Content uploaded by Aysit Tansel
Author content
All content in this area was uploaded by Aysit Tansel
Content may be subject to copyright.
The role of macroeconomic
instability in public and private
capital accumulation and growth:
the case of Turkey 1963–1999
Mustafa Ismihan
a
*, Kivilcim Metin-Ozcan
b
and Aysit Tansel
c
a
Atilim University, Kizilcasar Koyu, 06836, Incek-Golbasi, Ankara, Turkey;
b
Bilkent University, 06800 Bilkent, Ankara, Turkey and
c
Middle East Technical University (METU)
This study investigates the empirical relationship(s) between macro-
economic instability, public and private capital accumulation and growth
in Turkey over the period 1963–1999. Time series econometric techniques,
such as cointegration and impulse response analysis, are used. The results of
this paper suggest that the chronic and increasing macroeconomic instabil-
ity of the Turkish economy has seriously affected her capital formation
and growth. Furthermore, the Turkish experience indicates that chronic
macroeconomic instability seems to be a serious impediment to public
investment, especially to its infrastructural component, and shatters, or
even reverses, the complementarity between public and private investment
in the long run.
I. Introduction
Turkey and many other developing countries, such as
those in Latin America, experienced chronic macro-
economic instability by following unstable economic
policies, such as populist and myopic macroeconomic
policies, over extended periods of time.
1,2
During
their chronic instability episodes the typical develop-
ing country tends to exhibit excessive and persistent
budget deficit, high debt to GNP ratio and high
inflation rate.
Most of the countries suffering from chronic
macroeconomic instability register low and volatile
rates of capital formation and economic growth.
Furthermore, they tend to exhibit low level of
public capital spending
3
as a share of output. Many
economists nowadays believe that macroeconomic
instability
4
is detrimental to capital accumulation
and economic growth, and there is empirical evidence
that supports this view.
5
Theoretical arguments in
this line of research focused on the detrimental effects
of macroeconomic instability on private investment
* Corresponding author. E-mail: ismihan@atilim.edu.tr
1
Recently new political economy literature has emphasized the role of political factors, such as political instability
and polarization, on macroeconomic instability. See Drazen (2000) and Persson and Tabellini (2000) for an overview.
2
Developing countries may also experience macroeconomic instability as a result of structural characteristics such as
vulnerability to external shocks.
3
We will use the terms ‘public capital spending’ and ‘public investment’ interchangeably throughout this study.
4
We define macroeconomic instability in line with Fischer (1993a, 1993b) and Bleaney (1996). A rise in macroeconomic
instability means a rise in one or more policy-induced macroeconomic instability indicators, such as inflation rate, public
deficit to GNP ratio and external debt to GNP ratio.
5
See, for example, Fischer (1993a, 1993b), Cardosa (1993) and Bleaney (1996) among others.
Applied Economics ISSN 0003–6846 print/ISSN 1466–4283 online #2005 Taylor & Francis Group Ltd 239
http://www.tandf.co.uk/journals
DOI: 10.1080/0003684042000286115
Applied Economics, 2005, 37, 239–251
and productivity.
6
Moreover, early empirical studies,
which assessed the effects of macroeconomic instabil-
ity on investment, used either aggregate investment
(Bleaney, 1996) or private investment data (Cardosa,
1993) in their analyses.
Macroeconomic instability has negative effects on
both private and public investment, albeit through
different channels. While the rise in macroeconomic
uncertainty is the main cause for a reduction in pri-
vate investment, the reduction in the fiscal (as well as
political) ‘ability’ of the government is the principal
reason for the decrease in public investment. That is,
a rise in the level of macroeconomic instability leads to
(or aggravates) fiscal stringency due to the existence
of the budget constraint of the government. For
example, high inflation rate and/or excessive debt
accumulation lowers the overall public resources
otherwise available for public expenditures, namely
capital and current expenditures.
7
In turn, the incum-
bent government lowers public capital expenditures
rather than current expenditures when faced with
fiscal stringency since it is politically easer to cut the
former than the latter.
8
Many recent empirical studies found positive
effects of public capital spending, particularly infra-
structural spending, on private investment, produc-
tivity and growth.
9
These studies suggest that a
decrease in public capital spending can be harmful
for economic growth. More importantly, given the
detrimental effects of macroeconomic instability on
public investment, these studies imply that chronic
macroeconomic instability can be very costly in terms
of private capital accumulation and hence economic
growth if public and private investment are comple-
mentary (i.e. if public investment crowds-in private
investment). Thus there seems to be an additional
link between macroeconomic instability and economic
growth due to the possible complementarity between
public and private investment.
Currently there are two related strands of research
on the role of public capital spending in capital accu-
mulation and economic growth. The first one focuses
on the public capital spending and private investment
nexus. In this research area, many studies found
significant complementarity (crowding-in) effect, but
some studies were either inconclusive or found con-
tradictory results.
10
Blejer and Khan (1984) among
others suggest that this ambiguity might be the result
of using aggregate rather than disaggregate public
investment data, e.g. public investment in infrastruc-
ture. The evidence for the Turkish economy is also
ambiguous (see, for example, Anand et al., 1990;
Celasun and Tansel, 1993; Conway, 1990; Metin-
Ozcan et al., 2001 and Uygur, 1995). The second
approach analyses the public capital spending and
output (or growth) nexus. In this approach, the role
of public capital spending, especially, infrastructural
investment, is theoretically considered both in a
production function framework (Aschauer, 1989a)
and in a new growth theory framework (Barro,
1990). Considerable number of empirical studies,
using either a single-equation time series (Aschauer,
1989a) or a cross-section analysis (Easterly and
Rebelo, 1993) have found a positive effect of
(infrastructural) public investment on growth.
Nevertheless, early studies in these two groups of
literature were criticized on the empirical grounds.
The main empirical criticisms are related to the
reverse-causation, simultaneity, and ’spuriousness’
of the results (Sturm et al., 1998; Pereira, 2000). To
overcome these empirical problems, recent studies
used new time series techniques, such as multivariate
cointegration and impulse response analyses (Ghali,
1998; Pereira, 2000; Mittnik and Neumann, 2001).
11
6
For instance, it is widely argued that macroeconomic instability adversely affects the rates of productivity growth
and investment mainly by creating uncertainty about current and future macroeconomic environment. See, for example,
Fischer (1993b) and Bleaney (1996).
7
On the one hand, a rise in inflation rate usually rises the degree of dollarization and results in a loss of seigniorage revenue,
by reducing the demand for domestic currency. Furthermore, high inflation rate will also lower the revenues from ordinary
taxes due to the Olivera–Tanzi effect. On the other hand, high indebtedness leads to a high debt burden (principal plus interest
payment) and lowers the overall public resources available for other public expenditures. See, for example, Agenor and
Montiel (1996) for more detail.
8
This view is widely shared by many economists and there is empirical evidence that supports it. See De Haan et al. (1996)
for an overview and empirical evidence.
9
See, for example, Sturm et al. (1998), Pereira (2000) and Mittnik and Neumann (2001).
10
One of the earlier studies in the crowding-in, or more broadly, public capital spending-private investment nexus literature is
the work of Blejer and Khan (1984); however, this literature has exploded with the seminal work of Aschauer (1989b).
See Agenor and Montiel (1996) for an overview of earlier evidence for the developing countries.
11
Private investment, public investment and output were the commonly used variables in their analyses of the effects of public
capital spending on private capital spending and output.
240 M. Ismihan et al.
In this study we are mainly interested in the empiri-
cal assessment of the effects of macroeconomic
instability on public and private investment as well
as in the nature of their relationships (e.g. comple-
mentarity) and economic growth. Therefore we
extend the recent literature, on the role of public
capital spending in capital accumulation and eco-
nomic growth, to include the issue of macroeconomic
instability. We consider the Turkish experience in
our study, which we believe is a good case study
since Turkey has been suffering from chronic
macroeconomic instability over the last 25 years. To
accomplish this we estimate the long-run relationship
between public investment, private investment,
macroeconomic instability and output in Turkey
for the period 1963–1999 by using multivariate
cointegration analysis. The empirical analysis is
also extended by considering the infrastructural
component of public investment. Furthermore, the
generalized impulse response functions are used to
examine the dynamic effects of a rise in (i.e. a shock
on) a given variable on all the other variables in the
system. We are particularly interested in the analysis
of the dynamic effects of an increase in public
investment and macroeconomic instability, respec-
tively, on all the other variables.
This paper is organized as follows. Section II
provides a condensed overview of macroeconomic
instability, capital formation and growth in the
Turkish economy over the sample period (1963–
1999). Empirical results appear in Section III and
finally Section IV provides a summary and the policy
implications of the findings.
II. An Overview of Macroeconomic
Instability, Capital Formation and
Growth in the Turkish Economy,
1963–1999
In this section, we provide a condensed overview of
the Turkish economy for the 1963–1999 period. In
line with the aims of this paper, we will mainly
focus on capital formation, growth and macro-
economic instability.
12
Table 1 provides summary information on the
Turkish economy for the whole (1963–1999) and
two sub-periods, namely, the inward-oriented period
(1963–1979) and the outward-oriented period (1980–
1999). During the 1963–1979 period, Turkey followed
a state-led inward-oriented growth strategy by
following import substitution policies and economy-
wide planning by the State Planning Organization
(SPO). Besides the trade restrictions and financial
repression policies (e.g. regulated interest rates), the
state made use of heavy public investment, especially
in the manufacturing sector, to promote industrial-
ization and economic development.
During the inward-oriented period, Turkey enjoyed
a high rate of growth (real GNP grew at an annual
average rate of 5.1%) and a rapid rate of capita accu-
mulation. While real private investment
13
increased
at an average annual rate of 7%, public investment
increased at an average annual rate of 9.7%.
12
See, for example, Aricanli and Rodrik (1990), Celasun and Rodrik (1989), Ekinci (1990, 2000), Metin-Ozcan et al. (2001)
and Ozatay (1999) among others for more detailed analyses and discussions on these and related issues on the Turkish
economy.
13
It should be noted that investment series have been revised several times in Turkey during the last two decades (see, for
example, Conway, 1990). We reported our results in this section and elsewhere based upon the most recent series of the SPO
(see the data appendix for more detail).
Table 1. Selected indicators of the Turkish economy,
1963–1999
1963–99 1963–79 1980-99
I. Output and capital formation
I.A Annual average growth rate*
Real GNP 4.4 5.1 4.2
Real private fixed investment 5.8 7.0 6.1
Real public fixed investment 5.0 9.7 1.6
Real public fixed core
infrastructural investment
5.9 10.8 2.7
Real public fixed non-core
infrastructural investment
4.3 9.0 0.8
I.B Composition of public
investment**
Core infrastructural
investment (as % of total)
44.4 37.3 50.5
Non-core infrastructural
investment (as % of total)
55.6 62.7 49.5
II. Macroeconomic instability**
Macroeconomic instability
index (MII)
0.326 0.104 0.514
Inflation rate,*** % 41.7 18.1 61.8
* Annual average growth rate of variable X(e.g. real GNP)
is calculated by using the following formula: Annual
average growth rate of X(%) ¼((X
tþn
/X
t
)
1/n
1) 100,
where X
t
¼the value of Xin year t(initial year), X
tþn
¼
value of Xin year tþn(final year) and n¼no of
‘periods’ (years).
** Simple period average.
*** Percentage change in GNP deflator.
Source: See the data appendix
Macroeconomic instability in public/private capital accumulation and growth 241
During the 1960s the macroeconomic environment
was quite stable.
14
However, mainly due to foreign
exchange difficulties of the late 1960s, in 1970
Turkey introduced an IMF-based stabilization pack-
age, which involved a maxi devaluation. From 1973
to 1977, Turkey experienced an unprecedented
growth in investment, led by public sector invest-
ment, mainly in manufacturing and transportation.
Both public and private investment grew at high
rates of 20.4% and 8.4%, respectively, during this
period. However, macroeconomic instability signifi-
cantly increased during the mid-1970s mainly due to
the deterioration of the fiscal balances and the exces-
sive reliance on foreign borrowing. By the late 1970s
Turkey reached a state where it could no longer
service even the short-term debts and experienced
severe economic crisis.
In 1980, Turkey took a crucial decision to switch
its overall economic strategy from inward-oriented
growth to outward-oriented growth. The 1980
programme had both stabilization and structural
aspects (e.g. trade and financial liberalization), and
was strongly backed by IMF, World Bank and
OECD consortium. The role of state has crucially
changed with this programme; for example, the
state changed its investment strategy from manufac-
turing to infrastructure. During the outward-oriented
period of 1980–1999, real GNP of the Turkish
economy grew at an average annual rate of 4.2%.
Compared to the inward-oriented period, this
performance doesn’t seem impressive. Furthermore,
real GNP fluctuated less during the inward-oriented
period compared to the outward-oriented period.
Nevertheless, the economic growth rate was higher
during the 1980s (5.2% per year) compared to the
1990s (3.2% per year). Private sector’s capital
formation performance was better compared to
public sector’s during the outward-oriented period.
Real private (public) investment grew at an
average annual rate of 6.1% (1.6%), from 1980 to
1999. As we mentioned before, the crucial change
in this period was the changing role of the state
in the investment process. The share of core public
infrastructural (transport þcommunication þenergy)
investment in total public investment rose from
37.3% in the inward-oriented period to 50.5% in
the outward-oriented period. Nevertheless, while
private investment-GNP ratio (in current prices)
rose from 12.8% in the 1980s to 18.1% in the
1990s, public investment-GNP ratio dropped from
8.8% in the 1980s to 6.2% in the 1990s. The main
reason behind this fall is the rising macroeconomic
instability after the late 1980s, which has seriously
lowered the fiscal and political ‘ability’ of govern-
ments for making necessary investments, especially
infrastructural investments, due to budgetary
pressures.
15
Generally speaking, macroeconomic instability
has steadily increased since the mid-1970s and has
become a chronic problem for the Turkish economy.
During the early 1980s Turkey was successful in low-
ering the macroeconomic instability inherited from
the economic crisis of the late 1970s. Inflation rate
and macroeconomic instability index (MII) fell from
89.6% and 0.520 points in 1980 to 26% and 0.317
points in 1983, respectively. Similarly, macroeco-
nomic management was quite good during the
mid-1980s. However, starting from the late 1980s
macroeconomic instability rose again, mainly due to
the political factors (e.g. political instability and
polarization) and related populist and myopic
policies,
16
and associated problems of public sector
imbalances. Until late 1993, Turkey had managed
to maintain the populist and myopic policies mainly
with the help of capital inflows.
17
However, the cost
of this strategy was very high, real interest rate on
domestic debt had increased steadily during the
early 1990s and this further deteriorated the fiscal
balances; for instance, domestic interest payments
out of consolidated budget (as a percentage of
GNP) almost doubled from 1990 to 1993.
Turkey experienced a very severe financial crisis
in early 1994 mainly due to unsustainable fiscal
balances, the collapse of the domestic debt market,
monetization and the expectations of further moneti-
zation. Real GNP contracted by 6.1% during 1994,
which is the peak rate of contraction of the Turkish
economy over the 1963–1999 period. Similarly, real
public investment fell dramatically by about 40%,
from 1993 to 1994. This is a solid evidence of the
negative effect of macroeconomic instability on fiscal
‘ability’ of governments for making investment. Real
private investment, however, contracted only moder-
ately (about 5%). Both inflation and MII peaked
14
During the 1960s average inflation rate was only 5.2% compared to the 1970s (27%), the 1980s (50.4%) and the 1990s
(73.2%). Furthermore, macroeconomic instability index was 0.04 points and it was the lowest decade average compared to
other decades (see the data appendix for more detail on macroeconomic instability index).
15
In line with this argument, Conway (1990:82) stated that ‘[r]eal public investment growth appears to have [negatively]
responded to budgetary pressures’ in Turkey.
16
See, for example, Ozatay (1999) and Akyurek (1999) for more detail and empirical evidence.
17
Turkish Lira became fully convertible and capital account was fully-liberalized in 1989.
242 M. Ismihan et al.
in 1994, inflation rate was 107.3% and MII was
0.842 points in 1994. Furthermore, Turkish Lira
depreciated by more than 150% against US$ in
1994. In mid-1994, Turkey adopted an IMF-based
stand-by agreement, and managed to cool-down the
severe economic crisis. Inflation rate and MII fell
from 107.3% and 0.842 points in 1994 to 87.2%
and 0.563 points in 1995, respectively. However,
macroeconomic instability has continued until the
late 1990s, mainly due to reluctance of governments
(e.g. to avoid negative political consequences) to take
the necessary painful measures; in other words,
governments delayed stabilization.
18
During this
period, public sector balances were unsustainable
due to reliance on the domestic borrowing (e.g. real
interest rate on domestic debt almost doubled from
1994 to 1999). In December 1999, Turkey signed
a three-year IMF-based stand-by agreement, which
mainly aimed to solve the public sector imbalances.
19
Unfortunately, this programme had failed in early
2001 due to a major economic crisis (real GNP con-
tracted by 9.4% during 2001) and Turkey signed
another programme backed by the IMF and the
World Bank, which is still being implemented today.
III. Empirical Results
Data and unit root tests
The data used in this study are Turkish annual data
from 1963 to 1999. The sample period is determined
by the availability of official investment data.
Figures 1–3 show the time plots of the logarithms
of real GNP (y), real private fixed investment (ip),
real public fixed investment (ig), real public fixed core
infrastructural investment (igi) and macroeconomic
instability index (mii). The data appendix provides
18
See, for example, Veiga (2000) for well-documented reasons for delayed stabilizations.
19
See Ekinci (2000) and the references therein for a thorough overview of these problems and extensive assessment of the
aspects of this program.
9
10
11
12
13
14
15
1963 1967 1971 1975 1979 1983 1987 1991 1995 1999
ip
ig
igi
Fig. 2. Time plots of the logarithms of real fixed private investment (ip),
real fixed public investment (ig) and real fixed public core infrastructural investment (igi), 1963–1999
13.6
14
14.4
14.8
15.2
15.6
1963 1967 1971 1975 1979 1983 1987 1991 1995 1999
Fig. 1. Time plot of the logarithm of real GNP ( y), 1963–1999
Macroeconomic instability in public/private capital accumulation and growth 243
the details on the definitions and the sources of
the data.
Visual inspection of the data suggests that all these
series have a unit root(s). However, we also provide
the formal unit root test results in Table 2. As
expected, for all the variables the null hypothesis
of a unit root are not rejected at the 95% level.
Furthermore, the null hypothesis of a unit root
for the first differences of all variables are rejected
at the 95% level. Therefore, all the variables under
consideration are integrated of order 1. However, it is
well-documented that if we neglect the level and/or
the trend shift (e.g. due to a structural break) in the
unit root tests we could possibly obtain ‘spurious’
unit root results (Franses, 1998). Therefore, since
we know the break date quite well from the evidence
Table 2. Unit root tests
Variables
ADF test Perron test
Level First difference Innovation
outlier
model
f
Without
trend
a
With
trend
b
Without
trend
a
y1.0696 (0)
c
2.4769 (0) 4.9665 (0)*
d
2.6439(0)
h
ip 1.2921 (1) 3.0168 (3) 3.3808 (0)* 2.0148(1)
ig 2.2886 (3) 2.4310 (1) 4.5756 (0)* 4.4852(2)
g
igi 1.9448 (0) 1.8378 (0) 4.4798 (1)* 3.8352(1)
mii 1.2578 (1) —
e
8.0355 (0)* —
e
Notes:
a
ADF regressions include an intercept but not a linear trend (see, Pesaran and Pesaran, 1997, p. 53).
b
ADF regressions include both an intercept and a linear trend (see, Pesaran and Pesaran 1997, p. 53).
c
Numbers in parentheses are the order of augmentations ( p*) chosen by the Akaike Information Criterion
(AIC). Note that unit root test results also hold when p*(s) are chosen by Schwarz Bayesian Criterion
(SBC). Due to a size-power trade-off in the determination of the order of augmentation (p) of ADF tests,
we choose to select p* by AIC, which is a common practice in the applied works (see, Pesaran and Pesaran,
1997, p. 213). Therefore, in line with Pesaran and Pesaran (1997:213), first we estimated ADF regressions
for p¼0top¼4 and selected the order of augmentation ( p*) based on AIC. Then, we performed the ADF
tests (see the text). Note that the same sample period (1969–1999) is used in calculations.
d
An asterisk (*) represents the rejection of the unit root null hypothesis at the 95% level (MacKinnon,
1991, Table 1).
e
Since MII is bounded between 0 and 1 due to its construction (see the data appendix), we did not include
trend for mii (see, for example, Ahmet and Rogers, 2000). Furthermore, linear trend in mii is not mean-
ingful from the economic point of view.
f
This model is within innovation outlier framework and allows for both a change in the level and trend
(see, Franses, 1998, p. 150–1, for this test).
g
ig rejects the null hypothesis at the 95% level but not at the 99% level (see Franses, 1998, Table 6.6, for
critical values).
h
Numbers in parentheses are the order of augmentations ( p*) chosen by the Akaike Information Criterion
(AIC). Note that test results also hold when p*(s) are chosen by Schwarz Bayesian Criterion (SBC) or if we
just use the same p*(s) of the third column. We use the same procedure as in note (c) for determining the
order of augmentation ( p*). Note that the same sample period (1969–1999) is used in calculations.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
1963 1967 1971 1975 1979 1983 1987 1991 1995 1999
Fig. 3. Time plot of the logarithm of macroeconomic instability index (mii), 1963–1999
244 M. Ismihan et al.
reported in Section II, which is 1980, we also
performed a Perron test, which allows for a change
in the level and the trend. As can be seen from
Table 2 all variables except ig cannot reject the null
hypothesis of a unit root at the 95% level. However,
ig cannot reject the null at the 99% level. We
conclude that all the variables are integrated of
order 1, which is a pre-condition for the cointegration
analysis.
Cointegration analysis
This section provides the cointegration analysis for
investigating the long-run relationship(s) between
public investment, private investment, macro-
economic instability and output in Turkey over the
period 1963–1999. We provide results for total public
investment and its infrastructural components. In
line with this, we form two cointegration systems:
System 1 [ip,y,ig,mii] and System 2 [ip,y,igi,mii].
We used the Johansen multivariate technique in our
cointegration analyses (Johansen, 1988). Following
Hendry and Juselius (2001) and Pesaran and Smith
(1998) among others, we first performed a cointegra-
tion analysis with constant term entering unrestric-
tively but with a trend term restricted to lie in the
cointegration space. However, the trend term was
found to be insignificant in the cointegration
relation(s);
20
hence, following Hendry and Juselius
(2001) we performed a cointegration analysis with
the constant term entering unrestrictively and
without the trend term. Following Juselius (2001)
among others, we also included a step (intervention)
dummy in each cointegration system to account for
the structural break of 1980. It entered restrictively to
the cointegration space. However, this step dummy
was found to be insignificant in the cointegration
relation.
21
This might be because the 1980 structural
break could have affected several variables similarly
22
and hence caused the intervention effects to cancel
out (Hendry and Juselius, 2001). Therefore, we did
not include this dummy variable in our cointegration
analysis.
Analysis using total public investment
First we form the system with variables [ip,y,ig,mii]
and test for cointegration. Table 3 provides cointe-
gration result for this system with lag length of
VAR ¼1.
23
We also included an impulse dummy
for 1994 (D94) unrestrictively in our cointegration
analysis.
24
The trace and max statistics suggest one
cointegration relation,
25
which seems to be the follow-
ing simple long-run private investment relation:
26
ip ¼3:24y4:67mii 0:29ig ð1Þ
This equation suggests that private investment is
positively affected by output, negatively affected by
macroeconomic instability and public investment for
the period under study. These results are consistent
with theory and the descriptive analysis of the
Turkish economy provided in Section II. The stan-
dard errors of the cointegration vector in Table 3
show that all variables except ig are statistically
significant. We also formally tested the significance
of the variables by the exclusion test, which confirm
the previous results (Table 3). However, if we con-
sider the cointegration result for System 1 without
D94, we have the following simple long-run private
investment relation:
ip ¼3:44y5mii 0:38ig ð2Þ
Equations 1 and 2 are quite similar to each other
but when we examine standard errors (not reported)
all variables seem to be significant in Equation 2.
20
Underlying trends of the variables under consideration possibly cancelled out in the cointegration relation (Hendry
and Juselius, 2001).
21
Similarly, we also include a step dummy in each cointegration system to account for the effect of the full-financial
liberalization in 1989. This step dummy is also found to be insignificant in the cointegration relation.
22
As can be seen from Figs 1–3, there is some visual evidence on this.
23
Note that the lag length of the VAR for each system is determined by the Schwarz Bayesian Criterion.
24
When we examine the regression results for each equation in VAR(1) model, ig equation has non-normal residuals.
This is clearly evident in the residual plot of that equation in which 1994 is an outlying observation. (Note that this is
consistent with the evidence in Section II.) Therefore, following Hendry and Juselius (2001), we include impulse dummy for
1994 (D94) in our cointegration analysis for System 1. After including D94 in VAR(1) unrestrictively, all equations have
normal distributions and none of them show autocorrelation and heteroscedasticity (see Pesaran and Pesaran, 1997
for detailed information on these tests). Due to the same considerations we also include an impulse dummy for 1994 in
our cointegration analysis for System 2.
25
It should be noted here that the trace and max statistics for System 1 without the impulse dummy (D94) also suggest one
cointegration relation. Therefore, our results are not an artefact of D94.
26
Note that this is a simple investment relation since other determinants of investment, such as real interest rate, are absent
in Equation 1 due to the purpose of the study, or data availability and/or limitations of cointegrated VAR analysis
with relatively small sample size (Pesaran and Pesaran, 1997).
Macroeconomic instability in public/private capital accumulation and growth 245
The result of significance (exclusion) test provides a
p-value of 0.059 for ig (yand mii both have
p-value ¼0); therefore, there is some (but marginal)
evidence of long-run crowding-out effect.
Analysis using infrastructural public investment
We now provide the cointegration analysis for an
alternative version by substituting public core infra-
structure investment (igi) for total public investment
(ig). Table 4 provides the cointegration results for
this system with lag length of VAR ¼1.
27
The
evidence favours one cointegration relation and it is
also interpreted as the private investment relation.
The crucially different result in this system is that
even though igi has a negative effect (crowding-out)
on ip, its coefficient is not significant, as indicated
by the long-run exclusion test. Furthermore, this is
also the case without D94.
28
After imposing the long-
run exclusion restriction (and the normalization
restriction), the investment relation becomes:
ip ¼3:15y5:20mii ð3Þ
This simple investment equation suggests that
private investment is positively affected by output
and negatively affected by macroeconomic instability.
Again, these results are consistent with theory and the
descriptive analysis of the Turkish economy provided
in Section II.
Finally, we would like to note that our main results
in this section also hold when we use inflation rate
as a proxy
29
for macroeconomic instability. This is
provided in Ismihan (2003). In the next section,
we examine the dynamic effects of a rise in (i.e. a
shock on) a given variable on all the other variables
in the system, by using impulse response analysis.
Impulse response analysis
Lutkepohl and Reimers (1992) and Pesaran and
Pesaran (1997) among others pointed out the impor-
tance of impulse response analysis in cointegrated
systems. In this section, we provide the generalized
impulse response (IR) functions
30
to examine the
dynamic effects; that is, short- and medium-run
effects of a shock on a given variable on all the
27
Due to similar considerations D94 enters unrestrictively to the cointegration analysis.
28
Note that, in System 2 without D94, the private investment relation becomes: ip ¼3.21y0.23igi 4.45mii. Furthermore,
all variables are significant except igi (p-value ¼0.11).
29
According to Fischer (1993b:487), inflation rate is the best single indicator [of policy-induced macroeconomic instability]
and ‘serve as an overall ability of government to manage the economy’.
30
We prefer to use generalized IR functions since, unlike to the orthogonalized IR functions, the generalized IR functions
do not depend on the ordering of the variables within the system (Pesaran and Shin, 1998).
Table 3. Cointegration analysis of System 1
Tests of cointegration rank
Eigenvalues 0.60355 0.21840 0.12638 0.03471
Null hypotheses r¼0r1r2r3
Max statistic 33.31 8.87 4.86 1.27
95% critical value
a
27.07 20.97 14.07 3.76
Trace statistic 48.31 15.0 6.13 1.27
95% critical value
a
47.21 29.68 15.41 3.76
Cointegration results (r¼1)
ip y mii ig
(0)
b
13.2364
(0.48324)
c
4.6669
(1.2104)
0.289
(0.18182)
(0)
d
0.0919 0.0161 0.14499 0.0280
Hypotheses tests
e
X
2
(u)up-value
Test of significance of y17.3 1 0.00
Test of significance of mii 24.3 1 0.00
Test of significance of ig 2.2 1 0.14
Notes:
a
Critical values are from Osterwald-Lenum (1992, Table 1).
b
Standardized eigenvector.
c
Asymptotic standard errors are in parentheses.
d
Adjustment coefficients.
e
Test of long-run exclusion (see, Hendry and Juselius, 2001).
246 M. Ismihan et al.
other variables in the system. Below, we present this
analysis for the System 1 only since the impulse
response analysis of System 2 is quite similar to
that of System 1.
31
In order to assess the dynamic effects of a rise in
macroeconomic instability on other variables in the
system, we examine the generalized IRs to a positive
unit [one standard error (SE)] shock in macroeco-
nomic instability (mii) equation. These generalized
IRs are provided in panel (a) of Fig. 4. As expected,
short- and medium-run responses are negative. That
is, private investment, public investment and output
are negatively and permanently affected from a rise in
macroeconomic instability. However, the effect on
private investment was larger compared to the effect
on output. The effect on public investment was also
large. This is probably due to a negative effect of a rise
in macroeconomic instability on the public resources.
Hence, governments tend to cut public investment
rather than current (or populist) spending in the
case of fiscal stringency since it is politically easer to
cut public investment than current spending. All these
results are consistent with our analysis in Section II.
Finally, as can be seen from panel (a) of Fig. 4, the
impact effects of a rise in macroeconomic instability
on both private and public investments are smaller
compared to the medium-term effects; that is, the
effect of an increase in macroeconomic instability
has an accelerating negative effect on investment,
especially, on private investment.
We next examine the dynamic effects of a rise in
(a positive unit shock to) public investment on other
variables in the cointegration system. As can be seen
from the panel (b) of Fig. 4, responses of private
investment and output are positive; however, the
response of the former is much larger. These results
31
There is only one considerable difference. In the System 2, infrastructural public investment is more seriously affected from
macroeconomic instability (shock) compared to total public investment in the System 1. This is consistent with the observa-
tion that Turkey failed to make necessary infrastructural investment due to the constraining effects of macroeconomic
instability (and associated fiscal problems) on the incumbent goverments’ budgets during the late 1990s, and hence experi-
enced infrastructural bottlenecks, such as energy bottlenecks, during the late 1990s and the early 2000s. The results of
generalized IR analysis for the System 2 can be requested from the first author.
Table 4. Cointegration analysis of System 2
Tests of cointegration rank
Eigenvalues 0.638 0.17509 0.10778 0.042736
Null hypotheses r¼0r1r2r3
Max statistic 36.68 6.93 4.11 1.57
95% critical value
a
27.07 20.97 14.07 3.76
Trace statistic 49.19 12.61 5.68 1.57
95% critical value
a
47.21 29.68 15.41 3.76
Cointegration results (r¼1)
ip y mii igi
(0)
b
13.1551
(0.39148)
c
4.3892
(1.0593)
0.20909
(0.12639)
(0)
d
0.0905 0.0185 0.1499 0.1393
Hypotheses tests
e
X
2
(u)up-value
Test of significance of y17.7 1 0.00
Test of significance of mii 29.5 1 0.00
Test of significance of igi 2 1 0.16
Restricted cointegration analysis
ip y mii
(0)
b
13.1539
(0.50615)
c
5.2016
(1.4605)
(0)
d
0.0548 0.0198 0.1290
Notes:
a
Critical values are from Osterwald-Lenum (1992, Table 1).
b
Standardized eigenvector.
c
Asymptotic standard errors are in parentheses.
d
Adjustment coefficients.
e
Test of long-run exclusion (See, Hendry and Juselius, 2001).
Macroeconomic instability in public/private capital accumulation and growth 247
suggest a complementarity between public and
private investment in the short and the medium-run.
Note that, public and private investment moved or
‘wandered’ together, implying complementarity, until
the late 1970s (see Fig. 2) but after the late 1970s
this relationship started to shatter possibly due to a
negative effect of chronic macroeconomic instability
on both private and public investment but via dif-
ferent channels (and with different magnitudes), as we
mentioned in the introduction section. Furthermore,
this relationship seems to be reversed after the late
1980s, possibly due to rising macroeconomic instabil-
ity and associated deterioration in fiscal balances,
which has affected both public and private invest-
ment. Therefore, in the case of Turkey, chronic and
increasing macroeconomic instability and associated
fiscal problems seems to shatter or even reverse the
complementarity between public and private invest-
ment in the long run.
32
Furthermore, response of macroeconomic instabil-
ity to a rise in public investment is initially negative
but over the medium term it diminishes towards zero.
This result suggests that the rise in public investment
does not contribute to the macroeconomic instability
over the short and the medium term. One potential
explanation for this seemingly counterintuitive result
is that an increase in public investment in case of
chronic macroeconomic instability and associated
fiscal stringency signals a decisive change in fiscal
policy, e.g. from populist to productive spending,
and could have immediate political credibility
and expectation effects which will lower expected
32
Metin-Ozcan et al. (2001) pointed out the negative effects of domestic debt financing (due to a rise in interest rates), after
1989, on the crowding-in effects of public investment on private investment and output in Turkey.
(b) Generalized IR(s) to one S.E. shock in the equation for ig
(a) Generalized IR(s) to one S.E. shock in the equation for mii
(c) Generalized IR(s) to one S.E. shock in the equation for y
-0.09
-0.08
-0.07
-0.06
-0.05
-0.04
-0.03
-0.02
-0.01
0
0 2 4 6 8 101214
Horizon (Years)
ip
y
ig
-0.03
-0.02
-0.01
0
0.01
0.02
0.03
0.04
0.05
0
2
4
6
8
10
12
14
Horizon (Years)
ip
y
mii
-0.03
-0.01
0.01
0.03
0.05
0.07
0.09
0.11
0
2
4
6
8
10
12
14
Horizon (Years)
ip
ig
mii
Fig. 4. Generalized IR(s) to one SE shock in the equation for (a) mii, (b) ig, and (c) y
248 M. Ismihan et al.
inflation, inflation risk on borrowing, and hence
macroeconomic instability (see, for example, Alesina
et al., 1998; and Perotti, 1999, and the references
therein for similar arguments).
33
Furthermore, public
(and also private) investment affects both the demand
and the supply-side of the economy. The rise in public
investment increases expenditures of government but
the rise in public investment also increases national
income and output due to its dual role. The rise in
national income will, in turn, increase the revenues of
government, e.g. tax and seigniorage revenue, and
help to reduce the fiscal deficit and inflation over
some period, but with diminishing effects.
Finally, we examine the dynamic effects of a rise in
(a positive unit shock to) output on all the other
variables. As panel (c) of Fig. 4 reveals, the short and
the medium-run responses of private investment to a
rise in output is positive as expected. Similarly, the
response of public investment is also positive. As
panel (c) of Fig. 4 makes clear, the impact effects
of a rise in output on both private and public invest-
ment are only slightly different from medium-term
effects. The short-run responses of macroeconomic
instability is negative (e.g. due to the positive effect
of a rise in output on revenues of government and,
hence, on budget deficit and inflation); however
over the medium term this response approaches
towards zero.
34
IV. Conclusions and Policy Implications
This study investigated the empirical relationships
between macroeconomic instability, public and
private capital accumulation and growth in Turkey
over the period 1963–1999. The main conclusion is
that the chronic macroeconomic instability of the
Turkish economy has seriously affected the capital
formation and growth. Even though we found some
evidence of crowding-out effect of total public invest-
ment on private investment, there was no significant
effect of infrastructural public investment on private
investment in the long run. However, we found
some evidence of complementarity between private
and public investment (and with its infrastructural
component) over the short and medium-run. Our
results suggest that the chronic macroeconomic
instability seems to become a serious impediment to
the public investment, and has shattered, or even
reversed, the long-run complementarity. This result
may also shed some light on the ambiguity concern-
ing the empirical evidence on the complementarity
(crowding-in) effect for the Turkish economy.
Moreover, given the evidence on the short- and
medium-term complementarity between public and
private investment, these results imply that macro-
economic instability has been very costly in terms of
private capital accumulation and economic growth
during the chronic instability episode of Turkey.
The policy implications are straightforward when
we consider these results. Generally speaking, over
the last twenty-five years, governments in Turkey
either delayed or did not continue with the stabiliza-
tion programmes. The barriers to stabilization, such
as political instability and polarization, are well docu-
mented in Veiga (2000) and Drazen (2000) among
others. Nevertheless, as this study shows, macroeco-
nomic instability has an adverse impact on capital
accumulation and economic growth in Turkey.
Therefore, the government should continue the cur-
rent stabilization programme to restore macroeco-
nomic stability, as soon as possible. This is the first
policy implication. The second policy implication
is that policy makers have to be careful in their deci-
sions concerning the components of public spending
that would bear the burden of fiscal adjustment in
the process of the restoration of macroeconomic
stability. If government reduces public capital spend-
ing (especially, infrastructural spending) instead of
current and populist spending; then, this would
harm capital accumulation, economic growth and
development.
35
Furthermore, as opposed to the
conventional view that fiscal adjustments are reces-
sionary, there is growing evidence that some types of
fiscal adjustments may be expansionary (Perotti,
1999; Alesina et al., 1998). This new line of research
argues that composition of adjustment matters. For
example, fiscal adjustments based on current spend-
ing may be expansionary under certain conditions,
such as high level of debt to GNP ratio (Perotti,
1999).
In sum, Turkish experience has shown that macro-
economic instability not only deters economic growth
but it may also reverse the complementarity between
public and private investment in the long-run. In
order to shed more light on this result, this study
33
According to Perotti (1999:1400), ‘... in times of fiscal stress the economy’s response to fiscal shocks changes qualitatively.’
34
The dynamic effects of private investment shock are similar to public investment shock on all the other variables (simply
replace ip with ig in panel (b) of Fig. 4); therefore, it is not separately explained.
35
See, for example, World Bank (1994) and the references therein, for the crucial importance and the multi-dimensional roles
of infrastructure in economic development.
Macroeconomic instability in public/private capital accumulation and growth 249
can be further extended to other developing countries
suffering from chronic instability like Turkey and this
is left for future research.
Acknowledgements
This paper is based on Mustafa Ismihan’s PhD thesis
(Ismihan, 2003) prepared under the co-supervision of
Aysit Tansel and Kivilcim Metin-Ozcan in the
Department of Economics, METU. The first author
would like to thank Merih Celasun and Nazim Ekinci
for helpful comments and discussions. Substantial
revisions of this paper were made during Mustafa
Ismihan’s stay at the University of York as a
Visiting Scholar; he thanks the Turkish Academy of
Sciences (TUBA) for financial support and the
Department of Economics and Related Studies of
the University of York for their hospitality. Our
special thanks go to Gulcin Ozkan whose comments
and discussions have greatly improved this paper.
We are also grateful to Karim Abadir, Fikret
Senses and Fatma M. Utku-Ismihan for helpful
comments. The first draft of this paper was presented
in the ERC/METU International Conference in
Economics V, in September 2001. We would also
like to thank conference participants and, especially
Yakup Kepenek, for helpful comments. The usual
disclaimers apply.
References
Agenor, P. R. and Montiel, P. J. (1996) Development
Macroeconomics, Princeton University Press,
Princeton, NJ.
Ahmed, S. and Rogers, J. (2000) Inflation and the great
ratios: long term evidence from the U.S., Journal of
Monetary Economics,45, 3–35.
Akyurek, C. (1999) Fiscal policy, seigniorage, and myopic
governments: theory and evidence from Turkey, Yapi
Kredi Economic Review,10(1), 45–61.
Alesina, A., Perotti, R. and Tavares, J. (1998) Political
economy of fiscal adjustments, Brookings Papers on
Economic Activity, 197–266.
Anand, R., Chhibber, A. and Van Wijnbergen, S. (1990)
External balance and growth in Turkey: can they be
reconciled?, in The Political Economy of Turkey: Debt,
Adjustment, and Sustainability (Eds) T. Aricanli and
D. Rodrik, Macmillan, London.
Aricanli, T. and Rodrik, D. (1990) The Political Economy
of Turkey:Debt,Adjustment,and Sustainability,
Macmillan, London.
Aschauer, D. A. (1989a) Is public expenditure productive?,
Journal of Monetary Economics,23, 177–200.
Aschauer, D. A. (1989b) Does public capital crowd-out
private capital?, Journal of Monetary Economics,
24, 171–88.
Barro, R. J. (1990) Government spending in a simple model
of endogenous growth, Journal of Political Economy,
98, S103–25.
Bleaney, M. F. (1996) Macroeconomic stability, investment
and growth in developing countries, Journal of
Development Economics,48, 461–77.
Blejer, M. I. and Khan, M. S. (1984) Government policy
and private investment in developing countries, IMF
Staff Papers,31, 379–403.
Cardosa, E. (1993) Private investment in Latin America,
Economic Development and Cultural Change,41,
833–48.
Celasun, M. and Rodrik, D. (1989) Debt, adjustment and
growth: Turkey, in Developing Country Debt and
Economic Performance,Country Studies (Eds) J. Sachs
and S. M. Collins, University of Chicago Press, Chicago.
Celasun, M. and Tansel, A. (1993) Distributional effects
and saving-investment behavior in a liberalizing
economy: the case of Turkey, METU Studies in
Development,20, 269–98.
Conway, P. (1990) The record on private investment in
Turkey, in The Political Economy of Turkey:Debt,
Adjustment,and Sustainability (Eds) T. Aricanli and
D. Rodrik, Macmillan, London.
De Haan, J., Sturm, J. E. and Sikken, B. J. (1996) Govern-
ment capital formation: explaining the decline,
Weltwirtschaftliches Archiv,132(1), 55–74.
Drazen, A. (2000) Political Economy in Macroeconomics,
Princeton University Press, Princeton, NJ.
Easterly, W. and Rebelo, S. (1993) Fiscal policy and
economic growth: an empirical investigation, Journal
of Monetary Economics,32, 458–93.
Ekinci, N. (1990) Macroeconomic developments in Turkey:
1980-1988, METU Studies in Development,17(1–2),
73–114.
Ekinci, N. K. (2000) The IMF and reforming the public
sector in Turkey, METU Working Papers in
Economics, No. 00/04, METU, Ankara.
Fischer, S. (1993a) Does macroeconomic policy matter?
Evidence from developing countries, ICEG Occasional
Papers, No. 27.
Fischer, S. (1993b) The role of macroeconomic factors in
growth, Journal of Monetary Economics,32, 485–512.
Franses, P. H. (1998) Time Series Models For Business and
Economic Forecasting, Cambridge University Press,
Cambridge.
Ghali, K. H. (1998) Public investment and private capital
formation in a vector error-correction model of
growth, Applied Economics,30, 837–44.
Hendry, F. D. and Juselius, K. (2001) Explaining
cointegration analysis: part II, The Energy Journal,
22(1), 75–120.
Ismihan, M. (2003) The role of politics and instability
on public spending dynamics and macroeconomic
performance: theory and evidence from Turkey,
Unpublished PhD thesis, METU, Ankara, December
2003.
Johansen, S. (1988) Statistical analysis of cointegration
vectors, Journal of Economic Dynamics and Control,
12, 231–54.
Juselius, K. (2001) Unit roots and the demand for cigarettes
in Turkey: pitfalls and possibilities, Discussion
paper, No.01-02, Institute of Economics, University
of Copenhagen.
Lutkepohl, H. and Reimers, H. E. (1992) Impulse response
analysis in cointegrated systems, Journal of Economic
Dynamics and Control,16(1), 53–78.
MacKinnon, J. G. (1991) Critical values for cointegration
tests, in Long-run Economic Relationships:Readings in
250 M. Ismihan et al.
Cointegration (Eds) R. F. Engle and C. W. J. Granger,
Oxford University Press, Oxford.
Metin-Ozcan, K., Voyvoda, E. and Yeldan, E. (2001)
Dynamics of macroeconomic adjustment in a global-
ized developing economy: growth, accumulation and
distribution, Turkey 1969–1998, Canadian Journal of
Development Studies,22(1), 219–53.
Mittnik, S. and Neumann, T. (2001) Dynamic effects of
public investment: vector autoregressive evidence
from six industrialized countries, Empirical Economics,
26, 429–46.
Osterwald-Lenum, M. (1992) A note with quantiles of the
asymptotic distribution of the maximum likelihood
cointegration rank test statistics, Oxford Bulletin of
Economics and Statistics,54(3), 461–71.
Ozatay, F. (1999) Populist policies and the role of economic
institutions in the performance of the Turkish
economy, Yapi Kredi Economic Review,10(1), 13–26.
Pereira, A. (2000) Is all public capital created equal?,
Review of Economics and Statistics,82, 513–18.
Perotti, R. (1999) Fiscal policy in good times and bad,
Quarterly Journal of Economics,114, 1399–436.
Persson, T. and Tabellini, G. (2000) Political Economics:
Explaining Economic Policy, The MIT Press,
Cambridge, MA.
Pesaran, M. H. and Pesaran, B. (1997) Working with
Microfit 4.0:Interactive Econometric Analysis, Oxford
University Press, Oxford.
Pesaran, M. H. and Shin, Y. (1998) Impulse response
analysis in linear multivariate models, Economic
Letters,58, 17–29.
Pesaran, M. H. and Smith, R. P. (1998) Stuctural analysis
of cointegrating VARs, Journal of Economic Surveys,
12, 471–505.
State Planning Organization (SPO) (1997) Economic and
Social Indicators (1950–1998), SPO, Ankara.
State Planning Organization (SPO) (2001) Main Economic
Indicators,May 2001, SPO, Ankara.
Sturm, J., Kuper, G. H. and De Haan J. (1998) Modeling
government investment and economic growth on a
macro level: a review, in Market Behaviour and
Macroeconomic Modelling (Eds) S. Brakman, H. van
Ees and S. K. Kuipers, MacMillan, London.
United Nations Development Programme (UNDP) (1992)
Human Development Report 1992, UNDP, New York.
Uygur, E. (1995) Recent estimates of investment and
private investment behaviour in Turkey, in Investment
and The Labour Market in Turkey:Proceedings of A
Seminar Held in Ankara 7 December 1995 (Ed.)
T. Bulutay, SIS, Ankara.
Veiga, F. J. (2000) Delays of inflation stabilizations,
Economics and Politics,12, 275–95.
World Bank (1994) World Development Report 1994:
Infrastructure for Development, Oxford University
Press for the World Bank, New York.
Data Appendix
yis the (natural) logarithm of real GNP in 1994
prices (billion TL).
Source:SPO (1997) and SPO (2001).
ip is the (natural) logarithm of real private fixed
investments in 1994 prices (billion TL). Nominal
private fixed investment series were deflated by
private fixed investment deflator series.
Source:SPO (1997) and SPO (2001). Note:deflators
were provided by the SPO.
ig is the (natural) logarithm of real public fixed
investments in 1994 prices (billion TL). Nominal
public fixed investment series were deflated by public
fixed investment deflator series.
Source:SPO (1997) and SPO (2001). Note:deflators
were provided by the SPO.
igi is the (natural) logarithm of real public fixed core
infrastructural investments in 1994 prices (billion
TL). Nominal sectoral public fixed investment series
were deflated by relevant sectoral public fixed invest-
ment deflator series. In line with Ekinci (1990) among
others we define core infrastructural investment as
the total of the public energy, transportation and
communication sectors’ fixed investments. See World
Bank (1994:2) for broad definition of infrastructure.
Source:SPO (1997) and SPO (2001). Note:Sectoral
deflators were provided by the SPO.
mii is the (natural) logarithm of the macroeconomic
instability index (MII), i.e. mii ¼ln(1 þMII). This
index is calculated by using human development
index (HDI) methodology (UNDP, 1992) and it is
based on macroeconomic instability indicators, such
as inflation rate, public deficit to GNP ratio, external
debt to GNP ratio and change in exchange rate,
identified by previous researchers (see, for example,
Fischer, 1993a, 1993b; Bleaney, 1996). It is a simple
average of the four sub-indices obtained from these
four variables. MII is bounded between 0 and 1 due
to its construction. We use this index (MII) as a
proxy for macroeconomic instability.
Source: Ismihan (2003).
Macroeconomic instability in public/private capital accumulation and growth 251
