner

2009; O’Reilly–Parker 2013). In addition, industry experts from leading chemical plants

and factories were sought on professional networking platforms (e.g., LinkedIn). When

129

the thematic saturation reached a minimum, the search for experts was stopped

(Montrone et al. 2021; O’Reilly–Parker 2013). However, it should be noted that the

number of experts in the field of Azerbaijan’s chemical industry is extremely low;

therefore, considering time and location constraints, 10 was deemed the optimal number

of interviewees for the qualitative analysis.

A final sample of 16 experts was analyzed separately in two different groups,

namely the industry expert group and the economist group. Two industry experts (AB6

and EA1) were the directors of former production plants in the chlorine and soda

subsectors. One industry expert (MH1) was a former synthetic rubber production

engineer who also worked for SOCAR and Azerikimya PU. The remaining industry

experts were currently working in the industry, mainly in state-owned enterprises or

public–private partnerships (TG1, NN7, MH4, etc.). Only one respondent was from the

private sector of the chemical industry (OA9). The experts’ experience ranged from 1.5

years to 40 years. The qualitative data of 9 of the 10 interviewees came from the

recorded video and audio interviews, which were transcribed. The data from one

interviewee came from the interview notes, which were used to reconstruct the

interview. Most of the virtual face-to-face interviews were conducted using

communication platforms such as Zoom, WhatsApp, or Skype. However, one

interviewee (ISH8) shared his ideas through voice messages on WhatsApp. Azerbaijani

was the main language of the interviews, but two experts in the field were interviewed

in English (TG1 and MH4), as was a written interview with one economist (ERS8). All

non-English recordings were translated into English.

4.3.4. Methodology of the quantitative analysis

First, the chemical industry subsectors were analyzed with the pool of explanatory

variables using stepwise/OLS regression to create an initial baseline model with two

regressors. The sample size was small and based on annual data; therefore, for

feasibility, a stepwise regression algorithm was ordered to select only the two most

important regressors for each chemical industry subsector. The econometric formulation

of the first model is presented as follows:

      , (13)

where y is the output of the specific subsector of the chemical industry i in natural value

(tons or thousands of tons) at time t;  is the intercept coefficient;  and  are the

130

coefficients of the selected explanatory variables from the vector of Z at time t; and is

the error term of the model i at time t.

Second, FMOLS regression was applied to the baseline model (Model 2), which

provided an optimal estimate of the cointegrating regressors. FMOLS modifies the

stepwise/OLS regression methods to unfold the serial correlation and endogeneity

effects in the regressors resulting from the cointegration relationship (Mehmood–Shehid

2014). Third, the CCR technique of Park (1992) was applied to the baseline model to

check for robustness (Model 3). CCR has several advantages over OLS and FMOLS,

namely that it not only yields consistent constants and error terms but also an

asymptotically efficient and unbiased median (Ogaki et al. 1996). Moreover, the

obtained asymptotic distributions of the coefficients are free of nuisance parameters and

normally conditioned (Ogaki et al. 1996). Ogaki et al. (1996) claimed that this property

is an integral part of hypothesis testing and allows conventional interpretations of

standard errors. Most critically, CCR is better suited to annual data and a small sample

size, thus providing a method for testing robustness (Mehmood–Shehid 2014).

However, as Nazemzadeh et al. (2015) argued, FMOLS and CCR rely on the very

strong assumption that the cross-sectional error terms are independent. However, since

this analysis was based on only one country, this concern was not an issue with the

time-series data.

In Model 4 of the quantitative examination, the REER was included as an

“essential variable” variable in the stepwise algorithm. A re-estimation of the baseline

models was performed to determine how the initial model changed. A formal

representation of Model 4 is presented as follows, where the variables are the same as in

Formula 1:

         (14)

To identify the long-run relationship, robust least squares (RLS) was applied to

the level data with outliers. Essentially, two models were created using RLS. One was

the long-run estimate of the baseline model specified with stepwise/OLS (Models 1–3),

and the second RLS estimation included the adjusted model with the REER specified as

the “essential variable” regressor (Model 4). The last strategy allowed the impact of DD

on the specific lagging subsectors of Azerbaijan’s economy to be examined more

closely.

Tables A5.1, A5.2, and A5.3 in the Appendix present the unit root tests,

correlation, and Johansen cointegration analysis, respectively. According to the results,

131

the variables were nonstationary in their level form. Therefore, their first difference was

used for the regression estimates. Pearson’s R correlation coefficients revealed a strong

negative correlation between the chemical industry subsectors and DD-related variables.

Finally, long-run models could be built for the analysis because the variables of interest

displayed a cointegration relationship.

4.3.5. Methodology of the qualitative analysis

The methodology for the qualitative analysis consisted of semistructured and structured

expert interviews. A total of 14 questions were asked to the industry experts, which

were divided into five categories. The first category was aimed at understanding the

integration of Soviet-era technologies into Azerbaijan’s post-Soviet production

capacity; the second category concerned market structure, labor supply, domestic and

foreign investment, and competitiveness; the third category concerned oil prices and the

impact of the oil industry on chemical subsectors; the fourth category covered specific

subsectors of the chemical industry, with the main objective of understanding why

certain subsectors have collapsed while others have developed; lastly, the fifth category

collected data on the challenges, problems, and sustainable production potential in the

chemical industry.

103

The interviews with the economists was based on both structured and

semistructured interview techniques, as some economists agreed to be interviewed

through online communication tools, while others only allowed written interviews.

According to Montrone et al. (2021), the results of qualitative research based on expert

interviews may contain subjective elements as they depend on the authors’ own

judgment. To strive for maximum objectivity and reproducibility, only experts who

would be able to answer the questions completely and objectively were selected. The

written interviews were structured, while the video interviews were semistructured,

which allowed for interactive discussions.

The questions posed and the theoretical framework for the qualitative analysis

referred to DD theory and the process of de-industrialization provided by Corden and

Neary (1982) and Corden (1984). The chemical industry was considered a non-oil

manufacturing industry in this study, although it is closely related to the oil and gas

sector.

103

For the complete list of the interview questions, see Table A5.4 in the Appendix section.

132

The most critical part of the qualitative analysis was the coding. Linneberg and

Korsgaard (2019) mentioned that a code allows large amounts of qualitative data, such

as paragraphs, sentences, or words, to be broken down into small pieces to prepare the

data for analysis. A code is “a word or short phrase that symbolically assigns a

summative, salient, essence-capturing, and/or evocative attribute for a portion of

language-based or visual data” (Saldaña 2015: 3, as cited in Linneberg–Korsgaard

2019). On average, 44.6% of all text data from the expert interviews were coded.

Coding was both inductive and deductive, meaning that some codes were

designed before the coding process began while some emerged during the reading and

coding of the interview data. The software package used for the coding and analysis

was Quirkos version 2.4.1. Quirkos allows users to work interactively with data by

iteratively coding and recoding research-related themes. To establish coherence, each

new transcript was reviewed and compared with those already classified. To present the

results clearly, ideas expressed only once were not coded, while statements that

occurred repeatedly in the interview transcripts were coded. This is a standard and

widely used approach to coding (see Read et al. 2020, who used the same coding

procedure in the case of a health care study). The codes and their groupings are

presented in Figures A5.8 and A5.9 in the Appendix.

104

The analysis of the economists involved adopting a slightly different coding and grouping strategy

because the economists also offered their opinions about the non-oil industrialization policy of the

Azerbaijan government. The codes and their groupings can be seen in Figure 5.2A.

133

CHAPTER 5

RESULTS

This section reports the results from examining the collected data through quantitative

and qualitative methods. Section 5.1 reports on the analysis of NRC in the Azerbaijani

economy. Section 5.2 reports on the economic explanation of NRC, namely DD as a

precondition for de-industrialization. Finally, 5.3 analyzes a specific sector of the non-

oil industry, namely the chemical industry. Each section begins with a brief description

of the general analytical process that should be expected from that section and

concludes with a summary subsection.

5.1. The Analysis of Natural Resource Curse Theory in the Azerbaijan’s

Economy

The focus of this section is the NRC phenomenon in Azerbaijan. This was analyzed by

figure analysis, t-test, PCA, regression estimates, and interpretations of WVS.

5.1.1. Figure analysis

Figure 5.1 indicates that indices such as control of corruption, government

effectiveness, and rule of law experienced either a downward trend or a slowdown as

soon as the oil boom started in 2005. However, political stability dramatically improved

starting from 2006 but fell between 2011 and 2013. Interestingly, political stability

values during the post-boom period were lower than in the first half of the oil boom

period. Next, regulatory quality started to decline in 2009 but recovered after 2012.

Among the selected institutional variables, voice and accountability display a strong

negative trend starting in 2000. Lastly, it seems that there were positive developments

in the rule of law index in 2006 and a recovery after 2012. All of these observations

pointed to the relevance of the adverse effects of the oil boom on Azerbaijan’s

economy. This led this study to systematically investigate oil-related variables in

connection with institutional quality.

134

Figure 5.1: Worldwide governance indicators for Azerbaijan, 1996–2020.

Source: Author’s own calculations based on World Bank Worldwide Governance Indicators.

Notes: Red denotes the oil boom period between 2005 and 2014.

Figure 5.2: Distribution of year-over-year average growth rates for institutional quality,

based on the development phases of Azerbaijan’s economy (index values).

Source: Author’s own calculations based on World Bank Worldwide Governance Indicators.

The year-over-year growth rates illustrated in Figure 5.2 indicate that of the six

institutional variables, four were associated with lower development during the oil

boom period, which is 50% more compared with the results in Figure 5.1. Specifically,

the rule of law, control of corruption, regulatory quality, and the voice and

accountability indices displayed a lower average growth rate compared with the catch-

up period of Azerbaijan’s economy. During the post-boom period, only one indicator—

-12,0

-7,0

-2,0

3,0

8,0

13,0

18,0

1996-2004 2005-2014 2015-2019

135

the political stability index—had a severe deterioration. Further analyses of the

collected data uncovered more information about institutional quality in Azerbaijan’s

economy.

5.1.2. Results of one-sample t-test and international comparison

According to the results of a one-sample t-test, Azerbaijan’s mean institutional quality

was lower than those of oil-poor Armenia, Georgia, and the Baltic region during the oil

boom years (see Figure 5.3). Overall, the largest difference was with the Baltic region;

with Georgia the difference was large; while with Armenia it was the smallest. Only in

terms of the political stability index did Azerbaijan perform better than Georgia with a

mean difference of 0.12. The following subsections present the results of the PCA and

OLS regression and discuss them in detail.

Figure 5.3: Comparison of institutional quality of Azerbaijan with non-resource post-

Soviet countries during the oil boom period (2005–2014).

Source: Author’s own calculations based on the collected data.

Notes: (1) The values are mean differences obtained from a one-sample t-test; (2) all mean differences are

statistically significant at the 5% level, excluding the mean difference with Georgia in the case of political

stability.

5.1.3. Results of PCA

PCA allows researchers to reduce large data sets into more manageable principal

components that account for the most variation in the variables. Before the PCA, the

relevance of the data set for PCA had to be analyzed, for which the Kaiser–Meyer–

Olkin (KMO) measure of sampling adequacy and Bartlett’s test of sphericity were

applied. To produce optimal principal components, the data set was analyzed in its

-0,59 -0,53 -0,68 -0,4 -0,47 -0,59

-1,22 -0,94 -0,83 -0,61 -1,24

0,12

-2,23

-1,12 -1,57

-1,72

-1,67

-1,24

-4,50

-4,00

-3,50

-3,00

-2,50

-2,00

-1,50

-1,00

-0,50

0,00

0,50

Voice and

Accountability Government

Effectiveness Regulatory

Quality Rule of Law Control of

Corruption Political stabiliy

Armenia Georgia The Baltic region

136

original form, and then irrelevant variables were dropped (see Jaba et al. 2009 for

similar PCA adjustments). If KMO values are higher than 0.300, then PCA is

recommended (Kaiser 1974). As presented in Table 5.1, the KMO value was 0.772 in

the first analysis phase; moreover, Bartlett’s test of sphericity revealed high

significance, suggesting that at least one correlation was significant among the

variables. In the second phase of the analysis, the KMO value dropped to 0.624, but it

was still higher than the expected threshold values and still highly significant according

to Bartlett’s test of sphericity.

Table 5.1: Kaiser–Meyer–Olkin (KMO) values and Bartlett’s test results.

1st phase

KMO measure of sampling adequacy

0.772

Bartlett’s test of sphericity

Approx. chi-square

303.784

df

55

Sig.

0.000

2nd phase

KMO measure of sampling adequacy

0.624

Bartlett’s test of sphericity

Approx. chi-square

142.479

df

21

Sig.

0.000

Source: Author’s own calculations based on the collected data.

The applicability of PCA is heavily dependent on communalities (i.e., common

features). In PCA, a variable’s communality value reveals how much of the variation is

explained by the extracted component. A number larger than 0.35 is suitable for PCA

analysis to reach a significance level of 0.05 and a level of power of 80% (Tsiouni et al.

2021). The greater the communality value, the more it explains the variance of the

original variable of interest. The extraction was high in variables such as control of

corruption, rule of law, and government effectiveness indices (see Table 5.2.). Oil rents

and the oil boom had values of 0.764 and 0.766, respectively. EDI and the government

integrity index had the lowest extraction values, but they still exceeded the level of

0.600.

137

Table 5.2: Communalities of the variables related to institutional quality and the oil

sector in Azerbaijan’s economy.

Communalities

Initial

Extraction

COC

1

0.920

ROL

1

0.944

GOVEFF

1

0.927

GOVINT

1

0.634

OIL_RENTS

1

0.764

EDI

1

0.660

OIL_BOOM

1

0.766

Source: Author’s own calculations based on the collected data.

Notes: Extraction method = principal component analysis.

The first component accounted for 47.7% of the variation based on rotation

sums of squared loadings. The second component individually accounted for 32.6% but

cumulatively 80.2% of the variation in the data set. Although the main variables were

reduced to two principal components, the fact that these numbers are high indicates that

enough information was stored (see Table A5.1, Appendix section).

Figure 5.4: Scree plot of the variables related to institutional quality and the oil sector in

Azerbaijan’s economy.

Source: Author’s own calculations based on the collected data.

Next, the scree plot in Figure 5.4 indicates that the optimal number of

components out of the original variables is 2 because the eigenvalues drop below 1 if

the number of the components is higher than 2.

Table 5.3 presents the main results of the PCA, including the component matrix

and rotated component matrix. From both matrices, it became clear that the first

138

component covers the variation among variables such as control of corruption, rule of

law, government effectiveness, and government integrity, as they loaded high and

positively on it. Similarly, the second component was the most optimal subset of the oil-

related variables, such as oil rents, EDI, and oil boom. Therefore, the first component

should be called “institutional quality” and the second component should be called “oil

factor.” Visual representations of the loadings are depicted in Figure 5.5.

Table 5.3: Component matrices of the principal component analysis (PCA) related to

institutional quality and the oil sector in Azerbaijan’s economy.

Component Matrixa

Rotated Component Matrixb

Component

1

2

1

2

COC

0.929

0.24

COC

0.959

0.04

ROL

0.913

0.331

ROL

0.963

0.132

GOVEFF

0.959

0.088

GOVEFF

0.956

−0.115

GOVINT

0.792

−0.08

5

GOVINT

0.756

−0.249

OIL_RENTS

−0.245

0.839

OIL_RENTS

−0.063

0.872

EDI

−0.201

0.787

EDI

−0.031

0.812

OIL_BOOM

−0.211

0.849

OIL_BOOM

−0.028

0.875

Extraction method: PCA

Extraction method: PCA. Rotation

method: Varimax with Kaiser

normalization.

a – two components extracted.

b – rotation converged in three iterations.

Source: Author’s own calculations based on the collected data.

Figure. 5.5: Component plot in rotated space of institutional quality and the oil sector in

Azerbaijan’s economy.

Source: Author’s own calculations based on the collected data.

139

5.1.4. Results of the DOLS and OLS analyses

Before performing a regression analysis of the extracted components, the stationarity of

the variables of interest was checked. Table A5.2 (Appendix section) presents the

augmented Dickey–Fuller (ADF) unit root test results of the principal components. All

variables became stationary at their first difference; therefore, the DOLS models also

used their first difference form.

The DOLS model of the principal components with one lead and one lag

identified a statistically significant and negative impact of the oil factor on institutional

quality in Azerbaijan (see Table 5.4). The sign of the coefficient related to the oil factor

was always negative in the DOLS model and the intercept was positive and statistically

significant.

Table 5.4: Dynamic ordinary least squares (DOLS) results of the oil factor and

institutional quality in Azerbaijan’s economy.

(1)

(2)

(3)

(4)

(5)

C

0.13**

(2.18)

0.11**

(1.75)

0.15**

(2.86)

0.15**

(2.62)

0.15**

(2.35)

Oil factor

−0.23

(−1.68)

−0.30

(−1.68)

−0.42**

(−2.48)

−0.28

(−1.30)

−0.30

(−1.51)

R-squared

0.11

0.14

0.24

0.30

S.E. of regression

0.32

0.31

0.26

0.27

0.29

Long-run variance

0.08

0.05

0.06

0.08

Jarque–Bera

1.04

[0.595]

0.55

[0.759]

0.34

[0.843]

0.51

[0.777]

0.55

[0.757]

Wald test – F-stat.

3.55**

2.63*

6.35***

4.01**

3.30*

Source: Author’s own calculations based on the collected data.

Notes: Model 1: without lags and leads; model 2: one lag, zero leads; model 3: one lag, one lead; model

4: two lags, one lead; model 5: two lags, two leads.

The next part of the regression analysis included some individual indicators that

were definitely related to the NRC doctrine (see Table 5.5). They were out-of-pocket

expenditures on healthcare (OP_EXP_HC), total government expenditure on education

(TGEE), and human rights (HUM_RIGHTS). These variables were regressed against

the following oil-related variables: oil rents, share of oil exports in GDP, EDI, economic

shocks, oil FDI, mining industry’s share of overall industrial production, and share of

SOFAZ in the state budget.

The human rights scores provided unambiguous results regarding the NRC as

EDI, oil FDI, mining industry’s share of overall industrial production, and share of

SOFAZ in the state budget exhibited negative and statistically significant coefficients.

Next, oil rents and EDI negatively and statistically significantly influenced TGEE.

140

However, the share of oil exports in GDP and economic shocks positively impacted

TGEE. Lastly, out-of-pocket expenses on health care tended to rise when EDI rose and

economic shocks occurred, but oil rents and oil exports as a share of GDP negatively

affected out-of-pocket expenses on health care.

All of the models were statistically significant according to significant F

statistics, moderate R-squared values, and no multicollinearity issues as the variance

inflation factors (VIFs) were less than 10.0. Moreover, CUSUM and CUSUMSQ tests

indicated that the models were stable. Furthermore, the models were functionally

correct, without any serial correlation and heteroscedasticity problems. Lastly, the Wald

test indicated that all coefficients differed from zero in a statistically significant manner.

Table 5.5: OLS results of individual NRC-related indicators against oil-related

variables.

Dep. Var.

OP_EXP_HC

TGEE

HUM_RIGHTS

C

13.34**

(2.88)

−0.17***

(−3.25)

0.01

(0.15)

Oil Rents

−2.84***

(−3.02)

−0.01

(−0.74)

Oil Exp/GDP

−108.24**

(−1.89)

1.15*

(1.77)

EDI

5.11*

(1.84)

−0.19**

(−2.78)

−0.01**

(2.24)

Econ. Shocks

33.38**

(2.82)

0.43***

(3.21)

Oil FDI

−3.44***

(−3.84)

Mining Industry

−0.01**

(−2.76)

SOFAZ’s Share

−0.01*

(−1.94)

R-squared

0.76

0.64

0.58

Adj. R-squared

0.71

0.54

0.46

F-stat.

12.10

6.23

4.76

F-stat. prob.

0.00

0.01

Variance inflation

factors

All <10.00

CUSUM

Within 5% sig.

CUSUMSQ

Within 5% sig.

Ramsey reset test

Functional spec. is true

Wald test (F-stat.)

16.04***

7.40***

6.60***

Wald test (2)

80.22***

37.92***

33.02***

JBN test

0.48

0.11

1.80

JBN test Prob. value

0.79

0.94

0.41

Serial corr. (F-stat.)

0.17

1.77

0.76

Serial corr. (Obs*R2)

0.53

4.33

2.13

Heteros. (F-stat.)

1.82

2.02

1.76

Heteros. (Obs*R2)

6.48

6.96

6.37

Source: Author’s own calculations based on the collected data.

Notes: (1) Dep. var = dependent variable; (2) OP_EXP_HC = out-of-pocket expenditure on health care;

(3) TGEE = total government expenditure on education; (4) HUM_RIGHTS = human rights; (5) *, **,

141

and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively; (6) the figures were

rounded to two decimal places for compactness; (7) values inside parentheses indicate standard errors and

those inside brackets are t-statistics.

5.1.5. Results of the WVS analysis

The NRC doctrine emphasizes the importance of political, institutional, and governance

factors in resource-rich countries when it comes to managing their resource wealth. The

increasing body of literature—including this study—is focused on macroeconomic

indicators that reflect institutional quality, political and governance trends, and human

capital and education. However, although it is difficult to diagnose the exact cause–

effect relationship between institutional failures and resource wealth, their reflection on

society among citizens’ perception of NRC-related channels may offer insight into the

presence of NRC in a qualitative manner. In other words, if members of society are not

interested in managing the politics of the country, democracy and governance quality

will be endangered and the opportunistic behavior of politicians will be readily

supported through rent-seeking activities.

The levels of trust and confidence that society has in government transparency

determine the institutional capacity to manage resource wealth. If citizens have little

faith in political and governance institutions, resource wealth will not contribute to the

country’s economic development. These findings are documented in the NRC literature,

but they have been based on quantitative data. Therefore, qualitative data are worth

examining even if they are limited (e.g., by the number of respondents) and to some

extent prone to bias (e.g., bad answers to survey questions). An evaluation of WVS

results could provide insight into the relevance of NRC syndrome in Azerbaijan’s

economy and fill some gaps in the literature.

First, survey results related to the importance of and interest and confidence in

politics and political parties in Azerbaijan were examined, mainly between 1994 and

2020. Then, political actions such as voting, signing of petitions, and boycotting were

examined.

This study found that the importance of politics for Azerbaijani citizens

decreased from 10% to 4% between 1994 and 2020, as measured by the “very

important” category of the survey question that identified the importance of politics (see

Table 5.6). The number of respondents evaluating politics as “rather important”

decreased during the first and second halves of the oil boom period compared with the

recession period (27% between 1994 and 1998); however, this figure rose to 31%

142

between 2017 and 2020 from lows such as 24% and 18% between 2005 and 2014.

Notably, the number of citizens who were uninterested in politics increased during the

oil boom period. However, a decrease occurred in the number of people who viewed

politics as “not very important” or “not at all important” during the same period (from

43% between 2010 and 2014 to 35% in 2017 and 2020). Lastly, the percentage of

respondents who were uncertain about the importance of politics in their lives increased

from 2% during the oil boom period (2005–2014) to 3% between 2017 and 2020.

Overall, not many citizens regarded politics as important during the oil boom period;

most citizens actually viewed politics as unimportant.

Table 5.6: Importance of politics among Azerbaijani citizens (in %), Question wording:

For each of the following aspects, indicate how important it is in your life – Politics.

Would you say it is very important, rather important, not very important, or not

important at all?

1994–1998

2005–2009

2010–2014

2017–2020

Very important

10

19

8

4

Rather important

27

24

18

31

Not very important

42

30

43

35

Not at all important

21

26

31

26

Don’t know

1

2

0

3

(N)

2,002

1,505

1,002

1,817

Source: World Values Survey (WVS)

Notes: Due to technicalities in the rounding process for the data section of the

WVS webpage, the summary of each column exceeds 100% in some cases.

While the importance of politics may be high for many citizens, their interest in

it may be less certain. Political actions are required to engage with the government

apparatus because resource-rich countries are notorious for repressing the political will

of their citizens. Although interest in politics exhibited a modest increase of 1

percentage point between 2005 and 2009 compared with 1994–1998, the survey results

indicated that only 6% and 4% of respondents were “very interested” in politics in

2010–2014 and 2017–2020, respectively (see Table 5.7). Additionally, the percentage

of respondents who were “somewhat interested” rose to 37% during the first half of the

oil boom compared with the recession period. However, the smallest number of people

expressed an interest in politics during the second half of the oil boom period (2010–

2014). Conversely, most of the responses fell into the “not very interested” and “not at

all interested” categories. The percentage of respondents who were “not at all

interested” experienced the sharpest and steepest increase after 1994–1998, rising from

143

24% to 36% in 2017–2020. Moreover, uncertainty among survey participants declined

from 2% and 5% during the 1994–1998 and 2005–2009 periods, respectively, to 0%

and 1% during the 2010–2014 and 2017–2020 periods, respectively.

Table 5.7: Interest in politics among Azerbaijani citizens (in %), Question wording:

How interested would you say you are in politics?

1994–1998

2005–2009

2010–2014

2017–2020

Very interested

8

9

6

4

Somewhat interested

33

37

18

27

Not very interested

32

47

44

32

Not at all interested

24

1

32

36

Don’t know

2

5

0

1

(N)

2,002

1,505

1,002

1,817

Source: World Values Survey (WVS)

Note: Due to technicalities in the rounding process for the data section of the WVS

webpage, the summary of each column exceeds 100% in some cases.

The importance of and interest in politics would partially reflect the social

reality if the NRC is present in Azerbaijan. Actions such as signing petitions and

participating in boycotts demonstrate the real intentions of citizens when they are

dissatisfied with the government. Therefore, people who have already undertaken or

may undertake such actions can serve as a proxy indicator for the suppressed political

will of a society. Table 5.8 indicates that the categories of “have done” and “may do”

fluctuated from 1994 to 2020, corresponding to the low shares in the overall answers.

However, the percentage of respondents who said that they would never sign a petition

reached 91% between 2010 and 2014—the highest since the recession period. However,

this decreased to 63% during the post-boom period. There was also a dramatic increase

of 10 percentage points regarding uncertainty about signing a petition between 2017 and

2020 compared with the oil boom period. Thus, the oil boom period weakened the will

of citizens to politically act if necessary through tools such as petitions and

demonstrations.

Table 5.8: Intensity of political action in Azerbaijan, as measured by willingness to sign

a petition (in %), Question wording: Now I’d like you to look at this card. I’m going to

read out some different forms of political action that people can take, and I’d like you to

tell me, for each one, whether you have actually done any of these things, whether you

might do it or would never, under any circumstances, do it.

1994–1998

2005–2009

2010–2014

2017–2020

Have done

9

13

4

7

144

May do

15

31

6

19

Would never do

69

56

91

63

Don’t know

6

0

10

No answer

0

2

(N)

2.002

1.505

1.002

1.817

Source: World Values Survey (WVS)

Note: Due to technicalities in the rounding process for the data section of the

WVS webpage, the summary of each column exceeds 100% in some cases.

Similarly, people who participated in a boycott as a political action declined

from 19% between 2005 and 2009 to 0% between 2010 and 2014; there was a moderate

increase of 1 percentage point from 2017 to 2020 (see Table 5.9). Between 1994 and

2014, the percentage of respondents who said that they might participate in a boycott

exhibited a downward trend (from 12% to 5%); however, between 2017 and 2020, 14%

of respondents said they would agree to participate in a boycott. The percentage of

respondents who said that they would never participate in a boycott peaked between

2010 and 2014 (95%). Uncertainty about whether to participate in a boycott increased

during 2017–2020, with 2% and 9% of respondents choosing “no answer” and “don’t

know,” respectively. Like signing petitions, a decreased propensity to join a boycott

was also negatively associated with the oil boom period.

Table 5.9: Intensity of political action in Azerbaijan, as measured by willingness to

participate in a boycott (in %), Question wording: Now I’d like you to look at this card.

I’m going to read out some different forms of political action that people can take, and

I’d like you to tell me, for each one, whether you have actually done any of these things,

whether you might do it or would never, under any circumstances, do it.

1994–1998

2005–2009

2010–2014

2017–2020

Have done

2

19

0

1

May do

12

7

5

14

Would never do

80

72

95

74

No answer

0

1

0

2

Don’t know

6

1

0

9

(N)

2,002

1,505

1,002

1,817

Source: World Values Survey (WVS)

Note: Due to technicalities in the rounding process for the data section of the

WVS webpage, the summary of each column exceeds 100% in some cases.

Regarding confidence in political parties, the highest expression of confidence

on the survey was captured through the answer category of “a great deal” (see Table

5.10). In Azerbaijan, confidence in political parties sharply declined from 16% (1994–

145

1998) and 26% (2005–2009) to 10% and 7% during 2010–2014 and 2017–2020,

respectively. However, 26% of respondents indicated that they had confidence in the

government from 2005 to 2014. Similarly, those who had confidence in politicians (i.e.,

those who chose “quite a lot” as a response) rose from 31% during the recession period

(1994–1998) to 39% between 2005 and 2009; however, the figure fell by 7 and 15

percentage points during the second half of the oil boom and the post-boom period,

respectively. Meanwhile, the percentage of respondents who did not trust political

parties peaked at 36% between 2010 and 2014; this figure was 26% between 1994 and

1998 and 17% between 2005 and 2009. During the post-boom period, the number of

citizens who had little trust in political parties decreased to 28%. Similarly, the

percentage of respondents who completely distrusted political parties (i.e., those who

chose “none at all” in the survey) increased to 23% between 2010 and 2014; this

remained the same for 2017–2020. In addition, the number of people who were

uncertain or did not know how to feel toward political parties (i.e., the sum of the “don’t

know” and “no answer” categories) rose to 17% between 2017 and 2020. Thus, it can

be argued that, compared with the recession period and the first half of the oil boom,

citizens’ confidence in the government decreased during the second half of the oil

boom. In addition, a significant group of people remained uncertain about their

confidence in the government during the post-boom period.

Table 5.10: Confidence in political parties in Azerbaijan (in %), Question wording: I am

going to name a number of organizations. For each one, could you tell me how much

confidence you have in them: is it a great deal of confidence, quite a lot of confidence,

not very much confidence, or none at all? (Political parties)

1994–1998

2005–2009

2010–2014

2017–2020

A great deal

16

26

10

7

Quite a lot

31

39

32

24

Not very much

26

17

36

28

None at all

14

16

23

Don’t know

14

2

0

15

No answer

0

2

(N)

2,002

1,505

1,002

1,817

Source: World Values Survey (WVS)

Note: Due to technicalities in the rounding process for the data section of the

WVS webpage, the summary of each column exceeds 100% in some cases.

Voting remains one of the most critical indicators of the political involvement of

citizens. Therefore, voting data at both the local and national levels can provide

146

information about ongoing trends related to the actuality of NRC syndrome in

Azerbaijan. As seen in Table 5.11, the percentage of respondents who “always” voted at

the local level increased by 2 percentage points between 2017 and 2020 compared with

2010–2014. However, at the national level, a decrease occurred from 39% to 34% for

the same category. Respondents who “usually” voted comprised the largest group

between 2017 and 2020, accounting for 40% and 41% of the sample at the local and

national levels, respectively. However, it is interesting that 43% of respondents never

voted at the local level and 32% never voted at the national level between 2010 and

2014. Moreover, there was an overall increase of four and five percentage points

(measured by summing the categories “don’t know” and “no answer”) in the voting

behavior of citizens between 2017 and 2020 compared with 2010 and 2014. Although it

is difficult to argue that high levels of low voting behavior are directly related to the

NRC phenomenon, the percentage of people who voted during the post-boom period

increased compared with the oil boom period.

Table 5.11: Voting participation of Azerbaijani citizens in local and national elections

(in %), Question wording: Vote in elections: National level; Vote in elections: Local

level

Local level

National level

2010–2014

2017–2020

2010–2014

2017–2020

Always

30

32

39

34

Usually

28

40

29

41

Never

43

23

32

20

Not allowed to vote

0

Don’t know

0

1

0

3

No answer

0

3

0

2

(N)

1,002

1,817

1,002

1,817

Source: World Values Survey (WVS)

Note: Due to technicalities in the rounding process for the data section of the WVS webpage, the

summary of each column exceeds 100% in some cases.

Next, Azerbaijani citizens’ perception of the government reflects the political

mood, which might be relevant when seeking to observe signs of the NRC. Although

confidence in the government decreased from 37% to 22% between the late recession

and catch-up period and the first half of the oil boom (measured through the answer

category of “a great deal”), confidence grew to 47% during the second half of the oil

boom, followed by a decrease of 6 percentage points between 2017 and 2020 (see Table

5.12). Meanwhile, the percentage of respondents who answered “quite a lot” exhibited a

147

gradual drop during the oil boom period (from 49% in 1994–1998 to 36% and 33% in

2005–2009 and 2010–2014, respectively) but recovered to 49% during the post-boom

period. The responses “not very much” and “none at all” indicated low to no confidence

in the government. During the first half of the oil boom period, people’s lack of

confidence in the government increased from 7% to 40% (as measured by the sum of

the categories “not very much” and “none at all”) but decreased at the end of the oil

boom and post-boom periods. There was also a slight increase of 3 percentage points in

uncertainty among citizens between 2017 and 2020, as measured by the sum of the

categories “don’t know” and “no answer.”

Table 5.12: Level of confidence in the government among Azerbaijani citizens (in %),

Question wording: I am going to name a number of organizations. For each one, could

you tell me how much confidence you have in them: is it a great deal of confidence,

quite a lot of confidence, not very much confidence, or none at all? The government (in

your nation’s capital).

1994–1998

2005–2009

2010–2014

2017–2020

A great deal

37

22

47

41

Quite a lot

49

36

33

49

Not very much

5

22

10

6

None at all

2

18

10

1

Don’t know

7

2

0

2

No answer

0

1

(N)

2,002

1,505

1,002

1,817

Source: World Values Survey (WVS)

Note: Due to technicalities in the rounding process for the data section of the

WVS webpage, the summary of each column exceeds 100% in some cases.

Another part of the WVS examined the perception of democracy among citizens

(see Figure 5.5, panel a). Democracy is among the main set of values of people

worldwide and is an essential indicator of social and economic dynamics. The survey

found that 87.2% of respondents valued democracy in their lives. Moreover, 66.5% of

respondents between 2017 and 2020 viewed the government as the main provider of or

as responsible for individual well-being in Azerbaijan (see Figure 5.5, panel b).

As an organizing tool that enables workers to defend their rights, labor unions

are critical. In the survey, the highest confidence level in labor unions was represented

by the answer category of “a great deal.” The percentage of respondents who selected

this answer notably increased by 18 percentage points between 2005 and 2009 from 4%

148

between 1994 and 1998 (see Table 5.13). This increase was followed by a decrease to

9% and 10% from 2010 to 2014 and 2017 to 2020, respectively. Citizens increased their

confidence in labor unions from 1994 to 2014—first from 23% to 34%, and then to

36%. However, a decrease of four percentage points occurred between 2017 and 2020,

as measured by responses in the “quite a lot” category. The highest distrust in labor

unions (44%) occurred between 1994 and 1998. The percentage of respondents who

answered “not at all” peaked at 24% during the second half of the oil boom period, but

then decreased by 14 percentage points during the post-boom period. Interestingly, the

percentage of people who responded “don’t know” also dramatically increased from

lows of 0–1% during the oil boom period to 23% between 2017 and 2020. Thus, the

state of confidence in labor unions in Azerbaijan was ambiguous.

Figure 5.6: Democracy and government responsibility in Azerbaijan, 2017–2020.

a. How important is it for you to live in a

country that is governed democratically?

On this scale, where 1 means it is “not at

all important” and 10 means “absolutely

important,” what position would you

choose?

b. Government vs. individual

responsibility (1 = the government

should take more responsibility to ensure

that everyone is provided for or 2 =

people should take more responsibility to

provide for themselves).

Source: World Values Survey (WVS).

Furthermore, Azerbaijani citizens’ confidence in the justice system and courts

rose during the first half of the oil boom compared with the recession period (from 5%

to 27%, see Table 5.14). However, this upward trend was interrupted during the second

half of the oil boom (decreasing from 27% to 23% in 2005 and 2009 and to 10% in

2010 and 2014), as measured by the percentage of respondents who answered “a great

149

deal.” Most respondents (41%) answered “quite a lot” between 2005 and 2009; then,

this declined to 26% between 2010 and 2014 before rising again to 48% between 2017

and 2020. Although many respondents were not confident in the justice system and

courts during the late recession and early transition periods (42% did not have much

trust), 20–23% of respondents indicated slight distrust—as measured by the “not very

much” category—toward the justice system and courts in Azerbaijan between 2005 and

2020. The second half of the oil boom saw the highest percentage of respondents who

had little or no confidence in the justice system (19%), which declined by a further 13

percentage points from 2017 to 2020. A noticeable increase occurred in the uncertainty

level among citizens who were asked about their confidence in the justice system. The

categories of “don’t know” and “no answer” rose to 16% and 1%, respectively, between

2017 and 2020 from 0% in 2010 and 2014. Overall, the majority of respondents had

confidence in the justice system and courts, although the percentage with a doubtful

perception of the justice system is rising.

Table 5.13: Confidence in labor unions among Azerbaijani citizens (in %), Question

wording: I am going to name a number of organizations. For each one, could you tell

me how much confidence you have in them: is it a great deal of confidence, quite a lot

of confidence, not very much confidence, or none at all? (Labor unions)

1994–1998

2005–2009

2010–2014

2017–2020

A great deal

4

22

9

10

Quite a lot

23

34

36

32

Not very much

44

31

23

None at all

20

12

24

10

Don’t know

10

1

0

23

No answer

0

1

(N)

2,002

1,505

1,002

1,817

Source: World Values Survey (WVS).

Note: Due to technicalities in the rounding process for the data section of

the WVS webpage, the summary of each column exceeds 100% in some

cases.

5.1.6 Summary of the section

In this chapter, the typical signs of NRC syndrome in Azerbaijan’s economy were

examined through figure analysis, PCA, DOLS, OLS regressions, and an evaluation of

WVS results. The use of both quantitative and qualitative methods enabled an analysis

of the underlying government- and society-related dynamics of the NRC to relate it to

economic concepts such as DD. To this end, data related to institutions, governance,

and human capital in Azerbaijan were collected, mainly covering the period 1996–2019.

150

The figure and survey analyses allowed a visualization of the national economy and

society-wise patterns in institutional quality concerning the economic development

phases. Moreover, PCA and variable-specific modeling enabled this study to capture the

typical NRC signs to estimate them for the hypothesis testing.

Table 5.14: Confidence in the justice system and courts among Azerbaijani Citizens (in

%), Question wording: I am going to name a number of organizations. For each one,

could you tell me how much confidence you have in them: is it a great deal of

confidence, quite a lot of confidence, not very much confidence, or none at all? (Justice

system/courts)

1994–1998

2005–2009

2010–2014

2017–2020

A great deal

5

27

23

10

Quite a lot

39

41

36

48

Not very much

42

20

23

20

None at all

8

11

19

6

Don’t know

5

2

0

16

No answer

0

1

(N)

2,002

1,505

1,002

1,817

Source: World Values Survey (WVS).

Note: Due to technicalities in the rounding process for the data section of the

WVS webpage, the summary of each column exceeds 100% in some cases.

Moreover, a figure analysis of selected institutional variables related to

Azerbaijan’s economy revealed negative trends and slowdowns in institutional quality,

as measured by variables such as control of corruption, government effectiveness, voice

and accountability, and the rule of law as soon as the oil boom period started. In

addition, year-over-year and periodic averages of the growth rates revealed a systematic

decline in institutional quality during the oil boom years. For example, the period 2005–

2014 had lower year-over-year growth rates for the rule of law, control of corruption,

regulatory quality, and voice and accountability indices compared with the recovery

phase. Azerbaijan’s average institutional quality is lower than that of its resource-poor

post-Soviet counterparts, such as Georgia, Armenia, and the Baltic region. Furthermore,

Azerbaijan is the most resource-dependent FSU country with low levels of institutional

quality. This descriptive exploration allowed this study to go further and apply

empirical methods to test for the presence of noneconomic sides of NRC syndrome.

Therefore, the PCA indicated that institutional quality and oil-related variables

can be explained by a few key variables, principal components, and a DOLS analysis.

The latter demonstrated that the oil sector negatively affected the institutional quality in

151

Azerbaijan between 1996 and 2019. Variables such as out-of-pocket expenses on health

care and total government expenditures on education and human rights exhibited

statistically significant and negative associations with oil-related variables, and they

captured the negative nexus between human capital channels of the NRC and the oil

sector.

Lastly, substantive and pervasive institutional failures in Azerbaijan that

impeded the efficient management of oil wealth were traced with the help of data from

the large-scale, cross-national WVS. Results related to the main survey questions about

politics and government were compared to demonstrate the societal dimensions of

Azerbaijan’s institutional climate during the last years of the oil boom and post-boom

periods. The main results illustrated the developmental origins of the NRC doctrine in

Azerbaijan’s economy. They indicated that, despite trusting the government and

believing that officials are responsible for maintaining citizens’ individual welfare and

well-being, respondents were not interested in changing the government through

political parties supported by citizens’ voting behavior and engaging in political actions

such as petitions and boycotts. In other words, although respondents mentioned

democracy as the main value for society, their interest in elections and voting remained

low during the second half of the oil boom period. Bidirectional expectations from the

government, a lack of participation, and an unwillingness to create change created a

fertile ground for the NRC to develop. This affects the lives of Azerbaijani citizens in

the form of low institutional, governance, human capital, and education quality.

Although the WVS proved to be useful to this study, Braun et al. (2014) highlighted a

main limitation in terms of the long-term validity of cross-national surveys, namely that

it is threatened by the intercultural nature of the surveys. This means that over time, the

same questions can be interpreted differently by different cultures following certain

social changes. Moreover, due to some methodological artifacts, the results of the same

survey questions may be noticeably different from time to time.

5.2. Analysis of DD

This chapter aims to examine DD signs and effects on Azerbaijan's economy by

applying various empirical techniques that capture the impact of the oil boom. Although

the oil boom allowed Azerbaijan to escape extreme poverty, complaints regarding the

negative effects of lopsided industrial production were voiced by Arvanitopoulos

(1998), Hoffman (1999), and Laurila (1999) as early as 1998.

152

Critical challenges considered typical symptoms of DD are as follows:

appreciation of the REER, lower competitiveness in non-oil sectors (including

agriculture), low production, low employment in manufacturing, and high domestic

price levels. Although several studies have attempted to establish signs of DD in

Azerbaijan’s economy, there is no overwhelming empirical evidence for their existence.

Therefore, this study examined the signs of DD using various linear models, such as

vector autoregressive (VAR), Bayesian VAR (BVAR), ordinary least squares (OLS),

and fully modified OLS (FMOLS). The resource movement and spending effects of the

theory were addressed separately. However, before examining the specific effects of

DD on Azerbaijan’s economy, the sectoral impact (in three sector models—booming,

lagging, and non-tradeable sectors) of DD-related indicators was analyzed. Specifically,

an Azerbaijan-specific literature review (Section 4.1) was conducted in three directions,

namely on early studies that expressed initial concerns about DD in the medium and

long term; direct studies on DD that have attempted to prove the existence of the

phenomenon; and indirect studies that have focused on the components of DD rather

than the symptoms. Section 4.2 discusses the data and methodology of the chapter.

Section 4.3 then presents the results of the study. Lastly, Section 4.4 draws some

pertinent conclusions.

5.2.1. Appreciation of REER

Unrestricted standard VAR was used to discover how the REER changes in the short

term when the price of oil changes. However, prior to the analysis, an ADF test was

conducted to determine whether the variables of interest were stationary (see Table

A5.3 in the Appendix). According to the results, the REER and oil prices are not

stationary in their level form, but they become stationary in their first difference.

Therefore, the VAR model of REER appreciation was run using the first difference of

both the REER and oil prices.

Then, the optimum lag order was selected based on the AIC, which was 12 (see

Table A5.4 in the Appendix). The LR and FPE criteria also confirmed this. However,

one or two lags according to the SC and HQ criteria were not empirically meaningful

for the large data set based on monthly data. Finally, a 12-month lag was chosen and

included in the model.

Figure 5.7 depicts the impulse response functions (IRFs). The REER positively

responds to the oil price shocks after the first and second months (see Figure 5.7, panel

153

b). In the 13th month, the REER exhibits a spike in its positive response. Moreover, the

cumulative responses of the REER to oil price shocks also exhibit a significant positive

trend (see Figure 5.7, panel d). The REER’s response to its own shocks exhibits

fluctuations (see panels a and c in Figure 5.7).

Figure 5.7: Impulse response functions of the VAR model, 1995M01–2020M12.

a

b

0

1

2

2 4 6 8 10 12 14 16 18 20 22 24

Response of D(R EER) to D(REER )

0

1

2

2 4 6 8 10 12 14 16 18 20 22 24

Response of D(R EER) to D(OIL_P)

c

d

0

1

2

3

4

5

2 4 6 8 10 12 14 16 18 20 22 24

Accumulated R esponse of D(R EER) to D(R EER)

0

1

2

3

4

5

2 4 6 8 10 12 14 16 18 20 22 24

Accumulated R esponse of D(REER ) to D(OIL_P)

Source: The author’s own calculations based on the collected data.

Notes: Response to Cholesky one S.D. Innovations ± 2 S.E; D(REER) represents the real effective

exchange rate and D(OIL_P) denotes the oil prices.

To verify the validity of the aforementioned findings, the results of VAR needed

to be subjected to a series of stability tests. For example, using the roots of a

characteristic polynomial, the dynamically stable model could be tested for the

requirement of model stability. The VAR model would be stable if all of its eigenvalues

fell within the unit circle. The inverse roots of the autoregressive (AR) characteristic

polynomial are illustrated in Figure A5.1 (Appendix). All eigenvalues lie inside the

circle.

Other diagnostics of the VAR model included tests of residual autocorrelation

and residual serial correlation LM. There should be no significant deviations of the

154

residual autocorrelation values from two standard error bounds. The test results

indicated that all residuals up to 12 lags were within two standard error bounds (see

Figure A5.2 in the Appendix). The tests of the serial correlation of the residuals LM

revealed that the estimated VAR model was free from serial correlation up to 12 lags

(see Table A5.4 in the Appendix).

Table 5.15 lists the forecast error (SE) as a percentage of the error from the

VAR error, while each column indicates how much of this error was explained by each

variable. In this case, the variance decomposition technique was used because it

provided the necessary information about the relative importance of each random shock

(Ayadi 2005). The oil prices could explain up to 17.99% of the forecast error for the

next 24 months, while the REER could explain 82.01%.

Table 5.15: The variance decomposition of DREER.

Period

S.E.

D(REER)

D(OIL_P)

1

2.46

100

0

2

2.57

96.49

3.51

6

2.73

91.98

8.02

10

2.85

85.27

14.73

14

2.97

82.00

18.00

18

2.99

82.21

17.79

22

3.00

82.13

17.87

24

3.01

82.01

17.99

Source: The author’s own calculations based on the collected data.

Notes: Here, the figures were rounded to the second decimal place for compactness.

Furthermore, the individual and joint significance of oil prices in the VAR

Granger causality test, as well as a unidirectional causal effect from oil prices to REER,

were consistent with the IRFs (see Table 5.16).

Table 5.16: VAR Granger causality/block exogeneity Wald tests and pairwise Granger

causality tests between REER and oil prices, 1995M01–2020M12.

VAR Granger Causality/Block Exogeneity Wald Tests

Excluded

Chi-sq.

df

Prob.

D(OIL_P)

64.16

12

0.000

All

64.16

12

0.000

Pairwise Granger Causality Tests

Null Hypothesis:

Obs.

F-Statistic

Prob.

OIL_P does not Granger Cause REER

300

4.02

0.001

REER does not Granger Cause OIL_P

0.46

0.939

Source: The author’s own calculations based on the collected data.

Notes: Twelve lags were included in the pairwise Granger causality tests; figures were rounded to the

second decimal place for compactness (excluding the probability values).

155

The long-run relationship between the REER and oil prices was estimated by

applying the FMOLS and canonical cointegration regression (CCR) techniques (see

Table 5.17). The long-run relationship revealed that a 1% increase in oil prices led to a

15% appreciation of the Azerbaijani REER. Exactly the same result was obtained using

the CCR technique.

Table 5.17: FMOLS and CCR results of the analysis of oil prices on the REER,

1995M01–2020M12.

Fully Modified OLS (FMOLS)

Variable

Coefficient

Std. Error

t-Statistic

Prob.

LOG(OIL_P)

0.15

0.03

4.82

0.000

C

4.06

0.12

33.38

0.000

R-squared

0.25

Mean dependent var.

4.64

Adjusted R-squared

0.25

S.D. dependent var.

0.19

S.E. of regression

0.16

Sum squared resid.

8.29

Long-run variance

0.13

Canonical Cointegration Regression (CCR)

LOG(OIL_P)

0.15

0.03

4.86

0.000

C

4.06

0.12

33.65

0.000

R-squared

0.25

Mean dependent var

S.D. dependent var

4.64

Adjusted R-squared

0.25

0.19

Source: The author’s own calculations based on the collected data.

Notes: The figures were rounded to the second decimal place for compactness (excluding the probability

values).

5.2.2. Sectoral effects of EDI, REER, NEER, oil prices, and oil rents

According to the results of the Johansen cointegration test (reported in Table A5.5 in

the Appendix), all variables of interest displayed cointegration relationships. This

allowed the estimation of long-run linear models to capture the sectoral effects of the

REER, NEER, oil prices, and oil rents.

While the results of the models presented here reveal said sectoral effects of the

EDI, REER, NEER, oil prices, and oil rents, they do not necessarily reflect the precise

effects of DD, namely resource movement and spending effects. The sectoral effects of

the DD-related variables were previously studied in a more focused approach on

resource movement and spending by Mironov and Petronevich (2015).

Table 5.18 indicates that the EDI positively affected output in the booming

sectors (Eqs. 1 and 2), while the manufacturing output and value-added were negatively

affected by the EDI (Eqs. 3–6).

Moreover, the association between the NEER and SB was negative and

statistically significant. The same direction of the relationship between the NEER and

manufacturing output as well as value-added was positive. However, since the REER

156

was the second control variable for the exchange rate, Eqs. 4 and 6 demonstrate that

manufacturing output and value-added have significantly negative links with the REER.

Table 5.18: Sectoral implications of REER, NEER, oil prices, and oil rents for 1990–

2019.

Equation Name:

1

2

3

4

5

6

Dependent

Variable

SB_OUT_S

H

SB_OUT_SH

SM_OUT_SH

SM_VA_SH

EDI

4.31***

3.07***

−2.71***

−2.22***

−0.44

−0.25

(1.45)

(0.61)

(0.87)

(0.65)

(−0.26)

(−0.2)

[2.98]

[5.00]

[−3.09]

[−3.41]

[−1.69]

[−1.27]

NEER_66

−0.19***

0.09**

0.03***

(0.06)

(0.04)

(0.01)

[−3.01]

[2.36]

[2.88]

OIL_BOOM

10.93**

4.20*

−6.69**

−2.92

−0.92

0.20

(4.68)

(−2.1)

(2.83)

(−2.23)

(−0.85)

(−0.67)

[2.34]

[2.00]

[−2.36]

[−1.31]

[−1.09]

[0.30]

OIL_PRICES

−0.03

−0.19***

0.02

0.11**

−0.01

0.02*

(−0.09)

(0.04)

(−0.06)

(0.04)

(−0.02)

− 0.01)

[−0.32]

[−5.20]

[0.40]

[2.80]

[−0.59]

[1.76]

OIL_RENTS

0.40*

0.72***

−0.10

−0.28***

−0.02

−0.08***

(−0.21)

(0.08)

(−0.13)

(0.09)

(−0.04)

(0.03)

[1.89]

[8.53]

[−0.76]

[−3.06]

[−0.49]

[−2.91]

C

54.29***

18.35***

20.36***

38.88***

8.68***

14.92***

(7.88)

(2.14)

(4.77)

(2.28)

(1.43)

(0.69)

[6.89]

[8.56]

[4.27]

[17.07]

[6.08]

[21.66]

@TREND

0.95***

0.62***

−0.33**

−0.12

−0.21***

−0.16***

(0.25)

(0.11)

(0.15)

(−0.12)

(0.05)

(0.04)

[3.78]

[5.45]

[−2.18]

[−0.97]

[−4.68]

[−4.32]

REER_66

0.29

−0.16***

−0.05***

(0.03)

(0.01)

[10.37]

[−5.49]

[−5.54]

Observations:

29

R-squared:

0.88

0.95

0.72

0.81

0.83

0.88

Adj-R-squared

0.85

0.93

0.64

0.75

0.79

0.85

Long-run

variance

35.52

6.81

12.99

7.69

1.17

0.70

JB Normality

1.65

[0.44]

0.68

[0.71]

0.04

[0.98]

3.29

[0.19]

1.25

[0.54]

0.20

[0.90]

Centered VIF

All <10.00

Wald Test (F-

stat.)

529.87***

2755.85***

117.45***

200.90***

200.78***

334.91***

Wald Test

(Chi-Square)

3709.09***

19290.93***

822.15***

1406.31***

1405.48***

2344.37***

Source: The author’s own calculations based on the collected data.

Notes: 1) Included observations: 29 after adjustments; 2) cointegrating equation deterministics: C and

@Trend; 3) the long-run covariance estimate (prewhitening with lags = 3 from AIC maxlags = 3, Bartlett

kernel, Newey–West fixed bandwidth = 4.0000); 4) *, **, and *** indicate statistical significance at the

10%, 5%, and 1% levels, respectively; 5) the figures were rounded to two decimal places for

compactness; 6) the values inside parentheses show the standard errors, while those in brackets are the t-

statistic; 7) no d.f. adjustment for standard errors & covariance; 8) the coefficients are presented in bold

to make a clear distinction between the standard errors and t-statistic; and 9) SB_OUT_SH means the

share of output of booming sectors in the overall output, SM_OUT_SH means the share of the

manufacturing sector in the overall output, and SM_VA_SH means the share of manufacturing value-

added in the overall value-added.

Equations 1 and 2 indicate that the oil boom period boosted output in the

booming sectors while negatively affecting manufacturing output. However, this

157

relationship was not stable; that is, the statistical significance disappeared after the

NEER was replaced by the REER, but the coefficient remained negative. The same

situation was found for manufacturing value-added, namely a negative association with

the oil boom period (Eq. 5).

In addition, oil prices exhibited contradictory associations with the booming and

manufacturing sectors. They had statistical significance along with the REER only in

Eqs. 2 (negative), 4 (positive), and 6 (positive).

Moreover, oil rents had a statistically positive effect on booming sectors;

however, they had a statistically negative effect on manufacturing value-added and

output (Eqs. 3–6).

The next phase of clarifying the sectoral effects of the DD-related variables

involved an analysis of value-added in agriculture (as the second part of the lagging

sectors in the three-sector approach) and in non-tradeable sectors. Equations 7–10 in

Table 5.19 present the resulting estimates.

The EDI was negatively and statistically significantly associated with value-

added in agriculture (Eqs. 7 and 8), while it was positively but not significantly

associated with the non-tradeable sectors (Eqs. 9 and 10). The NEER exhibited only one

statistically significant association with value-added in the non-tradeable sector (Eq. 10

– negative). The REER had a negative and statistically significant effect on value-added

in the agriculture sector (Eq. 8). The REER positively impacted non-tradeable value-

added (Eq. 10). Oil prices and oil rents had a negative and statistically significant effect

on both agriculture and non-tradeable value-added (Eqs. 7–10). Oil prices yielded

significant results only when combined with the REER variable (Eqs. 8 and 10 –

positive and negative, respectively). Finally, all of the intercepts and trend components

were statistically significant.

Table 5.19: Sectoral implications of REER, NEER, oil prices, and oil rents, for 1990–

2019 (continued).

Equation Name:

7

8

9

10

Dependent Variable:

SA_VA_SH

SNT_VA_SH

EDI

−0.97*

−0.81**

0.49

0.19

(−0.54)

(0.38)

(−0.41)

(−0.30)

[−1.80]

[−2.15]

[1.18]

[0.62]

NEER_66

0.03

−0.05**

(−0.02)

(0.02)

[1.11]

[−2.64]

OIL_BOOM

−3.54*

−2.24*

−2.82**

−4.26***

(−1.75)

(−1.29)

(1.34)

(1.03)

[−2.02]

[−1.73]

[−2.11]

[−4.14]

158

OIL_PRICES

0.03

0.06**

−0.02

−0.06***

(−0.03)

(0.02)

(−0.03)

(0.02)

[0.87]

[2.66]

[−0.66]

[−3.10]

OIL_RENTS

−0.15*

−0.20***

−0.44***

−0.35***

(−0.08)

(0.05)

(0.06)

(0.04)

[−1.88]

[−3.86]

[−7.29]

[−8.51]

C

27.63***

33.45***

45.2***

36.54***

(2.95)

(1.32)

(2.25)

(1.05)

[9.36]

[25.34]

[20.08]

[34.79]

@TREND

−0.88***

−0.75***

0.21***

0.18***

(0.09)

(0.07)

(0.06)

[−9.36]

[−10.77]

[3.01]

[3.14]

REER_66

−0.07***

0.06***

(0.02)

(0.01)

[−3.80]

[4.37]

Observations:

29

R-squared:

0.93

0.95

0.82

0.85

Adj-R-squared

0.93

0.78

0.81

Long-run variance

4.98

2.58

2.90

1.64

JB Normality

0.67

[0.72]

9.09

[0.01]

1.23

[0.54]

0.05

[0.98]

Centered VIF

All <10.00

Wald Test (F-stat.)

201.53***

382.56***

1591.72***

2820.04***

Wald Test (Chi-

Square)

1410.70***

2677.92***

11141.02***

19740.29***

Source: The author’s own calculations based on the collected data.

Notes: 1) Included observations: 29 after adjustments; 2) cointegrating equation deterministics: C and

@Trend; 3) the long-run covariance estimate (prewhitening with lags = 3 from AIC maxlags = 3, Bartlett

kernel, Newey–West fixed bandwidth = 4.0000); 4) *, **, and *** indicate statistical significance at the

10%, 5%, and 1% levels, respectively; 5) the figures were rounded to two decimal places for

compactness; 6) the values inside parentheses indicate the standard errors, while those in brackets are the

t-statistic; 7) no d.f. adjustment for standard errors & covariance; 8) the coefficients are presented in bold

to make a clear distinction between the standard errors and t-statistic; and 9) SA_VA_SH means the share

of agriculture value-added in the overall value-added, while SNT_VA_SH means the share of non-

tradeable value-added sectors in the overall value-added.

Table 5.20 summarizes employment between 1990 and 2019. Employment in

the booming sectors was positively and statistically significantly affected by the EDI,

NEER, and oil rents (Eqs. 11 and 12). Oil prices and REER negatively affected

employment in the lagging sectors, but the EDI, NEER, and oil rents had positive

connections. Then, the EDI, NEER, and oil rents negatively affected employment in

non-tradeable sectors, but oil prices and REER had a significantly positive impact (Eqs.

15 and 16). Finally, all constants in Eqs. 11–16 were highly significant, but the trend

component of Eqs. 11 and 12 did not exhibit significance.

Table 5.20: Sectoral implications of REER, NEER, oil prices, and oil rents for 1990–

2019 (continued).

Equation Name:

11

12

13

14

15

16

Dependent Variable:

SB_EMP_SH

SL_EMP_SH

SNT_EMP_SH

EDI

0.02

0.05**

0.61***

0.99***

−0.62***

−1.02***

(0.01)

(0.02)

(0.16)

(0.28)

(0.17)

(0.29)

[1.40]

[2.50]

[3.77]

[3.55]

[−3.68]

[−3.49]

NEER_66

0.01***

0.05***

−0.05***

(0.00)

(0.01)

159

[9.15]

[7.07]

[−7.17]

OIL_BOOM

−0.05

0.02

−0.65

0.09

0.67

−0.11

(0.04)

(0.06)

(0.52)

(0.95)

(0.55)

(1.00)

[−1.54]

[0.35]

[−1.24]

[0.09]

[1.22]

[−0.11]

OIL_PRICES

−0.01***

0.01

−0.05***

−0.01

0.05***

0.01

(0.01)

(0.02)

(0.01)

(0.02)

[−3.23]

[0.66]

[−4.48]

[−0.85]

[4.44]

[0.79]

OIL_RENTS

0.01***

0.01**

0.16***

0.10***

−0.17***

−0.10***

(0.01)

(0.02)

(0.04)

(0.02)

(0.04)

[8.41]

[2.80]

[6.89]

[2.50]

[−7.02]

[−2.54]

C

0.32***

0.98***

35.07***

42.30***

64.37***

56.72***

(0.06)

(0.07)

(0.88)

(0.97)

(0.92)

(1.02)

[5.26]

[14.88]

[39.70]

[43.47]

[69.85]

[55.54]

@TREND

−0.01

0.09***

0.05***

−0.09***

−0.05

(0.01)

(0.03)

(0.05)

(0.03)

(0.05)

[−0.48]

[−0.80]

[3.18]

[0.99]

[−2.92]

[−0.83]

REER_66

−0.01***

−0.03***

0.03***

(0.01)

[−3.53]

[−2.20]

[2.22]

Observations:

29

R-squared:

0.65

0.55

0.66

0.55

0.66

0.54

Adj-R-squared

0.56

0.43

0.57

0.42

0.57

0.42

Long-run variance

0.01

0.45

1.40

0.49

1.55

JB Normality

3.63

[0.16]

0.44

[0.80]

6.92

[0.03]

0.42

[0.81]

7.01

[0.03]

0.40

[0.82]

Centered VIF

All <10.00

Wald Test (F-stat.)

1763.25***

564.87***

17418.09***

5525.06***

26556.38***

8344.05***

Wald Test (Chi-

Square)

12342.78***

3954.11***

121926.6***

38675.41***

185894.7***

58408.37****

Source: The author’s own calculations based on the collected data.

Notes: 1) Included observations: 29 after adjustments; 2) cointegrating equation deterministics: C and

@Trend; 3) the long-run covariance estimate (prewhitening with lags = 3 from AIC maxlags = 3, Bartlett

kernel, Newey–West fixed bandwidth = 4.0000); 4) *, **, and *** indicate statistical significance at the

10%, 5%, and 1% levels, respectively; 5) the figures were rounded to two decimal places for

compactness; 6) the values inside parentheses indicate the standard errors, while those in brackets are the

t-statistic; 7) no d.f. adjustment for standard errors & covariance; 8) the coefficients are presented in bold

to make a clear distinction between the standard errors and t-statistic; and 9) SB_EMP_SH,

SL_EMP_SH, and SNT_EMP_SH denote the share of employment of booming sectors, lagging sectors,

and non-tradeable sectors, respectively.

Finally, Table 5.21 reports the export side of the distribution of the effects of the

EDI, REER, NEER, oil prices, and oil rents. Initially, the exports of SB were

significantly and positively associated only with the EDI. However, after the NEER

(Eq. 17) was replaced by the REER (Eq. 18), oil rents and the REER were also

significantly positively associated with SB exports. Moreover, oil prices and the oil

boom period had a significantly negative coefficient. The reason for this was related to

the fact that production and, consequently, exports in Azerbaijan rose and fell rapidly

after the transition period. Even though Azerbaijan’s development phase (2005–2014)

can be described as an oil boom, it was not associated with higher export shares.

Moreover, the negative sign for oil prices but positive sign for oil rents meant that

exports of the booming sectors increased when profits also increased.

160

Equation 19 indicates that the EDI had a significantly negative effect on lagging

sectors’ exports. However, after using the REER, oil rents and the REER itself also had

negative and significant coefficients. Interestingly, the oil boom and oil prices had a

positive and statistically significant effect on exports of the SL. This could indicate that

the non-oil sectors are heavily subsidized by oil revenues. However, the appreciation of

the REER and the increasing dependence on the extractive industry jeopardize the

competitiveness of SL.

Table 5.21: Sectoral implications of EDI, REER, NEER, oil prices, and oil rents for

1990–2019.

Equation Name:

17

18

19

20

Dependent Variable:

SB_EXP_SH

SL_EXP_SH

EDI

7.36***

5.29***

−5.88***

−5.56***

(2.18)

(0.90)

(1.10)

(0.72)

[3.38]

[5.88]

[−5.33]

[−7.73]

NEER_66

−0.32***

0.06

(0.10)

(0.05)

[−3.35]

[1.22]

OIL_BOOM

0.09

−10.03***

2.91

6.60**

(7.04)

(3.08)

(3.57)

(2.46)

[0.01]

[−3.26]

[0.82]

[2.68]

OIL_PRICES

−0.04

−0.31***

0.05

0.11**

(0.14)

(0.05)

(0.07)

(0.04)

[−0.26]

[−5.70]

[0.68]

[2.61]

OIL_RENTS

0.25

0.79***

−0.19

−0.32***

(0.32)

(0.12)

(0.16)

(0.10)

[0.80]

[6.36]

[−1.18]

[−3.23]

C

72.29***

14.32***

40.26***

54.94***

(11.86)

(3.15)

(6.02)

(2.52)

[6.10]

[4.55]

[6.69]

[21.84]

@TREND

1.60***

1.15***

−1.22***

−0.94***

(0.38)

(0.17)

(0.19)

(0.13)

[4.25]

[6.87]

[−6.41]

[−7.02]

REER_66

0.44***

−0.16***

(0.04)

(0.03)

[10.83]

[−5.01]

Observations:

29

R-squared:

0.84

0.93

0.89

0.92

Adj-R-squared

0.80

0.91

0.86

0.90

JB Normality

1.60

[0.45]

0.74

[0.69]

1.04

[0.59]

0.85

[0.65]

Centered VIF

All <10.00

Wald Test (F-stat.)

329.16***

1797.59***

107.17***

236.36***

Wald Test (Chi-

Square)

2304.10***

12583.14***

750.17***

1654.51***

Source: The author’s own calculations based on the collected data.

Notes: 1) Included observations: 29 after adjustments; 2) cointegrating equation deterministics: C and

@Trend; 3) the long-run covariance estimate (prewhitening with lags = 3 from AIC maxlags = 3, Bartlett

kernel, Newey–West fixed bandwidth = 4.0000); 4) *, **, and *** indicate statistical significance at the

10%, 5%, and 1% levels, respectively; 5) the figures were rounded to two decimal places for

compactness; 6) the values inside parentheses show the standard errors, while those in brackets are the t-

statistic; 7) no d.f. adjustment for standard errors & covariance; 8) the coefficients are presented in bold

to make a clear distinction between the standard errors and t-statistic; and 9) SB_EXP_SH denotes the

161

share of the booming sectors in total exports, while SL_EXP_SH denotes the share of the lagging sectors

in total exports.

Regarding the results of the goodness-of-fit and stability tests of the estimated

models, all models had high values for R-squared and adjusted R-squared. Moreover,

the intercept and trend components of the estimates were statistically significant. In

addition, the equations had normally distributed residuals, with the exception of Eqs. 8,

13, and 15.

Furthermore, according to the centered VIF factors, all models were free from

the problem of multicollinearity. The Wald test indicated that the coefficients obtained

from the estimates contributed to the models being significantly different from zero.

Finally, most models were free of autocorrelation and partial correlation problems up to

16 lags, as indicated by the correlogram Q-statistic results (available upon request).

5.2.3. The resource movement effect

Table 5.22 presents the OLS results for the resource movement effect in Azerbaijan’s

economy between 2000 and 2018. The estimated models had high R-squared and

adjusted R-squared values, especially the models where output growth rates were the

dependent variables. Furthermore, they had significant F-statistics (at both the 5% and

10% levels), and homoscedasticity and normally distributed residuals (except for Eq.

31, which has serial correlation). The sample size was 17 for all models.

Moreover, a statistically significant intercept was not found in any model, but

more than half of the intercept coefficients found were negative. This indicated an

overall downward linear trend in output and employment in the lagging and non-

tradeable sectors. While the growth rate of output in SB had a negative effect on output

in SL (Eq. 25, ß = −0.21, t = −0.50), that of employment in SB negatively affected

employment in SNT (Eqs. 31 and 32, ß = −0.01, t = −0.45, and ß = −0.01, t = −0.10,

respectively).

The estimates also indicated that the growth rates of output and employment in

SL are negatively related to the growth rate of output in SB (Eq. 21, ß = −0.10, t =

−0.02). The growth rate of employment in SL only had a statistically significant effect

on SNT employment growth rates (Eqs. 31 and 32, ß = −0.28, t = −2.58, and ß = −0.34, t

= −2.90, respectively).

162

Table 5.22: Resource movement effect of Dutch Disease in Azerbaijan’s economy for

2000–2018.

Dependent variable

Output growth rate:

Employment growth rate:

Exp.var

SB

SL

SNT

SB

SL

SNT

(21)

(22)

(23)

(24)

(25)

(26)

(27)

(28)

(29)

(30)

(31)

(32)

Intercept

0.10

[0.02]

−0.09

[−0.02]

−0.35

[−0.27]

−0.36

[−0.25]

−0.48

[−0.16]

−0.24

[−0.14]

0.68

[0.76]

0.64

[0.76]

0.01

[0.02]

−0.01

[−0.03]

−0.01

[−0.08]

−0.01

[−0.09]

SB-GR

−0.09

[−0.25]

−0.21

[−0.50]

1.19

[1.50]

0.23

[0.46]

8.91

[0.01]

0.02

[0.56]

−0.01

[−0.45]

−0.01

[−0.10]

SL-GR

−0.10

[−0.01]

2.12

[0.19]

−9.45*

[−2.07]

−3.63

[−1.29]

0.01

[0.01]

0.94

[0.56]

−0.28**

[−2.58]

−0.34**

[−2.90]

SNT-GR

−41.66**

[−2.19]

−43.59**

[−2.31]

−6.03

[−1.62]

−9.04**

[−2.27]

−1.36

[−0.45]

−0.29

[−0.10]

−0.99**

[−2.58]

−0.96**

[−2.90]

Inc.AZN

0.78

[1.24]

0.46***

[3.30]

0.67**

[2.257]

0.06

[0.62]

−0.02

[−1.14]

−0.01**

[−2.07]

Inc.USD

0.47

[1.10]

0.23**

[2.17]

0.89***

[6.90]

0.10

[1.56]

−0.02**

[−2.21]

−0.01**

[−1.83]

R2

0.44

0.43

0.56

0.43

0.44

0.80

0.06

0.16

0.28

0.40

0.41

0.38

Adj. R2

0.31

0.29

0.45

0.30

0.31

0.77

−0.15

−0.03

0.12

0.26

0.27

0.24

Obs.

17

F-stat.

3.36

3.22

5.40

3.25

3.42

18.37

0.30

0.84

1.72

2.90

2.98

2.66

Prob(F-

stat)

0.05

0.45

0.06

0.04

0.00

0.83

0.49

0.21

0.07

0.09

Het.F stat

1.04

[0.35]

0.46

[0.65]

0.81

[0.65]

0.16

[0.97]

0.76

[0.47]

1.04

[0.77]

0.74

[0.83]

0.56

[0.58]

1.68

[0.56]

0.72

[0.49]

1.04

[0.35]

1.15

[0.31]

JBN

0.55

[0.76]

0.62

[0.73]

0.31

[0.86]

0.24

[0.89]

0.22

[0.99]

0.57

[0.75]

0.47

[0.79]

0.34

[0.84]

0.23

[0.89]

0.38

[0.83]

1.57

[0.46]

1.54

[0.46]

LM test

0.77

[0.36]

0.21

[0.79]

1.77

[0.12]

0.77

[0.37]

0.55

[0.50]

1.92

[0.10]

0.51

[0.52]

0.55

[0.49]

0.89

[0.31]

1.13

[0.23]

4.06

[0.03]

1.56

[0.14]

Source: The author’s own calculations based on the collected data.

Notes: 1) GR means growth rate; 2) the bold coefficients highlight the significant results; 3) *, **, and

*** indicate statistical significance at the 10%, 5%, and 1% levels, respectively; 4) the figures in brackets

are the corresponding t-statistics; 5) the left-hand side of the table employs output growth rates, such as

SB-GR, SL-GR, and SNT-GR as independent variables; 6) the right-hand side of the table employs

employment growth rates as independent variables, such as SB-GR, SL-GR, and SNT-GR; 7) the

estimations do not include any degree-of-freedom adjustment for standard errors and covariance; 8) all

figures were rounded to the second decimal place for compactness; 9) Het.F stat denotes the

heteroscedasticity test based on the Breusch–Pagan–Godfrey method, JBN denotes Jaque–Bera normality

test results, and the LM test is the Lagrange multiplier test for serial correlation. The values in brackets

indicate p values.

The growth rate of output in SNT had several statistically significant effects, as

depicted in Table 5.22. The presence of a strong negative relationship between the

growth rates of output of SB and SNT (Eqs. 21 and 22, ß = −41.66, t = −2.19 and ß =

−43.59, t = −2.32, respectively) and SL and SNT (Eq. 24, ß = −9.04, t = −2.27) indicated

that for Azerbaijan’s economy, output of SNT increased, SB decreased over time, and SL

contracted. In this regard, a clear pattern of output spillovers exists. The growth rate of

employment in SNT also negatively affected SL (Eqs. 29 and 30, ß = −0.99, t = −2.58, ß

= −0.96, t = −2.90, respectively).

Furthermore, population income in this study captured the demand side of the

economy in terms of wage levels, which would also affect the anticipated resource

movement effect of DD. Therefore, Eqs. 23–26 revealed that population income had a

163

positive relationship with output growth in the lagging and non-tradeable sectors in both

AZN and USD. However, population income was negatively related to employment

growth rates in the lagging and non-tradeable sectors, as clearly indicated by equations

30–32.

Figure 5.8 presents the IRFs of manufacturing employment relative to itself, oil

prices, the REER, mining employment, and services employment. All estimations were

carried out in the first difference of the variables of interest according to ADF test (see

Table A5.6, Appendix). In the first four quarters, manufacturing employment can be

observed to respond favorably to self-induced shocks, but from the fifth quarter it

fluctuates until the tenth quarter.

The response of manufacturing employment to oil price shocks was initially

positive (second quarter), but in the third, fourth, and fifth quarters it was either zero or

negative. After the positive response to oil price shocks in the sixth quarter, the next

three quarters exhibited negative responses.

Furthermore, manufacturing employment responded negatively to REER shocks

over the periods examined. This suggests that the REER had a large impact on

manufacturing, consistent with the symptoms predicted by the DD hypothesis. In

addition, manufacturing employment responded unfavorably to mining employment

shocks in the second quarter but positively from the third quarter to the ninth. This

element of manufacturing and mining employment reflects the capital intensity of

Azerbaijan’s oil industry, as workers leaving the oil industry sector are theoretically

absorbed by non-oil manufacturing. However, manufacturing employment responded

negatively to the mining employment shocks that began in the eleventh quarter.

Regarding the response of manufacturing employment to shocks in the service sector, it

was positive only in the fifth quarter and negative in all other quarters.

Figure 5.8: Impulse response functions of manufacturing employment in VAR models.

-0.5

0.0

0.5

1.0

1.5

2 4 6 8 10 12

Response of D(MAN) to D(MAN)

-0.5

0.0

0.5

1.0

1.5

2 4 6 8 10 12

Response of D(MAN) to D(OIL_PRICE)

164

-0.5

0.0

0.5

1.0

1.5

2 4 6 8 10 12

Response of D(MAN) to D(REER)

-0.5

0.0

0.5

1.0

1.5

2 4 6 8 10 12

Response of D(MAN) to D(MINING)

-0.5

0.0

0.5

1.0

1.5

2 4 6 8 10 12

Response of D(MAN) to D(SERVICES)

Source: The author’s own calculations based on the collected data.

Notes: 1) Response to Cholesky one S.D. Innovations ± 2 S.E; 2) MAN denotes manufacturing

employment, MINING is mining employment in thousands of persons, and SERVICES is the

employment figure in the service sector in thousands of persons. Similarly, REER is the real effective

exchange rate and OIL_PRICE denotes oil prices.

Finally, VAR Granger causality tests were conducted to determine whether the

time series used in the study could predict each other, particularly for manufacturing

employment. The results did not support a unidirectional causal relationship between

the variables of interest and manufacturing employment (see Table 5.23). Nevertheless,

the REER and mining employment could be predicted based on manufacturing

employment and oil prices because of their individual and combined statistical

significance.

Table 5.23: Results of VAR Granger causality tests.

Dep. Var.

ΔMAN

ΔOIL_P

ΔREER

ΔMIN

ΔSERV

ΔJS

ΔMAN

–

5.31

(0.379)

6.09

(0.298)

0.83

(0.975)

1.99

(0.851)

14.38 (0.811)

ΔOIL_P

4.76

(0.446)

–

2.97

(0.705)

4.81

(0.440)

2.30

(0.807)

8.87

(0.984)

ΔREER

9.85

(0.080)*

74.41

(0.000)***

–

13.55

(0.019)**

10.13

(0.071)

90.86

(0.000)***

ΔMIN

21.34

(0.000)***

9.80

(0.081)*

6.11

(0.295)

–

3.98

(0.552)

48.21

(0.000)***

ΔSERV

3.17

(0.674)

1.06

(0.957)

3.60

(0.678)

1.77

(0.879)

–

8.22

(0.990)

Source: The author’s own calculations based on the collected data.

Notes: 1) *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively; 2)

figures were rounded to the second and to third decimal place for compactness; 3) figures in parentheses

are the corresponding probability values; and 4) MAN, OIL_P, REER, MIN, SERV, and JS denote

manufacturing employment, oil prices, real effective exchange rate, mining employment, services

employment, and joint significance, respectively.

165

The VARs were stable and passed all standard diagnostic tests. The Appendix

contains the results of the characteristic polynomial AR, residual autocorrelations, VAR

residual serial correlation LM tests, and variance decomposition of manufacturing

employment (Table A5.7–5.9, Figure A5.3 and A5.4). For all IRFs see Table A5.5 in

the Appendix section.

5.2.4. The spending effect

Compared with the equations for the resource movement effect, the estimated models

for the spending effect yielded higher R-squared values, statistically significant F-

statistics, and normally distributed and homoscedastic residuals (see Table 5.24;

excluding Eq. 36, which has serial correlation).

MPC, population income in USD, and government spending in billions of USD

and as a percentage of GDP had positive and statistically significant relationships with

CPI growth rates. Moreover, non-tradeable output had a negative relationship with CPI

(see Eqs. 33 and 34). In addition, both the REER and the NEER were significantly

positively related to the income of the population in USD. However, government

spending was negatively related to the REER and the NEER.

Table 5.24: Spending effect of Dutch Disease in Azerbaijan’s Economy for 2000–2018.

Dependent variable

CPI-GR

REER-GR

NEER-GR

Exp.var.

(33)

(34)

(35)

(36)

(37)

(38)

Intercept

−0.23

[−0.27]

−0.40

[−0.24]

−0.01

[−0.01]

0.15

[0.07]

−0.17

[−0.07]

−0.02

[−0.01]

MPC

0.11***

[5.13]

0.09*

[2.17]

0.01

[0.34]

0.04

[0.78]

0.06

[1.02]

0.08

[1.32]

SNT output

−1.07***

[−7.61]

−1.26***

[−4.83]

−0.05

[−0.15]

0.04

[0.11]

−0.18

[−0.45]

−0.07

[−0.17]

Population income in

USD

0.67***

[4.18]

1.35***

[4.91]

1.00**

[2.49]

0.50

[1.42]

1.00**

[2.18]

0.48

[1.14]

Gov. spen. – billions

USD

0.48***

[8.64]

−0.32**

[−2.31]

−0.34*

[−2.17]

Gov. spen. – % share

of GDP

0.30**

[2.91]

−0.34**

[−2.61]

−0.31*

[−1.96]

R2

0.91

0.67

0.63

0.65

0.51

0.49

Adj. R2

0.88

0.56

0.51

0.54

0.35

0.32

Obs.

17

F-stat.

30.0

6.18

5.16

5.70

3.11

2.87

Prob(F-stat)

0.00

0.01

0.06

0.07

Het. F stat

0.41

[0.82]

0.27

[0.98]

0.35

[0.97]

0.56

[0.94]

0.29

[0.97]

0.47

[0.68]

JBN

3.24

[0.20]

1.23

[0.54]

1.57

[0.45]

1.47

[0.48]

3.31

[0.19]

2.26

[0.32]

LM test

1.03

[0.23]

1.23

[0.17]

1.70

[0.10]

2.86

[0.04]

2.33

[0.06]

2.55

[0.05]

166

Source: The author’s own calculations based on the collected data.

Notes: 1) CPI means consumer price index, REER denotes real effective exchange rate, and NEER means

nominal effective exchange rate; 2) GR denotes growth rates; 3) MPC is the marginal propensity to

consume; 3) Gov. spen. means government spending; 4) the bold coefficients highlight the significant

results; 5) *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively; 6)

values in brackets are the corresponding t-statistics; 7) the figures were rounded to the second decimal

place for compactness; 8) Het. F stat denotes the heteroscedasticity test based on the Breusch–Pagan–

Godfrey method, JBN denotes the Jaque–Bera normality test results, and LM test denotes the Lagrange

multiplier test for serial correlation. The values in brackets indicate p values.

Lastly, a more thorough analysis of the spending effect using the BVAR model

revealed that the CPI positively responded to government spending over a 15-month

period (see Figure 5.9, panel a). The cumulative responses of the CPI to government

spending also indicated a strong positive association between government spending

shocks and the CPI (see Figure 5.9, panel b). The BVAR model that captured the

spending effect of DD in Azerbaijan’s economy was stable and passed all diagnostic

and stability tests (see Appendix, Figure A5.6 and A5.7 for results of AR root plots and

residual correlograms, respectively and Table A5.10 for variance decomposition).

Long-term modeling between the CPI and government spending could not be performed

because the sample size was small and the data had no cointegration relationship.

Figure 5.9: Impulse response functions of the CPI used in the Bayesian VAR model.

a. Responses of CPI to government

spending

b. Accumulated responses of CPI to

government spending

.0

.1

.2

.3

.4

2 4 6 8 10 12 14 16 18

0.0

0.5

1.0

1.5

2.0

2 4 6 8 10 12 14 16 18

Source: The author’s own calculations based on the collected data.

Notes: 1) Response to Cholesky one S.D. Innovations ± 2 S.E.

5.2.5. Summary of the section

The results of this study were consistent with the dictionary definition of DD syndrome.

As seen above, the REER in Azerbaijan’s economy increased in parallel with the

increase in international oil prices between 1995M01 and 2020M12. This result was

obtained using methods such as VAR, FMOLS, and CCR. The increase in REER

negatively affected manufacturing output and value-added as well as agriculture (so-

called lagging sectors in DD theory). All of this was observed in the context of

167

increasing dependence on the oil industry and prices measured by the EDI. In other

words, manufacturing and agriculture value-added, as well as manufacturing value-

added, were negatively affected by increasing oil dependence. In parallel, the higher the

EDI, output and employment in the booming sectors increased significantly.

Surprisingly, employment in the lagging sectors also exhibited a significantly positive

relationship with the EDI; however, employment in the non-tradeable sectors was

negatively affected by the EDI. This could indicate the impact of subsidies (based on oil

revenues) on non-oil employment in the case of lagging sectors. However, the negative

coefficients for non-tradeable sectors prompted the author to further examine the

relevance of the spending effect of DD, since the expected relationship between

employment in the non-tradeable sectors and the EDI was positive.

Oil prices were statistically significant when the REER was used in a model

instead of the NEER. This indicated the ability of inflation effects in Azerbaijan’s

economy to reflect initial DD signs and symptoms. In other words, it means that oil

prices and other related variables more accurately predict the sectoral effects of the

aforementioned variables when inflation effects are removed from the exchange rate

indicator. Put differently, inflationary effects mask the worsening negative impact of the

oil boom on the non-oil sectors. This leads to unsustainable economic growth and

development; that is, there is oil-driven growth, but non-oil sectors are crowded out of

the economy.

DD theory does not require the effects of DD-related variables on exports from

booming and lagging sectors to be tested; however, out of curiosity, this study tested

this nexus. As expected, oil dependence, the REER, and oil rents had a negative impact

on non-oil exports, whereas the opposite occurred for the booming sectors. Surprisingly,

this was not the case during the oil boom, when oil prices had a positive impact on the

non-oil sectors and a negative impact on the booming sectors.

The export volumes of the oil and gas industry in Azerbaijan are regulated by

international treaties and agreements. An interesting interpretation of the positive

impact of oil prices on non-oil exports is to attribute it to the economic characteristics of

the main trading partners. The main international buyers of non-oil products from

Azerbaijan are also oil-rich countries, whose purchasing power increases when oil

prices spike. Chubrik and Walewski (2010) first argued that Azerbaijan exports non-oil

products mainly to Russia, where falling oil prices reduce domestic demand for

168

imported goods. Consequently, when oil prices increase in Russia, demand for

Azerbaijani non-oil products also increases, creating a positive relationship.

In terms of resource movement and spending effects of DD in Azerbaijan’s

economy between 2000 and 2019, the following can be noted: There is evidence both

for and against the resource movement effect. The growth rates of output and

employment of SB did not cause statistically significant deteriorations of SL or SNT.

However, the growth rates of output and employment in SNT had a significantly

negative effect on the employment dynamics in SL. This suggests the outcome of the

indirect de-industrialization of the resource movement effect. The proxy variables for

domestic demand—income in AZN and USD—also exhibited positive effects on the

output growth rates in each sector, but employment growth rates declined in SL and SNT

when domestic demand increased. The estimated standard unrestricted VAR model

revealed that the REER, oil prices, and service sector jobs all had negative effects on

manufacturing jobs.

The test of the spending effect demonstrated that higher population income,

MPC, and government spending were associated with higher levels of CPI. Moreover,

the CPI responded positively to shocks in state budget spending during the 16 months

covered by the BVAR model. However, the OLS approach based on annual data was

unable to capture the expected nexus between DD-related variables and the REER and

NEER. More advanced and country-specific linear modeling techniques may be able to

capture the spending effect of DD in the case of Azerbaijan’s economy.

5.3. Oil-Led De-Industrialization of Non-Oil Manufacturing in Azerbaijan’s

Economy: An Analysis of the Chemical Industry

In this chapter, Azerbaijan’s economy is studied from a de-industrialization point of

view, which seems to be the result of DD syndrome. While the NRC and DD have been

studied in the case of Azerbaijan’s economy, the de-industrialization of certain non-oil

subsectors has been largely ignored as a separate analysis. As a result, little is known

about the possible reasons for the production collapse in the tradeable non-oil sectors

after 2005 and 2006 (i.e., the beginning of the oil boom). The chemical industry is a

good example of a non-oil manufacturing sector to study because, although it has

169

accounted for most of Azerbaijan’s non-oil exports since 1991, some of its branches

have ceased production while others have grown since the transition period.

105

The NRC and DD theories were discussed in the previous chapter, and the

results suggested that the NRC and DD can be found in Azerbaijan’s economy.

According to these theories, the negative effects of the oil boom could lead to the de-

industrialization of some non-oil manufacturing sectors. Identifying de-industrialization

in specific subsectors would enable the design of more robust policies for addressing

industrial policy challenges. Thus, unsustainable economic growth in Azerbaijan could

be transformed into more sustainable, long-term economic development. Therefore, the

contribution of this study is to fill the gap in the current documentation on the de-

industrialization patterns of non-oil manufacturing sectors. Both quantitative and

qualitative research methods were applied in this study. The results presented later in

this chapter clearly demonstrate the oil-led de-industrialization of certain chemical

subsectors in Azerbaijan’s economy, such as chlorine and soda (liquid and caustic).

5.3.1. Descriptive analysis

The SSCRA provides production data in a natural form (i.e., in thousands of tons) for

specific subsectors of the chemical industry. To understand which specific subsectors

were affected by the presumed negative impact of the oil boom on Azerbaijan’s

economy, a brief figure analysis was first required. Figures 5.1 and 5.2 present the

subsectors that experienced a decline in production during the oil boom as well as those

that did not but either fluctuated or developed. An analysis of the data in Figure 5.1

revealed that six subsectors—caustic soda, chlorine, hydrochloric acid, isopropyl

alcohol, liquid soda, and sulfuric acid—experienced production declines that

overlapped with the start of the oil boom in 2005–06. In addition, caustic soda

production began in 2000, lasted only seven to eight years, and has not been significant

since 2009.

The subsectors in Figure 5.1 are closely related. For example, the most popular

industrial technique for producing chlorine is to combine cooking salt and sulfuric acid

to transform hydrochloric acid into chlorine (Kragh 2017). A decline in demand for

105

The other non-oil tradeable sectors, such as textiles and machinery, were also considered for the

analysis here. However, the textile industry does not show any consistent and systematic output collapses,

but rather fluctuations. Machinery production almost completely collapsed following the breaking of

Soviet Union. Even some machinery subsectors, such as the production of information technologies, have

been industrialized. Therefore, only the chemicals industry in Azerbaijan can scientifically be analyzed

against oil-led de-industrialization.

170

chlorine or a disruption in production capacity could also have led to a collapse in

demand for hydrochloric acid and sulfuric acid. Furthermore, liquid soda and caustic

soda are typical feedstocks for soap and detergents. Isopropyl alcohol is also a major

ingredient in numerous cosmetics, hand lotions, antiseptics, and medicines. It is also

used to convert ethanol (ethyl alcohol) into a less harmful form. Whatever causes the

breakdown of one can also cause the collapse of the other (i.e., the domino effect).

Figure 5.10: Subsectors of the chemical industry that experienced output de-

industrialization for 1995–2020.

Source: SSCRA (2022)

Notes: The red shaded areas denote the oil boom period in Azerbaijan between 2006 and 2020.

As Figure 5.2 indicates, certain subsectors of the chemical industry did not

experience significant and continuous declines in production during the oil boom. For

example, as barium sulfate is used extensively in oil and gas production and in the paint

and construction industries (Choudhury–Cary 2001), domestic demand was high and

production developed positively between 2000 and 2020. For bitumen, production was

also consistently positive between 2000 and 2020, with the exception of 2013 and 2014.

In 2013 and 2014, several reconstruction works occurred at the State Oil Refinery,

where bitumen is produced (Azertag 2013).

106

The main reason for the strong

development of bitumen in Azerbaijan could be associated with renovation works in

2000; at that time, the Austrian company Parner and the US-based company Perfokart

used Biturox technology to improve the quality of bitumen production (Abbasov et al.

2015). This helped to meet international standards and improve export potential.

106

Repair work carried out at bitumen production facility No. 401 (Azertag 2013).

171

Moreover, ethylene production exhibited a positive trend between 1995 and

2006 but slumped between 2007 and 2009. After the merger with Azerikimya PU—a

unit of SOCAR—ethylene production recovered rapidly in 2010. Overall, ethylene,

polyethylene, and propylene production recovered quickly from the energy and raw

material crises of the 1990s due to the purchase of a $95-million state-guaranteed

steam-generator complex from Japan. This will be remembered as one of President

Haydar Aliyev’s greatest contributions to the country’s petrochemical industry (Trend

News Agency 2015). Ethylene production was brought online with project capacity,

production was increased, raw materials were saved, and balance losses were

minimized, which helped to reduce production costs and increase profitability (Trend

News Agency 2015).

The privately owned Baku Steel Company was the largest supplier of oxygen

and nitrogen in Azerbaijan and played a crucial role in the metallurgical industry

(Azertag 2014). Between 2011 and 2014, it spent $200 million to modernize oxygen

and nitrogen production facilities to meet its own needs (Azertag 2014). A contract

worth €15 million was signed with the French company Air Liquide, resulting in a

modern 50-ton electric arc furnace and a furnace from German company Siemens VAI

being installed (Azertag 2014).

107

Moreover, liquid air serves as an industrial source of

oxygen and nitrogen, so the plants that produce oxygen and nitrogen most likely also

produce liquid air for the air separation process. Finally, the increasing nitrogen

production was due to a urea plant project implemented by SOCAR and the South

Korean Samsung Engineering Co. This was possible due to the State Program on Food

Security of the Population in Azerbaijan for 2008–2015 (Trend News Agency 2015).

108

All of these factors were reflected in the production levels of the selected subsectors,

which did not record a continuous and significant decline in production, as indicated in

the official statistics.

107

Baku Steel Company also had equipment manufactured by leading companies in the United States,

France, Italy, Turkey, India and other major countries. In addition, the number of nozzles increased by

reconstructing a modern continuous casting unit manufactured by the Canadian company STEL-TEK.

108

The Engineering, Procurement and Construction Agreement was signed between Samsung and

Azerbaijani government. The construction of the plant with a daily production capacity of 1,200 tons of

ammonia and 2,000 tons of granular urea was completed in 2019 (Trend News Agency 2015).

172

Figure 5.11: Subsectors of the chemical industry that did not experience output de-

industrialization for 1995–2020.

Source: SSCRA (2022)

Notes: The red shaded areas indicate the oil boom period in Azerbaijan between 2006 and 2020.

The production dynamics of the chemical industry depend on several factors,

such as available domestic and foreign markets, the level of production forces, and

domestic demand. Figure 5.12 indicates that chemical exports peaked in 2019, although

the industry’s share of total exports has declined. In the late 1990s, the chemical

industry’s share declined from 6.8% in 1996 to 1.1%, but the oil boom caused its share

of total exports to decline even further to 0.4% in 2008. Nevertheless, chemical exports

increased by up to 1.8% by 2020, mainly because the post-boom period brought lower

oil exports.

109

For information on the specific subsectors of the chemicals industry that are not listed here, see Figure

A5.10 and Figure A5.11 in the Appendix section.

173

Figure 5.12: Exports of the chemical industry (1996–2020).

Source: SSCRA (2022)

Moreover, the subsectors that experienced a slowdown at the beginning of the

oil boom appeared to be more export-oriented than the so-called developed subsectors

during the recovery or transition period (see Table 5.25). Both the trade value and share

of total exports of the collapsed subsectors were high during the recovery period, with

the exception of sulfuric acid and liquid soda. However, the developed subsectors had

little or no share in total exports during the recovery period. Notably, their share

increased during the oil boom period. Developed subsectors also belonged to the private

sector (oxygen, nitrogen, and liquid air were produced mainly by the Baku Steel

Company CJSC) and public–private partnerships (e.g., the production of ethylene

polymers by SOCAR Polymer LLC). However, the plants for chlorine, soda, and other

chemicals date back to the Soviet era and were unsuccessfully privatized in accordance

with the 2001 presidential decree (Aliyev 2001). According to data from the

Observatory of Economic Complexity (OEC), the so-called developed subsectors of the

chemical industry had more diverse export destinations than the collapsed subsectors

(see Table 5.25 for the two largest export destinations of each subsector). Thus, the

impact of DD could be seen at the subsectoral level in the chemical industry, especially

in subsectors that were more export-oriented at the beginning of independence.

0,0

1,0

2,0

3,0

4,0

5,0

6,0

7,0

8,0

0,0

50 000,0

100 000,0

150 000,0

200 000,0

250 000,0

300 000,0

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

Exports of chemical products, in current AZN

Share of chemical industry in total exports, in %

174

Table 5.25: Trade value of subsectors of the chemical industry (in current USD) and

share in total exports (in %) for 1997–2020.

Product

1996–2004

2005–2014

2015–2020

Main Export Destination

Collapsed subsectors

Isopropyl alcohol

5,701,203

(14.1%)

13,541,105

(8.4%)

6,044,366

(2.9%)

Turkey and Russia

Caustic soda

1,034,192

(2.6%)

377,668

(0.9%)

503

(0.0%)

Georgia and Russia

Sulfuric acid

425,812

(1.1%)

1,064,373

(2.6%)

70

(0.0%)

Georgia

Chlorine

378,463

(0.9%)

55,622

(0.1%)

120

(0.0%)

Turkmenistan and Georgia

Liquid soda

254,024

(0.6%)

1,042,723

(2.6%)

1,521

(0.0%)

Georgia

Hydrochloric acid

58,973

(0.1%)

8,753

(0.0%)

1,212

(0.0%)

Turkmenistan and Georgia

Developed subsectors

Oxygen

125

(0.0%)

2,824

(0.0%)

35,563

(0.1%)

Georgia and Kazakhstan

Nitrogen

207

(0.0%)

54,501

(0.1%)

11,679

(0.1%)

Georgia and the United Kingdom

Liquid air

1,279

(0.0%)

6,682

(0.0%)

56

(0.0%)

Georgia and Turkey

Paint products

437,691

(1.1%)

1,116,648

(2.8%)

2,408,672

(6.0%)

Georgia and Uzbekistan

Propylene

564,342

(1.4%)

12,963,562

(32.1%)

15,556,814

(28.6%)

Poland, Russia and Turkey

Bitumen

11,020,53

(2.7%)

4,808,735

(11.9%)

4,563,044

(11.3%)

Georgia and Turkey

Ethylene polymers

14,727,494

(36.5%)

63,133,566

(39.1%)

77,619,905

(37.5%)

Turkey and China

Source: The Observatory of Economic Complexity (OEC) and SSCRA

5.3.2. Econometric estimations

A stepwise regression analysis revealed that the oil boom period and the one-year lag of

the REER negatively affected caustic soda production (see Table 5.26). While FMOLS

supported these results, CCR did not capture statistical significance; however, the signs

of the coefficients remained the same (see Models 1, 2, and 3). Model 4 retained the

REER as an “essential variable”, and the algorithm continued to capture the oil boom as

a statistically significant and negative determinant of caustic soda production. In

addition, the stepwise algorithm included the service sector employment in Model 4.

Models 5 and 6 produced long-run RLS results. They indicated that the significantly

negative impact of the oil boom dropped in the long run, while the one-year lag of the

REER maintained its statistical significance. These results were robust even after

including employment in the service sector (see Model 6, Table 5.26).

175

Table 5.26: Regression results for the production of caustic soda (solid form) for 1995–

2020.

Dependent Variable: Production of Caustic Soda (Solid)

(1)

(2)

(3)

(4)

(5)

(6)

Stepwise +

OLS

FMOLS

CCR

Stepwise +

OLS

(REER)

Long-Run

RLS of 1–3

Long-Run

RLS of 4

Constant

41.22

8.26

1.95

185.79

7804.67***

10519.42***

Oil Boom

−2211.86*

−1838.85*

−1585.74

−2072.13**

568.46

474.28

REER(−1)

−53.02

−37.48

−34.82

−54.26*

−66.22***

−47.50*

Services Emp.

−8.34

−4.06

R2

0.24

0.23

0.22

0.28

0.26

0.31

Adj-R2

0.17

0.16

0.14

0.18

0.19

0.21

Rw-squared

0.32

0.39

F-Stat

3.38*

2.70*

DW

1.96

1.81

AIC

17.62

17.64

19.69

22.54

Rn-squared

stat.

8.77**

11.35***

Source: The author’s own calculations based on the collected data.

Notes: 1) *, **, and *** denote statistical significance at the 10%, 5%, and 1% levels, respectively; 2) the

figures were rounded to the second decimal place for compactness; 3) in the stepwise regression process,

the lag was set to 1 because the data set was yearly; 4) OLS stands for ordinary least squares; FMOLS

stands for fully modified OLS; CCR stands for canonical cointegration regression; and RLS stands for

robust least squares; 5) RLS was applied to level data containing outliers; 6) Model 4 employed the

REER variable as an “essential” regressor; 6) Model 5 was an RLS estimation of Models 1–3; 7) Models

1–4 were short-run estimates, while Models 5 and 6 were long-run regression models; and 8) Rw-squared

denotes residuals that were weighted and then squared.

From the results presented in Table 5.27, the oil boom appeared to have a

significantly negative impact on chlorine production with a one-year lag (Models 1–3).

However, in the long run, this relationship proved to be positive and statistically

significant (see Model 6). Meanwhile, oil prices had a significantly positive impact on

chlorine production in both the short- and long-run models. Including REER in the

analysis revealed that it had a negative impact on chlorine production in both the short

and long run. When service sector employment was included in the equation, a

significantly negative relationship with chlorine production was observed.

Although the production of hydrochloric acid is closely related to that of

chlorine, the analysis did not reveal statistically significant coefficients in the short run

(see Table 5.28). However, among the long-run regressors, employment in the service

sector had a significantly negative impact on the production level of hydrochloric acid.

Furthermore, the inclusion of additional variables in Model 6, such as the REER,

revealed many significant relationships. Together, the REER, one-year lag of the oil

boom, and service sector employment have had a statistically significant impact on

hydrochloric acid production—but only in the long run.

176

Table 5.27: Regression results for the production of chlorine for 1995–2020.

Dependent Variable: Production of Chlorine

(1)

(2)

(3)

(4)

(5)

(6)

Stepwise +

OLS

FMOLS

CCR

Stepwise +

OLS

(REER)

Long-Run

RLS of 1–3

Long-Run

RLS of 4

Constant

−203.53

−223.36

−222.28

208.91

5547.93**

51422.75***

Oil Boom(−1)

−4480.93**

−5048.18**

−5477.41*

1660.11

9522.98***

Oil prices

88.24**

83.07**

83.85*

96.54***

−40.33

75.34**

REER

−110.70*

−373.80***

Services Emp.

−23.35*

−14.10***

R2

0.33

0.32

0.35

0.02

0.62

Adj-R2

0.26

−0.06

0.55

Rw-squared

0.04

0.87

F-Stat

5.23**

3.82**

DW

2.22

2.27

AIC

18.85

18.52

42.46

41.73

Rn-squared

stat.

0.44

81.43***

Source: The author’s own calculations based on the collected data.

Notes: 1) *, **, and *** denote statistical significance at the 10%, 5%, and 1% levels, respectively; 2) the

figures were rounded to the second decimal place for compactness; 3) in the stepwise regression process,

the lag was set to 1 because the data set was yearly; 4) OLS stands for ordinary least squares; FMOLS

stands for fully modified OLS; CCR stands for canonical cointegration regression; and RLS stands for

robust least squares; 5) RLS was applied to level data containing outliers; 6) Model 4 employed the

REER variable as a “essential” regressor; 6) Model 5 is an RLS estimation of Models 1–3; 7) Models 1–4

were short-run estimates, while Models 5 and 6 were long-run regression models; and 8) Rw-squared

denotes residuals that were weighted and then squared.

Table 5.28: Regression results for the production of hydrochloric acid for 1995–2020.

Dependent Variable: Production of Hydrochloric Acid

(1)

(2)

(3)

(4)

(5)

(6)

Stepwise

+ OLS

FMOLS

CCR

Stepwise +

OLS (REER)

Long-Run

RLS of 1–3

Long-Run

RLS of 4

Constant

−0.14

−0.26

−0.30

−0.15

20.48***

27.18***

Oil Boom(−1)

−1.03

−1.33

−1.70

−1.12

0.12

2.92***

Services Emp.

−0.01

0.01

−0.01

−0.01***

−0.12***

REER

0.01

−0.10***

R2

0.11

0.08

0.03

0.11

0.64

0.76

Adj-R2

0.02

−0.02

−0.01

−0.2

0.61

0.72

Rw-squared

0.69

0.82

F-Stat

1.32

0.84

DW

2.13

AIC

3.46

3.54

22.92

Rn-squared stst.

38.93***

73.51***

Source: The author’s own calculations based on the collected data.

Notes: 1) *, **, and *** denote statistical significance at the 10%, 5%, and 1% levels, respectively; 2) the

figures were rounded to the second decimal place for compactness; 3) in the stepwise regression process,

the lag was set to 1 because the data set was yearly; 4) OLS stands for ordinary least squares; FMOLS

stands for fully modified OLS; CCR stands for canonical cointegration regression; and RLS stands for

robust least squares; 5) RLS was applied to level data containing outliers; 6) Model 4 employed the

REER variable as an “essential” regressor; 6) Model 5 was an RLS estimation of Models 1–3; 7) Models

1–4 were short-run estimates, while Models 5–6 were long-run regression models; and 8) Rw-squared

denotes residuals that were weighted and then squared.

The picture for isopropyl alcohol was similar to that for hydrochloric acid

production; that is, only one significant short-run relationship existed between oil prices

177

and isopropyl alcohol production (see Table 5.29). This relationship was positive and

only captured using FMOLS. Moreover, in the short run, the inclusion of the REER did

not change the picture for isopropyl alcohol (see Model 4); however, in the long run, oil

prices had a negative impact on this subsector with a one-year lag (Model 5). In the

long-run model, represented by Model 6, no statistically significant results were

obtained.

Table 5.29: Regression results for the production of isopropyl alcohol for 1995–2020.

Dependent Variable: Production of Isopropyl Alcohol

(1)

(2)

(3)

(4)

(5)

(6)

Stepwise

+ OLS

FMOLS

CCR

Stepwise +

OLS (REER)

Long-Run

RLS of 1–3

Long-Run

RLS of 4

Constant

−491.22

−476.62

−504.00

−463.35

13061.93***

23390.85**

Oil prices

80.70

90.92*

229.04

74.77

115.78

116.79

Oil prices (−1)

−87.88

−85.68

−218.52

−80.69

−131.22*

−67.57

REER

−47.28

−131.76

R2

0.16

−0.29

0.16

0.12

0.14

Adj-R2

0.08

0.07

−0.42

0.04

0.02

Rw-squared

0.14

0.17

F-Stat

1.97

1.33

DW

2.54

2.55

AIC

19.89

19.97

20.14

24.02

Rn-squared stat.

2.82

3.52

Source: The author’s own calculations based on the collected data.

Notes: 1) *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively; 2)

the figures were rounded to the second decimal place for compactness; 3) in the stepwise regression

process, the lag was set to 1 because the data set was yearly; 4) OLS stands for ordinary least squares;

FMOLS stands for fully modified OLS; CCR stands for canonical cointegration regression; and RLS

stands for robust least squares; 5) RLS was applied to level data containing outliers; 6) Model 4 employed

the REER variable as an “essential” regressor; 6) Model 5 was an RLS estimation of Models 1–3; 7)

Models 1–4 were short-run estimates, while Models 5 and 6 were long-run regression models; and 8) Rw-

squared denotes residuals that were weighted and then squared.

The analysis of liquid soda demonstrated that, in the short run, oil prices

positively and statistically significantly affected production (see Table 5.30, Models 1–

3). In the long run, however, this relationship was reversed, as indicated by Model 5.

Moreover, the oil boom period had a significantly negative effect on liquid soda

production in the short run. This relationship disappeared in the long run, as captured by

Model 6, becoming nonsignificant but positive. The inclusion of the REER in the

regression equations was fruitful, as Models 4 and 6 produced statistically significant

results. Model 4 implied a negative short-run relationship while Model 6 implied a

long-run negative relationship between the REER and liquid soda production. Service

sector employment also had a negative long-run effect on liquid soda production.

178

Table 5.30: Regression results for the production of liquid soda for 1995–2020.

Dependent Variable: Production of Liquid Soda

(1)

(2)

(3)

(4)

(5)

(6)

Stepwise +

OLS

FMOLS

CCR

Stepwise +

OLS (REER)

Long-Run

RLS of 1–3

Long-Run

RLS of 4

Constant

−1511.25

−1192.62

−1196.47

−1834.85*

28579.04***

9963.40***

Oil prices

145.01**

137.05**

137.86*

157.37**

−351.86***

98.42

Oil boom (−1)

−7816.93**

−8905.02**

−9696.47**

12465.14

REER

−226.15**

−211.11*

Services Emp. (−1)

34.52

−63.08***

R2

0.31

0.34

0.33

0.35

0.30

0.70

Adj-R2

0.25

0.27

0.26

0.24

0.65

Rw-squared stat.

0.39

0.78

F-Stat

4.81

3.64**

DW

2.13

2.63

AIC

19.98

20.00

35.16

28.70

Rn-squared stat.

8.93**

55.21***