Content uploaded by Vincent Charles
Author content
All content in this area was uploaded by Vincent Charles on Apr 01, 2014
Content may be subject to copyright.
Volume 7, Issue 1, 2014
Volume 7, Issue 1, 2014
ISSN 1851-6599
Among the 100 early adopters schools of the PRME
Indexed/listed by
CONTENTS
Articles
1. Currency Choices in Valuation: An Approach for Emerging Markets
Guillermo L. Dumrauf
2. Minimum Wage and Job Mobility in Peru
Nikita Céspedes Reynaga, Alan Sánchez
3. Internationalization of Firms’ Activities and Company Union Wage Strategies
Domenico Buccella
4. Positions on Regulations Affecting Auditing and Nonauditing Activities
Rosario López Gavira, José Ángel Pérez López, José Enrique Romero García
5. Competitiveness among Higher Education Institutions: A Two-Stage Cobb-Douglas Model
for Efficiency Measurement of Schools of Business
Sonia Valeria Avilés-Sacoto, Wade D. Cook, David Güemes-Castorena
The Business and Economics Research Journal
Volume 7 • Issue 1 • March 2014
ISSN 1851-6599
The Business and Economics Research Journal
Journal of
CENTRUM
Cathedra
JCC
ISSN 1851-6599
Journal of
CENTRUM
Cathedra
JCC
Published by Cengage Learning Argentina S.A.
Rojas 2128 - (C1416CPX) Buenos Aires, Argentina • clientes.conosur@cengage.com • www.cengage.com
in coedition with CENTRUM Católica Graduate Business School
Pontificia Universidad Católica del Perú
Editor-in-Chief
Vincent Charles, Ph.D., CENTRUM Católica Graduate Business School, PUCP, Peru
Associate Editor
John Kuiper, Doctorandus, CENTRUM Católica Graduate Business School, PUCP, Peru
Regional Associate Editors
Mark Goh Keng Hock, Ph.D., National University of Singapore, Singapore
Paul Rouse, Ph.D., University of Auckland, Auckland, New Zealand
John Ruggiero, Ph.D., University of Day ton, Dayton, OH, USA
Leonardo Trujillo, Ph.D, National University of Colombia, Colombia
Advisory Committee
Anand Asthana, Ph.D., CENTRUM Católica Graduate Business School, PUCP, Peru
Ian Chaston, Ph.D., University of Plymouth, United Kingdom
Sergio Chión, Ph.D., CENTRUM Católica Graduate Business School, PUCP, Peru
Wade D. Cook, Ph.D., Schulich School of Business, York University, Toronto, Ontario, Canada
Rocky Dwyer, Ph.D., St. Paul University, Canada
Carlos Montoro, Ph.D., Bank for International Settlements, Mexico
Yasar A. Ozcan, Ph.D., Virginia Commonwealth University, USA
John Parnell, Ph.D., University of North Carolina at Pembroke, North Carolina, USA
Fred Y. Phillips, Ph.D., Alliant International University in San Diego, California, USA
Luc Soenen, Ph.D., TiasNimbas Business School, Tilburg University, The Netherlands
Vicente Tuesta, Ph.D., CENTRUM Católica Graduate Business School, PUCP, Peru
Frederick Wallace, Ph.D., Universidad de Quintana Roo, Mexico
Joe Zhu, Ph.D., Worcester Polytechnic Institute, Worcester, MA, USA
Editorial Office
Managing Editor E-mail: jcc@pucp.edu.pe
Contact: Clara Rosselló, CENTRUM Alianzas
The Journal of CENTRUM Cathedra is indexed/listed by:
· EBSCO · http://www.ebscohost.com
· Econlit · http://www.aeaweb.org/econlit/index.php
· Gale · http://www.gale.cengage.com
· Pro Quest · http://www.proquest.com
· SSRN · http://www.ssrn.com/link/journal-of-centrum-cathedra.html
· Cabell · http://www.cabells.com
· IndexCopernicus · http://www.indexcopernicus.com
· Ulrichsweb · http://www.ulrichsweb.com
Editing
Language Online Academic Editing (LOL): http://www.languageonline.us
Information for prospective authors may be found at the journal’s editorial web site: centrumwebs.pucp.edu.pe/jcc
Volume 7 • Issue 1 • March 2014
Journal of
CENTRUM
Cathedra
JCC
The Business and Economics Research Journal
Volume 7 • Issue 1 • March 2014
Contents
9 Foreword
Articles
11 Currency Choices in Valuation:
An Approach for Emerging Markets
Guillermo L. Dumrauf
Universidad del CEMA, Buenos Aires, Argentina
23 Minimum Wage and Job Mobility in Peru
Nikita Céspedes Reynaga
Banco Central de Reserva del Perú, Lima, Peru
Alan Sánchez
Grupo de Análisis para el Desarrollo, Lima, Peru
51 Internationalization of Firms’ Activities and Company
Union Wage Strategies
Domenico Buccella
Leon Kozminski University, Warsaw, Poland
75 Positions on Regulations Affecting Auditing and
Nonauditing Activities
Rosario López Gavira, José Ángel Pérez López, and José Enrique Romero García
University of Seville, Seville, Spain
Contents
91 Competitiveness among Higher Education Institutions:
A Two-Stage Cobb-Douglas Model for Efciency
Measurement of Schools of Business
Sonia Valeria Avilés-Sacoto
Instituto Tecnológico y de Estudios Superiores Monterrey, Monterrey, NL,
Mexico
Wade D. Cook
Schulich School of Business, York University, Toronto, Canada
David Güemes-Castorena
Instituto Tecnológico y de Estudios Superiores Monterrey, Monterrey, NL,
Mexico
117 Information for contributors
9
Foreword
The aim of the Journal of CENTRUM Cathedra (JCC): The Business and Economics Research Journal is
to become an evergreen, favorable journal through disseminating high quality scholarly research articles to
the pool of knowledge seekers in the field of business and economics; as well as play a vital role as a medium
of exchange for transmitting and simulating the frontiers of thought and enhancing business and economics
research between Latin American and non-Latin American countries with its well-balanced research framework.
This issue of the Journal of CENTRUM Cathedra (JCC): The Business and Economics Research Journal
includes five research articles written by authors from Argentina, Canada, Mexico, Peru, Poland, and Spain.
The articles cover a spectrum of research areas such as competitiveness, valuation and arbitrage, foreign direct
investment and international trade, labor unions, auditing and non-auditing activities, and the minimum wage
and labor mobility. This issue of the journal, like earlier issues, maintains the aim to present a global perspective.
One of the most common yet important matters that multinational companies face is deciding which currency
(domestic currency or US dollars) to value an investment or an acquisition in when working in a foreign or
emerging market because of how it affects the inputs. In “Currency Choices in Valuation: An Approach for
Emerging Markets,” Guillermo L. Dumrauf describes the equivalence of a discounted cash flow value using
different currencies, followed by a discussion of the method used to forecast the exchange rate using a yield
spread in market bonds. The adopted approach provides a framework to assess the consistency of the macro-
economic variables of the financial projections. This in turn allows the model to be extended to explore the
effect on the business value of a company as currencies depreciate or appreciate.
In “Minimum Wage and Job Mobility in Peru,” Nikita Céspedes Reynaga and Alan Sánchez revisit the
effect of the minimum wage on the Peruvian labor market. As the authors themselves highlight within the
paper, the study of the dynamic effects of the minimum wage in the context of the Peruvian economy is of great
interest since this country has experienced a remarkable transformation over the last two decades, including a
period of persistent economic growth and of economic labor growth. In consequence, authors study the effects
of the minimum wage on employment and income in Peru by considering a monthly database that captures
seven minimum wage changes registered between 2002 and 2011. The analysis includes an exploration of the
modifications in the minimum wage observed throughout the last decade; the effect of the minimum wage on
a range of outcomes, including employment status, job mobility, informality, and workers’ income; and the
short- and long-run effects of the minimum wage.
In “Internationalization of Firms’ Activities and Company Union Wage Strategies,” Domenico Buccella
discusses and illustrates the consequences and implications of the process of international market integra-
tion. The author explores how trade costs or undertaking direct investment in a foreign country is affected in
firms that are fully unionized. This is identified as a gap in the existing literature as few authors have to date
researched how this process affects the unions’ strategic behavior and how the unions’ behavior may in turn
affect the firms’ strategic choices in terms of international activities. Buccella develops a non-cooperative
three-stage game of international duopoly in the presence of unionized workforces at the company level. The
paper concludes with a set of managerial implications and the author makes some final remarks and sugges-
tions for future research in the area.
One of the main concerns that has emerged after a number of financial scandals over the past few decades
is about the execution of multiple and various services by auditors. In the article “Positions on Regulations
Affecting Auditing and Non-auditing Activities,” Rosario López, José Ángel Pérez, and José Enrique Romero
investigate the positions taken by academics and auditors on the legal aspects that regulate the execution of
auditing services. Hence, the purpose of this study is to assess whether changes made in the auditing legisla-
tion are likely to contribute to a reduction of the existing controversy surrounding the execution of various
services by auditors. The study’s target population comprised of auditors belonging to the Registry of Spanish
Auditors and academics belonging to the Spanish Accounting Professors Association.
Competitiveness has been a subject of study for decades and even today it continues being a constant concern
for countries, regions, companies, and organizations worldwide. In their paper titled “Competitiveness Among
Higher Education Institutions: A Two-Stage Cobb-Douglas Model for Efficiency Measurement of Schools
of Business,” Sonia Valeria Avilés-Sacoto, Wake D. Cook, and David Güemes-Castorena focus on a study
of competitiveness, benchmarking, and efficiency in the context of higher education. The authors pinpoint
10
data envelopment analysis (DEA) as an important tool for identifying best practice in both competitive and
noncompetitive settings. In order to meet the goal of their paper, Avilés-Sacoto, Cook, and Güemes-Castorena
develop a two-stage DEA methodology based on the Cobb-Douglas function. The authors also apply the devel-
oped model to a data set of undergraduate business programs and conclude the paper with their final insights.
The efforts of many academics and researchers who contributed articles and the knowledge of the experts
within the field who reviewed the articles have made this issue of the journal possible. We thank you. We further
extend our gratitude to the administrative and editorial staff of CENTRUM Católica Graduate Business School,
CENGAGE Learning, and Language Online Editing (www.languageonline.us). Special recognition goes to
Professor Fernando D´Alessio, the Director General of CENTRUM Católica Graduate Business School; and
Professor Percy Marquina, the Deputy Director General of CENTRUM Católica Graduate Business School
for their support.
We believe that the articles published in this issue of the JCC should be of considerable interest to our
readers. Thus, we wish you, our readers, informative reading.
Vincent Charles
Currency Choices in Valuation:
An Approach for Emerging Markets
Guillermo L. Dumrauf
Universidad del CEMA, Buenos Aires, Argentina
Abstract
One of the common decisions that multinational companies face is whether to value an investment or an
acquisition in a foreign or an emerging market in the domestic currency or U.S. dollars. The choice of
currency affects the inputs. Since these investments generate sales, expenses, and cash flows in domestic
currency, senior management is required to express expected cash flows in a strong currency, usually U.S.
dollars. Therefore, it is necessary to forecast the exchange rate for the investment horizon. In this paper, we
demonstrate the equivalence value independent of the currency used in the valuation. This is done through
using an arbitrage-free pricing model to obtain a fair value for the business. Users of the model assume the
simultaneous fulfillment of Interest Rate Parity theory (IRP) and Purchasing Power Parity theory (PPP),
which implies a constant real exchange rate (RER). This model can be extended to explore the effect on the
business value as a consequence of a depreciation or appreciation of the currency.
Keywords: Valuation, emerging markets, arbitrage-free, interest parity theory, purchasing power parity,
discounted cash flow, sovereign bond yield spread, forward exchange rate
JEL Classification codes: F23, G24, G30, G31
http://dx.doi.org/10.7835/jcc-berj-2014-0093
One of the common choices that multinational companies face when valuing an investment or an acqui-
sition in a foreign or an emerging market is the choice of the currency because of how it affects the inputs.
In particular, the issue is whether an analyst should value a foreign company in domestic currency such as
pesos, reais, or rupees or in U.S. dollars. Each of these choices carries certain implications. By definition,
the currency used to forecast cash flows should not affect the value, since the cash flow does not change as a
consequence of being expressed in another currency. The key rule here is that the cash flows and the cost of
capital have to be expressed in the same currency. In other words, if an analyst decides to perform the analysis
in U.S. dollars, then the cash flows and the cost of capital have to be expressed in U.S. dollars. This leaves
two alternative methods that can be used to obtain a fair value1 for the company:
1. Forecast both cash flows and cost of capital in U.S. dollars, estimating the Discounted Cash Flow
(DCF) value in U.S. dollars.
2. Forecast both cash flows and cost of capital in domestic currency, estimating the DCF value in domestic
currency. Afterwards, this can be converted to U.S. dollars using the spot exchange rate.
The first is used frequently in emerging markets where the domestic currency is depreciated against the
U.S. dollar or other strong currencies. The second is used when the domestic inf lation rate is only two or
three percentage points higher than the international inf lation rate, like in Colombia. If the exchange rate and
Journal of
CENTRUM
Cathedra
JCC
JCC: The Business and Economics Research Journal ● Volume 7, Issue 1, 2014 ● 11-22
12 JCC: The Business and Economics Research Journal
inflation rate are forecasted consistently and the inflation differential rate is considered in the cost of capital
expressed in domestic currency, both methodologies yield identical, fair values.
Although some corporate finance textbooks give examples of this equivalence using the Interest Rate
Parity theory (IRP) and Purchasing Power Parity theory (PPP), they always assume constant, nominal interest
rates (Berk & DeMarzo, 2007; Brealey, Myers, & Marcus, 2008; Emery & Finnerty, 2007; Graham, Smart, &
Megginson, 2010; Ross, Westerfield, & Jaffe, 2009). This leads to a flat yield curve, which is rarely observed
and transitional. Brigham and Erdhardt (2010) used the IRP to forecast the exchange rate but do not offer proof
of the equivalence. Damodaran´s (2012) opinion is similar:
If we assume purchasing power parity then differences in interest rates reflect differ-
ences in expected inflation rates. Both the cash flows and the discount rate are affected
by expected inflation; thus, a low discount rate arising from a low risk free rate will be
exactly offset by a decline in expected nominal growth rates for cash flows and the value
will remain unchanged. If the difference in interest rates across two currencies does not
adequately ref lect the difference in expected inflation in these currencies, the values
obtained using the different currencies can be different. In particular, projects and assets
will be valued more highly when the currency used is the one with low interest rates
relative to inf lation. The risk, however, is that the interest rates will have to rise at some
point to correct for this divergence, at which point the values will also converge. (p. 156)
Koller, Goedhart, and Wessels (2010) proposed an approach assuming constant, real interest rates and a
slightly increasing inflation rate to obtain the nominal interest rates.
Since in all cases the use of the interest rates across two countries is proposed, some confusion remains.
In reality, if the expected market exchange rate has to be obtained, it is more appropriate to use the sovereign
bond returns from the same issuer. This is because the only risk difference between a bond nominated in
domestic currency and a bond nominated in U.S. dollars is the exchange risk.
Since most financial statements are richer and more detailed in domestic currency, tax charges are based on
nominal financial statements. Managers are familiar with this information; therefore, they could estimate the
cash flows in domestic currency first and then convert them into U.S. dollars. It is usually better to estimate
the expected cash flows in domestic currency to incorporate the senior management’s forecasts of quantities
and prices. Once this is done, it is easy to then convert the amounts into U.S. dollars using the expected market
exchange rate. This can be obtained by comparing the yields spread in the market bonds issued in domestic
currency and U.S. dollars by the foreign or emerging country, assuming the simultaneous fulfillment of IRP
and PPP theories. Once the cash flows are expressed in U.S. dollars, managers can estimate the fair value of
the business using a DCF model, with the cost of capital expressed in U.S. dollars. Valuing in U.S. dollars or
domestic currencies yields identical fair values when the simultaneous fulfillment of the IRP and the PPP is
assumed. The Weighted Average Cost of Capital (WACC) and the long-term growth rate used to estimate the
Terminal Value of the business is adjusted by the inflation differential rate.
Despite empirical tests for both, IRP and PPP theories have shown poor results. These play instrumental
roles in an arbitrage-free valuation model, since both are used to forecast the exchange rate and inflation rate.
Once the projections are made, analysts have a framework to analyze a strongly linked triad of interest rates,
inflation rates, and exchange rate.
The simultaneous fulfillment of IRP and PPP implies that the real exchange rate (RER) remains constant
for the explicit forecast period. However, users can extend the model to analyze different economic scenarios
such as devaluation, different pass-through rates, etc. This is particularly important in emerging markets, since
the exchange rate has observed periods ranging from strong appreciation to periods with strong depreciation.
The remainder of this paper is organized as follows. First, the focus is on describing the equivalence of the
DCF value using free cash flows (FCF) nominated in U.S. dollars and domestic currency. The second section
focuses on the method used to forecast the exchange rate using yields spread in market bonds. The next section
focuses on describing an example. The fourth section includes a summary of the works done on the effect of
an unexpected change in foreign exchange rates on the market value of businesses where it is suggested that
the model can be extended to consider such effects. The final section contains some concluding remarks.
13
Currency Choices in Valuation: An Approach for Emerging Markets
The Equivalence of the DCF Value Using Different Currencies
In an ideal world, the prices of goods and services that companies buy or sell should reflect the average
inflation rate. Valuing a company in this ideal world using nominal FCF and discounting them with a nominal
cost of capital should yield the same value as using real FCF and real cost of capital. Financial theory already
recognizes this issue. Since taxes are based upon nominal income and senior management is used to work-
ing with nominal rather than real categories on a daily basis, making the analysis in nominal terms is more
appropriate.
Another issue that analysts often engage with when valuing an acquisition in a foreign country is whether
the analysis should be done in domestic or foreign currency. When a valuation is performed in an emerging
market, there are important reasons for performing it using both nominal FCF and cost of capital expressed
in U.S. dollars. First, as the currency of emerging markets has observed a tendency towards depreciation and
higher inflation rates than developed countries, senior management of a multinational is required to forecast the
FCF in U.S. dollar amounts. This creates a consistent set of macroeconomic assumptions regarding exchange
rate, domestic inflation rate, interest rates, and gross domestic product (GDP) growth. Second, all inputs of the
Capital Asset Pricing Model (CAPM) are nominated in U.S. dollars. A risk free rate is commonly the expected
return on U.S. Treasury Bonds and the market risk premium for emerging markets is commonly assumed as
the American risk premium because of the lack of representativeness or the absence of the emerging market
stock index. If a country’s risk premium is added to the cost of capital, this is usually estimated as the spread
of a long-term Bond issued by the relevant country and nominated in U.S. dollars rather than a long-term
U.S. T-Bond. Third, a business sale transaction is performed in U.S. dollars and not in domestic currency.
To be consistent, the cost of capital and cash f lows should be expressed in the same currency. By definition,
the company value does not depend upon what currency is used for creating the analysis; the value should not
change whether the analysis is done in domestic currency or foreign currency. If the company value is USD
100 million, in Argentinian pesos it will be ARS 434 million, in euros it will be EUR 79 million, and so on
(quotes from Reuters on 1/19/2011). The important fact is that cash f lows and the cost of capital have to be
estimated consistently in the same currency.
In order to estimate the fair value of the company´s operations indicated with the term V, the DCF model
is used, which consists in calculating the present value of FCF. FCF reflects the cash flow that is generated
by a company’s operations. This cash flow should also be available to all the company’s capital providers,
both debt and equity.
For consistency with the cash f low definition, the discount rate applied to the FCF should reflect the
opportunity cost to all the capital providers. In other words, this should reflect the WACC. The formula used
to estimate the fair value of operations, including the Terminal Value, is as follows:
V
FCF
WACC(1 )
t
t
t1
∑
=+
=
∞. (1)
The WACC can be expressed as the weighted average of the after-corporate-tax required return for debt
and the required return for equity:
WACC kd t
D
DE
ke
E
DE
(1
).
=-
+
+
+
. (2)
In this equation:
kd = required return for debt,
t = marginal corporate tax rate,
D = debt market value,
E = equity market value,
ke = required return for equity.
The required return for equity is usually estimated using the formula of the famous CAPM model:
ke rf Ermrf() .
b
[]
=+ - (3)
14 JCC: The Business and Economics Research Journal
The CAPM model expresses the required return for equity as the riskless return, rf, plus a risk premium
(Sharpe, 1964). This is calculated as the difference between the market portfolio´s expected return E(rm) and
the riskless return adjusted by the common stock´s beta coefficient. The beta coefficient plays a critical role
in asset pricing, since it measures how much an individual asset contributes to the standard deviation of the
market portfolio.
An additional challenge in valuing a business is its explicit forecast period and its indefinite life. The
value after the explicit forecast period is referred to as the Terminal Value. This is generally calculated using
the growing perpetuity formula, which assumes that the company’s free cash will grow at a constant rate, g,
beyond the explicit forecast period2. The formula to estimate the terminal value is as follows:
TV
FCFg
WACC g
(1 )
()
.
T
=
+
-
. (4)
In this formula, FCF
T
(1+g) is the FCF in the first year after the explicit forecast period and g is the nominal
growth rate at which the FCF is expected to grow for perpetuity. This technique provides the same result as a
long explicit forecast when the company’s cash flow is forecasted to grow at the same rate and the company
earns a constant rate of return on all new capital invested during the continuing-value period. Terminal Value is
added to the last forecasted cash f low before all cash f lows are discounted to obtain the fair value of operations.
Analysts can perform a valuation using nominal FCF in U.S. dollars and then discount them with a WACC
expressed in U.S. dollars to obtain the fair value of the business in U.S. dollar amounts:
V
FCF
WACC(1 )
.
USDt
USD
t
US
Dt
t1
∑
=+
=
∞. (5)
Otherwise, analysts could perform the valuation discounting the FCF in domestic currency using a WACC
expressed in domestic currency to obtain the fair value in domestic currency:
V
FCF
WACC(1 )
.
Dt
D
t
Dt
t1
∑
=+
=
∞. (6)
To obtain the fair value in U.S. dollar amounts, analysts can divide Equation 6 by the spot exchange rate
SD/US D
:
V
V
S
.
USD
D
DUSD/
=. (7)
The inf lation rate observed in emerging countries is higher compared to the international inflation rate
observed in developed countries. Therefore, the WACC expressed in U.S. dollars must be adjusted by the
inflation rate differential to obtain the discount factor expressed in domestic currency:
(1 +WACCt
D)=(1 +WACCt
USD)
(1
+πt
D
)
(1 +π
t
*)
.
. (8)
In this equation:
π
t
D
= t–period domestic inflation rate,
πt
*
= t–period international inflation rates.
For the same reason, the inflation differential rate must be used to adjust the long-term growth rate in order
to calculate the terminal value for a company that operates in the foreign market3:
gD=g*
(1
+πT
D
)
(1 +πT
*)
.
. (9)
In this equation:
gD = nominal long-term growth rate in the emerging country,
15
Currency Choices in Valuation: An Approach for Emerging Markets
g* = nominal long-term growth rate in the developed country,
πT
D
= last forecast period domestic inf lation rate,
πT
*
= last forecast period international inf lation rate.
Then, the Terminal Values in U.S. dollars and domestic currency are expressed as:
TVUS D =
FCF
T
US D
(1
+
g
*
)
WACCT
US D −g*
,
, (10)
TV D=
FCF
T
D
(1
+
g
D
)
WACCT
D−gD
.
. (11)
IRP states a relationship between the interest rates and the currency exchange rates for the two countries.
In the absence of arbitrage opportunities, the interest rate differential is equivalent to the differential between
the spot and forward rates, so the IRP must hold for every period of t in equilibrium. To illustrate this for a
single year:
F
t
D
/
US D
SD/US D =
1
+
i
t
D
1+it
US D
.
. (12)
In this equation:
it
D= one period forward domestic interest rate,
it
US D = one period forward U.S. dollar interest rate,
F
t
D
/
US D = forward exchange rate for one period forward.
Therefore, the forward exchange rate can be expressed as:
F
t
D/US D =SD/USD
1
+
i
t
D
1+it
US D
.
. (13)
Uncovered IRP refers to the condition where exposure to unanticipated changes in exchange rates is
uninhibited. On the other hand, covered IRP refers to the condition where a forward contract has been used
to cover risks concerning the exchange rate. The use of a forward market eliminates the investor´s risk, and
as a result there are no interest rate arbitrage opportunities.
On the other hand, the relative PPP theory is formally stated in terms of inflation rates. It requires that
the inflation rate differential between two countries be equal to the change in the foreign exchange rate. The
expectations theory of forward exchange rates maintains that the expected spot exchange rate t periods in the
future equal the t-period forward rate, so the PPP must hold for every period of t in equilibrium:
F
t
D/USD
SD/USD=
1
+πt
D
1+πt
*
.
. (14)
Therefore, the forward exchange rate for one period forward is:
F
t
D/USD=SD/USD
1
+πt
D
1+πt
*
.
. (15)
IRP, coupled with PPP and the expectations theory of forward exchange rates, implies the international
Fisher effect (Emery & Finnerty, 2007). These relationships are mutually consistent, so both theories should
be predicting the same forward exchange rate:
F
t
D/USD=SD/USD
1
+
i
t
D
1+i
t
USD=SD/USD
1
+πt
D
1+π
t
*
.
. (16)
Since the inf lation rate fluctuates much less and is lower in developed countries than developing countries,
staying around 2.2% and 5.5%, respectively, analysts can assume the international inf lation rate remains
16 JCC: The Business and Economics Research Journal
constant along the forecast period (Central Intelligence Agency, 2012). Thus, analysts can obtain the expected
domestic inf lation rate from Equation 16 for one period forward:
F
t
D/USD
SD/USD(1+πt
*)−1=πt
D. (17)
The forecasted FCF expressed in domestic currency is equal to the forecasted FCF in U.S. dollars multiplied
by the forward foreign exchange rate:
FCF
t
D
=
FCF
t
US D F
t
D/US D .
(18)
Now, analysts can use these equations to prove the equivalence of valuing a company in domestic or foreign
currency. This is done through replacing Equations 7, 8, and 18 in Equation 6 to create:
VUSD=
V
D
SS/USD=
1
SS/USD
FCF
t
USD
F
t
D/USD
1+WACCt
USD
()
t1+πt
D
()
t
1+πt
*
()
t
t=1
∞
∑
.
. (19)
Equations 5 and 6 assume a flat yield curve. It is common that practitioners consider a constant cost of
capital along the investment horizon; if the interest rates change, as is the general case, the WACC must be
recalculated for each period. In this case, the WACC expressed in U.S. dollars for the t-period will be equal
to the WACC expressed in U.S. dollars for the t-1 period plus the change in the interest rate:
WACCt
US D =
WACCt
−
1
US D +
(it
US D −
it
−
1
US D
).
(20)
As is shown in Equation 8, the WACC expressed in domestic currency will be equal to the WACC expressed
in U.S. dollars multiplied by the inflation rate differential. The forward exchange rate is estimated using IRP
rather than PPP. This is because one can observe forward interest rates in the market, but one cannot directly
observe the expected inf lation rates. In the next section, the data of market bonds is used to obtain the forward
interest rates. With the forecast for the forward exchange rate, analysts can obtain the domestic inflation rate
assuming the fulfillment of the PPP.
If both theories predict the same exchange rate, the RER can be subtracted from Equation 15. Rearranging
terms, if the PPP holds, it implies that the RER remains constant along the projection. To illustrate for one
single year:
RER0
D/USD=F
t
D/USD
1
+πt
*
1+πt
D
.
. (21)
Use of the Sovereign Bond Yield Spread to Forecast the Exchange Rate
When valuing a company in an emerging market, analysts usually find that the forward exchange rates are
not liquid or are not available beyond 12/18 months. To obtain the company fair value using the DCF method,
analysts need to forecast the forward exchange rate for a longer horizon, usually ten years. To fill the gap,
analysts can use the data of market bond yields in an emerging country and forecast the forward exchange rate
for the explicit forecast period, assuming that the IRP holds. Emerging market bor rowers able to issue bonds in
local currency generally pay a premium for bonds issued in a strong currency like U.S. dollars. The premium
reflects the expected depreciation of the domestic currency. Since the issuer and the jurisdiction are the same
and differ solely in the currency, the yield spread ref lects the market’s opinion about the forward exchange rate.
Figure 1 shows the yield curves in domestic and foreign currency for the Sovereign Mexican Bonds on June
2011 (Source: Reuters). While there are a number of sophisticated techniques for constructing yield curves,
these are beyond the scope of this paper. We obtain the best-fit curves adjusting a logarithmic trend-line to
illustrate expected returns for both sovereign bonds in domestic currency and in U.S. dollars.
17
Currency Choices in Valuation: An Approach for Emerging Markets
y = 0.0195ln(x) + 0.0025
R2 = 0.9363
y = 0.0069ln(x) + 0.0542
R2 = 0.7988
-6%
-4%
-2%
0%
2%
4%
6%
8%
10%
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
Yield
Modified duration
USD Domestic currency
Figure 1. Yield curves for sovereign Mexican bonds.
As can be seen, bonds nominated in domestic currency have higher yields than bonds nominated in U.S.
dollars. Since the market requires a currency premium for the devaluation risk, the yield spread reflects the
market’s opinion about the expected exchange rate. Analysts can easily obtain the required returns for a
specific year using the yield curve equations. For example, here is an equation to calculate the required return
in domestic currency and U.S. dollars for one year from now:
i0.0542 0.0069 ln(1) 5.42%,
D
1
=+ =
i0.0025 0.0195 ln(1) 0.25%.
USD
1
=+ =
To obtain the expected yields for subsequent years, analysts only have to change the year number in the
equation. For example, to obtain the expected yields for three years from now:
i0.0542 0.0069 ln(3) 6.18%,
D
1=+ =
i0.0025 0.0195 ln(3) 2.39%.
USD
1
=+ =
To forecast the forward exchange rate for a specic year, analysts can extrapolate what some market partici-
pants refer to as the market´s consensus of forward interest rates. Given the two-year spot rate, there could be a
rate on a one-year instrument one year from now that will make the investor indifferent between two alternatives:
(1
+
i1
US D )(1
+
f1
US D )
=
(1
+
i2
US D )2.
(22)
Solving Equation 22 for f1USD, there is this equation:
f1
US D =
(1
+
i
2
US D
)2
(1+i1
US D )−1. (23)
For example, to forecast the expected exchange rate for two years from now:
F
2
D/US D =SD/USD
(1
+
i
1
US D
)(1
+
f
1
US D
)
(1+i
1
D)(1 +f
1
D)
. (24)
18 JCC: The Business and Economics Research Journal
A Hypothetical Example
“Foreign investor” is a company that has operations in several countries and is considering an acquisition
in México. As the currency of México has observed a higher inflation rate than developed countries, senior
management is required to forecast the FCF in U.S. dollar amounts. This forecast creates a consistent set of
macroeconomic assumptions regarding the exchange rate, the domestic inflation rate, the interest rates, and
the GDP growth.
The target acquisition´s FCF is expected to grow at a rate of 3% per year plus the domestic inflation rate.
The explicit forecast period is projected until the company reaches a stable state by the end of the forecast
period of ten years. Beyond this period, FCF is expected to grow at a rate of 2% per year plus the domestic
inflation rate. A Terminal Value in foreign and domestic currency is estimated using Equations 10 and 11, the
WACC expressed in U.S. dollars is estimated to be 15%, and the spot exchange rate is SD/USD = 11.72.
Table 1 shows the forecasted FCF values for the Target Company in domestic currency and U.S. dollars,
domestic and U.S. dollar interest rate, domestic and international inf lation rate, and the forward exchange
rate. The last row shows the RER, which, as mentioned earlier, if the IRP and the PPP hold, remains constant
along the explicit forecast period.
Table 1
DCF Equivalence With Domestic and Foreign Currencies
Yr 0 Yr 1 Yr 2 Yr 3 Yr 4 Yr 5 Yr 6 Yr 7 Yr 8 Yr 9 Yr 10
FCF D100.00 109.50 119.30 129.50 140.19 151.42 163.23 175.69 188.82 202.67
Terminal Value D1 565.42
FCF D100.00 109.50 119.30 129.50 140.19 151.42 163.23 175.69 188.82 1 768.09
FCF USD 8.11 8.52 8.96 9.41 9.89 10.39 10.91 11.46 12.04 12.65
Terminal Value USD 97.73
FCF USD 8.11 8.52 8.96 9.41 9.89 10.39 10.91 11.46 12.04 110.38
F D/USD 11.72 12.32 12.85 13.32 13.76 14.18 14.58 14.96 15.33 15.68 16.02
i D 5.42% 5.90% 6.18% 6.38% 6.53% 6.66% 6.76% 6.85% 6.94% 7.01%
i USD 0.25% 1.60% 2.39% 2.95% 3.39% 3.74% 4.04% 4.30% 4.53% 4.74%
p D7.26% 6.31% 5.77% 5.39% 5.10% 4.86% 4.66% 4.49% 4.34% 4.21%
p*2.00% 2.00% 2.00% 2.00% 2.00% 2.00% 2.00% 2.00% 2.00% 2.00%
Disc. factor USD 0.87 0.75 0.65 0.56 0.48 0.42 0.36 0.32 0.27 0.24
PV Cash Flow USD 70.73 7.06 6.37 5.78 5.26 4.78 4.36 3.97 3.62 3.30 26.24
Disc. factor D0.83 0.68 0.57 0.48 0.40 0.34 0.28 0.24 0.20 0.17
PV Cash Flow D828.95 82.69 74.67 67.75 61.59 56.06 51.05 46.52 42.40 38.66 307.57
Real exchange rate 11.72 11.72 11.72 11.72 11.72 11.72 11.72 11.72 11.72 11.72 11.72
Using Equations 5, 6, and 7, these are the equivalence values:
V USD = 70.73,
V D = 828.95,
V USD =
828.95
11.72
= 70.73.
The approach used here differs from the work of Koller et al. (2010) in that they forecast the nominal inter-
est rates assuming that the real interest rate will remain constant and the inflation rate will slightly increase
19
Currency Choices in Valuation: An Approach for Emerging Markets
along the projection. In the model used for this study, the interest rates are observed in the market bonds. The
domestic inf lation rate is then subtracted from the PPP arbitrage formula. This proposal can be summarized
in the following steps:
1. Estimate the FCF in domestic currency. This allows senior management to work with prices and
quantities that they deal with every day.
2. Build the yield curves of the foreign or emerging country using sovereign bonds nominated in domestic
currency and U.S. dollars for a similar, modified duration4.
3. Assuming that IRP holds, forecast the forward exchange rate.
4. Convert the FCF expressed in domestic currency to U.S. dollars using the forward exchange rate and
then discount that amount with the cost of capital expressed in U.S. dollars. If a valuation in domes-
tic currency is performed, the expected inflation rate needs to be estimated using the PPP, and both
the WACC in U.S. dollars and the long-term growth rate for the inflation differential rate to obtain
the WACC in domestic currency has to be adjusted. Finally, the cash flow must be discounted and
expressed in domestic currency and the value must be divided by the spot exchange rate.
Since this model assumes the simultaneous fulfillment of IRP and PPP, it is pertinent to examine the
empirical evidence of the theories. Both IRP and PPP have been highly researched since the 70s, and an intense
debate has been generated on their fulfillment both in the short and long-term. The conventional view is that
uncovered interest parity (UIP) and PPP are appealing in theory but rejected empirically.
Early empirical studies tested the null hypothesis that the RER does not mean revert, but instead, follows
a random walk. They argued this to be an implication of the efficiency of international markets (Adler &
Lehmann, 1983). Generally, the hypothesis of mean reversion has not been successful.
Meese and Rogoff (1983) suggest that the random walk model outperforms a range of fundamentals-based
models of exchange rate determination at horizons of up to one year. Beyond the one year period, however, the
random walk model does not yield the minimum forecast errors. Mark (1997) arrived at a similar conclusion.
Abuaf and Jorion (1990) examined the evidence on PPP and in the long term discredited the random walk
hypothesis. Deviations from PPP, while substantial in the short term, appear to decrease over time. Cerrato
and Sarantis (2003) have also found that the mean reversion hypothesis does not hold when testing the PPP
in emerging markets.
Taylor (2002) and Taylor and Taylor (2004) made an overview and concluded that the new econometric
methods permit some degree of confidence in the long-term PPP again. Bansal and Dahlquist (2000) and in
the short term, Flood and Rose (2001) found evidence supporting UIP for emerging market currencies.
Using bonds with maturities ranging from f ive to ten years, Chinn and Meredith (2004), Chinn (2006), and
Zhang (2006) showed that yield differentials explain future currency movements. They suggested that UIP
tends to hold for financial instruments of longer maturities. Cheung, Chinn, and Garcia Pascual (2005) found
that UIP performs well in predicting exchange rate movements at long horizons, relative to other structural
models of the exchange rate.
Singh and Banerjee (2006) tested the real interest parity in emerging markets and found significant deviations
between short-term, emerging market, real interest rates, and world real interest rates. Their paper suggests
that real interest rates in the emerging markets show some convergence in the long term.
Mehl and Cappiello (2009) found support in favor of UIP for U.S. dollar rates vis-à-vis major mature
economy currencies, but far less against emerging market currencies.
Burnside, Eichenbaum, Kleshchelski, and Rebelo (2010) suggest that the forward exchange rate is a biased
forecaster of the future spot exchange rate. They argue that the explanation for the higher average payoff to
the carry trade is that it reflects the presence of a peso problem, which refers to a rare event where there are
negative payoffs to the carry trade. Even though the losses of an unhedged carry trade in the peso state are
moderate, the investor attaches great importance to those losses. This results in a higher value of the stochastic
discount factor in the peso state.
Skinner and Mason (2011) found that while covered IRP holds for large and small triple-A rated economies,
it does not hold for longer maturities for Brazil, Chile, Russia, and South Korea.
20 JCC: The Business and Economics Research Journal
Currency’s Effect on Business Value
It is a common belief among practitioners that unexpected changes in foreign exchange rates should affect
the market value of certain firms. Shapiro (1975) focused on cash flow sensitivity. He stated that the major
factors affecting a multinational firm’s exchange risk include the distribution of its sales between domestic
and export markets, the amount of import competition it faces domestically, and the degree of substitutabil-
ity between local and imported factors of production. The economic exposure to exchange rate changes can
be interpreted as the regression coefficient of unexpected firm value changes on unexpected exchange rate
changes (Adler & Dumas, 1983; Adler & Dumas, 1984; Adler & Simon, 1986; Dumas, 1978; Hodder, 1982;
Jorion, 1990). However, empirical evidence suggests that firms are not systematically affected by foreign
exchange rate changes (Allayannis, 1996; Bartov & Bodnar, 1994; Bodnar, 1993; Choi & Prasad, 1995; Jorion
1990; Miller & Reuer, 1998).
Several studies show that using various functional forms, such as quadratic and cubic functions, can more
effectively capture the degree of exposure when a linear model cannot, at least for some firms (Allayannis &
Ihrig, 2001; Bodnar & Wong, 2003; Priestley & Ødegaard, 2007). Bartram and Bodnar (2007) suggested
that the hedging behavior of firms could explain the lack of exposure. Nevertheless, documenting a strong
and systematic contemporaneous relation between stock returns and exchange rate exposure continues to
be puzzling.
Although the empirical evidence is puzzling, managers should be interested in a model that provides a
framework to estimate the possible effect on the business value as a consequence. For example, they should
be interested in a model that could factor in a sudden devaluation which could affect the GDP and probably
the business sales. Senior management could explore different economic scenarios and different combina-
tions of devaluation and pass-through rates. Appreciated currencies in emerging countries generally lead to
sharp devaluations that generate a fall in the GDP and sales measured in domestic currency. In this case,
there would be an effect on the business value, although the value measured in domestic or foreign currency
would continue to be the same.
Conclusions
Valuation in foreign currency or in domestic currency yields identical values when the simultaneous fulfill-
ment of both interest rate parity theory and purchasing power parity theory is assumed. This study is focused
on obtaining fair value for a company using an arbitrage-free valuation model based on the simple fact that
an asset cannot be sold for more than one price in the market. If one assumes that the IRP does not hold, one
is immediately valuing an asset assuming arbitrage opportunities. The use of a forward market eliminates the
risk for an investor, and as a result there are no interest rate arbitrage opportunities. If no forward exchange
rates were available for more than one or two years in emerging markets, their prices would certainly be
governed by the IRP should they exist.
The concept of arbitrage is at the heart of the financial economic analysis, just like the notion of general
equilibrium is at the heart of the macroeconomic analysis. While the conclusion is obviously theoretical,
since a given asset cannot consistently sell at more than one price in the market, establishing the correct
assumptions is required. At the practitioner level, one way to forecast the expected exchange rate is to adjust
the spot rate by the yield spread observed in market sovereign bonds, for bonds issued in domestic and U.S.
dollar currencies. In this way, the exchange rate is forecasted and reflects the appreciation or depreciation
rate expected by the market.
While our approach is essentially a suggestion for valuation in emerging markets, it can also be used by
multinationals which have businesses in developed countries. It provides a framework to assess the consistency
of the macroeconomic variables of the financial projections. This can be extended to consider the effect of
something like a sudden devaluation within a set of economic assumptions to assess the effect on the fair value.
21
Currency Choices in Valuation: An Approach for Emerging Markets
Endnotes
1 Fair value is understood to be the price at which the company would change hands when both parties have reasonable
knowledge of the relevant facts and no one is under the effects of compulsion.
2 The terminal value is sometimes estimated as a multiple of EBITDA or as a liquidation value of the firm´s assets in the
terminal year, estimating what others would pay for the assets that the firm has accumulated at that point.
3 If a high-inflation currency is used to estimate cash flows and discount rates, the long-term growth rate will be much
higher, since the expected inflation rate is added on to real growth. For instance, the long-term growth rate that would
be used to value a company that operates in an emerging market will be much higher if the valuation is done in domestic
currency than in U.S. dollars.
4 Modified Duration is a measure of the price change as a consequence of a change of one percentage point in the interest
rate.
References
Abuaf, N., & Jorion, P. (1990). Purchasing power parity in the long run. Journal of Finance, 45(1), 157-174. dx.doi.
org/10.1111/j.1540-6261.1990.tb05085.x
Adler, M., & Dumas, B. (1983). International portfolio choice and corporate nance: A synthesis. Journal of Finance,
38(3), 925-984. dx.doi.org/10.1111/j.1540-6261.1983.tb02511.x
Adler, M., & Dumas, B. (1984). Exposure to currency risks: Denition and measurement. Financial Management, 13(2),
41-50. dx.doi.org/10.2307/3665446
Adler, M., & Lehmann, B. (1983). Deviations from purchasing power parity in the long run. Journal of Finance, 38(5),
1471-1487. dx.doi.org/10.1111/j.1540-6261.1983.tb03835.x
Adler, M., & Simon, D. (1986). Exchange risk surprise in international portfolios: Some equity markets are super-nominal!
Journal of Portfolio Management, 12(2), 44-53. dx.doi.org/10.3905/jpm.1986.409044
Allayannis, G. (1996). Exchange rate exposure revisited (Working paper DSWP-97-06). Charlottesville, VA: Darden
Graduate School of Business Administration, University of Virginia.
Allayannis, G., & Ihrig, J. (2001). Exposure and markups. Review of Financial Studies, 14(3), 805–835. dx.doi.org/10.1093/
rfs/14.3.805
Bansal, R., & Dahlquist, M. (2000). The forward premium puzzle: Different tales from developed and emerging econo-
mies. Journal of International Economics, 51(1), 115-144. dx.doi.org/10.1016/S0022-1996(99)00039-2
Bartov, E., & Bodnar, G. M. (1994). Firm valuation, earnings expectations, and the exchange-rate exposure effect. Journal
of Finance, 49(5), 1755-1785. dx.doi.org/10.1111/j.1540-6261.1994.tb04780.x
Bartram, S. M., & Bodnar, G. M. (2007). The exchange rate exposure puzzle. Managerial Finance, 33(9), 642-666. dx.doi.
org/10.1108/03074350710776226
Berk, J., & DeMarzo, P. (2007). Corporate nance: The core (2nd ed.). Boston, MA: Pearson.
Bodnar, G. M. (1993). Exchange rate exposure and industry characteristics: evidence from Canada, Japan, and the USA.
Journal of International Money and Finance, 12(1), 29-45. dx.doi.org/10.1016/0261-5606(93)90008-Y
Bodnar, G. M., & Wong, M. (2003). Estimating exchange rate exposures: Issues in model structure. Financial Management,
32(1), 35-67. dx.doi.org/10.2307/3666203
Brealey, R., Myers, S., & Marcus, A. (2008). Fundamentals of corporate nance (9th ed.). New York, NY: McGraw-Hill.
Brigham, E., & Erdhardt, M. (2010). Financial Management, Theory and Practice. Mason, OH: South-Western Cengage
Learning.
Burnside, C., Eichenbaum, M., Kleshchelski, I., & Rebelo, S. (2010). Do peso problems explain the returns to the carry
trade? (Working Papers). Durham, NC: Duke University, Department of Economics, 10-44.
Cerrato, M., & Sarantis, N. (2003, January). Does the purchasing power parity hold in emerging markets? Evidence from
black market exchange rates. Retrieved from http://repec.org/res2003/Cerrato.pdf
Cheung, Y-W., Chinn, M. D., & Garcia Pascual, A. (2005). Empirical exchange rate models of the nineties: Are any t to
survive? Journal of International Money and Finance, 24(7), 1150-1175. dx.doi.org/10.1016/j.jimonn.2005.08.002
Chinn, M. D. (2006). The (partial) rehabilitation of interest rate parity in the oating rate era: Longer horizons, alternative
expectations, and emerging markets. Journal of International Money and Finance, 25(1), 7–21. dx.doi.org/10.1016/j.
jimonn.2005.10.003
Chinn, M. D., & Meredith, G. (2004). Monetary policy and long-horizon uncovered interest rate parity. IMF Staff Papers,
51(3), 409-430.
Choi, J. J., & Prasad, A. M. (1995). Exchange risk sensitivity and its determinants: A rm and industry analysis of U.S.
multinationals. Financial Management, 24(3), 77-88. dx.doi.org/10.2307/3665559
22 JCC: The Business and Economics Research Journal
Central Intelligence Agency, T. W. (2012). The World Factbook. Retrieved from https://www.cia.gov/library/publications/
the-world-factbook/geos/xx.html
Damodaran, A. (2012). Investment valuation tools and techniques for determining the value of any asset (3rd ed.). New
York, NY: John Wiley & Sons.
Dumas, B. (1978). The theory of the trading rm revisited. Journal of Finance, 33(3), 1019-1030. dx.doi.
org/10.1111/j.1540-6261.1978.tb02041.x
Emery, D., & Finnerty, J. (2007). Corporate nancial management (3rd ed.). New Jersey, NJ: Prentice Hall.
Flood, R., & Rose, K. (2001). Uncovered interest parity in crisis: The interest rate defense in the 1990s (IMF Working
Paper WP/01/207). Retrieved from http://www.imf.org/external/pubs/ft/wp/2001/wp01207.pdf
Graham, J., Smart, S. B., & Megginson, W. L. (2010). Corporate nance: Linking theory to what companies do (3rd ed.).
Mason, OH: South-Western Cengage Learning.
Hodder, J. E. (1982). Exposure to exchange rate movements. Journal of International Economics, 13(3-4), 375-386.
dx.doi.org/10.1016/0022-1996(82)90065-4
Jorion, P. (1990). The exchange-rate exposure of U.S. multinationals. Journal of Business, 63(3), 331-345. dx.doi.
org/10.1086/296510
Koller, T., Goedhart, M., & Wessels, D. (2010). Valuation: Measuring and managing the value of companies (5th ed.).
Hoboken, NJ: JohnWiley & Sons, Inc.
Mark, N. (1997). Exchange rates and fundamentals: Evidence on long-horizon predictability. American Economic Review,
85(1), 201-218.
Meese, R. A., & Rogoff, K. (1983). Empirical exchange rate models of the seventies: Do they t out of sample? Journal
of International Economics, 14(1-2), 3-24. dx.doi.org/10.1016/0022-1996(83)90017-X
Mehl, A., & Cappiello, L. (2009). Uncovered interest parity at long horizons: Evidence on emerging economies. Review of
International Economics, 17(5), 1019–1037. dx.doi.org/10.1111/j.1467-9396.2008.00793.x
Miller, K. D., & Reuer, J. J. (1998a). Firm strategy and economic exposure to foreign exchange rate movements. Journal
of International Business Studies, 29(3), 493-513. dx.doi.org/10.1057/palgrave.jibs.8490004
Priestley, R., & Ødegaard, B. A. (2007). Linear and nonlinear exchange rate exposure. Journal of International Money and
Finance, Elsevier, 26(6), 1016-1037. dx.doi.org/10.1016/j.jimonn.2007.05.001
Ross, S. A., Westereld, R. W., & Jaffe, J. (2009). Corporate nance (9th ed.). New York, NY: McGraw-Hill/Irwin.
Shapiro, A. C. (1975). Exchange rate changes, ination, and the value of the multinational corporation. The Journal of
Finance, 30(2), 485–502. dx.doi.org/10.1111/j.1540-6261.1975.tb01824.x
Sharpe, W. F. (1964). Capital asset prices: A theory of market equilibrium under conditions of risk. Journal of Finance,
19(3), 425-442. dx.doi.org/10.1111/j.1540-6261.1964.tb02865.x
Singh, M., & Banerjee, A. (2006). Testing real interest parity in emerging markets (IMF Working paper WP/06/249).
Retrieved from https://www.imf.org/external/pubs/ft/wp/2006/wp06249.pdf
Skinner, F. S., & Mason, A. (2011). Covered interest rate parity in emerging markets. International Review of Financial
Analysis, 20(5), 355-363. dx.doi.org/10.1016/j.irfa.2011.06.008
Taylor, A. M. (2002). A century of purchasing-power parity. Review of Economics and Statistics, 84(1), 139-150. dx.doi.
org/10.1162/003465302317331973
Taylor, A. M., & Taylor, M. P. (2004). The purchasing power parity debate. The Journal of Economic Perspectives, 18(4),
135-158. dx.doi.org/10.1257/0895330042632744
Zhang, Y. (2006). Does the horizon matter? – The uncovered interest parity reconsidered. International Journal of Applied
Economics, 3(2), 61-79.
Author Note
Guillermo L. Dumrauf, Universidad del CEMA, Avenida Córdoba 374, C1054AAP, Buenos Aires, Argentina.
Correspondence concerning this article should be addressed to Guillermo L. Dumrauf, Email: gl24@cema.edu.ar
The author expresses his gratitude to the anonymous reviewer(s) for his/her very helpful comments on earlier drafts.
Minimum Wage and Job Mobility in Peru
Nikita Céspedes Reynaga
Banco Central de Reserva del Perú, Lima, Peru
Alan Sánchez
Grupo de Análisis para el Desarrollo, Lima, Peru
Abstract
We study the effects of the minimum wage over employment and income in Peru by considering a monthly
database that captures seven minimum wage changes registered between 2002 and 2011. We estimate that
about 1 million workers earn an income by main occupation in the neighborhood of the minimum wage.
Findings show that the minimum wage-income elasticity is statistically significant; the evidence also
suggests that those who receive low incomes and those working in small businesses are the most affected by
increases in the minimum wage. Employment effects are monotonically decreasing in absolute terms by firm
size: they are moderate in large firms and higher in small firms. Results are robust when assessing the job-
to-job transitions. Finally, we present evidence that supports the hypothesis that the minimum wage in Peru
is correlated with income. The movement of income distribution in the context of changes in the minimum
wage and the results provided by a model that captures the drivers of income justify this finding.
Keywords: Minimum wage, labor mobility, income dynamics, informality
JEL Classification codes: E24, E26, J20, J21, J61
http://dx.doi.org/10.7835/jcc-berj-2014-0094
The minimum wage in Peru was first introduced in 1962.1 Over time, it has gone under different names.
Currently, it is called minimal vital remuneration (remuneración mínima vital - RMV). The study of the dynamic
effects of the minimum wage in the context of the Peruvian economy is of great interest since this country has
experienced a remarkable transformation over the last two decades, including a period of persistent economic
growth (5.5% of average yearly GDP growth during the first decade of 2000) and of labor productivity growth
(Tello, 2012). The minimum wage has been raised several times in the last 10 years. It is difficult to assess
what the effect of the minimum wage policies has been for two reasons. First, the bonanza experienced by the
country might have facilitated the absorption of increases in the minimum wage; second, a labor market that
is still predominantly informal renders difficult the enforcement of regulatory changes.
In this paper, we revisit the effect of the minimum wage on the Peruvian labor market. Our study differs
from previous similar studies (Céspedes, 2006; Chacaltana, 2006; Del Valle, 2009; Jaramillo, 2012; Jaramillo &
López, 2006) in three ways. First, we tracked all the modifications in the minimum wage observed throughout
the last decade. Second, we examined the effect of the minimum wage on a range of outcomes, including
employment status, job mobility, informality, and workers’ income. Job mobility is an important aspect to
Journal of
CENTRUM
Cathedra
JCC
JCC: The Business and Economics Research Journal ● Volume 7, Issue 1, 2014 ● 23-50
24 JCC: The Business and Economics Research Journal
consider, given that it tends to be high in countries with a large proportion of low-skilled workers such as Peru
(Romero & Cruthirds, 2009). Third, we calculated both short- and long-run effects.
In particular, we analyzed seven changes in the minimum wage from 2003 to 2011 by using a comparable
database that records the working status of workers as well as the duration of employment and unemployment
in the context of changes in the minimum wage. Our identification strategy enabled us to capture the changes
in the employment status as well as the income of workers who are directly affected by changes in the mini-
mum wage. This identification was based on the employment status of a panel of workers and their duration
of employment and unemployment.2 This method also provided some evidence of the indirect effects of the
minimum wage on both employment status and income. Hence, our comparable database and our identif ication
strategy enabled us to perform a comprehensive evaluation of the various effects of the changes in minimum
wage over employment, informality, and the workers’ income.
Our purpose was to provide answers to the following relevant questions: Are changes in the minimum
wage important in the job market (in terms of employment and income)? Has the importance of the minimum
wage changed over the last decade? How significant is the minimum wage in terms of job mobility? Does the
minimum wage foster informality? As mentioned earlier, the available studies cover specific periods over
one decade. They also cover a relevant database to identify the effects of the minimum wage and determine
whether their importance has changed over the last decade.
According to the Permanent Employment Survey (Encuesta Permanente de Empleo - EPE), about 20% of
employed workers register job-to-job transitions towards a quarter, after having experienced short spells of
unemployment or short spells of inactivity (out of the labor market) within a quarter. Therefore, we identified
the effects of the minimum wage on employment status (or income) within a quarter when we took job mobil-
ity into account. One issue that is of interest in the context of the Peruvian economy concerns the relationship
between the minimum wage and labor informality. Our procedure enabled us to capture this: those who change
jobs induced by the change in the minimum wage can move from a formal to an informal job within a quarter.
If this change is statistically significant, we can then suggest that the minimum wage fosters informality in
the labor market. Similarly, we considered the heterogeneous effects of the changes in the minimum wage
according to the size of companies and different categories in the job market.
We estimated that about 1 million workers earn an income by main occupation in the neighborhood of
the minimum wage, with a greater participation in some sectors and/or job categories (textiles, manufactur-
ing, construction, trade, house workers, etc.). Findings show that minimum wage changes have statistically
significant effects on employment and income. These results are robust after controlling for observable micro
heterogeneity, aggregate macro variables, and seasonality of employment and income. Our procedure also
enabled us to identify the heterogeneous effect of minimum wage changes according to firm size, employ-
ment status, and income ranges.
This paper is organized as follow. First, we briefly discuss the international evidence and the evidence
available for Peru. Second, we present the data used in the study and provide a profile of those individuals who
earn around the minimum wage in Peru. Third, we illustrate the effects of the minimum wage over income
and/or salaries. Fourth, we study the effects of the minimum wage on employment. Finally, we draw conclu-
sions regarding the effects of the minimum wage over employment and income in Peru.
Literature Review
The minimum wage literature is abundant worldwide. One of the first studies is that by Stigler (1946),
who discussed the potential effect of increasing the post war U.S. minimum wage on labor market outcomes
and on welfare measures. Brown, Gilroy, and Kohen (1982) provided a survey of the early literature. Flinn
(2011) presented a synthesis of more recent contributions from a methodological perspective. Among the most
representative empirical studies are those of Bell (1997), Brown, Gilroy, and Kohen (1982, 1983), Campolieti,
Fang, and Gunderson (2005), Card and Krueger (1994), DiNardo, Fortin, and Lemieux (1996), Eckstein and
Wolpin (1990), Meyer and Wise (1983a, 1983b), Neumark, Schweitzer, and Wascher (1999), Pereira (2003),
and Van den Berg and Ridder (1998).
Prior to the 1990s, most of the empirical evidence suggested that increases in the minimum wage were
harmful for employment. This was the expected outcome of such measures in a competitive labor market.
Brown, Gilroy, and Kohen (1982) focused on the U.S. segment of the population earning the minimum wage
25
Minimum Wage and Job Mobility in Peru
(teenagers and young adults) and presented a synthesis of early studies based on a time-series analysis. They
concluded that for workers 16-19 years of age, a 10% increase in the minimum wage tends to reduce employ-
ment by 1% to 3% (elasticity between -0.1 and -0.3). The elasticity for workers 20-24 years of age was found
to be considerably smaller. Meyer and Wise (1983b) reached a similar conclusion using micro-level data. In
the early 1990s, this evidence was contested in a series of studies summarized in Card and Krueger (1997) and
best exemplified in the case study of the fast-food industry in New Jersey and Pennsylvania (Card & Krueger,
1994). In this case, the authors were not able to detect a negative effect on employment of a marginal increase
in minimum wages. Earlier, Katz and Krueger (1992) had detected a positive effect on employment for this
industry. Studies by Card (1991, 1992) showed similar findings.
A positive effect or a non-effect of a minimum wage increase on employment is theoretically possible
in the context of firms with monopsony power. For instance, Van den Berg (2003) argued that if firms have
monopsony power and there are job search frictions, firms can pay wages that are below the productivity
level of the workers because it takes time for them to find a better paying job. Under those circumstances, the
adoption of a (or an increase in the) minimum wage reduces the degree to which employers can exploit their
monopsony power without necessarily harming employment. Flinn (2006) reached a similar conclusion. The
evidence collected by Card and Krueger (1997) has been influential, and their results can be reconciled with
theory. However, in a review of 102 studies published between 1990 and 2006 collectively known as the “new
minimum wage research,” Neumark and Wascher (2006) noted that in about two thirds of the studies, the
traditional result of a negative effect on employment was still found. For instance, Neumark and Wascher (1991,
1992) exploited variation across states and over time in the United States to find elasticities that corroborate
the findings obtained by Brown, Gilroy, and Kohen (1982). Pereira (2003) used microlevel data and a quasi-
experimental setting for Portugal and found that, for workers 18-19 years of age, the elasticity is between
-0.2 and -0.4. Also exploiting a quasi-experimental setting, Orazem and Mattila (2002) obtained elasticities
between -0.06 and -0.12 for all workers and much larger (between -0.31 and -0.85) for low-wage employees.
Overall, two features emerge from consideration of the international literature. First, most of the evidence
seems to be consistent with the prediction that increases in minimum wages lead to reductions in employment
among the segment of the population that earns a salary close to the minimum wage. Second, there is aware-
ness that the specific effect of an increase in the minimum wage depends on the context. For instance, the
magnitude of the elasticity might vary according to the point in the economic cycle that the country is facing
or according to the proportion of the labor force population that earns an income close to the minimum wage.
The effects of the minimum wage in the Peruvian labor market have been analyzed using a variety of
empirical methods. A key aspect to bear in mind is that in Peru, there is a high concentration of workers whose
earnings are located in the neighborhood of the minimum wage; hence, studies do not need to focus exclu-
sively on the population of teenagers and young adults.3 Chacaltana (2006) provided a survey of the studies
by Céspedes (2006), Jaramillo and López (2006), and Del Valle (2009). Céspedes used aggregated monthly
employment data from the EPE and applied dynamic panel data techniques to calculate the average impact
of the minimum wage over employment, exploiting the changes observed between 1997 and 2003. Jaramillo
and Lopez used individual-level data from the EPE to study the impact of the change in the minimum wage
observed in 2003. They estimated a probability linear model of employment status conditional on having been
employed three months previously. The estimation controlled for individual characteristics, firm character-
istics, month-fixed effects, and quarterly GDP growth. Del Valle used the same database and implemented a
difference-in-difference analysis of the changes observed in the minimum wage in 2003 and 2006, using the
changes observed in the year of no change as counterfactual.
Although the authors of the three studies mentioned above used different techniques, they reached a
qualitatively similar result: increases in the minimum wage lead to reductions in average employment levels.
Céspedes (2006) estimated an average elasticity of -0.13, whereas Del Valle (2009) and Jaramillo and Lopez
(2006) obtained a larger average elasticity (around -0.75 in both cases). Both Del Valle and Jaramillo and
Lopez allowed in their estimation for heterogeneous effects according to the position of the individual in the
wage distribution prior to the policy change. Del Valle found that the increase in the minimum wage has a
larger effect on those earning below or around the minimum wage, whereas in Jaramillo and Lopez, the effect
is larger on those that earn around or above the minimum wage.
One limitation of the studies by Del Valle (2009) and Jaramillo and Lopez (2006) is that in both cases, the
empirical identification relied on only one change in the minimum wage.4 However, there have been several
changes in the minimum wage in the last decade. Since there has also been a persistent economic growth
26 JCC: The Business and Economics Research Journal
during the same time period, it is unclear whether the previous result ought to hold. Another aspect to bear in
mind is that these studies examined only the short-run effect of the change in the minimum wage. Specifically,
they considered as treated (i.e., affected by the policy change) only those individuals observed one or two
months after the change. However, it is possible to be affected beyond this time horizon. For instance, people
working under temporary contracts cannot be fired in the very short run, but eventually might not have their
contracts renewed.
Jaramillo (2012) updated Jaramillo and Lopez (2006) to account simultaneously for the changes observed
in 2003, 2006, 2007, and 2010. Interestingly, in this case, the nature of the conclusions changed. According
to Jaramillo, increases in the minimum wage are found to increase employment for a segment of the informal
workers (those earning slightly above the minimum wage) and to have no effect on formal workers. Given
that the sample was composed of workers who were employed the previous quarter, what this suggests is that
those who had an informal job in one quarter were less likely to lose this employment status in the next quarter
if an increase in the minimum wage was observed. One possibility is that these results could be significant
regarding the effect of changes in the minimum wage on job mobility (transitions from the formal to the
informal sector and vice versa).
The literature in Peru has also provided evidence of the effects of the changes in minimum wage on earn-
ings outcomes. Minimum wage changes can affect the income distribution by directly affecting the income
of formal workers and by indirectly affecting the income of informal workers. This is the so-called lighthouse
effect, which several studies worldwide have shown to be relevant. Kristensen and Cunningham (2006) inves-
tigated the situation in Latin America. For Peru, the relationship between income and minimum wage was
studied by Yamada and Bazán (1994), Jaramillo and López (2006), Jaramillo (2012), Céspedes (2006), among
others. However, as in the previous case, these studies based their identification on one specific increase in
the minimum wage observed in 2003. The exception was Yamada and Bazan (1994) and Céspedes (2006),
who used a time-series econometric approach. However, time-series analyses performed at the macro level
may not capture the distributional effects of minimum wage changes.
The Data
The data source for the study is the EPE; the survey is performed on a monthly basis by the Peruvian
National Bureau of Statistics (Instituto Nacional de Estadística e Informática - INEI). The EPE is a survey
specially conceived to trace labor market-related aspects in the Lima Metropolitan Area. This geographic
area includes 43 districts in the Province of Lima and six districts in the Constitutional Province of Callao.
One of the main characteristics of this survey is that the individuals who are interviewed each month
include a share of the sample of people who were interviewed three months previously. The panel sample
rotates partially each quarter in such a way that individuals in the panel sample are interviewed twice in
two consecutive quarters. In this study, we built a sequence of the quarterly unbalanced panel samples, from
the first quarter of 2003 to the first quarter of 2012. For the analysis, we considered only individuals who
had reported having a job in the previous interview. After missing values in some demographic and labor
market-related variables, the panel sample built in this way showed a total of 97 547 individuals, of whom
82 552 (84.6%) were employed at the time of the most recent interview. The rest were unemployed or inactive
workers. For the income analysis, in some instances, we focused on those individuals for whom an income
different from zero was observed in both occasions the individuals were interviewed. In this case, the sample
size reduces to 76 282 people.
The level of inference of the quarter panel data is statistically significant, since approximately 30% of the
total sample was part of the panel. For example, the size of the quarter sample in the EPE was 4 800 households
in the year 2011 and 1 500 households in 2001. Hence, the total quarter sample was of about 18 500 people
in 2011. Additionally, the size of the quarter sample of the EPE has being increasing over time. As a result,
the estimates obtained from EPE are currently more precise than they were at the beginning of the survey.
Descriptive Statistics and the Prole of Workers Earning Around the Minimum Wage
The data from the EPE was used to characterize those individuals with an income around the minimum
wage in the Lima Metropolitan Area. Table 1 shows that all workers, whether in the formal or the informal
sectors, were included. Hence, were considered not only those workers earning the minimum wage, but also
27
Minimum Wage and Job Mobility in Peru
those earning about the same level of monthly income by informal arrangement. As an operational definition,
workers who earn a monthly income that is between above and below the minimum wage (+/- 100 Nuevos
Soles, the currency of Peru) are considered as “workers around the minimum wage” and shown as Group B
in Table 1.5 We used data from a pooled sample of EPE surveys (from the first quarter of 2007 to the fourth
quarter of 2009) in order to increase the sample size.6
Table 1
Employed Population by Income Range
Number of individuals
(in thousands) %
Group A:
Below the minimum wage (wage earners and independents) 1495.3 26
Group B:
Around the minimum wage (wage earners and independents) 997.9 18
Group C:
Above the minimum wage (wage earners and independents) 3194.7 56
Group D:
Total (wage earners and independents) 5687.9 100
Source: EPE 2007-2009, INEI.
Note. Results correspond to the Lima Metropolitan Area. The population of reference is the average population extrapolated
from the EPE for the years 2007, 2008, and 2009. We used data from a pooled sample of years to produce these statistics in
order to increase the sample size. This is important for the analysis because the sample is divided into a large number of cells.
Table 2 shows the demographic profile of workers and the type of economic activities in which individuals
earning an income close to the minimum wage are involved.
Table 2
Prole of Workers
Group A Group B Group C Group D
Age (in years)
Average 34.9 34.0 36.7 36.1
Standard deviation 15.9 13.0 12.1 17.0
Gender (in %)
Male 35.6 53.7 65.3 55.4
Female 64.4 46.3 34.7 44.6
Access to health insurance
Has health insurance 21.8 27.6 52.5 40.1
No health insurance 78.2 72.4 47.5 59.9
Firm size (in %)
n < 100 95.0 81.7 64.2 75.3
n > = 100 5.0 18.3 35.8 24.7
Type of occupation (in %)
Independent 44.6 34.7 27.0
Blue-collar 13.2 24.6 19.1
White-collar 17.1 34.6 44.9
House worker 5.3 6.1 5.4
Others 19.7 0.0 3.7
Source: EPE 2007, 2008, and 2009, INEI.
Note. Results correspond to the Lima Metropolitan Area. We used data from a pooled sample of years to produce these
statistics in order to increase the sample size.
28 JCC: The Business and Economics Research Journal
Findings for Group B show that approximately 18% of the employed population (around 1 million people)
earns an income within the minimum wage of +/- 100 Nuevos Soles (see Table 1). Table 2 shows that these
individuals, on average, are younger compared to the population of reference (34 versus 36 years of age); 53.7%
are men; most of them (81%) work in relatively small firms and lack health insurance (72%). In terms of job
categories, 35% self-report as white-collar workers, 32% as independents, and 25% as blue-collar workers.
Table 3 indicates the economic sectors in which those earning around the minimum wage work. Group B
is well diversified within occupations, including independent workers in the retail sector, blue-collar workers
in the manufacturing sector, house workers, among others.
Table 3
Workers Around the Minimum Wage (Group B) by Type of Occupation and Economic Sector
Independent White-collar worker Blue-collar
worker
House-
worker
Primary 0.1 0.1 0.6 0.0
Manufacture 3.3 2.9 12.7 0.0
Electricity 0.0 0.0 0.0 0.0
Construction 2.0 0.2 2.7 0.0
Retail and wholesale 13.5 10.8 2.5 0.0
Hotels / restaurants 2.7 2.1 1.6 0.0
Transportation 5.8 2.8 1.5 0.0
Other services 4.8 16.8 3.8 6.3
Sub-total 32.3 35.8 25.5 6.3
Source: EPE, INEI.
Note. Results correspond to the Lima Metropolitan Area. We use data from a pooled sample of years to produce these
statistics in order to increase sample size.
Minimum Wage and Income
We used recent information that allowed us to identify some of the regularities of the effects of the mini-
mum wage over workers’ income, which helped to complement current knowledge regarding the effects of the
minimum wage in Peru. We also examined the lighthouse effect of the minimum wage, namely, the hypothesis
that the minimum wage in Peru is a benchmark in determining the income of individuals. The Peruvian data
suggests that the changes in minimum wage are related to future movements or adjustments in the workers’
monthly income. This could suggest that there is a statistical correlation linking the minimum wage to the
income of workers.
Minimum Wage and Mean Income
The minimum wage imposes a friction in the labor market and becomes a relevant variable when the
equilibrium wage and the minimum wage are close enough. This would be a particular case to bear in mind
for Peru where the value of the minimum wage represents 60% of the average income, or alternatively, 70%
of the median income (see Figure 1). This ratio has shown an upward trend during most of the 2000s. Data
from the Peruvian Ministry of Labor (Ministerio de Trabajo y Promocion Social - MTPS) shows that this
tendency has been registered since 1993. Nevertheless, at the end of the 2000s and at the beginning of 2010,
we find a slight reduction in this ratio, such that the levels are similar to those at the beginning of the 2000
decade. This characteristic is evident with different indicators of the salary such as the estimated income by
the EPE, or the gross domestic product (GDP) per capita, or the income and salaries estimated by the MTPS
for workers employed in companies of 10 or more workers.
29
Minimum Wage and Job Mobility in Peru
Note. INEI, Central Bank of Peru (Banco Central de Reserva del Peru, BCRP).
Figure 1. Ratio minimum wage - income.
This regularity can be explained as follows: during the first seven years of the 2000s, the minimum wage
policy was very active, and the changes were proportional to the average income increases. Between 2008
and 2010, no changes in the minimum wage were registered, and the significant growth in average income
drove the negative trend of this ratio. After two changes in minimum wage (2010 and 2011), there was a
slight growth in this ratio. On average, the ratio minimum wage/income in 2011 was similar to the ratio at
the beginning of the 2000s.
In what follows, we provide microeconomic evidence that comes from the last seven changes in the mini-
mum wage in Peru that suggest a significant correlation between minimum wage and the average income
in the economy. Even though the evidence comes from Lima, we claim that the minimum wage works as an
important benchmark in the determination of salaries because most individuals with formal jobs seem to earn
around the minimum wage.
At the end of the 2000s, the concentration of workers earning an income close to the minimum wage is
higher than that at the beginning of the 2000s. This increase implies that at the end of the decade, the changes
in the minimum wage had a larger effect on income, and this is particularly true for the formal workers.
INEI data show that these regularities are related to the increase of the number of salaried workers and to the
reduction of informality in the labor market during the 2000s (Rodriguez & Higa, 2010). Figure 2 compares
the income distribution around the minimum wage in 2003 and 2011. The distribution is narrower near the
neighborhood of the minimum wage in 2011, which may suggest that there is a tendency to earn salaries closer
to the minimum wage.
An additional element which illustrates the direct and/or indirect short-run effects of changes in the mini-
mum wage is measured by comparing the distribution of income before and after the changes in the minimum
wage. The panel sample from the EPE enabled us to identify the employment status and the income of workers
before and after the changes in the minimum wage. This procedure, however, helps to capture only the short-
run distributive effects of the minimum wage since only the income of two consecutive quarters are being
compared. For example, Figure 3 illustrates this comparison for the change in minimum wage in August 2011
and shows a displacement towards higher salaries in the neighborhood of the minimum wage, while the rest
of the distribution does not experience significant changes. The lack of changes is more pronounced amongst
formal workers while informal workers experience marginal changes. This analysis was repeated for the last
seven changes in the minimum wage, and similar results are found in six out of seven of the cases, as Figures
6 and 7 indicate (see Appendix A).
30 JCC: The Business and Economics Research Journal
The Peruvian labor market regulation allows a certain degree of indexation in the minimum wage with
some components in the salary, in such a way that the increases in the minimum wage have direct effects over
some workers, mainly workers in the formal sector, even if we consider that in the aggregate they earn more
than the current minimum wage.7 Among these salary components, the one which would have a larger cover
would be family compensation because it is not proportional to the income.
Note. Income frequencies (EPE, Lima Metropolitan Area). The vertical lines represent the minimum wage in 2003 or
2011, respectively. Kernel Epanechnikov function.
Figure 2. Main job income, frequencies 2003 and 2011b.
Note. Frequencies before and after the current minimum wage rise (EPE, Lima Metropolitan Area). The vertical line
represents the minimum wage in 2011. Kernel Epanechnikov function.
Figure 3. Main job income, frequencies 2011b.
According to the characterization of workers by income around the minimum wage, approximately 18%
of workers would be directly affected by changes in the minimum wage, while the rest of the workers, mostly
in the informal sector, would be indirectly affected. Figure 3 shows that there is no clear clustering of salaries
around the minimum wage in the informal market. The average informal income is close to the minimum
31
Minimum Wage and Job Mobility in Peru
wage, and the distribution of informal salaries is displaced in a similar proportion to the changes in the mini-
mum wage (see Figure 7 in Appendix A). This would alter the effects of the minimum wage in the long run.
Minimum Wage and Income: A Formal Model
In order to assess the relationship between minimum wage and income in a more robust way, an equation
of income determinants at the level of the workers was estimated. This equation includes several controls to
capture demographic characteristics, income heterogeneity of workers, income seasonality, and the business
cycle. In the EPE, a share of the individuals being surveyed is interviewed twice to condition the analysis on
some characteristics from the first interview. The specification is as follows:
log Yi,y,m|(Ei,y,m–3 = 1) = ay + am + blogRMVy,m + XiΩ + µi,y,m, (1)
where log Yi,y,m is the log of monthly income of individual i interviewed in year m, month m; Ei,y,m–3 is the
employment status of the individual three months ago (1 if employed, 0 otherwise); logRMVy,m is the log of the
minimum wage prevalent in the same time period; Xi is a vector of controls that include gender, educational
attainment, years of experience (including a quadratic term), a dummy for whether the individual is the head of
the household, and the following characteristics, observed three months previously: job category (independent,
white-collar worker, blue-collar worker, house worker, and other categories), number of employees in the fir m,
and individual income divided by the minimum wage. The last two variables and educational attainment are
included by categories. The model is estimated conditional if the individual reports having a job in an interview
three months previously. The model also includes yearly and monthly fixed effects (ay and am, respectively),
which allowed us to control for trends in income over time (possibly associated with business cycles) and for
the seasonality of economic activities. The coefficient of interest is b, which reflects the overall effect of a
change in the minimum wage over average income, not just the short-run effect.
Based on this specification, we estimated Equation 1 for all individuals reporting an income in both
periods; in other words, these individuals belong to Group D in both periods.8 The sample size was 76 282.
We obtained a statistically significant minimum wage to income elasticity with a point estimate of 0.25 (see
Table 8, Appendix B). In other words, a 10% increase in the minimum wage increases income by 2.5%. This
figure, however, reflects an average effect. Those individuals who earn significantly more than the minimum
wage are less likely to be affected by the increase. Similarly, informal workers might not benefit or might
benefit only partially from the increase.
Minimum Wage and Employment
In this section, we examine the relationship between minimum wage and employment. As mentioned before,
the general conclusion for the Peruvian case is that the minimum wage has a negative effect on employment.
In order to examine this relationship, we used the information provided by the EPE, which enabled us to track
labor transitions in the context of changes in the minimum wage. We were able to capture not only the transi-
tions from employment to unemployment and/or to inactivity but also those from employment to another job,
namely job-to-job transitions. We used the job duration data to estimate the short-term job-to-job transitions
in the context of a changing minimum wage.
The previous point is particularly important in Peru because the employment aggregate statistics cannot
capture adequately the short-term job mobility which may be driven by changes in the minimum wage. The
employment status of the same worker is observed with a lag of three months. These two observations of the
same worker do not allow us to determine whether this worker has experienced a short spell of unemploy-
ment. In a context of changes in the minimum wage, it is possible to observe the same individual working
before and after the change in minimum wage, and if we do not control for this short-term unemployment
spell, we cannot observe the job lost due to rise of the minimum wage. Given that the unemployment duration
in Peru is short, between 12-15 weeks (Céspedes, Belapatiño, & Gutiérrez, 2013; Chacaltana, 2000; Díaz &
Eduardo, 2000),
9
the quarterly separation between two consecutive observations of the employment status
does not enable us to identify the likely destruction (or not) of jobs due to a change in minimum wage. An
estimate of estimate job-to-job transitions is necessary in order to determine the role of the minimum wage
in employment transitions.
32 JCC: The Business and Economics Research Journal
The importance of job-to-job transitions to determine the short-run effects of the minimum wage on
employment is indicated in Figure 4, which shows the impact of the increase in the minimum wage in 2011.
This graphical analysis compares the transition of those individuals who are observed before and after the
change in the minimum wage (the treatment group shown by a dashed line) with the same transition observed
for a control group (shown by a continuous line) the previous year.
10
Only the short-run transition of 1-2
months after the policy change is captured because individuals are interviewed in two consecutive quarters.
The second graph of this figure shows the job-to-others category of transitions (unemployment, inactivity, or
other jobs) across the income range.11 This figure shows that the job-to-others category of mobility induced
by changes in the minimum wage does not seem to be significant for this indicator, as the difference between
both groups is small. Results are markedly different when considering only job-to-job transitions induced by
the change in the minimum wage in 2011. This situation is shown in the first graph of Figure 4, where higher
job mobility is observed in the treatment group compared to the control group across most of the income
range. It is worth noting that in the extremes of the income distribution, job mobility is similar for both the
treatment and the control group.
Source: EPE, INEI.
Note. The figure represents the proportion of employed workers who change to another labor category by income range
(panel a) and job-to-job transitions by income range (panel b) (EPE, Lima Metropolitan Area). The x axis shows frac-
tions of the current minimum wage. The dashed line represents the quarterly job mobility indicator of the treatment
group, before and after the current minimum wage increase, while the continuous line denotes the control group, which
is the quarterly job mobility indicator in the same months a year previously.
Figure 4. Job transitions, 2011b.
The procedure depicted in Figure 4 is applied to all the registered changes in the minimum wage during
the 2000s, and the results are consistent with the ones previously mentioned in the majority of the cases, with
the exception of 2008 as is shown in Figure 8 (see Appendix B). This reinforces the argument that the short-
run effects of the minimum wage over job mobility are registered mostly in the neighborhood of the current
minimum wage.
This analysis can be extended to other indicators of transitions of the labor market. For instance, in the
case of unemployment-to-employment transitions, an increase in the minimum wage may reduce the job
creation for those workers expecting to receive an income close to the minimum wage. There is no support
for this hypothesis, however. As shown in Figure 12 (see Appendix B), there is no strong movement in the
neighborhood of the minimum wage. Similarly, Figures 10, 11, and 13 (see Appendix B) show that minimum
wage changes may not have a clear effect in other employment transitions.
While the results of the graphical analysis are suggestive, they capture only the short-run impact of the
policy change, namely the impact of the increase in the minimum wage after one or two months. However, it is
possible that an increase in the minimum wage might affect employment or job mobility beyond this horizon.
Hence, the next step was to calculate the overall impact of changes in the minimum wage over employment
and job mobility in a more formal framework.
33
Minimum Wage and Job Mobility in Peru
Minimum Wage and Employment: A Formal Model
Using the previous results as motivation, we estimated a discrete response Probit model to capture the rela-
tionship between the minimum wage and the employment status. We considered the following functional form:
Pr(Ei,y,m = 1| Ei,y,m–3 = 1) = G(ay + am + … + ρ RMVy,m + XiΩ + µi,y,m), (2)
where Pr(Ei,y,m) takes the value of 1 if individual i is employed in month m of year y. G(.) is the cumulative
distribution function of the standard normal distribution. RMVy,m is the prevalent minimum wage in the same
time period; X
i
is a vector that contains the same control variables used in Equation 2. As in the model reported
in Equation 1, this model also includes yearly and monthly fixed effects (ay and am, respectively) and is esti-
mated conditional on the individual having had a job as reported in an interview three months previously.
The result of interest is the elasticity of the minimum wage to the probability of being employed, conditional
on having a job three months previously.12
Based on this specification, Equation 2 was estimated for all individuals fulfilling the condition of having
a job three months previously, namely those who belong to Group D in the first interview. The sample size
was 97 547. In Table 9, Column 1 (see Appendix B), the coefficients associated with the model described in
Equation 2 are shown using data from EPE (Lima Metropolitan Area). These results imply a negative, statis-
tically significant relationship between minimum wage and employment. In Column 2, the model allows for
differential effects according to job category: independent, blue-collar, white-collar, house workers, and other
categories. In this case, results suggest that the relationship initially found also holds for independent workers.
Table 4 shows the elasticities derived from these two models. The minimum wage- employment elasticity
for the average individual in the sample is -0.25. In other words, a 10% increase in the minimum wage reduces
employment by 2.5%. The highest values of elasticity are observed for those individuals who self-reported as
blue-collar and white-collar workers, whereas those who self-reported as independent workers are the ones
least affected by changes in the minimum wage.
Table 4
Minimum Wage and Employment Status: Elasticities (Lima Metropolitan Area)
Coef. Std. Err. t-stat p-value
Model 1
Average -0.256 0.057 -4.430 0.000
Model 2
By type of occupation (3 months previously):
Independent worker -0.199 0.049 -4.010 0.000
White-collar worker -0.317 0.068 -4.610 0.000
Blue-collar worker -0.332 0.082 -4.040 0.000
House worker -0.247 0.072 -3.430 0.001
Other categories -0.227 0.057 -3.970 0.000
Note. The coefficients from which these elasticities were estimated are reported in Table 9 (Column 1 for Model 1 and
Column 2 for Model 2). All the control variables are kept at their average levels. The sample size is 97 547. The data come
from the EPE (January 2003 to March 2012). The sample consists of all individuals who are observed twice in the EPE
and who were employed the first time they were observed.
The average effect of the minimum wage on employment is likely to mask some heterogeneity. A priori,
those individuals with a formal job are more likely to be affected because formal firms are required by law to
conform to minimum wage policies. Similarly, people who earn the minimum wage, or around it, are likely
to be the target of job cuts. To take into account these possibilities, we reestimated our employment model,
allowing for heterogeneous minimum wage effects according to the following characteristics three months
previously: (a) whether or not the individuals had health insurance in their job (a proxy of formal employ-
ment), (b) the position of the individuals in the income/minimum wage ratio distribution, and (c) the size of
34 JCC: The Business and Economics Research Journal
the firm. Results for (b) and (c) are shown graphically in Figure 5. Full results are reported in Table 10 (see
Appe ndix B).13 Findings show that workers without health insurance, with lower income levels, and who work
in small firms are the ones most affected by increases in the minimum wage. Both those individuals earn-
ing around the minimum wage and those earning less than the minimum wage are affected. In fact, results
suggest that those individuals earning less than the minimum wage are the ones most affected. In contrast,
those earning more than four times the minimum wage are not affected.
To check whether a similar relationship between increases in the minimum wage and changes in employ-
ment is found at the national level, we used data from the Peruvian National Household Survey (Encuesta
Nacional de Hogares - ENAHO) to produce estimates of this elasticity, distinguishing between rural areas,
urban areas (excluding Lima), and the Lima Metropolitan Area. For this exercise, we could not replicate the
model specified in Equation 2. The ENAHO provided only one observation for each individual; hence, it was
not possible to condition the analysis on individuals’ having been employed t months before, nor to control
for the characteristics of the occupation (firm size, income earned) at that moment in time. Thus, results are
not entirely comparable due to differences in the population of reference. Additionally, the data used for this
estimation was a pooled sample from ENAHO corresponding to the years 2003 to 2010,14 so two of the seven
changes in the minimum wage observed over the last 10 years were not included in the calculations. With
these caveats in mind, it is worth noting that we obtained qualitatively similar findings in this case. Results
are reported in Table 5. For the Lima Metropolitan Area, we obtained a negative and statistically significant
elasticity, albeit slightly smaller than that obtained using data from EPE: -0.16. An almost identical result was
obtained for urban areas (excluding the Lima area). In contrast, an elasticity not statistically different from
zero at standard confidence levels was obtained for rural areas. This result was expected since labor markets
are less formalized in those areas of the country.
a) By income range b) By firm size
Note. Both graphs show minimum wage-employment elasticities. In the graph on the left, elasticities are reported by
relative income groups (the relative income is the individual income reported three months before the interview divided
by the minimum wage prevalent then). In the graph on the right, individuals are classified according to the size of the
firm where they worked three months previously.
Figure 5. Heterogeneity of elasticities by rm size and relative income (Lima Metropolitan Area).
Table 5
Minimum Wage and Employment Status: Main Elasticities at the National Level
Coef. Std. Err. t-stat p-value
Lima Metropolitan Area (Lima) -0.162 0.099 -1.630 0.103
Urban areas (excluding Lima) -0.155 0.085 -1.810 0.070
Rural areas -0.066 0.051 1.280 0.202
Note. The coefficients from which these elasticities were estimated are shown in Table 11. All the control variables are
kept at their average levels. The ENAHO is the source of the data.
35
Minimum Wage and Job Mobility in Peru
Our methodology is not directly comparable with that used by Del Valle (2009), Jaramillo and Lopez
(2006), and Jaramillo (2012). In those studies, a treatment group was defined that included individuals observed
before and after the policy change. A characteristic of that strategy is that only the short-run impact of the
policy change is captured, since those who are treated are observed one to two months after the increase in
the minimum wage. However, the effects of the policy change are not necessarily restricted to the following
one to two months after the event.
In our estimations, we followed a different route to estimate how employment status changes as the mini-
mum wage increases for all individuals who had a job three months previously. Since no specific treatment
group was defined, findings show the overall impact of the policy change, not just the short-run impact. This
has consequences for the interpretation of the results. If jobs that are destroyed by the increase in the minimum
wage can be recovered relatively quickly, the short-run elasticity will be larger than our estimates (in absolute
terms). Conversely, if the increase in the minimum wage makes workers more likely to lose their job a few
months after the policy change, the short-run elasticity will be smaller than our estimates (in absolute terms).
To check whether the short-run elasticity is smaller or larger than our overall elasticity, we reestimated
our main specification, defining a treatment variable that takes the value of 1 for those individuals who are
observed one or two months after a change in the minimum wage and 0 otherwise (see Table 6). In so doing,
we obtained an average elasticity of -0.13. The point estimate is not statistically different from zero. When we
calculated the elasticity, allowing for heterogeneity by type of occupation, an average elasticity of -0.46 was
obtained for white-collar workers, a result that is statistically significant. For the other groups (independent
workers, blue-collar workers, and house workers), the elasticities obtained are not statistically significant. This
finding is markedly different from previous results, which showed a larger average elasticity as well as elas-
ticities that were statistically significant for all the subgroups by type of occupation. The difference between
the two sets of results suggests that an increase in the minimum wage has wider implications on employment
status that are not necessarily apparent in the short-run.
Table 6
Minimum Wage and Employment Status: Short-Term Elasticities (Lima Metropolitan Area)
Coef. Std. Err. t-stat p-value
Model 1
Average -0.129 0.096 -1.330 0.184
Model 2
By type of occupation (three months before):
Independent worker 0.092 0.117 0.780 0.437
White-collar worker -0.465 0.159 -2.920 0.004
Blue-collar worker 0.075 0.214 0.350 0.725
House worker -0.535 0.351 -1.520 0.128
Other categories -0.203 0.203 -1.000 0.318
Note. Elasticities are estimated from the following Probit model:
Pr(Ei,y,m = 1| Ei,y,m–3 = 1) = G(ay + am + … + ρ CHANGE y,m + XiΩ + µi,y,m),
where CHANGE takes the value of 1 for those individuals who are observed before and after a change in the minimum
wage and 0 otherwise; all the other variables are defined as before. The sample size is 97 547. The data come from the
EPE (January 2003 to March 2012). The sample consists of all individuals who are observed twice in the EPE and who
were employed the first time they were observed.
Minimum Wage and Labor Mobility
A change in the minimum wage might affect employment in ways that are not captured by the previous
definition of employment status (1 if employed at the time of the interview, 0 otherwise, conditional on having
had a job three months previously). People who lose their job could find a new one quickly. Depending on the
exact timing of the household survey interviews and the changes in the minimum wage, it is possible that people
36 JCC: The Business and Economics Research Journal
who lost their job because of the increase in the minimum wage could have found a new one by the time of the
interview. If this is the case, the previous results would be a lower bound of the true minimum wage - employ-
ment elasticity. To take this possibility into account, we estimated the change in the probability of retaining
the same job compared to the alternative of having a new job.15 Because this is a selected sample composed
of individuals who have a job in both periods, we also show results comparing the probability of retaining
the same job versus either having a new job, being unemployed, or being inactive. This second definition of
employment status makes comparison with previous results possible. These elasticities are reported in Table 7.
Table 7
Minimum Wage and Job Transitions: Elasticities Using Alternative Denitions of Employment Status (Lima
Metropolitan Area)
Coef. Std. Err. t-stat p-value
Dependent variable, alternative 1:
1 if retained the same job (compared to 3 months before), 0 if in a different job. -0.071 0.061 -1.160 0.245
Dependent variable, alternative 2:
1 if retained the same job (compared to three months before), 0 if in a different
job, unemployed or inactive. -0.304 0.084 -3.590 0.000
Note. Each elasticity is estimated from a different model where the definition of employment status changes slightly.
As expected, when using a definition similar to that presented in Equation 2, we obtained a larger elasticity
(in absolute value). In contrast, when restricting the comparison to those individuals who had a job in both
periods, the elasticity is smaller and becomes statistically insignificant. This difference might stem from the
fact that this is a selected sample of workers with higher job stability.
Minimum Wage and Informality
The previous model was modified to capture the transition from formal to informal employment. The
dependent variable was defined as the probability of maintaining a formal job compared to having an informal
job, being unemployed, or being inactive at the moment of the interview. The model was estimated conditional
on people having a formal job before the change in the minimum wage. The explanatory variables are the
same as before.
Lacking additional information about the type of contract a worker has, for practical purposes, we defined
formality as having health insurance (public or private). This is only a proxy for formality: Indeed, it is possible
for a worker with health insurance to work in the informal sector (e.g., a worker can buy health insurance).
When estimating the model using access to health insurance as a proxy for working in the formal sector, no
evidence was found to support the claim that an increase in the minimum wage leads to a reduction in the
average proportion of the population with formal jobs. In fact, the elasticity has a positive sign; however, it
is statistically insignificant.16 In other words, it does not seem to be the case that an increase in the minimum
wage leads to more informality. Given that our informal employment indicator is weak due to data limitations,
we treat this result with caution.
Conclusions
We have examined the effects of the minimum wage over income and employment in Peru by considering
the seven changes registered between 2002 and 2011. The source of the data comes was the EPE for the Lima
Metropolitan Area and the ENAHO for the national analysis. We merged the information provided by the
monthly household survey and were able to measure the job-to-job transitions as well as the income dynamics
due to minimum wage changes. We estimated that about 1 million workers have an income by main occupation
in the neighborhood of the minimum wage, with a greater participation in some sectors and/or job categories
(textiles, manufacturing, construction, trade, house workers, etc.).
37
Minimum Wage and Job Mobility in Peru
Using a model that explains the probability of being employed, we estimated statistically significant mini-
mum wage-employment elasticity for the average worker. Although on average, both formal and informal
workers are affected by minimum wage increases, those individuals seemingly engaged in formal activities
are hit harder. The evidence also suggests that those who receive low incomes and those working in small busi-
nesses are the most affected by increases in the minimum wage. Effects are monotonic, decreasing in absolute
terms by firm size; the effects of minimum wage changes are moderate in big firms and higher in small firms.
The minimum wage - employment elasticity is larger in absolute value (more negative) when assessing
the probability that individuals are working in the same job in both periods. This finding suggests that part of
the effect of the minimum wage changes on employment is cleared due to the ability of individuals to reinsert
quickly in a dynamic labor market; it is worth remembering the persistent economic growth during the decade
under consideration. When considering informality, findings show that the increases in the minimum wage
do not appear to reduce the probability of people being formally employed. However, this result needs to be
revisited with proper data, given that our informal employment indicator is weak due to data limitations.
Finally, we have presented evidence for the hypothesis that the minimum wage in Peru is a benchmark for
determining the income of individuals (i.e., the lighthouse effect). The movement of income distribution in
the context of changes in the minimum wage and the results provided by a model that captures the drivers of
income justify this result.
Endnotes
1 Source: Peru National Bureau of Statistics.
2 In order to identify the effects of the minimum wage on employment, we need to observe whether the employed work-
ers are still employed after the change in the minimum wage. The database that captures this effect is the EPE (Encuesta
Permanente de Empleo). This survey registers the employment status of a group of workers twice, two monthly based
observations with a 3-month lag. By using the duration of employment, we can determine whether the workers do not
experience job-to-job transitions. By using this database, once the minimum wage changes, we can observe the employ-
ment status of the workers after three months. The data cover only the Lima Metropolitan Area, which can be a limita-
tion since this area represents only 30% of the population. The data covering Peru are found in the ENAHO (Encuesta
Nacional de Hogares). However, this database has yearly base panel observations only; these may not be adequate to
capture the short-term effects of minimum wage changes. The EPE allowed us to capture both short- and long-term
effects of the minimum wage changes.
3 Approximately 1 million workers may be exposed to minimum wage changes in the Lima Metropolitan Area, in the
sense that their income is in the neighborhood of the minimum wage.
4 Del Valle (2009) performed separate estimations for 2003 and 2006.
5 We used this approach to deal with measurement error. The question about income in EPE does not distinguish between
gross income and income after taxes and other deductions.
6 Although data from other years are available, only from to 2007 to 2009 can we observe a harmonic International
Standard Industrial Classification of All Economic Activities (ISIC); hence, we used data for these years only to
produce these descriptive statistics.
7 Among these concepts, we find family compensation (10% of minimum wage), intern minimum wage (25% above
the current minimum wage), journalist minimum wage (three minimum wages), minimum wage for night (30% above
minimum wage), and Essalud payments (9% of minimum wage).
8 In practice, some workers reporting zero income also have to be excluded.
9 The duration of unemployment estimated from the EPE has similar values with a decreasing trend during most of the
decade (Céspedes, Belapatiño, & Gutiérrez, 2013).
10 This controls for the seasonality of job mobility in a simple manner.
11 The income range is defined according to the income prevalent prior to the change in the minimum wage.
12 In this set of estimations, the nonemployed status includes the unemployed as well as those individuals who self-report
as inactive (out of the economically active population). We considered both categories because we were already condi-
tioning the analysis to having had a job three months previously, which already excludes the structural proportion of
the population that is not actively looking for a job.
13 See coefficients of the Probit model in Table 10, Appendix B.
14 The data from ENAHO 2011 were not available at the time this analysis was produced.
38 JCC: The Business and Economics Research Journal
15 In the dataset, it is possible to know for how long individuals have been in their current job and whether they had a
job three months previously. If they had a job three months previously but have worked less than three months in their
current position, we assumed there was a job transition.
16 We obtained elasticity for the average worker of 0.07 with a standard error of 0.06.
References
Bell, L. A. (1997). The impact of minimum wages in Mexico and Colombia. Journal of Labor Economics, 15(S3),
S102-S135. dx.doi.org/10.1086/209878
Brown, C., Gilroy, C., & Kohen, A. (1982). The effect of the minimum wage on employment and unemployment. Journal
of Economic Literature, 20(2), 487-528. Retrieved from http://www.jstor.org/stable/2724487
Brown, C., Gilroy, C., & Kohen, A. (1983). Time-series evidence of the effect of the minimum wage on youth employment
and unemployment. The Journal of Human Resources, 18(1), 3-31. dx.doi.org/10.2307/145654
Campolieti, M., Fang, T., & Gunderson, M. (2005). Minimum wage impacts on youth employment transitions, 1993–1999.
Canadian Journal of Economics, 38(1), 81-104. dx.doi.org/10.1111/j.0008-4085.2005.00270.x
Card, D. (1991). Do minimum wages reduce employment? A case study of California, 1987-89. Industrial and Labor
Relations Review, 4(1), 38-54. Retrieved from http://www.jstor.org/stable/2524737
Card, D. (1992). Using regional variation in wages to measure the effects of the federal minimum wage. Industrial and
Labor Relations Review, 46(1), 22-37. dx.doi.org/10.2307/2524736
Card, D., & Krueger, A. B. (1994). Minimum wages and employment: A case study of the fast-food industry in New
Jersey and Pennsylvania. The American Economic Review, 84(4), 772-793. Retrieved from http://www.jstor.org/
stable/2118030
Card, D., & Krueger, A. B. (1997). Myth and measurement: The new economics of the minimum wage. Princeton, NJ:
Princeton University Press.
Céspedes, N. (2006). Efectos del salario mínimo en el mercado laboral peruano. Revista de Estudios Económicos, 13. Retrieved
from http://www.bcrp.gob.pe/docs/Publicaciones/Revista-Estudios-Economicos/13/Estudios-Economicos-13-5.pdf
Céspedes, N., Belapatiño, V., & Gutiérrez, A. P. (2013). Determinantes de la duración del desempleo en una economía
informal. Lima, Peru: Banco Central de Reserva del Perú.
Chacaltana, J. (2000). Un análisis dinámico del desempleo en el Perú. Lima, Peru: MECOVI-INEI.
Chacaltana, J. (2006). ¿Qué hacemos con el salario mínimo? Economía y Sociedad, 60(9), 12-21.
Del Valle, M. (2009). Impacto del ajuste de la remuneración mínima vital sobre el empleo y la informalidad. Revista de
Estudios Económicos, 16, 83-102.
Díaz, J. J., & Eduardo, M. (2000). La dinámica del desempleo urbano en el Perú: Tiempo de búsqueda y rotación laboral.
Lima, Peru: CIES.
DiNardo, J., Fortin, N. M., & Lemieux, T. (1996). Labor market institutions and the distribution of wages, 1973-1992: A
semiparametric approach. Econometrica, 64(5), 1001-1044. dx.doi.org/10.2307/2171954
Eckstein, Z., & Wolpin, K. I. (1990). Estimating a market equilibrium search model from panel data on individuals.
Econometrica, 58(4), 783-808. dx.doi.org/10.2307/2938350
Encuesta Nacional de Hogares. (2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010). Lima, Peru: INEI.
Encuesta Permanente de Empleo. (2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011, 2012). Lima, Peru: INEI.
Flinn, C. J. (2006). Minimum wage effects on labor market outcomes under search, matching, and endogenous contact
rates. Econometrica, 74(4), 1013-1062. dx.doi.org/10.1111/j.1468-0262.2006.00693.x
Flinn, C. J. (2011). The minimum wage and labor market outcomes. Cambridge, MA: The MIT Press. dx.doi.org/10.7551/
mitpress/9780262013239.001.0001
Jaramillo, M. (2012). Ajustes del mercado laboral ante cambios en el salario mínimo: La experiencia de la década de 2000
(Documentos de Investigación No. 63.). Lima, Peru: GRADE.
Jaramillo, M., & López, K. (2006). ¿Cómo se ajusta el mercado de trabajo ante cambios en el salario mínimo en el Perú?
Una evaluación de la experiencia de la última década (Documento de trabajo No. 50). Lima, Peru: GRADE.
Katz, L. F., & Krueger, A. B. (1992). The effect of the minimum wage on the fast food industry. Industrial and Labor
Relations Review, 46(1), 6-21. dx.doi.org/10.2307/2524735
Kristensen, N., & Cunningham, W. (2006). Do minimum wages in Latin America and the Caribbean matter? Evidence
from 19 Countries (Policy Research Working Paper Series No. 3870). Washington, DC: World Bank.
Meyer, R. H., & Wise, D. A. (1983a). Discontinuous distributions and missing persons: The minimum wage and unem-
ployed youth. Econometrica, 51(6), 1677-1698. dx.doi.org/10.2307/1912112
Meyer, R. H., & Wise, D. A. (1983b). The effects of the minimum wage on the employment and earnings of youth. Journal
of Labor Economics, 1(1), 66-100. dx.doi.org/10.1086/298005
39
Minimum Wage and Job Mobility in Peru
Neumark, D., Schweitzer, M. E., & Wascher, W. (1999). The effects of minimum wages throughout the wage distribution
(NBER Working Paper No.7519). Cambridge, MA: National Bureau of Economic Research. Retrieved from http://
www.nber.org/papers/w7519.pdf
Neumark, D., & Wascher, W. (1991). Evidence on employment effects of minimum wages and subminimum wage provi-
sions from panel data on state minimum wage laws (NBER Working Paper No. 3859). Cambridge, MA: National
Bureau of Economic Research. Retrieved from http://www.nber.org/papers/w3859.pdf
Neumark, D., & Wascher, W. (1992). Employment effects of minimum and subminimum wages: panel data on state mini-
mum wages. Industrial and Labor Relations Review, 46(1), 55-81. dx.doi.org/10.2307/2524738
Neumark, D., & Wascher, W. (2006). Minimum wages and employment: A review of evidence from the new minimum wage
research (NBER Working Paper No. 12663). Cambridge, MA: National Bureau of Economic Research. Retrieved from
http://www.nber.org/papers/w12663.pdf
Orazem, P. F., & Mattila, J. P. (2002). Minimum wage effects on hours, employment, and number of rms: The Iowa case.
Journal of Labor Research, 23(1), 3-23. dx.doi.org/10.1007/s12122-002-1014-6
Pereira, S. C. (2003). The impact of minimum wages on youth employment in Portugal. European Economic Review,
47(2), 229-244. dx.doi.org/10.1016/S0014-2921(02)00209-X
Rodriguez, J., & Higa, M. (2010). Informalidad, empleo y productividad en el Perú (Documento de Trabajo No. 282).
Lima, Peru: Ponticia Universidad Católica del Perú.
Romero, E. J., & Cruthirds, K. W. (2009). Understanding employee turnover patterns in Mexican maquiladoras. Journal
of CENTRUM Cathedra, 2(1), 63-72. dx.doi.org/10.7835/jcc-berj-2009-0023
Stigler, G. J. (1946). The economics of minimum wage legislation. The American Economic Review, 36(3), 358-365.
Retrieved from: http://www.jstor.org/stable/1801842
Tello, M. D. (2012). Labor productivity in Peru: 1997-2007. Journal of CENTRUM Cathedra, 5(1), 115-142. dx.doi.
org/10.7835/jcc-berj-2012-0071
Van den Berg, G. J. (2003). Multiple equilibria and minimum wages in labor markets with informational frictions and
heterogeneous production technologies. International Economic Review, 44(4), 1337-1357. dx.doi.org/10.1111/1468-
2354.t01-1-00112
Van den Berg, G. J., & Ridder, G. (1998). An empirical equilibrium search model of the labor market. Econometrica,
66(5), 1183-1222. dx.doi.org/10.2307/2999634
Yamada, G., & Bazán, E. (1994). Salarios mínimos en el Perú: ¿cuándo dejaron de ser importantes? Revista Apuntes,
35(2), 77-88.
Authors Note
Nikita Céspedes Reynaga, Pontificia Universidad Católica del Perú, Av. Universitaria 1801, Lima, Peru & Banco
Central de Reserva del Perú, Jr. Antonio Miró Quesada 441-445, Lima, Peru.
Alan Sánchez, Grupo de Análisis para el Desarrollo, Av. Miguel Grau 915, Lima, Peru.
Correspondence concerning this article should be addressed to Nikita Cespedes Reynaga, Email: ncespedes@pucp.pe
We thank Marcos Agurto, Juan Chacaltana, the editor, and numerous seminar participants for comments that are incor-
porated throughout the paper. We are grateful to Jorge Luis Guzmán Correa for excellent research assistance and to
participants to the Research Seminar at the Central Bank of Peru. The views expressed in this paper are those of the
authors and do not necessarily represent those of the Central Bank of Peru. All remaining errors are our own.
40 JCC: The Business and Economics Research Journal
Appendix A: Minimum Wage and Income
Note. Income frequencies before and after the current minimum wage rise (EPE, Lima Metropolitan Area). The X axis
represents the Logarithm of income. The vertical line represents the minimum wage before the current minimum wage
increase. Kernel Epanechnikov function.
Figure 6. Main job income, formal salaried workers: frequency 2003-2011b.
41
Minimum Wage and Job Mobility in Peru
Note. Income frequencies before and after the current minimum wage rise (EPE, Lima Metropolitan Area). The X axis
represents the Logarithm of income. The vertical line represents the minimum wage before the current minimum wage
increase. Kernel Epanechnikov function.
Figure 7. Main job income, informal workers: frequency 2003-2011b.
42 JCC: The Business and Economics Research Journal
Appendix B: Results of Model Regressions
Table 8
Minimum Wage and Income, Main Results (Lima Metropolitan Area)
Coef. Std. Err.
Log minimum wage 0.252000*** 0.083000
Job category in t-3
Independent
Blue collar 0.010000 0.009000
White collar -0.041000*** 0.009000
House worker 0.053000*** 0.011000
Other categories 0.227000*** 0.010000
Relative income in t-3
Below or equal to 0.3
< 0.3; 0.6] 0.321000*** 0.010000
< 0.6; 0.9] 0.623000*** 0.010000
< 0.9; 1.2] 0.833000*** 0.009000
< 1.2; 1.5] 0.978000*** 0.009000
< 1.5; 2.0] 1.139000*** 0.009000
< 2.0; 2.5] 1.290000*** 0.010000
< 2.5; 3.0] 1.436000*** 0.012000
< 3.0; 4.0] 1.609000*** 0.012000
< 4.0; 5.0] 1.844000*** 0.015000
Above 0.5 2.327000*** 0.013000
Education level
No education
Kinder -0.302000 0.261000
Incomplete primary 0.003000 0.021000
Complete primary 0.029000 0.021000
Incomplete secondary 0.005000 0.021000
Complete secondary 0.057000*** 0.021000
Incomplete technical college 0.069000*** 0.022000
Complete technical college 0.138000*** 0.021000
Incomplete university 0.125000*** 0.022000
Complete university 0.294000*** 0.022000
Firm size in t-3 (n. of employees)
One employee
Between 2 and 10 0.086000*** 0.008000
Between 10 and 50 0.138000*** 0.011000
Between 50 and 100 0.212000*** 0.017000
More than 100 0.182000*** 0.010000
Note. Dependent variable: log monthly income.
The method of estimation is ordinary least squares. The sample size is 76 282. The data come from the
EPE (January 2003 to March 2012) and includes all individuals who are observed twice and who are employed
in both periods. Robust standard errors reported; *, **, *** denote significance at 10%, 5%, and 1% levels.
Estimations include year of interview and month of interview fixed effects and the following control variables:
access to health insurance in t-3, dummy that takes the value of 1 if head of the household and 0 otherwise,
dummy that takes the value of 1 if male and 0 otherwise, years of experience and years of experience squared.
43
Minimum Wage and Job Mobility in Peru
Table 9
Minimum Wage and Employment, Main Results (Lima Metropolitan Area)
(1) (2)
Coef. Std. Err. Coef. Std. Err.
Minimum wage (in Soles) -0.003240*** 0.000729 -0.003030*** 0.000752
Employee x minimum wage -0.000486 0.000332
Worker x minimum wage -0.000116 0.000372
House worker x minimum wage -0.000180 0.000623
Other categories x minimum wage -0.000256 0.000450
Job category in t-3
Independent
Blue collar -0.405000*** 0.046700 -0.155000 0.177000
White collar -0.621000*** 0.046400 -0.562000*** 0.196000
House worker -0.202000*** 0.044900 -0.109000 0.324000
Other categories -0.062200 0.048900 0.068700 0.236000
Relative income in t-3
Below or equal to 0.3
< 0.3; 0.6] 0.408000*** 0.035100 0.408000*** 0.035100
< 0.6; 0.9] 0.692000*** 0.035400 0.692000*** 0.035500
< 0.9; 1.2] 0.981000*** 0.035300 0.981000*** 0.035300
< 1.2; 1.5] 1.139000*** 0.038000 1.139000*** 0.038000
< 1.5; 2.0] 1.221000*** 0.038500 1.219000*** 0.038500
< 2.0; 2.5] 1.246000*** 0.049100 1.245000*** 0.049100
< 2.5; 3.0] 1.276000*** 0.060500 1.275000*** 0.060500
< 3.0; 4.0] 1.286000*** 0.064100 1.285000*** 0.064100
< 4.0; 5.0] 1.158000*** 0.086800 1.159000*** 0.086800
Above 5.0 1.320000*** 0.070300 1.319000*** 0.070300
Education level
No education
Kinder -0.553000 0.870000 -0.548000 0.871000
Incomplete primary -0.237000*** 0.083700 -0.238000*** 0.083700
Complete primary -0.368000*** 0.082300 -0.370000*** 0.082300
Incomplete secondary -0.459000*** 0.083800 -0.461000*** 0.083800
Complete secondary -0.435000*** 0.082500 -0.436000*** 0.082500
Incomplete technical college -0.356000*** 0.092200 -0.358000*** 0.092200
Complete technical college -0.319000*** 0.087600 -0.320000*** 0.087600
Incomplete university -0.615000*** 0.091000 -0.616000*** 0.091000
Complete university -0.438000*** 0.088900 -0.439000*** 0.088900
Firm size in t-3 (n. of employees)
One employee
Between 2 and 10 0.450000*** 0.040300 0.451000*** 0.040300
Between 10 and 50 0.532000*** 0.055100 0.534000*** 0.055100
Between 50 and 100 0.775000*** 0.093400 0.776000*** 0.093400
More than 100 0.892000*** 0.053600 0.894000*** 0.053600
Note. Dependent variable: employment status.
Coefficients of a Probit model for employment. The sample size is 97 547. The data come from the EPE
(January 2003 to March 2012) and includes all individuals who are observed twice and who are employed the
first time they were observed. Robust standard errors reported; *, **, *** denote significance at 10%, 5%,
and 1% levels. Estimations include year of interview and month of interview fixed effects and the following
control variables: access to health insurance in t-3, dummy that takes the value of 1 if head of the household
and 0 otherwise, dummy that takes the value of 1 if male and 0 otherwise, years of experience and years of
experience squared.
44 JCC: The Business and Economics Research Journal
Table 10
Minimum Wage and Employment: Additional Estimations
PAR T A
Lima Metropolitan Area: Heterogeneity of Elasticities by Individual Characteristics
Coef. Std. Err.
Estimation A: Heterogeneity by health insurance:
Did not have health insurance three months previously -0.234000*** 0.061000
Had health insuranve three months previously -0.276000*** 0.058000
Estimation B: Heterogeneity by rm size:
One employee -0.327000*** 0.082000
Between 2 and 10 -0.264000*** 0.058000
Between 10 and 50 -0.230000*** 0.061000
Between 50 and 100 -0.172000** 0.081000
Above 100 -0.174000*** 0.042000
Estimation C: Heterogeneity by location in
the income distribution
(relative to minimum wage; three months previously):
< 0.3; 0.6] -0.543000*** 0.117000
< 0.3; 0.6] -0.421000*** 0.091000
< 0.6; 0.9] -0.320000*** 0.072000
< 0.9; 1.2] -0.174781 0.055000
< 1.2; 1.5] -0.202000*** 0.049000
< 1.5; 2.0] -0.119000*** 0.046000
< 2.0; 2.5] -0.204500 0.049000
< 2.5; 3.0] -0.219000*** 0.055000
< 3.0; 4.0] -0.194000*** 0.057000
< 4.0; 5.0] -0.122000 0.076000
Above 5.0 -0.094000 0.058000
PART B
National Level Elasticities
Coef. Std. Err.
Estimation D: Heterogeneity by type of location
Minimum wage 0.000600 0.000000
Minimum wage*urban -0.001520*** 0.000000
Minimum wage*Lima -0.001470*** 0.000000
Note. Elasticities for the Lima Metropolitan Area and National Level were obtained from Probit models where the
dependent variable is whether the individual is employed. Part A presents the results of three different models. Control
variables included are the same as those reported in Table 9 (including year of interview and month of interview). In
Part B, the model controls for the level of education of the individuals, whether the individuals are the head of the house-
hold, their gender, age, age squared, year of interview, and month of interview. Robust standard errors are reported; *,
**, *** denote significance at 10%, 5%, and 1% levels.
45
Minimum Wage and Job Mobility in Peru
Note. The graphs represent the proportion of employed people who change to another job by income range (EPE, Lima
Metropolitan Area). The X axis represents the income in fractions of the current minimum wage.
Figure 8. Job-to-job transitions by income ranges, 2003-2011.
46 JCC: The Business and Economics Research Journal
Note. The graphs represent the proportion of employed people who change to another labor category by income range
(EPE, Lima Metropolitan Area). The X axis represents the income in fractions of the current minimum wage.
Figure 9. Employment-to-other categories transitions by income ranges, 2003-2011.
47
Minimum Wage and Job Mobility in Peru
Note. The graphs represent the proportion of employed people who change to inactivity by income range (EPE, Lima
Metropolitan Area). The X axis r epresents the income in fractions of the current minimum wage.
Figure 10. Employment-to-inactivity transitions by income ranges, 2003-2011.
48 JCC: The Business and Economics Research Journal
Note. The graphs represent the proportion of employed people who change to unemployment by income range (EPE,
Lima Metropolitan Area). The X axis represents the income in fractions of the current minimum wage.
Figure 11. Employment-to-unemployment transitions by income ranges, 2003-2011.
49
Minimum Wage and Job Mobility in Peru
Note. The graphs represent the proportion of unemployed people who change to employment by income range (EPE,
Lima Metropolitan area). The X axis is income in fractions of the current minimum wage.
Figure 12. Unemployment-to-employment transitions by income ranges, 2003-2011.
50 JCC: The Business and Economics Research Journal
Note. The graphs represent the proportion of inactive people who change to employment by income range (EPE, Lima
Metropolitan Area). The X axis represents the income in fractions of the current minimum wage.
Figure 13. Inactivity-to-employment transitions by income ranges, 2003-2011.
Internationalization of Firms’ Activities and
Company Union Wage Strategies
Domenico Buccella
Leon Kozminski University, Warsaw, Poland
Abstract
Using a two-country duopoly model with homogeneous goods, firms’ decisions with respect to international
activities (trade vs. foreign direct investment - FDI) in the presence of company-wide unions are analyzed.
If firms export, they pay trade costs per unit of the goods exported. If firms invest and set up plants abroad,
they incur sunk costs. The full set of production structures that arise as sub-game perfect Nash equilibriums
are derived when internationalization is feasible. The interdependence of exogenous integration costs,
endogenous union wage strategies, and firms’ strategic interactions affect the equilibrium outcome: either
symmetric (intra-industry trade or reciprocal FDI) or multiple symmetric (intra-industry trade and reciprocal
FDI) equilibriums exist.
Keywords: Foreign direct investment, international trade, labor unions
JEL Classification codes: F16, F21, F23, J51, L13
http://dx.doi.org/10.7835/jcc-berj-2014-0095
Two results emerge as the most evident consequences of the process of economic integration occurring in
the European context. First, the completion of the Single Market Program in 1992, established the free move-
ment of goods, capital, services, and people among the member states of the European Union (EU). Second,
the creation of the European Monetary Union (EMU) concluded with the introduction of the Euro in 2002.
Increases in the degree of liberalization of capital markets and continuous removal of internal tariff and non-
tariff barriers, with a consequent reduction in trade costs in product markets, exemplify this course of action.
Further developments and improvements in the Financial Service Action Plan (FSAP) and financial market
integration within the EU itself have driven significant growth in the figures related to intra-industry trade
(IIT ) (see European Commission, 2008a) and intra-EU foreign direct investments (FDI) (Jovanović, 2006;
European Commission, 2008b).
The EU economic background offers ideal “humus” for the internationalization of firms’ activities. At the
same time, as product and capital markets become more integrated, major actors in European labor markets, as
trade unions, start considering a broader perspective in their activities. Some of the European Commission’s
legislative initiatives, such as the approval of the 1994 European Working Councils (EWC) directive1 and the
2001 European Company (Societas Europaea - SE) directive, which advanced the practice of informing and
consulting the workforce in transnational contexts, are shifting toward the company level, the key level of
collective bargaining, in many industrial sectors. However, depending on the degree of market integration
and the presence of productive activities organized internationally, remarkable distinctions between indus-
tries exist. Company-level negotiations are prominent in those sectors characterized by a high incidence of
multinational enterprise (MNE) operations.
Journal of
CENTRUM
Cathedra
JCC
JCC: The Business and Economics Research Journal ● Volume 7, Issue 1, 2014 ● 51-74
52 JCC: The Business and Economics Research Journal
The international dimension of MNEs, rulings by the European Court of Justice, the institutions of the
European Works Council, the practice of opting out from national/sector collective bargaining in favor of
company-wide agreements (European Foundation for the Improvement of Living and Working Conditions -
Eurofound, 2009) have had a deep impact on labor market outcomes. This framework caught the attention
of some labor unions because it offered both the prospect of moving their wage bargaining strategies to the
European level (the “horizontal Europeanization” of labor relations; see Pernicka & Glassner, 2012; Müller,
Platzer, & Rüb, 2013) and arranging transnational agreements at company level. Indeed, the figures related
to the cross-border company agreements steadily increased in recent years, from a few dozens in 2000 to 244
in 2011 (Müller, Platzer, & Rüb, 2013).2
As Horn and Wolinsky (1988) suggested, firms would like to take strategic advantage of an MNE organi-
zational structure to avoid the creation of an encompassing union. On the other hand, unions may coordinate
bargaining across countries but at present, transnational coordination activities are still in an embryonic state.
However, instead of explaining current transnational agreements, this work aims at going one step further
and asking the following questions. First, if unions in the future are able to coordinate their bargaining activi-
ties effectively across countries within the same company, making progress in the process of “horizontal
Europeanization” of labor relations, might the prospect of a unique workers’ representative body affect the
fir ms’ internationalization strategies? Second, to what extent may unions improve their positions in negotiations
with respect to firms involved in international business? Focusing precisely on these issues, the intention is to
develop a symmetric two-country duopoly model where organized, company-wide workforce representatives
seek to gain part of the rents generated in the product market.
In recent years, unions started exploiting the potential of the EWC more intensively during company-wide
bargaining processes. For example, in the banking sector, Danish trade unions received the mandate to nego-
tiate on behalf of all employees working in the Danske Bank Work Council (European Industrial Relation
Observatory Online - EIROnline, 2009). The European Metalworking’s Federation (EMF), the UNI Europa
Graphical (UEG) and the European Public Service Union (EPSU), three cross-border industry level federations,
devised a procedure to receive the mandate in representing the overall workers’ side throughout company-wide
transnational agreements. Since the formulation of this internal procedure, the EMF has implemented it with
at least five MNEs, including Areva, Schneider, Daimler-Chrysler, John Deere, and ArcelorMittal
(Eurofound,
2009; Gennard, 2009a), while the EPSU used it with Suez-Lyonnais des Eaux (Papadakis, 2010). The creation of
cross-border unions is another response to company-wide negotiations. In 2009, for instance, a trans-boundary
seafarers’ union, Nautilus International, was launched, based in the UK and the Netherlands. It represents a
wide range of personnel working in the shipping sector, at sea, on inland waterways, and ashore. The cross-
border union is the result of a merger process following several years of closer cooperation between Nautilus
NL and Nautilus UK, including joint industrial negotiations with companies employing British and Dutch
workers (Gennard, 2009b). Finally, transnational campaigns to support wage bargaining, either in selected
sectors or in MNEs, are further vehicles labor unions take advantage of to move closer to issues at the core
of traditional collective bargaining at an international level (Keune & Schmidt, 2009).
In the present work, firms’ decisions about international activities (trade vs. FDI) in the presence of
company-wide unions are analyzed. The model is a two-country, three-stage game duopoly model where
firms produce a homogeneous product. Product markets are segmented and both countries are characterized
by unionized labor markets. In the first stage of the game, firms choose autonomously whether to invest. Each
firm has two strategies: Not to invest, thus maintaining all productive activities in the domestic country; or to
invest abroad, thus setting up a new plant. If firms serve the other country through exports, they pay constant,
exogenous trade costs per unit of the commodity exported. Otherwise, firms engage in FDI and establish
a production plant in the foreign country, incurring an exogenous, positive sunk cost. In the second stage,
monopoly unions set their optimal wage strategies, competing with each other in the labor market. Finally, in
the third stage, fir ms compete à la Cournot,
choosing profit-maximizing quantities separately for each market.
The main results of the paper are as follows. A rich set of the productive structure regimes will take
place in equilibrium for the international oligopoly. While the presence of unique symmetric regimes (IIT or
reciprocal FDI - RFDI) is a natural consequence, a novel result is that, for some combinations of integration
costs, multiple symmetric equilibriums are possible: IIT and RFDI regimes may occur simultaneously. The
rationale for this result resides in the fact that different combinations of integration costs have different effects
on prices (due to product market competition) and wages (because of unions’ strategic behavior). These, in
turn, affect the level of firms’ profits and, therefore, their strategic choice concerning the start of international
activities. Furthermore, the RFDI regime is one equilibrium of the game also applicable for relatively low
53
Internationalization of Firms’ Activities and Company Union Wage Strategies
values of trade costs when IIT is feasible, provided the scale of the sunk costs is low enough. The reason is
that a decline in trade barriers has a positive effect on the profit level of foreign subsidiaries and this, in turn,
makes the investment option more attractive. Nevertheless, if the scale of the sunk costs is large enough, IIT
is the unique equilibrium of the game. This is because the company-wide unions set higher wage rates when
firms invest than in the case of exports: High wages and sunk costs are not sufficient to counterbalance the
trade cost savings.
The focus of this paper is related to a body of literature analyzing, within different contexts, the implica-
tions of international economic integration on labor market outcomes in the presence of unions. Few authors
have investigated how this process affects the unions’ strategic behavior and how the unions’ behavior may,
in turn, affect firms’ strategic choices related to international activities. The first group of authors who have
examined how international integration affects the wage formation in the presence of unionized countries is
Huizinga (1993), Sørensen (1993), Naylor (1998, 1999), Borghijs and Du Caju (1999), Straume (2002), Lommerud,
Meland, and Sørgard (2003), Glass and Saggi (2005), Strozzi (2007, 2008), and Ishida and Matsushima (2009).
Authors such as Huizinga (1993), making use of a monopoly union model, and Sørensen (1993), using a
more general right-to-manage model, concluded that product market integration leads to an enlargement of the
market size. Consequently, the number of firms operating in the market increases, intensifying the degree of
competition. This, in turn, implies a decrease in the level of prices and wages. Moreover, under the assumption
of linear demand and production functions, Huizinga (1993) claimed that the decrease in wage levels is more
than offset by the increase in employment so that net union utility increases. These two models, however, do
not take into account any interaction between the two economies before integration occurs.
Closely related to this work are the contributions of Naylor (1998, 1999). In these articles, two identical
firms initially produce a homogeneous product for their home markets and, under the assumption of perfect
symmetry in both product and labor markets, engage in reciprocal dumping when trade cost levels fall below
a threshold value. This implies a fall in the wage demands of labor unions: IIT, putting unions in competition
internationally in the labor market, erodes their monopoly power. As the degree of economic integration increases
(further reduction in trade costs), unions set higher wages due to higher profits for both firms, capturing part
of the firms’ rent. These works studies the effects of economic integration on wages and unions’ outcomes and
the interaction between the two economies, exemplified by the unions’ strategic behavior in labor markets.
Of interest to the analysis are the works of Lommerud, Meland, and Sørgard (2003), Glass and Saggi (2005),
and Ishida and Matsushima (2009). Lommerud, Meland, and Sørgard (2003) made use of a two-country
reciprocal dumping model of oligopoly with only one country unionized, focusing the analysis on how trade
liberalization and wage setting affected the firms’ location choice, and therefore, the way firms chose to serve
their relevant markets. Ishida and Matsushima (2009) likewise analyzed the same issue in a similar framework
when domestic competition occurs between firms located in a unionized country. Taking a different approach
from Lommerud, Meland, and Sørgard (2003), Glass and Saggi (2005) determined endogenously the equilibrium
FDI regime without considering the effects of trade liberalization. In their international duopoly model, trade
costs are sufficiently low such that firms could always export their products. The crucial assumption is that
both firms require one intermediate product that a local upstream monopolist supplier provides exclusively.
The authors show that under these circumstances, outward FDI can act as a cost-raising strategy. However, in
these works, the strategic interaction in the labor markets is absent; consequently, there is not the opportunity
to study trade union cooperation.
A second strand of the literature has analyzed the interaction between unionized labor markets and firm
activities related to the internationalization of production through FDI. The general approach is to investigate
the effect of FDI, examining the union-firm interaction using either a “right-to-manage” (Bughin & Vannini,
1995; Naylor & Santoni, 2003; Eckel & Egger, 2009) or an efficient bargaining model (Mezzetti & Dinopoulos,
1991; Zhao, 1995, 1998) to explore the effects on wages and employment, either in a partial or in a general
equilibrium framework. Like Naylor and Santoni (2003), Zhao (1995), and Eckel and Egger (2009), in the present
paper, intra-industry RFDI and the presence of unions in the labor market is accommodated. Notwithstanding
the different approaches, underlying hypotheses and purposes of analysis, these models achieve a common
result: If firms have the opportunity to invest abroad, they will cause a moderation in wage demands in the
bargaining process. Consequently, the position of unions appears to be weakened.
The contribution of the present paper to the previous literature is the following. It widens Naylor’s (1998, 1999)
analysis by allowing firms to undertake FDI, as is the case in Lommerud, Meland, and Sørgard (2003) and
Ishida and Matsushima (2009). However, differently from Lommerud, Meland, and Sørgard (2003) and Ishida
54 JCC: The Business and Economics Research Journal
and Matsushima (2009), unionized workforces in both countries are considered in the model in a more realistic
reflection of the characteristics of the EU labor market. In doing so, making a link between two issues that the
previous literature treated as separate subjects is attempted. Furthermore, the hypothesis of company-wide
negotiations conducted by a unique workers’ representative body is retained: This is a crucial difference with
respect to previous theoretical models. For the scope of the present work, this assumption captured the idea
of unions’ hoped-for developments in transnational company agreements in Europe.
The implications of this framework are far-reaching. First, like Lommerud, Meland, and Sørgard (2003),
it is shown that trade liberalization makes the investment strategy more profitable. In fact, the domestic
firm has easier access to the foreign market (product market expansion effect), which implies an increase in
the domestic labor demand. At the same time, competition in the domestic country becomes more severe.
Nevertheless, the net effect is positive, and the domestic union raises wages, capturing part of the oligopoly
rents. However, while for Lommerud, Meland, and Sørgard (2003) high domestic wages give a strong incen-
tive for FDI to success a distributional conf lict between unions and firms, in this paper it is shown that when
unions are organized at company level in both countries, FDI may occur even if the wage rate in the investing
firm is higher than the wage resulting from the export strategy. This is because for some combinations of
integration costs, the product market competition in the asymmetric regimes for the exporting firm is harsher
than in the case where both firms invest: The beneficial effect on profit of trade cost reductions is more than
offset by the adverse effects of market competition and wage increase. Therefore, to undertake FDI is a mutual
best-response strategy for firms.
Second, as in Ishida and Matsushima (2009), if firms invest, unions in their domestic countries benefit
from FDI because wage rivalry tends to be less intense. However, while union utility increases because wage
gains may offset employment reduction, in the present paper, unions gain both from wages and employment
increases because of their cross-border, company-wide nature. Furthermore, Ishida and Matsushima (2009)
show that, in the asymmetric regime, the union in the exporting firm is induced to decrease wages to facilitate
the company remaining in a competitive position in the foreign market. Because wages for the exporting firm
are lower for all workers, it will produce at low cost for the domestic market, thus improving its position in
the home country. Consequently, the union in the investing country cannot increase the wages of domestic
workers because domestic output would be reduced. This, in turn, lowers employment in the investing firm
and hence its union utility. Ishida and Matsushima’s (2009) results contrast with those in the present paper. In
fact, in the asymmetric regime, increasing economic integration (a reduction in trade costs) stimulates exports.
Thus, labor demand for the exporting firm increases. This, in turn, implies that the union in the exporting
firm raises wages. On the other hand, increasing economic integration implies both a decrease in the total
output of the investing firm and a wage reduction for its workers. Nevertheless, wages and employment in the
investing firm are higher than those in the exporting firm.
In exploring this topic, an attempt is made to depict the prospects concerning union coordination in MNEs
by attempting to give some predictions about the potential implications of transnational bargaining on the
development of international businesses in the EU environment. Moreover, given the advances of the new
technologies, the proposed framework can be applied to both the manufacturing and service sectors. In fact,
due to the development of internet and online technologies, the provision of several services – banking and
insurance, for instance – does not necessarily require the presence of a physical subsidiar y in a foreign country.
In other words, services can be exported towards other countries.
The remainder of the article is organized in the following way. Section 2 outlines the analytical framework.
It develops a non-cooperative three-stage game of international duopoly in the presence of unionized work-
forces at company level. Firms act as first movers, choosing independently whether to invest in the foreign
country and paying a certain level of sunk costs. If firms do not invest, they may either export to the foreign
country or produce for their domestic country exclusively. Then, in the second stage, company unions select
their optimal wage strategies. The usual backward induction method solves the model. Depending on sunk
and trade costs, and due to the strategic interaction between firms and unions, different productive structures
may arise in equilibrium. A brief discussion of the managerial implications of the model closes the section.
Finally, Section 3 brings the paper to its conclusion.
55
Internationalization of Firms’ Activities and Company Union Wage Strategies
The Basic Model
There are two symmetric countries, A and B. In each country, the economy presents two sectors: A perfectly
competitive sector and an imperfectly competitive sector characterized by the presence of a monopolist, Firm 1,
located in Country A, and Firm 2, located in Country B. The two firms produce homogeneous goods, denoted
x when produced in Country A and y when produced in Country B. Firms consider each country as a separate
market (market segmentation hypothesis). Labor is the unique factor of production with linear technology and
constant return to scale. By this normalization (without loss of generality), each worker produces one unit
of the product: Therefore, production and employment are interchangeable. The perfectly competitive sector
represents a buffer where workers can always find employment at the competitive wage (normalized to zero).
Table 1
First stage, the Firms’ Game
Firm 2
Firm 1 Invest Not Invest
Invest ∏+∏- ∏+
∏-FF;
A
II
B
II
B
II
A
II
11 22
∏+∏- ∏
F;
A
IN
B
IN
B
NI
11 2
Not Invest ∏∏+∏ -
F;
A
NI
B
IN
A
IN
12 2
∏
B
NN
1
;∏
B
NN
2
The representative consumer in each country maximizes the following quasi-linear utility function:
U=U
_
(x,y)+z=(xik +yjk )−
1
2
(xik
2+yjk
2+2xik yjk )+z,
with
i,j=1,2 i ≠j;k=A,B
, where U
_
(x,y) is the quadratic utility derived from the consumption of the
goods produced in the imperfectly competitive sector, and z is the linear utility derived from the consumption
of the competitive goods.
3
These consumers’ preferences imply that the demand schedules are linear. Company
level unions operate and organize their activities in the imperfectly competitive sector whose workers are
fully unionized.
The model is a three-stage game. In the first stage of the game, firms autonomously choose whether to
invest. Each firm has two strategies4 (see Table 1): not to invest, thus maintaining all productive activities in
the domestic country; and to invest abroad, thus setting up a new plant (Greenfield venture). If firms under-
take FDI and establish a Greenfield venture in the foreign country, they incur an exogenous sunk cost F ≥ 0.
Otherwise, firms may serve the other country through exports, paying a constant, exogenous cost
t∈[0, 1)
per unit of the commodity exported.5 Several regimes might arise as a consequence. First, both firms do not
invest: Depending on trade costs and the unions’ strategic decisions, firms may serve the other market through
exports, allowing for IIT.6 Second, both fir ms invest (RFDI). Third, only one firm invests (asymmetric regimes).
In Table 1, ΠNN denotes the profits when both firms do not invest; ΠII denotes RFDI profits; ΠIN (ΠNI ) denotes
profits when one firm invests abroad while the other does not (and vice versa). In the second stage, monopoly
unions (having full bargaining power, see Dowrick, 1989) set their optimal wage strategies, competing with
each other in the labor market. Finally, in the third stage, firms engage in a Cournot competition,7 choosing
profit-maximizing quantities separately for each market realizing output. Market segmentation, combined
with the constant marginal costs assumption, implies that the price of the goods in each country depends
exclusively on the total quantity available in the market.
The model is solved by the backward induction method to derive sub-game perfect equilibriums. The
following subsections inspect, for each regime of the productive structures, first the output game among firms
in the product market deriving the firms’ labor demand functions in terms of wages. Then, in the second stage,
given the firms’ labor demands, the analysis of the unions’ wage setting in the labor markets is presented.
Finally, turning back to the first stage of the game, the results of the sub-games allow an evaluation of the
firms’ payoff functions. According to realized profits, each firm chooses which internationalization strategy
should be adopted. After collecting the relevant results, it is possible to derive the conditions under which a
particular structure of production arises as equilibrium of the game.
56 JCC: The Business and Economics Research Journal
Because the aim of this paper is to derive the productive structure arising in equilibrium when both firms
undertake international business, the analysis focuses on a subset of the integration costs, that is, sunk and
trade cost levels. First, the scale of the sunk costs is assumed to be small enough that each firm can invest
abroad independently from the strategic choice of the rival firm.
8
Second, trade costs are such that international
activities are supported as sub-game perfect equilibriums in pure strategies of the two-stage game “unions’
wage determination‒firms’ quantity choices” independently from the firms’ strategic choices concerning
internationalization. To obtain well-defined solutions in pure strategies, therefore, the subsequent analysis
imposes the following restriction on the values of the parameters t and F.
Restriction 1. t ∈ [0, 0.310], F < 0.034.
Restriction 1 defines the range of trade and sunk costs where sub-game perfect equilibriums allow inter-
nationalization of firms’ activities. The restriction on trade costs limits the analysis to sub-game perfect
equilibriums in the two-stage sub-game “unions’ wage determination‒firms’ quantity choices” such that the
internationalization of firms’ activities is always possible. The parameter’s restriction on trade costs is given
by t ≤ 0.310 because, at this level, any union wage combination in the two-stage sub-game “unions’ wage
determination‒firms’ quantity choices” is consistent with every configuration of the firms’ strategies involving
international activities (IIT, RFDI, and asymmetric regimes). The meaning of the restriction on sunk costs is
as follows. The profits generated in the foreign market by the investing firm have to be greater than the size
of the sunk costs to undertake the investment abroad. The amounts of these profits differ according to the
strategy that the rival firm selects.
Regime 1: Both Firms Do Not Invest: Intra-Industry Trade
This subsection analyzes Regime 1, the situation in which both firms decide not to invest. These results can
be found in Naylor (1998, 1999) and Straume (2002), which are the sole references for this part of the paper.9
In the last stage of the game, firms compete à la Cournot in the product markets. The profit functions are
the following:
Π
1
=p
A
x
1A
+p
B
x
1B
−w
1
x
1A
−w
1
x
1B
−tx
1B, (1)
Π
2
=
pAy2A
+
pBy2B
−
w2y2A
−
w2y2B
−
ty2A
, (2)
where
p
A
=1−x
1A
−y
2A is the price in Country A, which depends both on the quantity produced by Firm 1
for the domestic market,
x
1A, and Firm 2’s exports,
y
2A. Similarly,
p
B
=1−x
1B
−y
2B is the price in Country
B, where
x
1B is the quantity produced for exports by Firm 1, and
y
2B is the quantity produced by Firm 2 for
the domestic market. Notice that both firms pay a cost of
t∈[0, 1)
per unit of the product exported, repre-
senting a basket of costs including tariffs, red tape and, in the case of manufactured goods, transportation
and logistics, etc.
The firms’ reaction functions are obtained from the first-order conditions for profit maximization. These
represent the output produced as well as the firms’ labor demands. Thus, in the second stage, each union maxi-
mizes its utility function by considering the specific labor demand schedules of the firms, and it is possible
to derive the unions’ reaction functions. For trade costs below or equal to the threshold value of t ≈ 0.310, it
can be shown that the Bertrand-Nash wage in equilibrium is the following:
wIIT =
1
3
−
1
6
t.
In the case of IIT, unions compete with each other over employment, causing a fall in wage levels compared
to the autarky regime. Hence, trade in this model decreases union power. Nevertheless,
dw
IIT
/dt <0
suggests an increase in economic integration (a reduction in trade costs) will induce trade unions to raise
wages. The intuition is the following. For values lower than the threshold, IIT occurs between the two
countries. A decrease in trade costs will induce harsher competition amongst the participants in the inter-
national oligopoly: Firms’ outputs rise because exports increase. Consequently, labor demand increases,
and therefore, unions will choose to set higher wages, capturing a higher share of oligopoly rents, while
firms may experience a loss in profits. Substituting the equilibrium wage into the output expressions, the
following values are obtained:
57
Internationalization of Firms’ Activities and Company Union Wage Strategies
x1A=y2B=
2
9
+
7
18
t;
x1B=y2A=
2
9
−
11
18
t.
These represent the Cournot quantities in equilibrium in the presence of IIT. Further substitutions lead to
the following union utility and firm profits:
ΩIIT =
1
27
(2 −t)2,
ΠIIT =
8
81
−
8
81
t+
85
162
t2.
Regime 2: Both Firms Invest: Reciprocal FDI
Stage 3, firms’ quantity choices and labor demands:
The
RFDI
regime is now considered. The firms’ profit functions are the following:
Π
1
=p
A
x
1A
+p
B
y
1B
−w
1
x
1A
−w
1
y
1B
−F
, (3)
Π2=p
A
x2
A
+p
B
y2
B
−w2x2
A
−w2y2
B
−F
, (4)
where
pA
=
1
−
x1A
−
x2A
is the price in Country A, which depends both on the quantity produced by Firm 1
in Country A,
x1A
, and the quantity produced by Firm 2’s branch located in the same country,
x2A
. Similarly,
pB
=
1
−
y1B
−
y2B
is the price in Country B. Companies, in theory, may still export to the foreign country
instead of serving the foreign market by producing locally. However, having borne the burden of a sunk cost
equal to
F
, firms do not export. The rationale is that firms incur additional costs of
t
for the quantities
exported, and this is less profitable than the option of serving the foreign market with local production alone.
It follows that the specification of the firms’ profit functions in the presence of RFDI is exactly as indicated
in Equations 3 and 4.
Note that, in the present model, multinational firms pay the same wage in both countries because, by
assumption, the MNE’s workers are organized at company level. Unions act on behalf of overall workers and
set a non-discriminatory wage independently from the fact that workers are located in different countries. This
hypothesis would reflect the situation of centralized negotiations among unions operating in Works Councils
and the general management of firms pursuing international business. It may be argued that this assumption
is extremely strong and, undeniably, it is. However, as underlined in the introduction, the purpose is to inves-
tigate how international business decisions may be affected if unions’ eventual development in transnational
company agreements occurs and they are able to coordinate bargaining across countries effectively. This
hypothesis is consistent with the idea, usually found in the literature, that a multinational pays a wage rate
different from that of domestic firms (see, for example, Leahy & Montagna, 2000).
Figure 1. Investment boundaries and unions’ reaction functions.
58 JCC: The Business and Economics Research Journal
The firms’ reaction functions are obtained from first-order conditions for profit maximization (see Appendix A).
These represent the output produced by each firm in each country. If
(w
1
<1, w
2
<1)
, it is possible to show
that the solution of the quantity game in Country A as function of the wage rates, call it
(x
1A
C,x
2A
C)
, is such that:
(x1A
C,x2A
C)=
1
+
w
2−
2w
1
3,
1
+
w
1−
2w
2
3
,iff w1≤
1
+
w
2
2,w2≤
1
+
w
1
2; (5a
)
(0, 1−w2
2), iff w1≥1+w2
2,w2<1; (5b
)
(1−w1
2,0), iff w2≥1+w1
2,w1<1. (5c)
while the solution of the quantity game in Country B, call it
(y1B
C
,y2B
C
)
, is such that:
=
+- +-
≤+≤+
-≥+<
-≥+<
yy
ww ww
iff w
w
w
w
a
wiff ww
wb
wiff ww
wc
(, )
12
3,
12
3,
1
2,
1
2
;(6)
(0,1
2), 1
2,1
;(6)
(1
2,0), 1
2,1
.(6)
B
C
B
C
12
21 12
1
2
2
1
2
1
2
2
1
2
1
1
If unions fix too high wage levels (Regions II and III in Figure 1), the firms do not find it profitable to
exploit the foreign plant, although they have already incurred the sunk cost. High wages set by unions also
price out each firm from the domestic market; each firm finds it inconvenient to produce there. The reason is
that, given
w1
(
w2
), for
w2
(
w1
) such that the point
(w1,w2)
lies on the boundary between Regions I and II
(III) or is internal to Region II (III), the wage rate is not lower than the price under domestic monopoly (the
autarky case).
Stage 2, unions’ wage setting
From the above discussion, it follows that, in Stage 2 of the game, each company-level union chooses a
wage, allowing firms to pursue both domestic production and exploitation of the plants abroad. Making use
of the optimal quantities the utility function for Union 1 is:
Ω
1
=
w1(x1A
+
y1B)
, (7)
and the utility function for Union 2 takes a similar form. Substitution of Equations 5a and 6a into Equation 7,
and solving the maximization problem, leads to the following expression:
RF
1(w2)=w1=
1
4
+
1
4
w2,
the reaction function for Union 1. A similar result (interchanging
w1
with
w2
) pertains for Union 2.
Sub-Game perfect equilibriums of the two-stage game “unions’ wage determination‒firms’ output choices”
Solving the linear system composed by the two unions’ reaction functions, the Bertrand-Nash equilibrium
wage level is:
wRF DI =
1
3
.
It follows that production levels are:
x1A=y1B=y2B=x2A=
2
9
.
59
Internationalization of Firms’ Activities and Company Union Wage Strategies
Comparing wIIT and wRFDI , and production outcomes (and thus employment levels), it is immediately clear
that, in the case of international production, both achieve higher values, and therefore, unions in equilibrium
have higher utility levels in the RFDI regime than in IIT. The firms’ production levels for the domestic market
decrease, while those for the foreign market increase. The rationale for the latter result is that, in the case
of investment, the marginal cost of serving the market abroad by local production is lower than in the case
of exports. Nevertheless, the expansion in the foreign market more than offsets the loss of market shares in
the domestic market: Total output (and, therefore, employment), rises. Labor demand increases as well, and
therefore, each company-level union may claim for higher wages than in IIT while firms may experience a
fall in profit levels. In addition, wage rates increase because the firms’ rents in RFDI are larger than those in
IIT (due to trade cost savings), and unions are able to capture a share of these enlarged rents. Thus, for unions
organized at company level, the investment strategy of the firms is advantageous. After subsequent substitu-
tions, the following expressions for the union utility and firm profits are obtained:
ΩRF DI =
4
27
,
ΠRF DI =
8
81
−
F
.
Regime 3: Only One Firm Invests: Asymmetric Regimes
Stage 3, firms’ quantity choices and labor demands
The evaluation of firms’ profits in asymmetric regimes (one firm invests while the other does not) requires
the establishment of a set of game equilibriums. In these asymmetric regimes, different configurations in both
the product and the labor markets are possible. Consider, for example, the case that Firm 1 does not invest
while Firm 2 undertakes a FDI; in the general case, the firms’ profit functions are the following:
Π
1
=p
A
x
1A
+p
B
x
1B
−w
1
x
1A
−w
1
x
1B
−tx
1B, (8)
Π
2
=p
A
x
2A
+p
B
y
2B
−w
2
x
2A
−w
2
y
2B
−F
, (9)
where
pA
=
1
−
x1A
−
x2A
is the price in Country A, which depends both on quantities produced by Firm 1
and the quantities produced by Firm 2’s subsidiary in the same country, while
pB
=
1
−
x1B
−
y2B
, the price in
Country B, depends on Country B’s imports from Firm 1 and the quantity produced by Firm 2 for its domestic
market. Notice that, in the case under examination, Firm 2 may export towards Country A. Nevertheless,
having undertaken the sunk cost of F, Firm 2 does not export because it incurs additional costs of t for the
quantities exported. Therefore, the choice of simultaneous export and local production is less profitable than
the choice of local production only.
Figure 2. Trade and investment boundaries and possible congurations in asymmetric regimes.
60 JCC: The Business and Economics Research Journal
First-order conditions for the maximization of firms’ profits lead to the expressions for the Cournot reac-
tion functions (see Appendix A). Then, if
(w
1
<1, w
2
<1)
, it is possible to show that the solution in terms of
wage rates of the quantity game in Country A, call it
(x
1A
C,x
2A
C)
, is such that:
(x1A
C,x2A
C)=
1
+
w
2−
2w
1
3,
1
+
w
1−
2w
2
3
,iff w1≤
1
+
w
2
2,w2≤
1
+
w
1
2; (10a)
(0, 1−w2
2), iff w1≥1+w2
2,w2<1; (10b)
(1−w1
2,0), iff w2≥1+w1
2,w1<1. (10c)
while the solution of the quantity game in Country B, call it
(x1B
C
,y2B
C
)
, is such that:
(x1B
C,y2B
C)=
1
−
2t
+
w
2−
2w
1
3,
1
+
t
+
w
1−
2w
2
3
,iff w1≤
1
+
w
2−
2t
2,w2≤
1
+
w
1+
t
2; (11a)
(0, 1−w2
2), iff w1≥1+w2−2t
2,w2<1; (11b)
(1−w1−t
2,0), iff w2≥1+w1+t
2,w1≤1−t. (11c)
Figure 2 depicts all the boundary conditions and possible asymmetric configurations. These boundary
conditions generate six qualitatively different regions in the
(w1,w2)
plane, three involving trade, and three
involving local production due to the FDI. In the interior of Region I, all quantities are positive. This region
relates to values of wages sufficiently low such that both firms may undertake international business, either
in the form of exports or using the foreign plant for local production.9
In Region II,
w
1 is high enough, given
t
, that Firm 1 cannot export: In this case only Firm 2 undertakes
international business because of FDI, while
w
1
is still sufficiently low to ensure that Firm 1 produces positive
quantities for the domestic market. In Region III, on the other hand,
w
2 is such that, given
t
, Firm 2 cannot
exploit the production plant located abroad. However, albeit prohibitive to the exploitation of the foreign plant,
w
2 is still sufficiently low to allow domestic production. In Region IV (and similarly in Region V),
w
1 (
w
2)
is so high that Firm 2 (Firm 1) establishes a monopoly in both markets. Regions I, II, III, IV, and V embrace
configurations where forms of international activities occur. In Region VI, in contrast, no international busi-
ness occurs, and firms produce only for the domestic markets.10
As will be shown in the next subsection, each union maximizes its utility function taking into account
specific firms’ labor demand schedules (see Appendix A): The best-reply functions of each union differ
according to the wage rate chosen by the rival. Given the purposes of the paper (equilibriums involving
international activities for both firms), the relevant candidate for sub-game equilibriums has to be found in
Region I. However, some preliminary considerations allow restriction of the field of analysis for the definition
of the relevant best-reply functions.
First, for
(w1,w2)
pairs along the boundary between Regions II and IV and in Region IV, Firm 1 neither
exports nor produces for the domestic market. A similar reasoning applies for
(w1,w2)
pairs along the bound-
ary between Regions III and V and in Region V: Firm 2 is priced out of the market, and it does not produce.
Instead, for
(w1,w2)
internal to Region VI and along the boundaries between Regions II and VI and Regions
III and VI, no international business occurs: Each firm produces only for the domestic market. Second, the
following result is derived.
Proposition 1: Given the assumption that Firm 2 invests, in asymmetric regimes, at any wage pair
(w
1
,w
2
)
internal to Region III or on the boundary between Regions I and III, Union 2 fails to make a best response.
Proof (See Appendix A)
According to Proposition 1, the best reply function of the union of the investing firm is sufficiently low to
allow the exploitation of the foreign plant: Union 2 does not play wage levels in Region III. In the case under
61
Internationalization of Firms’ Activities and Company Union Wage Strategies
examination, the rationale is that, given w1, for w2 such that the point (w1, w2) is on the boundary between
Regions I and III or internal to Region III, the labor demand function for Union 2 is relatively elastic. More
specifically, the percentage change in employment is greater than the percentage change in wage, so that in
absolute value
e=>dl dw wl()()1
III III
2222
, where
l
2III =
y
2B
C
=
(1 3)(1
+
t
+
w
1−
2w
2
)
.11 From this discus-
sion, the field of analysis concerning the determination of sub-game equilibriums in asymmetric regimes can
be restricted to Regions I and II.
Stage 2, unions’ wage setting
Given the labor demands for each significant region (see Appendix A), it is possible now to define the
unions’ payoff functions. Union 1’s relevant payoff function in asymmetric regimes is as follows:
Ω=
-+-
+-
wwwt
www
1
3(2 422)
,(
Region I
)(
12)
1
3
(1 2)
.(
Region II) (13)
1
112
121
This function is continuous over the range of Union 1’s wage rates, namely,
w1∈(0,1)
. Union 1’s utility
is increasing in w
2
, and for a given w
2
, it increases in Region I (Firm 1 exports) when
t
decreases, whilst
remaining unaffected by trade costs in Region II (production only for the domestic market). In both regions,
∂Ω
1
∂
w1
∂
w2
>
0
, wages are strategic complements.
Instead, Union 2’s relevant payoff function is:
Ω=
-++
-+
wwwt
www
1
3(2 42 ), (Region I
)(
14)
1
6
(5 72). (Region II) (15)
2
221
221
This function is also continuous over the range of Union 2’s wage rates, that is,
w2
∈
(0,1)
. Union 2’s
payoff function is increasing in w1. For a given w1, a reduction in t decreases Union 2’s payoff function in
Region I, where Firm 1 exports, while in Region II, Union 2’s utility function is not affected by trade costs.
In addition, in both regions,
∂Ω
2
∂
w2
∂
w1
>
0
, wages are strategic complements. Depending on w2 (w1) and t,
Union 1 (2)’s payoff function presents one maximum or more relative maxima. Therefore, the unions’ payoff
functions are analyzed in relation to the respective independent variables’ changes in order to derive the rela-
tive best-reply functions.
Proposition 2: Unions’ reaction functions, call them respectively
RF
1
and
RF
2
, are as follows:
=
<+≈
=+-∈-
+≤≤+≈
=+ ∈+-
=+-∈+--
RF w
fort
wwtiff wt
fort
ww iff
wt
wwtiff wtt
()
2(22) 0.290 :
1
4(1 ), [0,(5)7);
2(22)20(29 15 2) 0.390 :
1
4(1 ), [0,(22)1));
1
4
(1 ), [(22)1,(5)7).
12
12 2
12 2
12 2
62 JCC: The Business and Economics Research Journal
=
∀∈ +
=++∈-
=+-∈
--
=+ ≥-
RF w
fort
wwtiff wt
wwtiff wt t
wwiffw t
()
0, 20 (29152):
1
8(2 2),[0,(1015)14);
221, [(10 15 )14,(1928)26
);
1
14
(5 2),(19 28 )26).
21
21 1
21 1
21 1
where
t
≤
20 (29
+
15 2 )
≈
0.398
is the critical value above which intra-industry international activities
are not supported in a pure strategy equilibrium.
Proof (see Appendix A)
Figure 3 depicts the two unions’ reaction functions for some definite values of
t
. The left box depicts the
case of trade costs such that the two unions’ best-reply functions are continuous and intersect in Region I. The
center box shows the case of trade costs such that the reaction functions are discontinuous, and the switching
wage is lower than the wage at which the two unions’ best-reply functions intersect. The right box shows the
unions’ reaction functions at
t
=
20 (29
+
15 2 )
, the critical value of trade costs supporting intra-industry
international activities in a pure strategy equilibrium.
Figure 3. Trade and investment boundaries and unions’ reaction functions in asymmetric regimes.
Sub-game perfect equilibria of the two-stage game “unions’ wage determination - f irms’ output decisions”
For trade cost levels less than, or equal to, the threshold of t ≈ 0.390, the Bertrand-Nash wages in equi-
librium are as follows:
w1,As y =
1
3
−
7
30
t,
w2, A sy =
1
3
+
1
15
t.
Substitutions of equilibrium wages into quantity expressions yield:
x1A=
2
9
+
8
45
t;
x1B=
2
9
−
22
45
t;
y2B=
2
9
+
19
90
t;
x2A=
2
9
−
11
90
t.
It is immediately evident that
t
plays a different role in wage levels, depending on the inter national economic
activity the fir m undertakes. In fact, increasing economic integration (a reduction in barriers to trade) stimulates
exports for Firm 1. Consequently, labor demand for Firm 1 increases, and this, in turn, implies that Union 1,
which operates in the exporting firm, chooses to set higher wages: Wages in the exporting firm increase. On
the other hand, a higher degree of economic integration translates both to a decrease in the total production
of Firm 2, the investing firm, and to a wage reduction for its workers. Nonetheless, wages and total output
63
Internationalization of Firms’ Activities and Company Union Wage Strategies
in the multinational firm are always higher than those in the exporting firm, unless t = 0. The union in the
multinational captures higher shares of the firm’s rents generated by the savings in trade costs. However, trade
cost savings imply an expansion in the multinational’s output, and, therefore, an increase in its employment
levels. The rival firm’s decision to undertake FDI shifts the union’s reaction function in the exporting firm
downward. Despite the strategic effect due to wage complementarities, the labor demand effect outweighs
these gains. The rationale is that the exporting firm faces stronger competition in the domestic market when
the competing firm produces locally. It follows that, in asymmetric regimes, wages in the exporting firm are
lower with respect to the IIT case. Subsequent substitutions of equilibrium wages and quantities lead to the
values for union utility level and firms’ profit functions in the case of asymmetric regimes:
Ω1,A sy
NI =
1
675
(7t−10)2,
Π1,A sy
NI =
8
81
−
56
405
t+
548
2025
t2;
Ω2, A sy
IN =
4
675
(5 +t)2,
Π2, A sy
IN =
8
81
+
16
405
t+
241
4050
t2−F.
First stage: Firms’ Selection Strategy and Game Equilibriums
It is now possible to go back to the first stage of the game to investigate the firms’ strategies. Trade and
investment costs and the unions’ wage setting determine the different productive structures that might arise
as equilibriums of the game. In the sub-game defined by firms’ strategy profile (N; N), IIT is supported as the
Nash equilibrium in pure strategies in the two-stage sub-game “unions’ wage determination-firms’ quantity
choices” if trade cost levels are below t ≤ 0.310.
Firms’ payoffs in the RFDI regime depend on wage levels set by unions and the amount of sunk costs.
Conversely, in the two asymmetric sub-games, depending on t, F, and the unions’ wage strategies, inter-
national activities for both firms are supported as the Nash equilibrium in pure strategies in the two-stage
sub-game “unions’ wage determination‒firms’ quantity choices” if
t≤0.398
. The threshold value for the
size of sunk costs derives from
F
∈
0,min(
Πij ,RF DI
II
;
Πij ,As y
IN
)
)
with
i=1,2; j=B,A
, where
Πij ,RF DI
II
=
4 81
and
Π=
-
t(1120) 8100
ij Asy
IN
,
2
.
Figure 4. Production structures in equilibrium.
It can easily be checked that, over the range t ≤ 0.310 ,
Πij ,RF DI
II
≥ Πij ,As y
IN
, with the equality holding only for
t = 0. Hence, the relevant range of F is
)
∈Π=-
≈
=
Ft
0,((11 20) 8100) 0.034
ij Asy
IN
t,
2
0.310
. In fact, for t ≤ 0.310 ,
profits associated with asymmetric structures of international activities are the lowest for the investing firm.
Therefore, this restriction defines the set where the investment strategy can be played by each firm at every
value of t ≤ 0.310 independently from the rival firm’s choice as regards its internationalization strategy.12
Making use of the results of stages 2 and 3 and Restriction 1, the firms’ payoff structure in Stage 1 of the
game is
(
Π1,R FD I
II ;
Π2, R FD I
II )
,
(
Π1,A sy
IN ;
Π2, A sy
NI )
,
(
Π1,A sy
NI ;
Π2, A sy
IN )
,
(
Π1,IIT
NN ;
Π2, IIT
NN )
. As Figure 4 shows, these outcomes
generate three different regions in the relevant
(t,F)
-plane, which represent the equilibriums of the game.
64 JCC: The Business and Economics Research Journal
The set
∈= ≤≤ ≈∪≤= -
∗
tF tFFt tt(( 0) 0 140 471 0.297)()(56 405)(314 675)
2
def ines the first region.
Direct comparison of payoffs shows that, in this area,
Πi,RF DI
II
≥ Πi,Asy
NI
and
Πi,Asy
IN
≥ Πi,IIT
NN
,
i=1,2
. In other
words, to invest is a dominant strategy for both firms. Therefore, the RFDI regime arises in equilibrium.
The second region is defined by the following set of points in the
(t,F)
-plane:
∈= <≤ ∪<≤= -
∗∗∗
tF tFtFFt tt(( 0) 0.297 0.310)()()(56 405)(548 2025).
2
In this region, when firm i
plays the investment strategy I, the rival firm’s j best response is to play the strategy I because
Π≥Π
,
jRFDI
II
jAsy
NI
,,
and vice versa. That is, to invest is a mutual best response and, therefore, the RFDI regime is a Nash equilib-
rium. In contrast, when firm i plays the N strategy, the firm’s j best response is to play the strategy N given
that
Πj,IIT
NN ≥ Π j,Asy
IN . The same reasoning applies if firm j plays the N strategy; not to invest is a mutual best
response. Therefore, the IIT regime is a Nash equilibrium. Thus, there are combinations of the parameters t and
F such that IIT and RFDI regimes arise as simultaneous, symmetric multiple Nash equilibriums of the game.
The rationale for this result can be explained as follows. At first, consider the case that firm
i
plays
the
N
strategy. If firm
j
also plays strategy
N
, profits are
Πj,IIT
NN
. Analytical inspection reveals that
∂Π ∂>
<
t0
jIIT
NN
, if
t
>
<
8 85 ≈0.094
. As Naylor (1998) explains, when
t>0.094
, a decrease of the trade costs
implies that profits fall because the product price decreases due to increasing international market competi-
tion, while wages increase because of the unions’ strategic behavior. These adverse effects on profit more
than offset the benefits of the reduced trade costs. The opposite applies for
t≤0.094
: The cost reduction
effect dominates. On the other hand, if firm
j
plays the strategy
I
, profits are
Πj,Asy
IN
. Differentiation shows
that
∂Π j,A sy
IN
∂
t
>
0,
∀
t
∈
[0, 0.310].
Therefore, a fall of the trade costs implies that profits unambiguously
decrease. The price decreases because of increasing product market competition (a decrease in
t
stimulates
the exports of the rival firm); wages also decrease because the union faces a lower labor demand. However,
the former effect outweighs the latter. For F = 0 and 0.297 < t < 0. 310 ,
Πj,IIT
NN
≥ Π j,Asy
IN
because competition in
the IIT regimes is less fierce than in the asymmetric regime. Nonetheless, as the magnitude of the sunk costs
increases, the profit function associated with the FDI goes down. This implies that for F** (t) < F, besides the
area with relatively high trade costs, also for low trade costs the profits associated with the export strategy
may exceed those related to the investment strategy. The reason is that the investing firm in the asymmetric
regime has to cover the sunk costs with a higher negative competition effect on prices than in the case of IIT.
Consider now the case that f irm i plays the I strategy. If firm j replies by playing I, the profits are
Πj,RF DI
II
. It
is immediately evident that, in the RFDI regime, trade costs have no effect on firms’ profits, which are affected
only by the scale of the sunk costs. On the other hand, if firm
j
replies with the
N
strategy, the profits are
Πj,Asy
NI
. Analytical inspection reveals that
∂Πj,Asy
NI
∂
t
>
<
0
if ≈
>
<
t35 137 0.250.
Similarly to the previous case,
when t > 0.250, a decline in trade costs leads to a fall in profits due to the price decreases driven by product
market competition, while wages rise because of the union’s strategic behavior. The adverse effects on profit
in the product market competition and wage increase more than counterbalance the beneficial effects of the
decline in trade costs, and vice versa, for t ≤ 0.250, the cost reduction effect becomes dominant. However, for
∈= <≤ ∪<≤
∗∗
∗
tF tFtFFt(( 0) 0.297 0.310)()()
,
Πj,RF DI
II
≥ Π j,Asy
NI
because for these combinations of inte-
gration costs, product market competition in the asymmetric regimes is harsher than in the RFDI regime. It
is worth noting that, if sunk costs are low enough, namely
F
∗∗
(t)
<
F
, the RFDI regime arises not only when
trade costs are high (the so-called tariff jumping argument), but also when barriers to trade are low, and IIT
is a viable option.
The explanation is as follows. A reduction in trade costs makes the investment option more attractive. In
fact, differentiation reveals that
∂(
Πij ,As y
IN
)
∂
t
<
0
∀
t
∈
0,0.310
[ ]
as
t
decreases, the profits generated in the
foreign country for the investing firm in the asymmetric regime increase. That is, the investing firm may
disburse a large amount for the sunk costs to enter the foreign market. In other words, the set of the parameter
values for which the investment strategy is feasible enlarges.
Finally, the set of points
(F
∈
(t
=
0) 0
≤
F
≤
0.034)
∪
F
∗∗
(t)
<
F)
characterizes the third region. Straight
forward evaluation of payoffs reveals that, in this region,
Πi,IIT
NN
> Πi,Asy
IN
and
Πi,Asy
NI
> Πi,RF DI
II
, where i = 1,2 ,
not to invest is a dominant strategy for both firms. As the magnitude of the sunk costs increases, the profit
function associated with the investment strategy moves downward. Consequently, for
F
∗∗
(t)
<
F
and low
trade costs, the profits associated with the export strategy exceed those related to the investment strategy. The
rationale resides in the fact that the firm that does not invest in the asymmetric regime faces a less adverse
competition effect on prices than in the case of RFDI without the need to cover the sunk costs. Furthermore,
the combination of relatively large sunk costs and wage levels higher than in the case of exports does not
65
Internationalization of Firms’ Activities and Company Union Wage Strategies
overcome the trade cost savings for firms when they invest. Thus, the IIT regime is the only equilibrium of
the game. These results can be summarized in the following proposition.
Proposition 3: Under Restriction 1:
(a) for
F=0
, in the range
0<t≤0.297
, RFDI is the Nash equilibrium; (b) for
(t
∈
(F
=
0) 0.297
<
t≤0.31) ∪F∗(t)<F≤F∗∗ (t)=(56 405)t−(548 2025)t2
(t∈(F=0) 0.297 <t≤0.310)
∪
F
∗
(t)
<
F
≤
F
∗∗
(t)
=
(56 405)t
−
(548 2025)t2
, multiple equilibriums (simultaneous R FDI
and IIT) arise; and (c) for
(F
∈
(t
=
0) 0
≤
F
≤
0.034)
∪
F
∗∗
(t)
<
F)
, IIT is the unique equilibrium of
the game.
Managerial Implications
To begin with, the first, relevant result arising from this framework is that, in a highly integrated economic
environment characterized by falling trade barriers and slackening regulations in capital markets, the decision
to focus exclusively on domestic business in oligopoly sectors with the presence of unionized labor force is
not valuable. If integration costs are at a level such that an expansion of the firm’s activities abroad is possible
(as depicted in this model), the internationalization strategy is unquestionably more profitable than the simple
domestic business development. The rationale behind this result is straightforward: The internationalization
strategy allows for gaining market shares of the rival companies and, eventually, pricing them out if labor
unions do not moderate their wage demands.
Second, Horn, and Wolinsky (1988) suggested firms may aim at taking strategic advantage of an MNE
organizational structure to prevent the formation of an encompassing union within a company and thereby,
keep wages lower in some plants to reduce the cost of production and gain access to foreign markets. However,
if unionized workers are able to coordinate their bargaining activities at the company level effectively, the
overall picture changes. In fact, due to wage coordination, labor costs maintain relatively high levels and thus
managerial efforts to cut costs to penetrate into a foreign market of the integrated area should be directed
toward other sources.
Third, despite the fact that managers in a company dispose of several technical forms to expand business
such as licensing and franchising to host country firms, and mergers and acquisitions of an already operat-
ing firm into the targeted market, the model in this work focuses on the two alternative internationalization
strategies of exporting and establishing a new, wholly owned subsidiary. On one hand, the exporting strategy
avoids the substantial cost of setting up manufacturing operations into another country. On the other hand,
from a managerial viewpoint, a wholly owned subsidiary in the form of a Greenfield venture, as delineated
in the model, is justified by the minimization of the risk of losing control over technological competences
and tight control over operations in different countries. However, this entry mode into a foreign market is, in
general, the most costly from the point of view of the capital investment (capital costs and risks).
In the case of horizontal FDI, with replication of the same production process at home and in the host
country with the presence of company level unions, wage rates keep up high levels because of coordination.
In the case of the exporting strategy, declining trade barriers allow unions to raise wage demands because
of the effect of the firms’ product market expansion. Thus, whatever the internationalization strategy chosen
to expand cross-border business, labor costs play a major role. Nevertheless, the integration costs affect in
different ways the profitability of the selected strategy. The sunk costs have a direct impact on profits in the
case of the investment strategy. The trade costs have both a direct (increasing costs) and indirect impact (due
to their effect on the union wage strategy) on firms’ profitability in the case of the exporting strategy. The
findings in this work imply that, in a situation characterized by elevated wages and deep economic integra-
tion, the investment strategy is advantageous when the size of the sunk costs is small. Worth noting is the
fact that to invest may be more beneficial than to export for low values of trade costs. On the other hand,
for intermediate values of the sunk costs, both internationalization strategies arise in equilibrium while, for
excessive sunk costs, only the exporting strategy is gainful for firms. In other words, if lower-cost locations
for manufacturing are not available (in this model, labor costs, because of wage coordination), the choice of a
FDI should be taken when it is required to have strict control over core managerial competences (for example,
marketing skills), and the cost of building and technical equipment to set up a new plant are not extremely
large. In contrast, if trade (transportation and tariff ) costs are low enough while the cost of manufacturing a
plant is significant and core managerial competences can be controlled easily from the domestic country, the
exporting strategy should be preferred.
66 JCC: The Business and Economics Research Journal
Conclusion
In this paper, through the reduction in trade costs and the possibility to undertake direct investment in
a foreign country on firms and unions’ strategic behavior, the consequences of the process of international
market integration were dealt with and exemplified. A general framework, to analyze how these two aspects
of economic integration affect firms’ decisions concerning international business and the strategic behavior
of company-wide unions in the labor market, is developed. In the model, firms are allowed to choose their
internationalization strategy. The basic two-way IIT analytical framework of Naylor (1998, 1999) and the
FDI-autarky model of Naylor and Santoni (2003) are complemented. In a three-stage game, firms act as first
movers and choose independently whether to invest in a foreign country; monopoly labor unions select their
wage strategy in the second stage; in the third stage output is realized. The focus of the model is on a subset
of the integration costs (trade and sunk costs), such that firms can initiate international activities. Trade
costs affect the union’s wage strategy formation, and this, in turn, affects the strategic behavior of firms.
Considering the wage strategies of rival unions, the complete set of production structure regimes arising as
sub-game perfect Nash equilibriums for different combinations of trade and sunk costs is derived. The main
results are as follows.
Whenever a firm invests abroad to start international business, company unions cannot choose a prohibi-
tive wage rate condemning their workers to be priced out of the labor market. Nevertheless, labor unions gain
a larger share of the firms’ rents than in the IIT regime because of savings in trade costs: Company-wide
unions may welcome FDI.
Union wages exclusively influence the firms’ payoffs in the RFDI regime. The firms’ strategy profile
(N;N)
defines that IIT is supported as the Nash equilibrium in pure strategies if and only if the trade cost
level is below t ≤ 0.310. This result is also obtained in Naylor (1998). The two sub-games identified by the
firms’ strategy profiles
(I;N)
and
(N;I)
, if
t≤0.398
, define that the company-level union operating in the
exporting firm sets a wage level such that the firm would export in the other country. The union in the invest-
ing firm sets a wage such that the company can exploit the production facilities in both countries.
Because of the interdependence of t, F and unions’ strategic behavior, equilibriums involving different
configurations of international activities arise. Nonetheless, some noteworthy observations can be addressed.
First, the RFDI regime also arises as equilibrium for low values of trade costs: To invest is a viable strategic
option for a firm not only for the tariff jumping argument, but also when IIT is feasible. The reason lies in
the fact that increasing economic integration makes the investment option increasingly suitable. Second, the
interdependency between trade and sunk costs and the unions’ strategic behavior generates the conditions
such that multiple symmetric equilibriums (RFDI and IIT) may arise in the game. Finally, if sunk costs are
sufficiently large, IIT is the unique equilibrium of the game: If firms want to invest, the size of the sunk costs
and the wage levels higher than in the case of exports do not offset the trade cost savings.
However, caution is advised with respect to the general conclusion of this article. The analysis uses a basic
framework. The model presents a certain lack of robustness because of specific functional forms for utility,
production and cost f unctions. These represent all the drawbacks of the model. As Naylor (1999) suggested, a more
general right-to-manage model of wage bargaining is a suitable way to develop this work. It would be interesting
to test, empirically, if the prospect of company-wide negotiations conducted by unique workers’ representatives
affects firms’ strategic decisions related to international activities. This is left for research in the future.
Endnotes
1 Recently revised in 2009, Directive 2009/38/EC of the European Parliament and of the Council of 6 May 2009 on the
establishment of a European Works Council or a procedure in Community-scale undertakings and Community-scale
groups of undertakings for the purposes of informing and consulting employees, OJ EU L. 122 of 16.5.2009.
2 On collective bargaining in MNEs, see also European Trade Union Confederation (2007), Eurofound (2009), and
European Commission (2009, 2011).
3 Consumers’ utility will take similar forms in the case of production in the presence of FDI.
4 This paper considers only pure strategies.
5 The condition t < 1 represents a “viability condition”. In fact, for t > 1, exports will never occur.
6 In principle, there are two additional outcomes in this sub-case, namely the one-way trade regimes. As Naylor (1999)
shows, if countries are symmetric and both labor markets are unionized, one-way trade is not an equilibrium regime.
67
Internationalization of Firms’ Activities and Company Union Wage Strategies
7 As Brander (1981) pointed out, it might be argued that it is unrealistic to take the quantity rather than the price as
the firm’s strategic variable. The Cournot setting in the output market has been chosen in this paper for the sake of
simplicity.
8 In fact, profits generated in the foreign market by the firm that invests have to be greater than the size of the fixed costs
to undertake the investment in the foreign country. However, these profits differ according to the strategy selected by
the rival firm. Therefore, if this restriction does not hold, the investment strategy is not always practicable, and the
model collapses in Naylor’s (1999) analysis.
9 Additional analytical details are available upon request from the Author.
10 The frontier of the area for reciprocal intra-industry international business is the union of the following four sets of
points in the
(w1,w2)
plane: (a)
w1
=
(1
+
w2
−
2t) 2 , w1
≤
1
−
(4 3)t;
(b)
w2
=
(1
+
w1) 2 , w2
≤
1
−
(2 3)t;
(c) w1 = 0,
w2 ≤ 1/2; and (d)
w2
=
0, w1
≤
(1
−
2t) 2
; with
--tt(1 (4 3) ,1 (2 3) )
being the intersection (the upper vertex of
Region I in Figure 2) between the graphs representing the equations (a) and (b). Differentiation of the inequalities in (a)
and (b) leads to
dw1dt <0
and
dw2dt <0
: The graph of the equation in (a) shifts down to the right, while the graph
of the equation in (b) moves up to the left. That is, a decrease in trade costs expands Region I, increasing the opportuni-
ties for intra-industry international activities. It is worth noting that, with the presence of an investing firm, the range of
t such that intra-industry international activities take place in asymmetric regimes is wider than that in the case of IIT
found in Naylor (1999). To be more precise, for a given level of trade costs t, a given (w1, w2) pair may not be congruent
with positive exports in both countries in the case of IIT, while it may be consistent with a situation of Firm 1’s positive
exports and the foreign plant’s exploitation of Firm 2 in the asymmetric regime under examination. Further analytical
details are available upon request from the Author.
11 Theoretically, there are three other regions that are not depicted in Figure 2 (the axes’ length equals 1, and is therefore
outside of the surface of the box diagram). The first region is characterized by w1 > 1 and w2 < 1; the second region by
w1 < 1 and w2 > 1; the third region by w1 > 1 and w2 > 1. In the first two regions, wage rates are so high that in Country
A (B) there is neither production nor consumption, but w2 (w1) is sufficiently low so as to make production for the
domestic market worthwhile. On the other hand, in the third region wage rates are so high that, in both countries, there
is neither production nor consumption.
12 This result can be checked as follows. The elasticity of the labor demand in Region III, in absolute value, is
ε
l
2III =(2w
2
(1+t+w
1
−2w
2
))
. The inequality
ε
l
2III >1
holds if and only if
4w2>1+t+w1
. Taking into
account that
w1<1−t
(otherwise the point
(w1,w2)
is outside Region III), this condition may be rewritten as
4w2>1+t+1−t−δ = 2− δ
, where
δ ≡ 1−t−w1>0.
However,
4w2>2−δ
always holds true since
4w2≥2
: The
last result follows from the fact that
2w2
≥
1
+
w1
(otherwise the point (w1, w2) is certainly outside Region III).
3 If the restriction on sunk costs does not hold, the investment strategy is not always practicable. Therefore, the results of
the analysis are those obtained in Naylor (1998).
References
Borghijs, A., & Du Caju, P. (1999). Globalisation, EMU and European trade union cooperation (Department of Economics
Research Paper No. 99-013). Antwerp, Belgium: University of Antwerp (UFSIA).
Brander, J. (1981). Intra-industry trade in identical commodities. Journal of International Economics, 11(1), 1-14.
Bughin, J., & Vannini, S. (1995). Strategic direct investment under unionized oligopoly. International Journal of Industrial
Organization, 13(1), 127-145. dx.doi.org/10.1016/0167-7187(94)00447-A
Dowrick, S. J. (1989). Union-oligopoly bargaining. Economic Journal, 99(398), 1123-1142.
Eckel, C., & Egger, H. (2009). Wage bargaining and multinational rms. Journal of International Economics, 77(2),
206-214. dx.doi.org/10.1016/j.jinteco.2009.01.004
European Commission. (2008a). External and intra-European Union trade. Luxembourg, Luxembourg: Ofce for Ofcial
Publications of the European Communities.
European Commission. (2008b). European Union foreign direct investment yearbook. Luxembourg, Luxembourg: Ofce
for Ofcial Publications of the European Communities.
European Commission. (2009). Industrial Relations in Europe 2008. Luxembourg, Luxembourg: Ofce for Ofcial
Publications of the European Communities.
European Commission. (2011). Industrial Relations in Europe 2010. Luxembourg, Luxembourg: Ofce for Ofcial
Publications of the European Communities.
European Industrial Relation Observatory online (EIROnline). (2009). Cross-border cooperation at company level in
banking sector. Retrieved from http://www.eurofound.europa.eu/eiro/2009/07/articles/dk0907041i.htm
European Trade Union Confederation. (2007). The Coordination of Collective Bargaining 2008. Retrieved from http://
www.etuc.org/a/4459.
68 JCC: The Business and Economics Research Journal
European Foundation for the Improvement of Living and Working Conditions (Eurofound). (2009). Multinational compa-
nies and collective bargaining. Retrieved from http://www.eurofound.europa.eu/eiro/studies/tn0904049s/tn0904049s.
htm
Gennard, J. (2009a). Development of transnational collective bargaining in Europe. Employee Relations, 31(4), 341-346.
dx.doi.org/10.1108/01425450910965405
Gennard, J. (2009b). A new emerging trend? Cross border trade union mergers. Employee Relations, 31(1), 5-8. dx.doi.
org/10.1108/01425450910916788
Glass, A. J., & Saggi, J. (2005). Exporting versus direct investment under local sourcing. Review of World Economics,
141(4), 627-647.dx.doi.org/10.1007/s10290-005-0049-1
Huizinga, H. (1993). International market integration and the union wage bargaining. Scandinavian Journal of Economics,
95(2), 249-255.
Horn, H., & Wolinsky, A. (1988). Worker substitutability and patterns of unionisation. The Economic Journal, 98(391),
484-497.
Ishida, J., & Matsushima, N. (2009). Domestic competition and foreign direct investment in unionized oligopoly (The
Institute of Social and Economic Research Discussion Paper No. 757). Osaka, Japan: Osaka University. dx.doi.
org/10.2139/ssrn.1485240
Jovanović, M. N. (2006). The Economics of International Integration. Cheltenham, UK: Edward Elgar.
Keune, M., & Schmidt, V. (2009). Global capital strategies and trade union responses: Towards transnational collective
bargaining? International Journal of Labour Research, 1(2), 9-26. Geneva, Switzerland: International Labour Ofce.
Leahy, D., & Montagna, C. (2000). Unionisation and foreign direct investment: Challenging conventional wisdom? The
Economic Journal, 110 (462), 80-92. dx.doi.org/10.1111/1468-0297.00522
Lommerud, K. E., Meland, F., & Sørgard, L. (2003). Unionised oligopoly, trade liberalisation and location choice. The
Economic Journal, 113 (490), 782-800. dx.doi.org/10.1111/1468-0297.t01-1-00154
Mezzetti, C., & Dinopoulos, E. (1991). Domestic unionization and import competition. Journal of International Economics,
31(1-2), 79-100. dx.doi.org/10.1016/0022-1996(91)90057-D
Müller, T., Platzer, H.-W., & Rüb, S. (2013). Transnational company agreements and the role of European Works Councils
in negotiations: A quantitative analysis in the metalworking sector (European Trade Union Institute Report No. 127).
Brussels, Belgium: ETUI aisbl.
Naylor, R. (1998). International trade and economic integration when labour markets are generally unionised. European
Economic Review, 42(7), 1251-1267. dx.doi.org/10.1016/S0014-2921(97)00075-5
Naylor, R. (1999). Union wage strategies and international trade. The Economic Journal, 109(452), 102-115. dx.doi.
org/10.1111/1468-0297.00394
Naylor, R., & Santoni, M. (2003). Foreign direct investment and wage bargaining. Journal of International Trade and
Economic Development, 12(1), 1-18. dx.doi.org/10.1080/0963819032000049178
Papadakis, K. (2010). Transnational company agreements on enterprise restructuring (Dialogue in Brief No. 2, July
2010). Geneva, Switzerland: ILO Industrial and Employment Relations Department (DIALOGUE). Retrieved from
http://www.ilo.org/wcmsp5/groups/public/---ed_dialogue/---dialogue/documents/publication/wcms_158625.pdf
Pernicka, S., & Glassner V. (2012). Horizontal Europeanization through trade union strategies in wage bargaining?
A neo-institutional framework (DFG Horizontal Europeanization Research Unit WP 2012/2). Oldenburg, Germany:
University of Oldenburg. Retrieved from http://www.horizontal-europeanization.eu/downloads/pre-prints/PP_
HoEu_2012 02_pernicka_glassner_trade_union_strategies_0.pdf
Sørensen, J. R. (1993). Integration of product markets when labour markets are unionised. Recherches Economiques de
Louvain, 59(4), 485-502.
Straume, O. R. (2002). Union collusion and intra-industry trade. International Journal of Industrial Organization, 20(5),
631-652. dx.doi.org/10.1016/S0167-7187(01)00091-1
Strozzi, C. (2007). Product market integration and union collusion. Review of International Economics, 15(1), 17-36.
dx.doi.org/10.1111/j.1467-9396.2007.00647.x
Strozzi, C. (2008). Union coordination and economic integration. Saarbrücken, Germany: VDM Verlag Dr. Müller.
Zhao, L. (1995). Cross-hauling direct foreign investment and unionized oligopoly. European Economic Review, 39(6),
1237-1253. dx.doi.org/10.1016/0014-2921(94)00038-2
Zhao, L. (1998). The impact of foreign direct investment on wages and employment. Oxford Economic Papers, 50(2), 284-301.
69
Internationalization of Firms’ Activities and Company Union Wage Strategies
Author Note
Domenico Buccella, Department of Economics, Leon Kozminski University, Jagiellonska Street 57/59 – 03-301,
Warsaw, Poland.
Correspondence should be addressed to Domenico Buccella, Email: buccella@kozminski.edu.pl
I am indebted to Vincent Charles, Editor-in-Chief of this Journal. I also thank an anonymous referee for constructive
comments. I am also grateful to Massimo De Francesco, Paolo Pin, Chiara Strozzi, Michele Santoni, Adrian W. Risso,
Lapo Filistrucchi, Karl Farmer as well as seminar participants at the Universities of Siena, Pisa, Graz, Warsaw and Leon
Kozminski University in Warsaw for their helpful comments, discussions and suggestions. I would like to thank also
Marco Paolo Tucci for the encouragement, and Kate Webster who contributed to improve the style of this paper. The
usual disclaimers apply.
70 JCC: The Business and Economics Research Journal
Appendix
Firms’ Reaction Functions in the RFDI Regime
From first-order conditions for profit maximization of Equations 3 and 4, the following Cournot reaction
functions are derived:
=
-
-≤
-
xx
w
xwxx w
()
1
2
1
2,for ( ,):1;
0, otherwise.
AA
EA
E
A
E
A
E
12
1
2122 1 (A.1)
=
-
-≤
-
yy
w
ywyy w
()
1
2
1
2,for ( ,):1;
0, otherwise.
BB
EB
E
B
E
B
E
12
1
2122 1 (A.2)
=
-
-≤
-
xx
w
xwxx w
()
1
2
1
2,for ( ,):1;
0, otherwise.
AA
EA
E
A
E
A
E
21
2
1211 2 (A.3)
=
-
-≤
-
yy
w
ywyy w
()
1
2
1
2,for ( ,):1;
0, otherwise.
BB
EB
E
B
E
B
E
21
2
1211 2 (A.4)
From Equation A.1, it can be seen that Firm’s 1 production for the domestic market is certainly zero,
regardless of
x2A
E, whenever
w1
≥
1
. Each of the Equations A.1 to A.4 also provides an upper limit for the
wage facing a firm, a wage not to be exceeded in order for that firm’s best response not to be zero, even when
the rival is expected to offer zero output in the product market concerned. This upper bound is, for example,
w2
=
1
as far as Firm 2’s local production in Country A is concerned. These upper bounds for the wages so
identified are shown in Figure 3. Equations A.1 and A.3, together with the equations of the realization of expec-
tations,
x2A
=
x2A
E
and
x1A
=
x1A
E
, and Equations A.2 and A.4, with
y2B
=
y2B
E
and
y1B
=
y1B
E
, represent the two
independent systems whose solutions determine the Cournot quantities in Countries A and B, respectively.
Firms’ Reaction Functions in the Asymmetric Regime
First-order conditions for the maximization of firms’ profits of Equations 8 and 9 lead to these expressions
for the Cournot reaction functions:
=
-
-≤
-
xx
w
xwxx w
()
1
2
1
2,for ( ,):1;
0, otherwise.
AA
EA
E
A
E
A
E
12
1
2122 1 (A.5)
=
--
-≤
--
xy
wt
ywyy
wt
()
1
2
1
2,for ( ,):1 ;
0, otherwise.
BB
EB
E
B
E
B
E
12
1
2122 1 (A.6)
=
-
-≤
-
xx
w
xwxx w
()
1
2
1
2,for ( ,):1;
0, otherwise.
AA
EA
E
A
E
A
E
21
2
1211 2 (A.7)
=
-
-≤
-
yx
w
xwyx w
()
1
2
1
2,for ( ,):1;
0, otherwise.
BB
EB
E
B
E
B
E
21
2
1211 2 (A.8)
From Equation A.5, it is seen that Firm 1’s production for the domestic market will certainly be zero,
no matter
x
2A
E
, whenever
w
1
≥1
, and similarly Firm 1’s exports will certainly be zero, no matter
y
2B
E
, if
71
Internationalization of Firms’ Activities and Company Union Wage Strategies
w
1
≥1−t
. Likewise, Firm 2’s production for the domestic market (exports) will certainly be zero, no matter
x
1B
E
(
x
1A
E
) whenever
w
2
≥1
. The upper bounds for the wages so identified are shown in Figure 2. In this case,
Equations A.5 and A.7, with the relative equations for the expectations’ realization,
x
2A=
x
2A
E
and
x
1A=
x
1A
E
,
and Equations A.6 and A.8 with
y
2B=
y
2B
E
and
x
1B=
x
1B
E
, represent the two independent systems whose
solutions establish the Cournot outputs in Countries A and B, respectively.
Firms’ Labor Demand Schedules in the Asymmetric Regime
Under the assumptions adopted in subsection 2.3 (Firm 1 does not invest and Firm 2 invests), Union 1
faces the following labor demand functions:
+= -+-≤
+
≤
+-
xx wwtfor w
w
w
wt
1
3
(2 422),
1
2
,
12
2;(
Region I)
AB11 12 2
1
1
2 (A.9)
=+-
+-
<≤
+
≤
+
xwwfor
wt
w
w
w
w
1
3
(1 2),
12
2
1
2
,
1
2;(
Region II)
A121
2
1
2
2
1 (A.10)
while Union 2’s labor demand functions are:
+=
-++≤
+≤+-
-+ +-
<≤
+≤+
yx
wwtfor w
w
w
wt
ww forwt
wwww
1
3(2 42 ),
1
2,
12
2; (Region I)(A.11a)
1
6
(5 72), 12
2
1
2
,1
2
.(Region II) (A.11b)
BA22
21 2
1
1
2
21
2
1
2
2
1
With these elements, the analysis derives the unions’ best-reply functions and the value of the trade costs
allowing intra-industry international activities to be supported in equilibrium in pure strategies.
Proof of Proposition 1
To prove Proposition 1, Union 2’s payoff function is analyzed in relation to changes of the independent
variables
w1
and
t
in Region III. Union 2’s payoff in that region is the following:
Ω2=w2
1
3
(1+t+w1−2w2)
.
This function is continuous over the range
w2
∈
(0,1)
. Union 2’s payoff function in Region III is: (a) increas-
ing in
w1
; (b) for a given
w1
, a reduction in
t
decreases Union 2’s payoff; (c)
∂Ω
2
∂
w2
∂
w1
>
0
, namely
wages are strategic complements; and (d)
∂
2
Ω
2
∂
w2
2
<
0
, the payoff function is concave with respect to w2.
Suppose now that Union 2 chooses to set a wage rate in Region III. In this region, Firm 2 does not exploit
the foreign plant while Firm 1 exports. For
w
1
∈0,1 −(4 3)t
[ ]
, the left derivative of Union 2’s payoff func-
tion, evaluated at
w
2 such that the
(w
1
,w
2
)
pair is on the boundary between Regions I and III, is equal to:
∂Ω2III ∂
w
2w2=(1 2 ) (1+w1)= −
1
+
t
−
w
1<
0,
∀
t
∈
(0,1).
Since the derivative is non-positive at the boundar y between Region I and Region III, it will be non-positive
for any
w
2 such that
(w
1
,w
2
)
lies in Region III, given the concavity of
Ω2 with respect to
w
2 in Region III.
The payoff function is decreasing across Region III and, therefore, Union 2’s reaction function cannot be in
that region.
Proof of Proposition 2
a.0) First, let us consider Union 1’s payoff function. Notice that, for
t
<
3 4
,
1 2
<
1
−
(2 3)t
: The value
of the positive, vertical, intercept of the line
w2
=
(1 2)(1
+
w1)
(the boundary between Regions I and III in
Figure 2), lies below the value of
w2
at the point of intersection between the boundaries
w2
=
(1 2)(1
+
w1)
72 JCC: The Business and Economics Research Journal
and
w
1
=(1 2)(1+w
2
−2t)
(the upper vertex of Region I in Figure 2). Region I exists in the first quadrant
of the Cartesian plan if and only if
t<3 4
. This implies that for
w
2
∈0,1 2
[ ]
, every
(w
1
,w
2
)
pair belongs
neither to Region III nor to Region VI.
a.1) The analysis begins by taking into account
w2
∈
0,1 2
[ ]
and ≤=
t14 0.250
. For
w2
∈
0,1 2
[ ]
and
≤=
t14 0.250,
the reaction function cannot be in Region II so long as the right derivative of the Union 1
payoff function, evaluated at
w1
such that
(w1,w2)
lies along the boundary between Regions I and II, is
negative. This occurs for:
∂Ω1II ∂
w
1w1=(1 2 ) (1+w2−2t)= −
1
−
w
2+
4t
<
0
⇒
w
2>
4t
−
1
.
Thus, for
t
≤
1 4
, because the derivative is non-positive at the boundary between Region I and Region II,
it will be non-positive for any
w1
such that
(w1,w2)
lies in Region II; given the concavity of
Ω
1
with respect
to
w1
in Region II, no interior maximum exists in Region II. Instead, along the boundary between Regions I
and II, it can be checked that the left derivative
∂Ω
1
∂
w1
<
0
. Given the concavity of
Ω
1
with respect to w1,
a relative maximum in Region I exists if and only if this condition holds: For
(w1
=
0, w2
≤
1 2)
, namely for
the points below or equal to the value of the vertical intercept of the line representing the boundary between
Regions I and III,
∂Ω
1
∂
w1
>
0
. This can be confirmed easily to be always the case, however small
w2
may be.
a.2) Let us continue by considering
w
2
∈(1 2 ,1 −(2 3)t]
and
≤=t14 0.250.
For
w
2
∈(1 2 ,1 −(2 3)t]
,
some
(w
1
,w
2
)
pairs reside in the interior of Region III and along the boundary between Region I and Region
III, while some other
(w
1
,w
2
)
pairs are in the interior of Region I. An interior maximum in Region I exists if
and only if, for
w
1 such that the
(w
1
,w
2
)
pair is on the boundary between Region I and Region III, the right
derivative
∂Ω1∂
w
1>
0
. This holds for:
∂Ω1I∂
w
1w1=2w2−1=
5
−
7w
2−
t
( )
≥
0
⇒
w
2≤
(5
−
t) 7 .
Further analytical inspection reveals that
(5 −t) 7 <1−(2 3)t
as long as
t<6 11 ≈0.540.
Summarizing,
because for
w
2
∈[0,1 2]
and
≤=t14 0.250
every
(w
1
,w
2
)
pair is always outside Regions III, V, and
VI, it follows that the reaction function is
RF
1
(w
2
)=w
1
=(1+w
2
−t) 4
because, for these values of w2,
Ω1II
(w
1
,w
2
)
< Ω1I
(w
1=
(1
+
w
2−
t) 4 , w
2
)
(from part a.1). On the other hand, for
w
2
∈(1 2 ,1 −(2 3)t)
and
t≤1 4
, the reaction function is
RF
1
(w
2
)=w
1
=(1+w
2
−t) 4
for wage rates
w
2
∈(1 2 , (5 −t) 7))
. For
w
2
>(5 −t) 7
, Union 1 plays a wage level such that the resulting
(w
1
,w
2
)
pair will be interior to Region III.
Given that trade costs are
1 4 <6 11
, according to Proposition 1, this wage combination cannot be an equi-
librium of the unions’ wage setting sub-game because Union 2 fails in making a best-response in Region III.
a.3) For
w2
∈
0,1 2
[ ]
, as
t
increases marginally above
1 4
, there are levels of
w2
such that Union 1’s
utility function,
Ω
1
, has an interior relative maximum in Region II. In fact, for trade costs marginally above
t
=
1 4
, the right derivative of the Union 1 payoff function, evaluated at
w1
such that
(w1,w2)
lies along the
boundary between Regions I and II, is positive (or equal to zero) if:
∂Ω1II ∂
w
1w1=(1 2 ) (1+w2−2t)= −
1
−
w
2+
4t
≥
0
⇒
w
2≤
4t
−
1
.
Thus, for
w2
∈
[0, 4t
−
1]
, the first-order conditions of Equation 13 yield that, in Region II, a relative
maximum is reached at
w1
=
(1
+
w2) 4
. Nevertheless, it can be checked that, for trade barriers margin-
ally above
1 4
and
w2
∈
[0, 4t
−
1]
,
Ω
1II (w1
=
(1
+
w2) 4 , w2)
< Ω
1I(w1
=
(1
+
w2
−
t) 4 , w2)
. Moreover, for
w2
∈
(1 2 ,1
−
(2 3)t]
and trade costs marginally above
1 4
, the analysis conducted in a.2 for this range of
w2
remains unaffected. This, in turn, implies that for
w2
∈
[0, (5
−
t) 7))
, the reaction function is in Region I
and is equal to
RF
1(w2)
=
w1
=
(1
+
w2
−
t) 4
.
a.4) The discussion in part a.3 has shown that for trade barriers marginally above
1 4
, there are levels of
w2
such that
Ω
1
has an interior relative maximum in Region II. As trade cost levels increase, there comes a
point such that, for
w2∈0,1−(2 3)t
[ ]
, the relative maximum in Region II equals or is higher than the relative
maximum of the payoff function in Region I. That is:
Ω
1II (w1
=
(1
+
w2) 4 , w2)
=
(1 24)(1
+
w2)
2≥ Ω
1I(w1
=
(1
+
w2
−
t) 4 , w2)
=
(1 12)(1
+
w2
−
t)
2.
This occurs if and only if
w
2≤
(2
+
2 )t
−
1
, representing the switching wage level in the asymmetric
regime. This, in turn, implies that for
≥+≈
t1(22) 0.290,
there are levels of
w
2
sufficiently low that
73
Internationalization of Firms’ Activities and Company Union Wage Strategies
the relative maximum of
Ω1 in Region II is equal to or higher than the relative maximum of
Ω1 in Region I.
Differentiation of the switching wage shows that
dw
2
d t >0
: As trade cost decreases, the range of
w
2 such
that Union 1’s best reply is a wage allowing Firm 1 to export gets smaller. Summarizing, for
w
2
∈0,1 −(2 3)t
[ ]
and trade cost levels marginally above
+≈
1(22) 0.290,
the analysis shows that Union 1’s reaction function
is
RF
1
(w
2
)=(1+w
2
) 4
for
w
2∈
(0, (2
+
2 )t
−
1]
, while for
w
2∈
((2
+
2 )t
−
1, (5
−
t) 7))
, the reaction
function is
RF
1
(w
2
)=w
1
=(1+w
2
−t) 4
.
a.5) For
w2
∈
0,1
−
(2 3)t
[ ]
, as
t
rises and reaches
t
>
1 3
, the left derivative
∂Ω
1I
∂
w1
for Union 1’s
wages, such that the
(w1,w2)
pair is along the boundary between Regions I and II, is negative if:
∂Ω1I∂
w
1w1=(1 2 ) (1+w2−2t)= −
1
−
w
2+
3t
<
0
⇒
w
2>
3t
−
1
.
Given that
t
>
1 3
, there are sufficiently high levels of
w2
that the previous condition is satisfied. This
condition says simply that, in the range
w2
∈
(0, 3t
−
1)
, the function
Ω
1
is increasing in Region I. The analysis
carried out in parts a.3 and a.4 has shown that there is a range of values of
w2
such that the utility function
Ω
1
has a relative interior maximum in Region II. Thus, for
w2
∈
0,1
−
(2 3)t
[ ]
, if trade barriers are marginally
above
1 3
, an interior relative maximum exists in Region I for
w2
∈
(3t
−
1, (5
−
t) 7))
, and an interior maxi-
mum is in Region II for
w2
∈
(0, 4t
−
1)
. Moreover, because
0<t
, the switching wage
(2
+
2 )t
−
1
>
3t
−
1
.
Therefore, for
w2
∈
(0, 3t
−
1)
, the function
Ω
1
has a relative maximum in Region II, and direct payoff compari-
son shows that, for
w2
∈
(3t
−
1, (2
+
2 )t
−
1]
,
Ω
1II (w1
=
(1
+
w2) 4 , w2)
≥ Ω
1I(w1
=
(1
+
w2
−
t) 4 , w2)
; for
wages
w2
∈
((2
+
2 )t
−
1, (5
−
t) 7))
,
Ω
1II (w1
=
(1
+
w2) 4 , w2)
< Ω
1I(w1
=
(1
+
w2
−
t) 4 , w2)
. Thus, Union
1’s reaction function is
RF
1(w2)
=
(1
+
w2) 4
for
w2
∈
(0, (2
+
2 )t
−
1]
, and
RF
1(w2)
=
w1
=
(1
+
w2
−
t) 4
for
w2
∈
((2
+
2 )t
−
1, (5
−
t) 7))
.
a.6) For
w
2
∈0,1 −(2 3)t
[ ]
, as
t
increases, there comes a point when the values of
w
2
=3t−1
and
w
2
=(5 −t) 7
equal
w
2
=1−(2 3)t
. This occurs at
=≈t611 0.540.
For
t
above this level,
Ω1 is increas-
ing across Region I and thus this function has a relative maximum only in Region II. Therefore, the reaction
function is
RF
1
(w
2
)=(1+w
2
) 4
for
w
2
∈(0, 5 7 ]
. The value of
5 7
is the value of the intersection of the
reaction function with the boundary between Regions II and VI.
b.0) Let us now consider Union 2’s payoff function. The analysis starts by taking into account the case
of
w
1
∈0, (1−2t) 2
[ ]
, where
w
1
=(1−2t) 2
is the intercept of the line representing the boundary between
Regions I and II in Figure 2, given by
w
1
=(1 2)(1+w
2
−2t)
. Notice that, for
t∈(0,1)
,
-<-tt(1 2)21(4 3) .
This implies that, for
w
1
∈0, (1−2t) 2
[ ]
, every
(w
1
,w
2
)
pair does not belong to Regions II and VI. Notice
also that for
t<3 8
,
1 2 <1−(4 3)t
, where
w
1
=1 2
is the intercept of the line representing the boundary
between Regions II and IV in Figure 2, given by
w
1
=(1 2)(1+w
2
)
.
b.1) Proposition 1 has shown that the right derivative of Union 2’s payoff function, evaluated at
w2
such
that
(w1,w2)
lies along the boundary between Regions I and III, is negative
∀
t
∈
(0,1)
. Thus, because the
derivative is non-positive at the boundary between Region I and Region III, it will be non-positive for any
w2
such that
(w1,w2)
lies in Region III; given the concavity of
Ω
2
with respect to
w2
in Region III, no interior
maximum exists in Region III.
b.2) Let us consider, for
w
1
∈[0,1−(4 3)t)
and
t<3 8
, the behavior of the function
Ω2
due to changes in
w
2 for
(w
1
,w
2
)
pairs belonging to Region I . Given the concavity of
Ω2 with respect to
w
2,
Ω2 has a relative
maximum in Region I, if and only if the two following conditions hold: First, for
(w
2
=0, w
1
≤(1−2t) 2)
, that
is, for points to the left of or equal to the intercept on the horizontal axis of the boundary between Regions I
and II,
∂Ω2∂
w
2>
0
. This can be checked to always hold true. Second, for
w
1
∈((1−2t) 2 , 1−(4 3)t)
, the
right derivative of
Ω2 with respect to
w
2 is positive for
w
2 such that the
(w
1
,w
2
)
pair is along the boundary
between Region I and Region II. This holds for:
∂Ω2I∂
w
2w1=(1 2 ) (1+w2−2t)=
(10
−
14w
1−
15t)
≥
0
⇒
w
1≤
(10
−
15t) 14
.
Because
t<1
, it follows that
(10 −15t) 14 t<1−(4 3)t
. Thus, for
w
1
∈0, (1−2t) 2
[ ]
, an interior rela-
tive maximum exists for
Ω2 in Region I. For
w
1
∈((1−2t) 2 , (10 −15t) 14))
, a relative maximum of this
function at the interior of Region I exists. Given that for
w
1
>(10 −15t) 14
the right derivative
∂Ω2∂
w
2<
0
for
w
2 such that the
(w
1
,w
2
)
pair is along the boundary between Region I and Region II,
Ω2 has a relative
maximum on the boundary between Regions I and II. For
w
1
∈((10 −15t) 14 , 1 2)
, given the concavity of
74 JCC: The Business and Economics Research Journal
Ω2 with respect to
w
2, this function has an interior maximum in Region II if and only if the left derivative
∂Ω2∂
w
2 along the boundary between Regions I and II is negative. This occurs when:
∂Ω2II ∂
w
2w2=2w1−1+2t=
19
−
26w
1−
28t
( )
≤
0
⇒
w
1≥
(19
−
28t) 26
.
Because
t<1
, it follows that
(10 −15t) 14 <(19 −28t) 26 <1−(4 3)t
. Consequently, Union 2’s
reaction function is as follows: For
w1
∈
(0, (10
−
15t) 14))
, first-order conditions of Equation 14 lead to
RF
2(w1)
=
w2
=
(2
+
2w2
+
t) 8
; for wages in the range
w1
∈
((10
−
15t) 14), (19
−
28t) 26))
, the best-reply
function is
RF
2(w1)=w2=(2w1−1+2t)
; for
w1∈((19−28t) 26),1 −(4 3)t)
, first-order conditions of
Equation 15 lead to
RF
2(w1)
=
w2
=
(5
+
2w1) 14
.
b.3) For
t
>
3 8
,
1
−
(4 3)t
<
1 2
, the intercept of the line representing the boundary between Regions II
and IV in Figure 2, given by
w1
=
(1 2)(1
+
w2)
, is greater than the value of
w1
representing the upper vertex
of Region I in Figure 2. Nevertheless, the shape of the reaction function is as in part b.2, adding only that for
w1
∈
(1
−
(4 3)t,1 2)
the reaction function is still
RF
2(w1)
=
w2
=
(5
+
2w1) 14
. Indeed, it can be verified
that this is Union 2’s best-reply for
w1
∈
(1
−
(4 3)t,16 27)
, where the latter value represents the value of the
point of intersection of the segment of Union 2’s reaction function in Region II with the boundary between
Regions II and IV.
c.0) The two unions’ reaction functions in Region I,
RF
1
(w
2
)=w
1
=(1+w
2
−t) 4
for Union 1 and
RF
2
(w
1
)=w
2
=(2 +2w
1
+t) 8
for Union 2, intersect at:
w
1,A sy =
1 3
−
(7 30)t
,
w
2, A sy =
1 3
+
(1 15)t
.
These values represent the Bertrand-Nash wages in equilibrium allowing both firms to undertake inter-
national business in the asymmetric regime where Firm 2 invests. Intra-industry international activities are
supported as pure strategy equilibrium until trade costs are such that the level of
w
2 representing the switch-
ing wage for Union 1, is satisfied concurrently with the Bertrand-Nash equilibrium wage for Union 2, that is:
1 3
+
(1 15)t
≥
(2
+
2 )t
−
1
.
It follows that the critical value of
t
above which intra-industry international activities in asymmetric
regimes are not supported as pure strategy equilibrium is equal to:
t
≤
20 (29
+
15 2 )
≈
0.398
.
Positions on Regulations Affecting Auditing and
Nonauditing Activities
Rosario López Gavira, José Ángel Pérez López, and José Enrique Romero García
University of Seville, Seville, Spain
Abstract
The change in regulations that occurred in Spain in the domain of auditing has led to the analysis of regulations
according to the positions adopted by different groups involved in the auditing market. The purpose of this
study was to investigate the positions taken by professionals involved in this sector regarding those aspects
of the law that regulate the provision of services other than the auditing of annual accounts, with a view to
obtaining relevant conclusions for the regulation of the auditing activity. Findings show the existence of three
professional subgroups according to the level of global prohibition of the incompatibilities analyzed and the
level of importance assigned to the prohibitions in two important groups of prohibitions. The difference
between these professional groups is analyzed in terms of their level of prohibition in comparison with the
law. Other results show the most important variables for measuring a firm’s degree of independence.
Keywords: Auditing, independence, incompatibilities, regulation, non-auditing services
JEL Classification codes: M420, K220
http://dx.doi.org/10.7835/jcc-berj-2014-0096
The auditing services market today includes, in addition to the traditional auditing of accounts, a rela-
tively wide range of services depending on the prohibitions and exceptions established by the regulations of
the country in which auditing firms operate. This reality of the auditing market stands in sharp contrast to
the position adopted by some researchers on the strictest prohibition of services carried out by these firms
(Abbott, Parker, Peters, & Raghunandan, 2003; Ashbaugh, Lafond, & Mayhew, 2003; Bartlett, 1993; Basioudis,
Papakonstantinou, & Geiger, 2008; Davis & Hollie, 2008; Duh, Lee, & Hua, 2009; Felix, Gramling, &
Maletta, 2005; Gonzalo, 1995; Lowe & Pany, 1995; Pany & Reckers, 1988; Sharma, 2001). Regulators have
shown themselves aware of the controversial effects on auditing firms’ independence of those firms’ offering
both nonauditing services (NAS) and auditing services. Hence, there have been many legislative efforts and
actions to solve this conflict worldwide, such as the Sarbanes-Oxley Act in the United States, the reform of
the Auditing Law in Spain, and others.
The Spanish context provides an opportunity to study the effects of the change in rules produced by reforms
to Auditing Law in the years 2002 and 2010, one of the consequences of which is the increasing number of
incompatibilities with regards to the professional activities carried out by firms. Different investigations of this
issue have produced conf licting results. Some st udies have shown that these activities can harm the independence
of the auditors (Basioudis et al., 2008; Davis & Hollie, 2008; Duh et al., 2009; Frankel, Johnson, & Nelson,
2002; Ye, Carson, & Simnett., 2011), while others show opposite results (Antle, Gordon, Narayanamoorthy, &
Zhou, 2004; Ashbaugh et al., 2003; Chung & Kallapur, 2003; Monterrey & Sánchez, 2007).
Journal of
CENTRUM
Cathedra
JCC
JCC: The Business and Economics Research Journal ● Volume 7, Issue 1, 2014 ● 75- 9 0
76 JCC: The Business and Economics Research Journal
The purpose of this study was to analyze the provision of NAS through an empirical investigation of the
positions of academics and auditors on the legal aspects that regulate the execution of auditing services. In
particular, the purpose was to determine the degree of agreement or disagreement with the current legislation
and to provide relevant conclusions that could be of interest for future reforms to auditing regulations. The
research method used was a questionnaire sent to the professionals enrolled in the Registry of Spanish Auditors
(REA) and academics enrolled in the Spanish Accounting Professors’ Association (ASEPUC).
The study f irst reveals the existence of subgroups with similar perceptions about regulating incompatibility
regarding the level of importance they assign to such incompatibilities. These groups should be taken into
account in future regulations of auditing activities. Findings indicate that those incompatibilities should be
controlled by more regulations. The evidence is in accordance with other movements on an international level
toward stricter incompatibilities with auditing activities, such as the Sarbanes-Oxley Act in the United States.
In the second section, the review of the literature provides an analysis of the most important consequences
of auditors’ offering multiple services and the modifications made to auditing law which affect the joint provi-
sion of auditing and other additional services. The third section of the paper describes the methodology and
the research design used in the study. The next section describes the main results obtained from the empiri-
cal investigation. The final section outlines the conclusions obtained from the study and its most important
implications for future research and limitations.
Background of the Research
According to Beattie and Fearnley (2004), one of the main concerns that have emerged following a number
of financial scandals occurring in the last decade of the 20th century and the beginning of the 21st century is
related to the execution of multiple and varied services by auditors. Fees charged for these services grew even
faster than those charged for auditing services. All of this led to the general belief that the execution of other
services could cause these professionals to compromise their independence.
Two main concerns arose. On the one hand, auditors tend to avoid disagreements with the management
of companies in order to maintain the abundant income derived from the provision of services not related to
auditing (Ashbaugh et al., 2003; Basioudis et al., 2008; Nice & Trompeter, 2004; Ruddock, Taylor, & Taylor,
2006; Van Der Plaats, 2000). On the other hand, the offering of a wide array of services could lead auditors to
identify too closely with the management of businesses, thus ultimately losing the neutrality needed for audit-
ing functions (Cahan, Emanuel, Hay, & Wong, 2008; Caplan & Kirschenheiter, 2000; Firth, 1997; Myring &
Bloom, 2003; Ruddock et al., 2006).
The supply of NAS has been the most debated topic of all the threats to independence identified in the
literature (Bartlett, 1993; Canning & Gwilliam, 1999; Callagan, Parcas, & Singhal, 2009; Habib & Islam,
2007). Many authors have argued that the provision of services is a practice that has negative consequences
on the functioning of the auditing market (Ashbaugh et al., 2003; Bloomfield & Shackman, 2008; Quick &
Warming-Rasmussen, 2009; Windmöller, 2000). The following negative consequences of this practice have
been identified:
•
It increases the economic dependence of the client (European Commission, 2000 a,b, 2003; International
Federation of Accountants - IFAC, 2001a; Khurana & Raman, 2006);
• It provokes a loss in auditing quality (Felix et al., 2005; Francis, 2006; Gonzalo, 1995);
• It increases familiarity and trust with the client (Chen, Elder, & Liu, 2005; European Commission,
2000 a,b, 2003; Gul, Jaggi, & Krishnan, 2007; IFAC, 2001a,b);
• It creates complicated situations for self-revision (IFAC, 2001a,b; Myring & Bloom, 2003);
• It harms the prestige of the auditing profession (Francis & Ke, 2006; Gonzalo, 1995; Law, 2008).
However, other authors have also pointed out a number of positive effects of the practice of joint service
provision and the execution of other types of work by auditors:
• It increases knowledge of the client (Asare, Cohen, & Trompeter, 2005; Beck & Wu, 2006; Gul et
al. 2007; Seunghan, 2006);
• It improves competition within the market of auditing firms (Ruiz, 2002; Wu, 2006);
• It benefits auditors’ independence (Arruñada, 1999; Lennox, 1999; Myungsoo, 2005);
77
Positions on Regulations Affecting Auditing and Nonauditing Activities
•
It improves the satisfaction of clients of auditing firms (García, Garrido, Vico, Moizer, & Humphrey,
1999; Lee, Mande, & Son, 2009; Malley, 2000);
•
It increases the chances of attracting and retaining personnel in auditing firms (Hillison & Kennelley,
1988).
On the whole, though both negative and positive consequences exist, expressions of alarm and concern
are more frequent than those of praise for the positive consequences.
As far as legislation on incompatibilities within auditing activities is concerned, a comparative study of
the statements and measures taken by different international agencies shows that the agency adopting the
strictest and most severe position on prohibitions is the Securities Exchange Commission (SEC) through the
Sarbanes-Oxley Act (2002). Greater consensus exists between the positions of the International Federation of
Accountants (IFAC) and the General Accounting Office (GAO). Lastly, the American Institute of Certified
Public Accountants (AICPA) is the least stringent agency in this respect (López, 2005).
The modification of the legislation on auditing in Spain was a long-awaited event desired by all the groups
involved, as many topics required revision and updating in the context of the new panorama affecting the audit-
ing services market. This situation was especially urgent with regards to the provision of nonauditing services
by auditors because, given the evolution of the auditing market, it was a topic needing specific modifications
and, above all, broader and more precise regulations. The previous rules established only a few sparse refer-
ences on the topic of confronting the issue of joint provision of auditing and other services.
Through the terms of Law 44/2002, the legislation on auditing was modified, with the aim of resolv-
ing existing conflicts and deficiencies. Specifically, Article 8.2 indicates the following:
It is established that the auditor does not possess sufficient independence in the exercise
of his functions in relation with a business or entity, when he or she provides the follow-
ing services or when a series of circumstances occur: the execution of services of design
and launching of f inancial information technology systems, evaluation services, services
of internal auditing, maintaining business relations, advocacy services, participation
in the hiring of executives or key personnel for the auditing client, and the provision by
the signing partner of services other than auditing to the audited entity, as well as the
payment of fees for providing auditing and non-auditing services to the same client, if
the latter constitute an unduly high percentage of the total annual income of the account
auditor in relation to the average of the last five years. (Law 44/2002)
Moreover, in the same article, the law also establishes that the calculation period for incompatibilities will
include the year in which the work was carried out as well as the third year previous to the tax year to which
the financial statements being audited refer.
Methodology and Research Design
To carry out this investigation, a system of email surveys was chosen in order to compile the opinions of
auditors and the academic community. This procedure was chosen because it is a straightforward research
method for collecting opinions, and it allows researchers to reach quickly a large number of elements of the
population under study. In addition, it provides many other advantages, such as the rapid reception of responses
from those being surveyed, the possibility of broadening the study’s geographical scope, and a considerable
reduction of research costs. Nonetheless, it also presents some disadvantages, such as difficulty in obtaining
certain email addresses, the fact that some people do not use email, and the loss of some responses because
the survey arrives along with a large number of spam messages.
With regards to the participating population, the choice of participants was based on the twin concepts of
knowledge and professional work. Thus, auditors chosen had a direct interest in the regulated matter, together
with a high knowledge of auditing and accounting. The selection of academics was based on the fact that it is
logical to think that they have a good knowledge of auditing and that the regulated activity could influence
their professional work as they must incorporate changes in rules into the classes that they teach; moreover,
their opinion on those changes must be considered free of partisan bias. These two groups were thus considered
to be an excellent proxy for those involved in auditing functions, as they initially present disparate positions,
and both groups’ opinions are supported by their knowledge of the regulated area.
78 JCC: The Business and Economics Research Journal
In this study, the usual steps were followed for this type of research: definition, design of the study, selec-
tion and definition of variables, design of the questionnaire, selection of the sample, validation and testing of
the questionnaire (Ruiz et al., 1998). Next, the process carried out is briefly summarized.
Denition and Purpose of the Study
The purpose of the study was to assess whether changes made in the auditing legislation are likely to
contribute to a reduction of the existing controversy surrounding the execution of various services by audi-
tors. If this is not the case, the study findings may serve to ease the conflict by proposing alternatives. Hence,
the potential effect of the changes was investigated via the opinions of two groups of users involved in and
committed to auditing activities. The target population was composed of auditors belonging to the Registry of
Spanish Auditors (REA) and academics belonging to the Spanish Accounting Professors Association (ASEPUC).
The objective of this investigation was to raise a debate on the modifications to auditing law with the
purpose of reaching a consensus on such questions. With regards to those parts of the law which have under-
gone change, the aim of this study was to find empirical evidence of the level of acceptance shown by the
individuals involved. With regards to those parts of the law which have not been modified, the aim of the
study was to provide additional evidence related to matters not changed or treated in the reform but which
individuals believe should have been taken into consideration.
Selection and Denition of the Variables
The next step was the selection and definition of different items of interest to gather relevant informa-
tion to meet the aims of the study. Starting with the key auditing and legal concepts, a set of variables was
constructed that would ultimately constitute the complete questionnaire. The variables analyzed correspond
to the different incompatibilities that are described in the extract from Article 8.2 of the Auditing Law quoted
above. Table 1 shows a list of these variables and the modalities taken into consideration.
Table 1
Variables and Modalities under Consideration
Nomenclature Variables analyzed Modalities considered
IncD Incompatibility related with “Design Services and Implementation
of Financial Information Technology Systems”.
NP = No prohibition.
-E = Less strict than Law.
IL = In accordance with Law.
+E = Stricter than Law.
RP = Radical Prohibition.
IncAS Incompatibility related with “Assessment Services”. The same.
IncIA Incompatibility related with “Internal Auditing Services”. The same.
IncRM Incompatibility related with “Maintenance of Managerial
Relationships”. The same.
IncLS Incompatibility related with “Legal Services”. The same.
IncTM Incompatibility related with “Top Manager or Key Personnel
Recruiting”. The same.
IncSP Incompatibility related with “Signatory Partner of auditing report
carrying out any type of nonauditing service”.
NP = No Prohibition.
IL = Prohibition only signatory partner =
In accordance with Law.
RP = Prohibition all members = Radical
Prohibition.
The level of prohibition equal to the law, for each starting variable, was established as 0.5, the minimum
value as 0, and the maximum as 1. Other values of each variable were rescaled according to those values.
79
Positions on Regulations Affecting Auditing and Nonauditing Activities
Design of the Questionnaire and Selection of the Sample
The questionnaire used was of a mixed, structured type, using both open and closed questions. A codification
phase facilitated the subsequent statistical treatment of data obtained through this survey. The representative
sample was composed of 1 610 members of REA who were sent a questionnaire by email. The rate of response
was around 12.3%. In the case of the academics, the sample was composed of 900 individuals belonging to
ASEPUC. The index of responses received was approximately 10.4%. In both cases, the number of responses
achieved was satisfactory in relation to the minimum standards established in the literature for similar stud-
ies (Assessing the representativeness of public opinion surveys, 2012). Once the data were purged, the final
participants were 80 academics and 186 auditors.
Validation and Test of the Questionnaire
For the validation and final test of the survey, a pretest was administered to a group of 15 academics in the
Department of Accounting and Financial Economics of the University of Seville. Additionally, a pilot survey
was carried out with the following groups: students in a Master Degree Program in Bank Management and
auditing professionals, two from large auditing firms and one from a medium-size local firm.
Statistical Methodology
First, a statistical analysis of the variables included was carried out in order to make sure there were no
anomalies in the data. Next, a principal components analysis (PCA) was executed to achieve the segmentation
of the individuals involved and determine the number and type of groups into which the individuals could be
subdivided. This process enabled us to find groups of variables in such a way that the behavior of individuals in
each professional segment was similar in variables of the same group and different in those of different groups.
Once the segments of professionals and groups of variables were determined, such segments of profes-
sionals were characterized depending on their behavior in the original variables. For this characterization,
several confidence intervals were carried out. These are summarized as follows:
• Determination of the confidence intervals for the average level of prohibition of each professional
segment in each principal component (PC). Hence, it was then possible to see whether statistically
significant relationships existed between the degree of prohibition for each of the groups of incom-
patibilities studied and for each professional segment.
• Determination of the confidence intervals for the average level of prohibition of each professional
segment for each of the incompatibilities studied.
In accordance with those confidence intervals, the objective of the study was to determine the position of
each professional segment for each of the incompatibilities under study, in terms of agreement or disagreement
regarding the level of prohibition established in the regulation.
Results
A multivariate study was carried out in order to determine unobserved relationships between the variables.
For this purpose, a PCA was applied in order to construct latent variables to explain the joint behavior of the
variables IncD, IncAS, IncIA, IncRM, IncLS, IncTM, and IncSP (see Table 1).
Next, the existing correlation between the different variables was verified in order to discover whether it
was of interest to conduct the analysis. Table 2 shows the correlation matrix that reveals that all correlations
are highly significant; hence, common factors must be causing these high correlations, and, thus, PCA could
be carried out. Retaining the first three principal components, it was possible to retain 74.62% of the informa-
tion provided by the original variables.
80 JCC: The Business and Economics Research Journal
Table 2
Correlation Matrix
IncD IncAS IncIA IncRM IncLS IncTM IncSP
IncD 1 0.663** 0.416** 0.357** 0.360** 0.355** 0.441**
IncAS 0.663** 1 0.438** 0.409** 0.368** 0.355** 0.546**
IncIA 0.416** 0.438** 1 0.509** 0.446** 0.477** 0.327**
IncRM 0.357** 0.409** 0.509** 1 0.562** 0.447** 0.274**
IncLS 0.360** 0.368** 0.446** 0.562** 1 0.501** 0.379**
IncTM 0.355** 0.355** 0.477** 0.447** 0.501** 1 0.341**
IncSP 0.441** 0.546** 0.327** 0.274** 0.379** 0.341** 1
Note. (**) Significant Value for p < 0.05.
Table 3 shows the component score coefficient matrix: For each professional, the score in each component
is obtained by multiplying the standardized variables values for the case by the weights or component’s score
coefficients.
Table 3
Component Score Coefcient Matrix
Component
PC1 PC2 PC3
IncD 0.202 -0.423 -0.448
IncAS 0.213 -0.446 -0.244
IncIA 0.203 0.219 -0.498
IncRM 0.200 0.393 -0.305
IncLS 0.203 0.353 0.416
IncTM 0.194 0.324 0.375
IncSP 0.183 -0.419 0.806
% of Variance Explained 50.982 65.492 74.622
The first component is interpreted as a joint level of prohibition of all the concepts analyzed; in other words,
a new variable was obtained explaining the level of global prohibition for each item of the set of variables
analyzed. The second component represents a contrast between the level of prohibition manifested in the
variables IncD, IncAS, and IncSP, on the one hand, and the variables IncIA, IncRM, IncLS, and IncTM, on the
other. Thus, the first group of variables could be considered to represent additional services directly related
to the financial information verified by the auditing activity (SDA) whereas the second group of variables
would indicate services indirectly related to the financial information verified by the auditing activity (SIA).
Thus, the second principal component could be interpreted as a contrast between the importance granted to
prohibitions on SDA and the relevance granted to prohibitions on SIA.
In the third component, the variable whose weight far exceeds that of the others is the one that measures
which members of an auditing team are incompatible with the realization of any other type of service provided
by the firm (IncSP). In other words, the third component represents the level of importance assigned to the
prohibitions, namely which members of the auditing team are incompatible with the realization of other
services provided by the firm.
Figure 1 shows where the original variables are represented in the space of the first two principal compo-
nents. There are two groups of variables: on the one hand, variables IncD, IncAS, and IncSP and, on the other,
variables IncIA, IncRM, IncLS, and IncTM.
81
Positions on Regulations Affecting Auditing and Nonauditing Activities
Figure 1. Representation of the initial variables in the space of rst and second principal components.
Next, in Figure 2, the original variables are represented in the space of the first and the third principal
component. The variable which is at a greatest distance from the others and from the origin of the coordinates
is IncSP, which indicates that it is the most significant variable in this PC3.
Figure 2. Representation of the initial variables in the space of the rst and third principal components.
Moreover, analysis showed that among auditors, two subgroups could be defined according to the function
of the type of auditor and his or her experience. On the one hand, 147 partners with considerable experience
(five years or more) and individual auditors with even more experience (10 years or more), who have been
called consolidated auditors (CA). On the other hand, 39 partners who do not have considerable experience
(less than five years) and individual auditors who do not have much experience (less than 10 years), who have
been called nonconsolidated auditors (NCA).
Figure 3 shows the centroids of the three groups obtained in the spaces of the first and second principal
component, and Figure 4 shows the same centroids in the space of the first and the third principal component.
82 JCC: The Business and Economics Research Journal
Figure 3. Centroids in the space of rst and second principal components.
In Figure 3, consolidated auditors are located in the second quadrant, nonconsolidated auditors in the third
quadrant, and academics in the fourth. These positions show the following evidence: academics have a greater
tendency to prohibit than auditors (CA and NCA), and CA tend to prohibit more in variables IncIA, IncRM,
IncLS, and IncTM than in variables IncD, IncAS, and IncSP. In the case of NCA and academics, the tendency
is in the opposite direction: they tend to prohibit more or equally in variables IncD, IncAS, and IncSP than
in variables IncIA, IncRM, IncLS, and IncTM. Later, we show that NCA and academics prohibit to the same
degree in the two groups of variables.
Figure 4. Centroids in the space of rst and third principal components.
83
Positions on Regulations Affecting Auditing and Nonauditing Activities
In Figure 4, academics are located in the first quadrant and NCA and CA in the third quadrant. Considering,
as indicated previously, that PC3 comes to represent the behavior of variable IncSP, these positions show the
following evidence: Academics tend to prohibit much in the variable IncSP, but in the case of NCA and CA,
the tendency is in the opposite direction: they tend to prohibit little in the variable IncSP. Later, we show that
NCA have an intermediate degree of prohibition in IncSP and that CA have a very low level of prohibition
in this variable.
The overlap in the scores of academics, nonconsolidated auditors and consolidated auditors is
shown graphically in three box/plot diagrams, namely Figures 5, 6, and 7.
Figure 5. Box/plot graph of the principal component 1.
Figure 5 shows that the scores for the first component are higher in academics than in auditors (CA and
NCA). In addition, academics’ scores are asymmetric on the left; in other words, there are academics who
clearly differ from the general behavior of the group in the sense that they assign a lower prohibition. With
regards to auditors, there is no significant asymmetry.
Figure 6. Box/plot graph of the principal component 2.
84 JCC: The Business and Economics Research Journal
Figure 6 shows that for the second component, the auditors’ score (CA and NCA) is higher than that of the
academics. The auditors’ scores are asymmetric on the right: There are auditors who have a preference for SIA
as opposed to SDA prohibitions, which is more pronounced than in the case of other auditors in their group.
Figure 7. Box/plot graph of the principal component 3.
Figure 7 shows that in the third component, the scores are fundamentally positive for academics and negative
for auditors (CA and NCA). In other words, academics grant more importance to IncSP prohibition, whereas
auditors do the opposite, giving little priority to IncSP prohibition. In addition, academics’ scores are clearly
asymmetric on the right and auditor’s scores (CA and NCA) are clearly asymmetric on the left, which shows
that there are people in the three groups who have much more extreme opinions than the rest of their group.
The next step was to corroborate through a confidence interval (CI) the graphical observations made
earlier on academics and auditors, CA and NCA, in the three principal components. An alternative procedure
would be hypothesis tests. In general, a confidence interval for the parameter θ with 100(1 - a)% confidence
level will be defined as any interval that contains all numbers θ0 for which the corresponding null hypothesis,
H0 : θ0 = θ0, is not rejected with a significance level of 100a%. Thus, a 95% confidence interval for the mean
equals a significance level of 5% for the corresponding hypothesis test H0 : µ0 = µ0 . Hence, Tables 4 to 6 show
the analysis of confidence intervals for PC1, PC2, and PC3 (namely for IncSP), respectively:
1. CI for Average level of prohibition in each PC by academics;
2. CI for Average level of prohibition in each PC by consolidated auditors;
3. CI for Average level of prohibition in each PC by consolidated auditors.
Table 4
Confidence Interval for the Mean in PC1
Average 95% condence interval for the mean
Lower limit Upper limit
PC1
Academics 0.4937 0.2807 0.7067
CA -0.2546 -0.4081 -0.1010
NCA -0.1615 -0.5024 0.1795
Law -0.2848
85
Positions on Regulations Affecting Auditing and Nonauditing Activities
Table 4 shows the following consequences regarding PC1:
• The level of prohibition in the legislation is less than the lower limit of CI of the academics in PC1,
such that academics prohibit globally more than the legislation.
• The CI of CA and NCA are totally overlapping in PC1, such that NCA and CA globally prohibit on
a similar level.
•
The CI in PC1 of academics is superior to the CI of NCA and CA, such that academics prohibit
globally more than auditors (CA and NCA).
• The level of prohibition in the legislation is contained in the CI of CA in PC1, such that CA have a
level of global prohibition similar to that of the legislation.
• The level of prohibition of the legislation is in the CI of NCA in PC1, such that NCA have a level of
global prohibition similar to that of the legislation.
Table 5
Condence Interval for the Mean in PC2
Average 95% condence interval for the mean
Lower limit Upper limit
PC2
Academics -0.4086 -0.6030 -0.2143
CA 0.3443 0.1848 0.5037
NCA -0.2820 -0.5663 0.0024
Table 5 shows the following consequences regarding PC2:
• The upper limit of CI of academics is less than zero in PC2, such that academics prohibit more in
SDA than in SIA.
•
0 is within the CI of NCA in PC2; thus, NCA can be considered to prohibit on a similar level in
SDA as in SIA.
•
The lower limit of CI of CA is greater than zero in PC2; thus, CA can be considered to prohibit
more in SIA than in SDA.
Table 6
Condence Interval for the Mean in IncSP
Average 95% condence interval for the mean
Lower limit Upper limit
IncSP
Academics 0.8000 0.7216 0.8784
CA 0.2449 0.1806 0.3091
NCA 0.4359 0.2774 0.5944
Table 6 shows the following consequences regarding IncSP:
• The mean of IncSP for CA is less than the lower limit of the CI for NCA, such that CA prohibit less
than NCA in IncSP.
• The mean of IncSP for NCA is less than the lower limit of the CI for academics; thus, NCA prohibit
less than academics in IncSP.
The summary of these conclusions in that in the variable IncSP, CA prohibit less than NCA, and the latter
less than academics.
Next, the behavior in the initial variables of the three groups of professionals was analyzed. Specifically,
the level of prohibition of each of these variables for each group was compared with level 0.5 of the legisla-
tion. In other words, the 95% confidence interval for the mean was carried out for each starting variable and
for each professional segment.
86 JCC: The Business and Economics Research Journal
Table 7
Confidence Interval for Mean Level of Prohibition in Original Variables
Average 95% condence interval for the mean
Lower limit Upper limit
IncD
Academics 0.6728 0.586 0.7596
CA 0.4184 0.3543 0.4824
NCA 0.5385 0.413 0.6639
IncAS
Academics 0.7037 0.6171 0.7903
CA 0.3537 0.2887 0.4188
NCA 0.4744 0.3432 0.6056
IncIA
Academics 0.6563 0.5579 0.7546
CA 0.5986 0.5301 0.6672
NCA 0.6282 0.4917 0.7647
IncRM
Academics 0.7938 0.7177 0.8698
CA 0.6888 0.6254 0.7522
NCA 0.5833 0.4541 0.7126
IncLS
Academics 0.7531 0.6685 0.8378
CA 0.5799 0.5107 0.6492
NCA 0.5897 0.4513 0.7282
IncTM
Academics 0.7563 0.6700 0.8425
CA 0.631 0.5621 0.6998
NCA 0.4872 0.3558 0.6186
IncSP
Academics 0.8000 0.7216 0.8784
CA 0.2449 0.1806 0.3091
NCA 0.4359 0.2774 0.5944
The following conclusions may be drawn from Table 7:
• In the case of the academics, the level of prohibition of the legislation, 0.5, is less than the lower
limit of the seven CI. For this reason, it can be assumed that this group prohibits more than the
legislation in all variables.
• In the case of CA, the level of prohibition of the legislation, 0.5, is less than the lower limit of CI
for the variables IncIA, IncRM, IncLS, and IncTM (SIA) and greater than the upper limit for the
variables IncD, IncAS, and IncSP (SDA). Thus, it can be assumed that AC prohibit more than the
legislation in SIA and, in contrast, prohibit less than the legislation in SDA.
• In the case of NCA, the level of prohibition of the legislation, 0.5, is within the seven CI, and, thus,
NCA appear to prohibit on the same level as the legislation in all variables.
These findings provide a clear characterization, using the original variables, of the three professional
segments found through the analysis of principal components.
Conclusions and Avenues for Future Research
The research study was based on the firm conviction that the auditing profession is necessary and useful to
the economy of any country, given that it can provide an important added value to the economic and financial
information provided by firms. In this context, an analysis of the quality of independence of auditors was
carried out. Specifically, the investigation focused on an issue that has generated controversy in the auditing
profession during recent years: the regulation of the joint offering of auditing and other multiple services.
The research study shows the positions maintained by both auditors and academics regarding the legisla-
tion governing this type of activities. Within the two targeted professions, three groups of individuals were
identified: academics, nonconsolidated auditors, and consolidated auditors. Findings show a considerable
87
Positions on Regulations Affecting Auditing and Nonauditing Activities
difference in criteria between the two professions. The joint analysis of the variable of experience and the type
of professor/auditor indicates that CA diverge significantly from academics. However, NCA have a clearly
intermediate opinion on prohibitions, between academics and AC. In addition, the starting variables were
subdivided into two groups: SDA and SIA.
The characterization of the three groups of participants, using the original variables representing different
prohibited services, provides the following evidence:
• Academics tend to prohibit more than the legislation on all variables.
• NCA prohibit on the same level as the law regarding both SDA and SIA.
• The level of global prohibition of CA appears similar to the law even though it is not. This group
opts to prohibit more than the law in SIA and less in SDA, and both situations cancel each other out,
such that the final result is deceptive.
More specifically, the results demonstrate that academics show a high level of prohibition in SDA when
compared to SIA, CA show a high level of prohibition in SIA when compared to SDA, and NCA prohibit on
the same level for both SDA and SIA. In addition, in the case of IncSP, the relevant variable that measures
the level of importance assigned to the prohibitions, namely which members of the auditing team are incom-
patible with the realization of other services of the firm, findings show that academics show a high level of
prohibition in IncSP, CA show a low level of prohibition in IncSP, and NCA have an intermediate level of
prohibition in this variable.
It thus appears that the most important variables for the independence of auditing work are those that have
been grouped under the heading of services directly related to auditing (SDA), and the least important are
grouped under the heading of services indirectly related to auditing (SIA). This finding is consistent with the
professional reality of each group. Academics are in an impartial position that allows them to see the need to
reinforce auditors’ independence through regulation, a fact that is reflected in their high level of prohibition,
actually more elevated than the current legislation. In contrast, CA, probably influenced by their line of work,
consider that a high level of prohibition in regulation is detrimental to their professional activity. NCA are
located in an intermediate position between the other two groups, demonstrating agreement with the level of
prohibition stipulated in the regulations. Given that they have not yet consolidated their position in the auditing
profession, they share certain features of neutrality with academics, and they do not yet show a pessimistic
view of the influence of the regulations on their professional activity.
Finally, the results obtained in this investigation, convergent in great measure with statements made at
international level regarding stricter incompatibilities with auditing activity (for example, the Sarbanes-Oxley
Act in the United States), provide important conclusions that could usefully be taken into account for future
legislation in auditing markets.
The study showed a number of limitations:
• The use of a questionnaire as a method for obtaining empirical evidence has inherent limitations.
Notable among these limitations are the participation of people who give random responses, problems
in interpretation, and difficulties in responding to questions related to specific topics.
•
The conclusions obtained have full validity in reference to the two groups providing the sample data.
Thus, the criteria of other groups such as firms, financial analysts, and so on cannot necessarily be
extrapolated. For this reason, future research could incorporate the opinion of these groups to give
the results more perspective.
• The opinions shown in the questionnaire could contain a considerable amount of subjectivity, espe-
cially in reference to one of the groups surveyed: auditing professionals.
Future lines of investigation could focus on the following avenues:
•
Analyzing whether the current legislation serves to encourage the independence of the auditing
profession or, on in contrast, is too permissive.
• Investigating whether earning excessively high payment in NAS can affect the independence of the
auditing profession.
• Analyzing in depth and in more detail the role played by the variable IncSP in the independence of
auditing firms.
88 JCC: The Business and Economics Research Journal
References
Abbott, L. J., Parker, S., Peters, G. F., & Raghunandan, K. (2003). An investigation of the impact of audit committee
characteristics on the relative magnitude of non-audit service purchases. Contemporary Accounting Research, 20(2),
215-234. dx.doi.org/10.1506/8YP9-P27G-5NW5-DJKK
Antle, R., Gordon, E. A., Narayanamoorthy, G. S., & Zhou, L. (2004). The joint determination of audit fees, non-audit
fees, and abnormal accruals. Review of Quantitative Finance and Accounting, 27(3), 235-266. dx.doi.org/10.2139/
ssrn.318943
Arruñada, B. (1999). The economics of audit quality: Private incentives and the regulation of audit and non-audit services.
Dordrecht, Germany: Kluwer.
Asare, S., Cohen, J., & Trompeter, G. (2005). The effect of non-audit services on client risk, acceptance and stafng deci-
sions. Journal of Accounting and Public Policy, 24(6), 489-520. dx.doi.org/10.1016/j.jaccpubpol.2005.10.003
Ashbaugh, H., Lafond, R., & Mayhew, B. W. (2003). Do non-audit services compromise auditor independence? Further
evidence. The Accounting Review, 78(3), 611-639. dx.doi.org/10.2308/accr.2003.78.3.611
Assessing the representativeness of public opinion surveys. (2012). Pew Research Center. Retrieved from http://www.
people-press.org/2012/05/15/assessing-the-representativeness-of-public-opinion-surveys/
Bartlett, R. W. (1993). A scale of perceived independence: New evidence on an old concept. Accounting, Auditing &
Accountability Journal, 6(2), 52-68. dx.doi.org/10.1108/09513579310036378
Basioudis, I. G., Papakonstantinou, E., & Geiger, M. A. (2008). Audit fees, non-audit fees and auditor going-concern
reporting decisions in the United Kingdom. Abacus, 44(3), 284-309. dx.doi.org/10.1111/j.1467-6281.2008.00263.x
Beattie, V., & Fearnley, S. (2004). Auditor independence and non-audit services: A Literature Review (Working Paper).
Institute of Chartered Accountants in England and Wales (ICAEW). Retrieved from http://www.icaew.co.uk
Beck, P. J., & Wu, M. G. H. (2006). Learning by doing and audit quality. Contemporary Accounting Research, 23(1), 1-30.
dx.doi.org/10.1506/AXU4-Q7Q9-3YAB-4QE0
Bloomeld, D., & Shackman, J. (2008). Non-audit services fees, auditor characteristics and earnings restatements.
Managerial Auditing Journal, 23(2), 125-141. dx.doi.org/10.1108/02686900810839839
Cahan, S., Emanuel, D., Hay, D., & Wong, N. (2008). Non-audit fees, long-term auditor–client relationships and earnings
management. Accounting and Finance, 48(2), 181-207. dx.doi.org/10.1111/j.1467-629X.2008.00251.x
Callagan, J., Parcas, M., & Singhal, R. (2009). Going–concern audit. Opinions and the provision of non audit services:
Implications for audit independence of bankrupt rms, Auditing, A Journal of Practice and Theory, 28(1), 153-169.
Canning, M., & Gwilliam, D. (1999). Non-audit services and auditor independence: Some evidence from Ireland. The
European Accounting Review, 8(3), 401-419. dx.doi.org/10.1080/096381899335853
Caplan, D. H., & Kirschenheiter, M. (2000). Outsourcing and audit risk for internal audit services. Contemporary
Accounting Research, 17(3), 387-428. dx.doi.org/10.1506/8CP5-XAYG-7U37-H7VR
Chen, K. Y., Elder, R. J., & Liu, J-L. (2005). Auditor independence, audit quality and auditor-client negotiation outcomes:
Some evidence from Taiwan. Journal of Contemporary Accounting and Economics, 1(2), 119-146. dx.doi.org/10.1016/
S1815-5669(10)70006-0
Chung, H., & Kallapur, S. (2003). Client importance, non-audit services, and abnormal accruals. The Accounting Review,
78(4), 931-955. dx.doi.org/10.2308/accr.2003.78.4.931
Davis, S. M., & Hollie, D. (2008). The impact of nonaudit service fee levels on investors’ perception of auditor indepen-
dence. Behavioral Research in Accounting, 20(1), 31-44. dx.doi.org/10.2308/bria.2008.20.1.31
Duh, R-R., Lee, W-C., & Hua, C-Y. (2009). Non-audit service and auditor independence: An examination of the Procomp
effect. Review of Quantitative Finance and Accounting, 32(1), 33-59. dx.doi.org/10.1007/s11156-007-0080-5
European Commission. (2000a). Consultative paper on statutory auditors’ independence in the EU: A set of fundamental
principles. Retrieved from http://europa.eu.int/comm/internal_market/en/company/audit/news
European Commission. (2000b). The Commission’s recommendation on quality control in legal auditing in the European
Union: Minimum requirements. Ofcial Journal of the European Union (2001/256/CE). Retrieved from http://europa.
eu.int/comm/internal_market/en/company/account/news/index.htm
European Commission. (2003). Reinforcing the statutory audit in the European Union. Retrieved from http://europa.eu.int/
comm/internal_market/en/company/audit
Felix, W. L., Gramling, A. A., & Maletta, M. J. (2005). The inuence of nonaudit service revenues and client pressure
on external auditors’ decisions to rely on internal audit. Contemporary Accounting Research, 22(1), 31-53. dx.doi.
org/10.1506/JN7X-B51L-V45W-4U7R
Firth, M. (1997). The provision of nonaudit services by accounting rms to their audit clients. Contemporary Accounting
Research, 14(2), 1-21. dx.doi.org/10.1111/j.1911-3846.1997.tb00524.x
Francis, J. R. (2006). Are auditors compromised by nonaudit services? Assesing the evidence. Contemporary Accounting
Research, 23(3), 747-760. dx.doi.org/10.1506/4VD9-AE3K-XV7L-XT07
89
Positions on Regulations Affecting Auditing and Nonauditing Activities
Francis, J. R., & Ke, B. (2006). Disclosure of fees paid to auditors and the market valuation of earnings surprises. Review
of Accounting Studies, 11(4), 495-523. dx.doi.org/10.1007/s11142-006-9014-z
Frankel, R. M., Johnson, M. E., & Nelson, K. K. (2002). The relation between auditors’ fees for non-audit services and
earnings management. The Accounting Review, Special Issue on Quality of Earnings, 77 (Supplement), 71-105.
García, M. A., Garrido, P., Vico, A., Moizer P., & Humphrey, C. (1999). La calidad del servicio de auditoría: Los auditores
vistos por sus clientes. Revista Española de Financiación y Contabilidad, 28(102), 1005-1041.
Gonzalo, J. A. (1995). La auditoría, una profesión en la encrucijada de los noventa. Revista Española de Financiación y
Contabilidad, 24(84), 595-629.
Gul, F. A., Jaggi, B. L., & Krishnan, G. V. (2007). Auditor independence: Evidence on the joint effects of auditor tenure
and nonaudit fess. Auditing: A Journal of Practice & Theory, 26(2), 117-142. dx.doi.org/10.2308/aud.2007.26.2.117
Habib, A., & Islam, A. (2007). Determinants and consequences of non-audit service fees. Managerial Auditing Journal,
22(5), 446-469. dx.doi.org/10.1108/02686900710750748
Hillison, W., & Kennelley, M. (1988). The economics of nonaudit services. Accounting Horizons, 2(3), 32-40.
International Federation of Accountants. (2001a). IFAC code of ethics for professional accountants. Retrieved from http://
www.ifac.org/Store/Details.tmpl?SID=9560085866929
International Federation of Accountants. (2001b). Outsourcing: Proposed international guideline on information technol-
ogy. Retrieved from http://www.ifac.org
Iyer, V. M., & Rama, D. V. (2004). Clients’ expectations on audit judgments: A note. Behavioral Research in Accounting,
16(1), 63-74. dx.doi.org/10.2308/bria.2004.16.1.63
Khurana, I. K., & Raman, K. K. (2006). Do investors care about the auditor’s economic dependence on the client?
Contemporary Accounting Research, 23(4), 977-1016. dx.doi.org/10.1506/D171-8534-4458-K037
Law 44/2002 (2002, November). Retrieved from http://ec.europa.eu/internal_market/nances/actionplan/transposition/
spain/f_d7_es_en.htm
Law, P. (2008). An empirical comparison of non-Big 4 and Big 4 auditors’perceptions of auditor independence. Managerial
Auditing Journal, 23(9), 917-934. dx.doi.org/10.1108/02686900810908454
Lee, H-Y., Mande, V., & Son, M. (2009). Do lengthy auditor tenure and the provision of non-audit servic-
es by the external auditor reduce audit report lags? International Journal of Auditing, 13(2), 87-104. dx.doi.
org/10.1111/j.1099-1123.2008.00406.x
Lennox, C. S. (1999). Non-audit fees, disclosure and audit quality. The European Accounting Review, 8(2), 239-252.
dx.doi.org/10.1080/096381899336014
López, R. (2005). La regulación como medio para la mejora de la calidad de la auditoría nanciera: El caso de la
prestación de servicios adicionales (Unpublished doctoral dissertation). Seville, Spain: University of Seville.
Lowe, D. J., & Pany, K. (1995). CPA performance of consulting engagements with audit clients: Effects on nancial state-
ment users’ perceptions and decisions. Auditing: A Journal of Practice & Theory, 14(2), 35-53.
Malley, M. (2000). The war of Independence. Accountancy, 126(1288),70-71.
Monterrey, J., & Sánchez, A. (2007). Rotación y dependencia económica de los auditores: Sus efectos sobre la calidad del
resultado en las compañías cotizadas españolas. Investigaciones Económicas, 31(1), 119-159.
Myring, M., & Bloom R. (2003). ISB’s conceptual framework for auditor independence. The CPA Journal, 73(1), 31-35.
Myungsoo, S. (2005). Do non-audit services inuence audit quality? (Unpublished doctoral dissertation). Lincoln,
Nebraska: University of Nebraska-Lincoln.
Nice, J. M., & Trompeter G. M. (2004). The demise of Arthur Andersen’s one-rm concept: A case study in corporate
governance. Business and Society Review, 109(2), 183-207. dx.doi.org/10.1111/j.0045-3609.2004.00191.x
Pany, K., & Reckers, P. (1988). Auditor performance of MAS: A study of its effects on decisions and perceptions.
Accounting Horizons, 2(2), 31-38.
Quick, R., & Warming-Rasmussen, B. (2009). Auditor independence and the provision of non-audit services: Perceptions
by German investors. International Journal of Auditing, 13(2), 141-162. dx.doi.org/10.1111/j.1099-1123.2009.00397.x
Ruddock, C., Taylor, S. J., & Taylor S. L. (2006). Non-audit services and earnings conservatism: Is auditor independence
impaired? Contemporary Accounting Research, 23(3), 701-746. dx.doi.org/10.1506/6AE8-75YW-8NVW-V8GK
Ruiz, G. (2002). La crisis internacional de la auditoría potencia su papel en España. Emprendedores, 56, 32-35.
Sarbanes-Oxley Act. (2002). Retrieved from http://www.sarbanes-oxley.com
Ruiz, J., Izquierdo, M., and Piñera, J. T. (1998). El cuestionario estructurado como herramienta para la investigación de las
instituciones documentales. VI Jornadas Españolas sobre Documentación (Fesabid 98). Retrieved from http://www.
orida-uni.es/fesabid98/comunicaciones/j-/ruiz/ruiz1.htm
Seunghan, N. (2006). The impact of non-audit services on capital markets. (Working Paper). New York, NY: Department
of Accounting, Taxation & Business Law, New York University.
90 JCC: The Business and Economics Research Journal
Sharma, D. (2001). The association between non-audit services and the propensity of going concern qualications:
Implications for audit independence. Asia-Pacic Journal of Accounting & Economics, 8(2), 143-155.
Van der Plaats, E. (2000). Regulating auditor independence. The European Accounting Review, 9(4), 625-638. dx.doi.
org/10.1080/09638180020024061
Windmöller, R. (2000). The auditor market and auditor independence. The European Accounting Review, 9(4), 639-642.
dx.doi.org/10.1080/09638180020024016
Wu, M. G. H. (2006). An economic analysis of audit and nonaudit services: The trade-off between competi-
tion crossovers and knowledge spillovers. Contemporary Accounting Research, 23(2), 527-554. dx.doi.
org/10.1506/4DA7-D9YU-P9HQ-PX82
Ye, P., Carson, E., & Simnett, R. (2011). Threats to auditor independence: The impact of relationship and economic bonds.
Auditing: A Journal of Practice & Theory, 30(1), 121-148. dx.doi.org/10.2308/aud.2011.30.1.121
Authors Note
Rosario López Gavira, Department of Accounting, University of Seville, Avda. Ramón y Cajal, 1 41.018, Seville,
Spain.
José Ángel Pérez López, Department of Accounting, University of Seville, Avda. Ramón y Cajal, 1 41.018, Seville,
Spain.
José Enrique Romero García, Department of Applied Economy, University of Seville, Avda. Ramón y Cajal, 1 41.018,
Seville, Spain.
Correspondence concerning this article should be addressed to Rosario López Gavira, Email: rlopezgavira@us.es
This research was supported partially by the Registry of Spanish Auditors (REA) and the Spanish Accounting Professors’
Association (ASEPUC). The authors are grateful to the University of Seville for funding and the referees and translators
for helpful comments and suggestions.
Competitiveness among Higher Education Institutions:
A Two-Stage Cobb-Douglas Model for Efficiency
Measurement of Schools of Business
Sonia Valeria Avilés-Sacoto
Instituto Tecnológico y de Estudios Superiores Monterrey, Monterrey, NL, Mexico
Wade D. Cook
Schulich School of Business, York University, Toronto, Canada
David Güemes-Castorena
Instituto Tecnológico y de Estudios Superiores Monterrey, Monterrey, NL, Mexico
Abstract
In this paper, we present a methodology for evaluating competing organizations in order to identify best
practices among those organizations. We focus attention specifically on competitiveness in the context of
a set of business schools for the purpose of identifying those that appear to be most efficient relative to
their peers. One of the most widely recognized efficiency measurement methodologies is data envelopment
analysis (DEA). DEA literature has witnessed the expansion of the original concept to encompass a wide
range of theoretical and applied research areas, with one such area being network DEA, with two-stage DEA
in particular. This latter concept and its extensions to multi-stage situations have been particularly influential
in such settings as supply chain management. In the conventional two-stage serial model, it is assumed that in
each stage efficiency will be defined by the standard ratio of weighted outputs to inputs or inputs to outputs.
This depends on whether an input or output orientation is chosen. In terms of the model used, we develop
a two-stage approach where at each stage we define efficiency in terms of a Cobb-Douglas function. This
serves two important purposes. First, the data in this particular setting appears in the form of percentages or
ratings. Therefore, a geometric mean which the Cobb-Douglas function is based on might be deemed as more
appropriate than the arithmetic mean concept at the center of the conventional model. Second, unlike some of
the previous models that define the aggregate efficiency of the process as the simple product of the scores for
the two stages, the Cobb-Douglas structure permits one to define aggregate efficiency as a weighted product
of those scores. This permits one to place greater emphasis on one stage versus the other. This allows for a
sensitivity analysis on the effect of the “stage weights” on the aggregate score and on the individual scores
that make up that aggregate.
Keywords: Data envelopment analysis, competitiveness, Cobb-Douglas, percentage data
JEL Classification code: C02
http://dx.doi.org/10.7835/jcc-berj-2014-0097
In this paper, we present a methodology for evaluating competing organizations and for identifying best
practices among those organizations. While competitiveness is in evidence virtually everywhere, nowhere
Journal of
CENTRUM
Cathedra
JCC
JCC: The Business and Economics Research Journal ● Volume 7, Issue 1, 2014 ● 91- 115
92 JCC: The Business and Economics Research Journal
is it more prevalent than among higher education institutions. Universities and departments within those
universities compete annually for the best and brightest students. They must continually re-examine their
business models to identify new trends and the ever-changing needs of the incoming student population.
Herein, we focus attention specifically on competiveness and what we consider to be the natural extension
of competitiveness, namely relative performance measurement and benchmarking. Specifically, we examine
these in the context of a set of business schools for the purpose of identifying those schools that appear to be
most efficient relative to their peers. At the same time, we identify shortfalls in performance of those institu-
tions that are not performing up to the standard set by the benchmarks. We view efficiency relative to a set
of input and output factors.
The terms competitiveness analysis and relative benchmarking are often confused. Competitiveness studies
often set out to examine how firms compare with their direct competitors along a set of prescribed dimensions.
This leads to ranking those firms’ positions relative to their competitors. Competiveness is a subject that is
often the focus of studies in business strategy. Organizations are interested in distinguishing themselves from
their competition by establishing a reputation as leaders on cost, quality, service, or other dimensions. Thus,
many studies of competitiveness look at specific factors or strategies adopted by organizations, with the aim
being to examine how or if performance improvement results from the application of such factors or strategies.
For example, Kingsley and Malecki (2004) examine the effect of networking among small manufacturers as
a policy innovation to promote competitiveness. Their study specifically focuses on networking within 50
small manufacturing firms.
A number of methodologies have been developed for assessing the “health” of an organization relative to
its competitors. At least two traditional methodologies, the European Foundation for Quality Management
(EFQM) and the Balanced Score Card (BSC) models, have been developed for determining whether an orga-
nization meets certain standards (Lamotte & Carter, 2000). Both of these methodologies involve creating a
set of standards or targets along various criteria. Once this is done, the firms in question are viewed from the
perspective of those standards and what has to be done to meet those standards.
Relative benchmarking and performance measurement, on the other hand, generally refer to identifying
those “competitors” that are achieving best practice. This would be done by comparing performance among
a specific set or peer group of organizations. Models that identify best practice comparators can provide an
understanding of the processes and skills that create superior performance (Wireman, 2004).
On the other hand, EFQM and BSC models aim more at absolute rather than relative standards. While
these tools are very useful in certain settings, it is important to understand that one must be able to establish
the levels of various strategic indicators or targets. The organization under study has to achieve these targets
in order for it to realize its long term vision (Podobnik & Dolinsek, 2008). While competitive analyses have
helped companies understand their respective market positions, relative benchmarking can then take over
where this opportunity for improvement reaches its limit. By observing the best practices within their relevant
peer group, relative benchmarking enables companies to move from a parity business position to a superiority
position (Wireman, 2004).
One of the most widely recognized relative efficiency measurement and benchmarking methodologies is
data envelopment analysis (DEA). Developed by Charnes, Cooper, and Rhodes (1978) (denoted as CCR), DEA
is a model for evaluating the relative efficiencies of a set of comparable and often competing decision-making
units (DMUs). In the sections to follow, we develop a modified version of the conventional DEA model. This
is then used to analyze a set of business schools.
In the sections to follow, we present a methodology for evaluating the performance of business schools.
First, we review the relevant literature on DEA, literature relating to the efficiency evaluation in education,
and some of the pertinent literature on competitiveness. Second, the section following this describes an appli-
cation of the modified DEA model, where the DMUs are business schools. Third, we develop a DEA-based
methodology based on the Cobb-Douglas function as presented by the units-invariant multiplicative model of
Charnes, Cooper, Seiford, and Stutz (1983). This structure readily lends itself to viewing performance from
the perspective of a multiplicative combination of outputs and inputs, rather than an additive one. The result-
ing DEA frontier becomes piecewise Cobb-Douglas. Fourth and finally, the developed model is applied to a
set of data on undergraduate business programs. This is followed by concluding remarks.
93
Competitiveness among Higher Education Institutions
Literature Review
In this paper, we focus on a study of competitiveness, benchmarking, and efficiency in higher education.
This is viewed from the perspective of a two-stage DEA model. We begin by giving a brief review of some
of the pertinent literature in these areas:
Competitiveness
Competitiveness and performance development have been subjects of study for decades. Competiveness
is a common concern for many countries, regions within countries, and companies and organizations world-
wide. In the book Competitive Advantage of Nations, Porter (1998) provided a new point of view based on a
competitive strategy framework. He advocates providing the right competitive environment to foster success
in performance. In contrast to this, other studies have promoted having adequate market development and
institutions. Some of this work provides examples of companies successfully competing in international
markets despite having been created in unfavorable environments.
Some of the competitiveness studies have looked at particular tactics such as coaching as a tool for
developing personnel in order to enhance the organization’s performance (Vidal-Salazar, Ferrón-Vílchez, &
Cordón-Pozo, 2012). Kingsley and Malecki (2004) examine networking as a tool for performance enhance-
ment. The articles by Lubitz and Wickramasinghe (2006), Moingeon and Edmondson (1996), Spender (1996),
and Wongrassamee, Gardiner, and Simmons (2003) provide further examples. As discussed above, a number
of tools have emerged to aid management in undertaking performance improvement. EFQM and BSC meth-
odologies are discussed in several publications including Elliott (1992), Kaplan and Norton (1996), Lascelles
and Peacock (1996), Podobnik and Dolinsek (2008).
Regarding competitiveness specifically in education, the literature focuses mainly on the role higher educa-
tion has played in promoting economic competitiveness. Sum and Jessop (2013) provide a good example of
this is their work. The subject of comparative competitiveness addressed in our current paper has not been
addressed previously in the same sense.
Data Envelopment Analysis
An important tool for identifying best practice in both competitive and noncompetitive settings is DEA.
Arguably, this tool might be seen as a natural extension or augmentation of the EFQM and BSC methodolo-
gies. The concept of efficiency is generally linked to the work of Farrell (1957). Twenty years after Farrell’s
seminal work and building on his ideas, Charnes et al. (1978) (CCR), responding to the need for satisfactory
procedures to assess the relative efficiencies of multi-input multi-output production units, introduced this
powerful methodology. The original idea behind DEA was to provide a methodology whereby, within a set
of comparable decision-making units (DMUs), those DMUs exhibiting best practice could be identified and
would form an efficient frontier. Furthermore, the methodology enables one to measure the level of ineffi-
ciency of nonfrontier units and to identify benchmarks against which such inefficient units can be compared.
Given the application presented herein, it is important to note that the very first application of DEA was in
the education sector (Charnes, Cooper, & Rhodes, 1981).
It is relevant to point out here that most applications of DEA are in what could be regarded as noncompeti-
tive settings. For example, the branches of a given bank are not in direct competition with each other, strictly
speaking. However, efficiency analyses in the branches yield important insights about which of the branches
are the benchmarks against which other inefficient branches are compared. The same would be true for a set
of comparable hospitals. In the example of business schools discussed below, we argue that the peer units are
in direct competition.
The CCR model portrays the efficiency of a DMU as a ratio of weighted outputs to weighted inputs, a
benefit/cost ratio. This strict segregation of factors or indicators into these two distinct categories is perhaps
one of the characteristics that distinguish DEA from tools like the BSC methodology. Unlike BSC, DEA
attempts to explain outcomes, outputs, not in absolute terms but rather relative to the resources or circum-
stances, inputs, which the DMU has at its disposal. The original DEA model, based on a constant returns to
scale (CRS) situation, was later extended to allow for variable returns to scale (VRS) (Banker, Charnes, &
Cooper, 1984). Both of these radial models see efficiency from the perspective of projection to the frontier
94 JCC: The Business and Economics Research Journal
by way of either input reduction, an input-oriented model, or output expansion, an output-oriented model. In
Charnes, Cooper, Golany, Seiford, & Stutz (1985) these ideas were extended to allow for projections in terms
of both input reduction and output expansion, simultaneously. This “additive” model was later adapted to
provide for an efficiency score along the lines of the earlier radial measures (Tone, 2001).
Many extensions of these models have been advanced over the years since the appearance of Charnes et
al.’s (1978) article, including the Cobb-Douglas model of Charnes et al. (1983). The methodology presented
in the current paper is a variant of this work. One of the major directions taken in extending the early work of
Charnes et al. (1978) involves settings with multi-stage structures. Färe, Grosskopf, Norris, and Zhang (1994)
and Färe and Grosskopf (2000) investigated such structures under the heading “network DEA.” This work is
very relevant to the application herein which views the education process as consisting of two stages. Much
research has been carried out in the interim (Chen, Cook, Kao, & Zhu, 2013; Chen, Cook, Li, & Zhu, 2009;
Chen, Cook, & Zhu, 2010; Cook, Liang, & Zhu, 2010; Cook, Zhu, Bi, & Yang, 2010; Cook, Zhu, & Liang, 2011;
Kao & Hung, 2008; Liang, Chen, Cook, Du, & Zhu, 2011; Liang, Cook, & Zhu, 2008; Liang, Yang, Cook, &
Zhu, 2006). Cook and Liang et al. (2010) provide a comprehensive survey of network DEA.
A number of books and edited journal volumes have been published on DEA theory and its applications.
Authors like Cook and Zhu (2005); Cook, Green, and Zhu (2006); Cooper, Seiford, and Tone (2006) have all
published texts on DEA theory.
Efciency Measurement in Education
Within the broad scope of education, frontier efficiency measurement techniques have been applied to
many different types of institutions. The articles by Bessent, Bessent, Kennington, and Reagan (1982); Chalos
and Cherian (1995); Deller and Rudnicki (1993) apply these measurement techniques to include primary and
secondary schools. The article by Athanassopoulos and Shale (1997) applies these techniques to a university.
Finally, articles by Beasley (1995), Beasley (1990), Chang, Chung, and Hsu (2012), G. Johnes (1988), G. Johnes
(1990), G. Johnes and Johnes (1993), J. Johnes and Johnes (1995), J. Johnes and Yu (2008), Kao and Hung
(2008), Madden, Savage, and Kemp (1997), Sinuany-Stern, Mehrez, and Barboy (1994), and Tomkins and
Green (1988) apply the measurement techniques to university departments. As is clear from the literature, the
DEA approach is the primary frontier technique employed in evaluating the efficiency of education programs
(Chalos, 1997; Chang et al., 2012; Charnes et al., 1981; Diamond & Medewitz, 1990; McCarty & Yaisawarng,
1993; Ray, 1991).
Bessent et al. (1982) conducted what is perhaps the best-known and earliest work in the area of education.
Employing the well-known Charnes et al. (1978) constant returns-to-scale DEA model, Bessent et al. (1982)
examined the productive efficiency of Houston’s 241 school districts. The study by Bessent et al. (1982) was
one of the first to point out some of the advantages of DEA over techniques used before. These advantages
include the incorporation of multiple outputs, the fact that a parametric functional form does not have to be
specified for the production function, and the ability to identify reasons why individual schools are inefficient.
In addition, Bessent et al. (1982) were the first to use standardized test scores as the measure of educational
success; to incorporate issues relating to local, state, and federal funding; and to indicate the quality of teach-
ing inputs with teaching experience, training, and qualifications.
As university administrators seek to improve resource usage, efficiency analysis has become an important
concern in managing performance (Avkiran, 1999; Caballero, Galache, Gomez, Molina, & Torrico, 2004;
Caroline, Castano, & Cabanda, 2007; Chalos, 1997; Cohn, Rhine, & Santos, 1989; Fandel, 2007; Glass,
McKillop, & Hyndman, 1995). Most studies focus on how to allocate educational resource inputs more effi-
ciently to improve output performance. The input indicators are generally units of measurement that repre-
sent the factors employed in service delivery. Generally, these inputs include human, financial, and material
resources (Martin, 2006). Previous studies often evaluated the performance for the departments of educational
institutions using several inputs that include: number of teachers, operating expenses, equipment, and usable
floor space. Output indicators measure the yield or activity level of programs and services (Martin, 2006).
Based on previous studies, the output indicators of educational departments generally contain research and
development (R&D) outputs and teaching outputs (Kao & Hung, 2008; Tyagi, Yadav, & Singh, 2009). R&D
outputs include project income as well as the numbers of publications, projects, patents, and awards given to
teachers. Teaching outputs used in previous studies have included the numbers of graduates, teaching evalu-
ations, graduate salaries, and employer satisfaction.
95
Competitiveness among Higher Education Institutions
In most studies, inputs usually include the number of teaching staff and are sometimes also accompanied by
the number of support and administrative staff. In the context of university education, G. Johnes and Johnes,
(1993) and J. Johnes and Johnes, (1995) also made a distinction between different types of labor. They also
split total the staff into teaching/research staff and research-only staff. These are obvious attempts to capture
the differing functions of labor in the educational process. Even within categories of labor, most empirical
studies have attempted to incorporate differences in the quality of inputs that may occur across the sample,
omission of which would result in misspecification.
Another set of frontier efficiency measurement studies that deserves particular attention is the instances
where educational outputs are jointly produced with strictly noneducational outcomes. This is the case with
the small number of studies concerned with either universities or academic departments within universi-
ties. G. Johnes and Johnes’ (1993) and J. Johnes and Johnes’ (1995) studies of UK university departments of
economics and J. Beasley’s (1995) study of UK departments of physics and chemistry are good examples of
this line of inquiry. In all three cases, research grants awarded to teaching/research and research-only staff
members are the inputs; published works and refereed journal articles are the measured outputs. The G. Johnes
and Johnes’ (1993) and J. Johnes and Johnes’ (1995) approach does differ somewhat in that no allowance is
given for actual teaching outputs. Beasley’s (1995) study incorporates the number of undergraduates and
postgraduates. While the DEA approach used in these studies places no particular weighting on outputs, like
the “managerial” choice between teaching and/or research performance, the general finding of these studies
is that university departments with higher teaching loads have lower research outcomes. A study by Madden
et al. (1997) also examined the efficiency of university departments of economics but no attempt was made
to distinguish between teaching/research and research-only staff.
These studies vary enormously in their chosen contexts and overall results. On the other hand, there is
broad agreement regarding their conceptualization of the educational process itself. There is also a similarity
in the process by which the education process transforms selected inputs into desired outputs.
In a number of respects, the problem setting discussed herein differs from those examined in the literature.
First and foremost, this is not a study that focuses strictly on resource usage, like on how all the staff members
are utilized, as is the case with the vast majority of previous enquiries. Rather, this is a study focused on
students and on the institution’s ability to attract and serve students in terms of helping them meet their ultimate
goals. The second difference is in the analysis model. In nearly all previous studies, the approach involved
conventional applications of the standard CCR model of DEA (Charnes et al., 1978). Here, we instead argue that
the process of attracting students, meeting their needs while enrolled, and then viewing the results following
graduation is best modeled as a two-stage process. We do point out that one other study done by Chang et al.
(2012) viewed efficiency evaluation in a similar way, albeit from a perspective focused on resource-allocation.
Undergraduate Business Programs Viewed as Two-Stage Serial Processes
Each year, business schools around the world compete to attract the best and brightest students. The rapid
growth in what universities and colleges can offer students creates tremendous competition between those
institutions. Therefore, institutions within the higher education market need to adopt or increase their interests
in those features, characteristics, and competitive differentiators used to attract the best applicants.
In a recent study, a set of 41 undergraduate business programs was surveyed to gain an understanding of
best practices in terms of admissions criteria and various outcome factors (Avilés-Sacoto, 2012). This study
was the backdrop for an MSc thesis at Instituto Tecnológico y de Estudios Superiores Monterrey, Mexico
(ITESM). Forty of the business schools are in the USA, while the 41st is the business school Escuela de
Negocios, Ciencias Sociales y Humanidades (ENCSH) at ITESM. In the current study, we reduced the scope
of the study to 37 schools because the data is incomplete on the remaining four.
Various researchers undertook many previous studies in this area, and there is much debate about what
factors are most relevant for the purpose of evaluating efficiency. We have attempted to choose factors that
define the quality of the students, which influences the quality of the institution as well. We also focused on
factors that reflect the achievements of the students as a result of being at that institution. With this in mind,
the following discussion is about the evaluation factors used. We point out that the data for the US schools was
taken from the 2011 edition of Best 373 Colleges by Franek et al. (2010). The only other institution included
in the sample is ENCSH in Mexico, for which the data comes from an internal survey.
96 JCC: The Business and Economics Research Journal
Top 25%: This factor is defined as the percentage of students applying to a school who were in the top
quartile of their high school classes. Franek et al. (2010), in the Best 373 Colleges, points to this factor as
an important indicator of how much universities focus their recruitment efforts on academically orientated
students. Rinn & Plucker (2004) stated:
Conventional wisdom appears to be that, although the intellectual progress of all college
students is important, the attitudes and accomplishments of the most talented students
help to improve an institution’s academic atmosphere and differentiate a university from
its peer institutions. (p. 26)
Reject: Percentage of applicants rejected and are not sent offers of admission. Franek et al. (2010) explain
that this rating or its complement, the Admissions rating, is a reflection of a number of factors including test
scores and class rank of entering freshmen. As Franek et al. (2010) put it, evidence shows that, “…one of the
leading determinants of a good university is the quality of its incoming students” (p. 54). Gansemer-Topf and
Schuh (2006) point out that selectivity scores provide information on the general academic qualities needed
for admittance. Evidence shows that schools that are more selective through rejecting more applicants may
have higher retention and graduation rates.
AcRat: Academic rating of the school. This rating is a measure of how hard students work at the school
and how much they get in return for their efforts. Franek et al. (2010) include this rating for putting institu-
tions into the Best Colleges ranking. Academic rating or reputation has been labelled variously as prestige or
quality. This value is derived from interconnected factors that lead to competitive advantage and, ultimately,
performance superiority (Boyd, Bergh, & Ketchen, 2010).
Enroll: Percentage of invited students who enrolled after being sent acceptance letters. This rating reflects
of the quality of the institution because students want to be there. It also reflects well on the capability of the
admissions office to close deals.
Scholar: Percentage of students who have earned or hold scholarships. On the one hand, Franek et al.
(2010) sees this as being on par with factors that put institutions into the Top 25%. This is because it is a clear
reflection of the quality of incoming students.
Intern: Percentage of enrolled students who obtained overseas internships while in the program. Higher
education institutions have implemented various initiatives to promote the skills, knowledge, and intercul-
tural understanding for creating international agreements. Doyle et al. (2010) see international internships as
making students more competitive, giving them the opportunity to develop a wider perspective on the world,
and often allowing them to develop foreign language skills. In addition, internships give students a wider set
of skills to allow them to compete in the job market. In this sense, the employment prospects for students can
only be enhanced by internships.
Jobs: Percentage of students who obtained employment in their chosen fields within six months of
graduation. Boyd et al. (2010) argue that prospective students will seek to be admitted to the most prominent
programs in order to maximize future income. In short, Alumni are satisfied with their educations when they
are prepared for employment and when the costs of their educations give them greater returns on their invest-
ments (Delaney, 2008; Lauer, 2002).
The relationships among these variables are complex, and it is not at all obvious what an appropriate
production function might look like in this context. Viewing this setting as one to which a relative efficiency
model could be applied, clearly a conventional DEA model might be appropriate. This can be done if one
accepts the argument that university and student quality measures like Top 25%, Reject, and Acrat impact
outcomes such as Enrol, Scholar, Intern, and Jobs. In other words, one can look at this from the perspective
of cause and effect. However, one can clearly make the case as well, as backed up by Franek et al. (2010) and
others, that the final goal of most graduates is Jobs. Thus, performance in the program such as being enrolled
in the program, scholarships earned, and internships served have an important influence on success in the job
market. This would seem to speak to the argument that a more appropriate model structure is one that views
the student experience as consisting of two stages: namely accomplishments while in the program such as
getting enrolled, receiving internships, and earning scholarships, and the earned accomplishment at the end
of the program which is success in the job market.
Given this argument, we therefore take the position that efficiency can be modeled as a two stage process,
as displayed schematically in Figure 1:
97
Competitiveness among Higher Education Institutions
INPUTS (Stage 1)
Academic Rating
% of applicants rejected
OUTPUTS (Stage 2)
% of students who get
jobs
OUTPUTS (Stage 1)
INPUTS (Stage 2)
Stage 2
% receiving internships
% receiving institutional
scholarships
% students in top 25% of
their classes
% of accepted applicants
enrolled
Stage 1
S
f
%
%
%
s
S
%
Fig ure 1. Two-stage serial process.
Stage 1 can be looked upon as the process of attracting applicants to programs and realizing the resulting
outcomes once applicants are enrolled. The inputs to this stage are intended to reflect the combined quality
of the institution and the students who apply there. The factors Reject and AcRat are measures of institutional
reputation and quality. Top 25% is an indication of the quality of students applying. Outputs here are taken as
those factors that reflect the combined accomplishments of the institution and the enrolled students. Specifically,
Enroll measures the success of the institution in converting applicants to enrolled students. Scholar and Intern
are important indicators of both institutional and student achievements while the student is in the program.
Stage 2 captures the accomplishments of students and institutions following graduation. The inputs to this
stage are the outputs from Stage 1. From the students’ perspective, Jobs are the major output from this stage.
In the following section, we investigate the modeling of efficiency for the two-stage processes as presented
in Figure 1. We first examine the conventional serial process where the efficiency measures for each stage
are represented by the Charnes et al. (1978) radial projection model. This utilizes the added combinations of
outputs and inputs. This is followed by a multiplicative methodology following the Cobb-Douglas method.
This method directly extends the conventional version in permitting the imposition of weights on the stage-
specific measures.
Modeling Efciency in Two-Stage Processes
The Conventional Two-Stage Serial Process
One area of DEA research is network DEA, particularly two-stage DEA. This latter research area, includ-
ing its extensions to multi-stage situations, has been particularly influential in such settings as supply chain
management (Liang et al., 2006). Cook and Liang et al. (2010) provide a survey of network models.
One of the common two-stage str uctures investigated in the literature on DEA is the serial process wherein
the outputs from the first stage become the inputs to the second stage. Some variants of this permit outputs
from Stage 1 to leave the system and inputs to the second stage to enter the system at that point; other two
stage systems are closed in that nothing enters or leaves the system between the stages. It is this latter closed
system where nothing enters or leaves that is analyzed in this study. The usual objective, if one regards the two
stage process as the DMU, is to view that process as one where inputs enter the DMU at Stage 1 and outputs
exit from Stage 2. Various methods have been suggested for evaluating the sub-efficiencies of each of the two
stages and for then combining these to get the overall DMU efficiency. The importance to the organization
of having a measure of overall efficiency, as well as sub-efficiencies, is to allow it to detect where it is and
is not competitive.
In the conventional two-stage serial model, it is assumed that in each stage efficiency will be defined by the
standard CCR ratio of weighted outputs to weighted inputs or weighted inputs to weighted outputs, depending
on whether an input or output orientation is chosen. In the current paper, the DEA efficiency measurement
relating to performance is applied to a set of 37 undergraduate business programs. In terms of the model
used, we develop a two-stage approach where at each stage we define efficiency in terms of a Cobb-Douglas
function. This serves two important purposes. First, the data in this particular setting appear in the form of
percentages or ratings. Therefore, a geometric mean, on which the Cobb-Douglas function is based, might be
deemed as more appropriate than the arithmetic mean concept at the center of the CCR model. Second, the
98 JCC: The Business and Economics Research Journal
model of Kao and Hwang (2008) defines the aggregate efficiency of the process as the simple product of the
scores for the two stages. In contrast, the Cobb-Douglas structure permits one to define aggregate efficiency
as a weighted product of those scores. This permits one to place greater or lesser emphasis on one stage versus
the other. This allows for a sensitivity analysis on the effect of the “stage weights” on the aggregate score and
on the individual scores that make up that aggregate.
As background for the development of our model for the process pictured in Figure 1, we briefly review
the methodology for one of the most common structures for two-stage serial processes. For simplicity of
presentation, ignoring the application-specific variables for the moment, we assume there are n(j = 1, …, n)
DMUs to be evaluated. For DMU j, let =
x{}
ij i
I
1 denote the inputs to Stage 1, =
y{}
rj r
R
1the outputs from Stage 2,
and =
z{}
dj d
D
1the intermediate variables that serve simultaneously as outputs from the first stage and inputs to
the second stage. Their corresponding multipliers are denoted by ni, ur, hd, respectively.
To evaluate the overall efficiency of business schools, we use an input-oriented model. Furthermore, we
decided to use the VRS model of Banker et al. (1984), (denoted as BCC) as opposed to the original Charnes
et al. (1978) CRS model. This is because the data appears in the form of ratios and percentages. It is important
to point out that in the case of an output-oriented rather than an input-oriented model, the VRS methodology
ensures that projections to the frontier do not exceed values of the variables experienced by the frontier units;
specifically, percentages and ratings will not exceed 100% under VRS. This is not guaranteed under CRS.
In this setting, one can represent the input-oriented VRS efficiency for Stage 1 as the solution to the radial
projection model:
ez
ux
ˆma
x/
oddo
d
iio
i
11
∑∑
hn=-
,
Subject to:
zu
xj/1
,
ddj
d
iij
i
1
∑∑
hn
-
≤∀
;
(1)
nj, hd ≥ 0, u1 unrestricted in sign.
The stage 2 model is given by:
∑∑
h
=-
euyu z
ˆma
x/
orro
r
ddo
d
22
,
Subject to:
∑∑
h-
≤∀
uy uz j
/1
,
rrj
r
ddj
d
2;
(2)
uj, hd ≥ 0, u2 unrestricted in sign.
We note that in the conventional CCR and BCC models, multiple inputs and multiple outputs are replaced
by a single virtual input and output, respectively. This is through using additive or arithmetic combinations
of the variables.
The literature suggests a number of approaches for deriving an overall score for the two stages combined
and from these deriving scores for the individual stages. Two of the methodologies are those based on game
theoretic principles, namely the cooperative and noncooperative game models (Cook & Liang et al., 2010).
The noncooperative or noncentralized approach views the two stages as players in a game and adopts a leader-
follower methodology. The cooperative game or centralized methodology, the approach adopted in this study,
derives the best aggregate efficiency score for the two stages combined. Then, one sets out to derive scores
for the two stages separately, which are such that when put together yield the overall score.
One of the original models for the two stage process was put forward by Kao and Hwang (2008) and is
a form of the cooperative or centralized approach. In their model, the overall efficiency is calculated using
=⋅eee
ooo
12
. Specifically, their model, designed for CRS settings only, is given by:
ezxx uy zu
yx
ˆmax/
//
oddo
d
iio
i
rro
r
ddo
d
rro
r
ii
o
i
∑∑ ∑∑
∑∑
hn
hn
=
=,
99
Competitiveness among Higher Education Institutions
Subject to:
zx j//1,
ddj
d
iij
i
∑∑
hn
≤∀
;
(3)
∑∑
h
≤∀uy zj//1,
rrj
r
ddj
d
;
nj, hd, ur ≥ 0.
It is noted that the two sets of constraints in Equation 3 are required to ensure that any multipliers chosen
are feasible for each stage of the process. Specifically, the two ratios for each DMU are restricted to be at or
below unity. We further note that normally one would impose the additional set of constraints:
∑∑
≤∀uy vx j/1,
rrj
r
iij
i
.
However, these are redundant in the presence of the other restrictions.
To accommodate VRS settings, as is needed in the application addressed herein, Chen et al. (2009) proposed
an additive approach for combining the two stages, as opposed to the multiplicative formulation suggested by
Kao and Hwang (2008). Under the Chen et al. (2009) approach, the objective function of Equation 3 for the
VRS setting, is replaced by:
ewzu xw uy
uz
ˆmax/ /
oddo
d
iio
i
rro
r
ddo
d
1
1
2
2
∑∑
∑∑
hn h=⋅ -
+⋅ -
. (4)
In this function, w1, w2 are weights the user can specify.
We now turn to what might be viewed as a more appropriate methodology than those above, in the case
that data appear as percentages.
A Cobb-Douglas Model for Two-Stage Processes
Suppose DMUo is under evaluation. Then, with inputs xij and outputs yrj , the units-invariant multiplicative
model of Charnes et al. (1983), for an input orientation, is given by:
µ
∏
∏
=
ξ
δ
e
ey
ex
max
o
ro
r
io
v
i
*
r
i,
Subject to:
µ
∏
∏
≤∀
ξ
δ
ey
ex j1,
ro
r
ij
v
i
r
i; (5)
δ, ξ ≥ 0, µr, vi ≥ 1.
In this case, multiple outputs and inputs are replaced by a single virtual output and virtual input by way of
a weighted geometric rather than arithmetic aggregation as in Equation 1. We point out that the notation e in
eξ, for example, is the natural number whose normal logarithm is unity. Now, taking the previous logarithms,
Equation 5 can be transformed to the linear form:
µ
∑∑
ξδ
=+ --
eyvxlogˆmaxˆˆ
orro
r
ii
o
i
*
,
Subject to:
µ
∑∑
ξδ+-
-≤
∀
yvxj
ˆˆ
0,
rrj
r
iij
i
; (6)
ξ, δ ≥ 0, vi, µr ≥ 1.
100 JCC: The Business and Economics Research Journal
We point out that in Equation 5 the term
∏
∏
ξµ
δn
ey
ex
rj
r
ij
k
r
i
could be replaced by
∏
∏
b
n
ey
x
rj
r
r
ij
k
i where bis unrestricted
in sign. Hence, in Equation 6 the expression ξ - δ can be negative or positive and can therefore be replaced
by b. For purposes of the development below, we retain the two nonnegative variables ξ and δ.
The notation
xy
ˆ,ˆ
in Equation 6 denotes the logarithm of the original data. It can be observed that in the
solution of this linear programming problem, either ξ, δ, or both will equal 0, meaning that the projected point
on the Cobb-Douglas frontier will experience increasing, decreasing, or constant turns to scale.
Using Equation 5 as a backdrop, we propose the following two-stage Equation 7 for describing overall
efficiency of the DMU.
Comparing this formulation to the Kao & Hwang (2008) model, we note that Equation 7 provides for a
more general structure, allowing for differential weights p1, p2 on the Stage 1 and Stage 2 efficiency ratios,
respectively. It is assumed that these satisfy the condition p1 + p2 = 1.
e
ez
exx
ez
ez
ma
x,
o
do
r
io
v
i
p
ro
r
do
i
p
*
agg
d
i
d
d
12
∏
∏
∏
∏
=
ξh
g
gµ
δh
Subject to:
ez
ex j1, ;
dj
r
ij
v
i
d
i
∏
∏
≤∀
ξh
g
(7)
ey
ez j1, ;
rj
r
dj
d
d
d
∏
∏
≤∀
gµ
ξh
ξ, g, δ ≥ 0, vi, hd, µr ≥ 1.
Applying the logarithmic transformation to Equation 7, which is analogous to that used to convert Equation
5 to the linear form in Equation 6, that Equation 7 is equivalent to the linear programming problem:
ep zvxp yzlogˆmaxˆˆ ˆˆ
,
oddo
d
iio
i
rr
od
do
dr
12
agg
∑∑ ∑∑
ξhgδµξh=+--
++ --
Subject to:
zvxj
ˆˆ
0, ;
ddj
d
iij
i
∑∑
ξhg+-
-≤
∀
(8)
zzj
ˆˆ
0, ;
rrj
r
ddj
d
∑∑
δµξh+-
-≤
∀
ξ, g, δ ≥ 0, vi, hd, µr ≥ 1.
As mentioned earlier, the notation
xzy
ˆ,ˆ,ˆ
denotes logarithms of the original data x, z, y.
Equation 8 yields a measure of log efficiency of the overall two-stage DMU. Letting
ξgδn
hµ
ˆ,ˆ,ˆ,ˆ,ˆ,ˆ
idr
denote the optimal solution in Equation 8, the simplified efficiency measure is then taken from Equation 7:
e
ez
exx
ey
ez
ˆ
.
o
do
d
io
v
i
p
ro
r
do
d
p
ˆˆ
ˆˆ
ˆˆ
ˆ
agg
d
i
r
d
12
∏
∏
∏
∏
=
ξh
g
δµ
ξh
.
(9)
Given an optimal log linear efficiency score arising out of Equation 8, which yields the Cobb-Douglas
score
e
ˆ
oagg from Equation 9, one needs to derive efficiency scores for the individual stages such that these
are consistent with this aggregate score. One approach is to adopt a leader-follower methodology similar to
that discussed in Cook and Liang et al. (2010) for linear representations of two-stage systems. As a specific
101
Competitiveness among Higher Education Institutions
example, if Stage 1 is chosen as being the highest priority stage or the leader and stage 2 is chosen as the
follower, Equation 10 below can be used to get an efficiency measure for that stage. In this model, we maxi-
mize the efficiency score of the first stage for DMUo subject to the usual requirement that the corresponding
first and second stage scores for every other DMU do not exceed unity constraints 10b and 10c. At the same
time, any multipliers used must be such that the aggregate score is not compromising 10d.
e
ez
ex
ˆ,
o
do
d
io
v
i
1
d
i
∏
∏
=
ξh
g
(10a)
Subject to:
e
ez
ex j
ˆ1, ;
o
do
d
ij
v
i
1
d
i
∏
∏
=
≤∀
ξh
g
(10b)
e
ey
ez j
ˆ1, ;
o
rj
r
dj
k
1
r
d
∏
∏
=
≤∀
δµ
gh
(10c)
ez
exx
ey
ez e
ˆ;
do
d
io
v
i
p
ro
r
do
d
p
oagg
µ
d
i
r
d
12
∏
∏
∏
∏
=
ξh
g
δ
ξh (10d)
ξ, g, δ ≥ 0, vi, hd, µr ≥ 1. (10 e)
In linear form, Equation 10 becomes:
∑∑
ξh g=+ --
ezvxlogˆmaxˆˆ
,
oddo
d
iio
i
1
Subject to:
zvxj
ˆˆ
0, ;
ddj
d
iij
i
∑∑
ξhg+-
-≤
∀ (11)
yzj
ˆˆ
0, ;
rrj
r
ddj
d
µ
∑∑
δξh+-
-≤
∀
∑∑∑∑
ξh gδµξh+--
+--
=
pz vx py ze
ˆˆ ˆˆ
logˆ;
ddoiio
id
rroddo
dr
oagg12
ξ, g, δ ≥ 0, vi, hd, µr ≥ 1.
Let
ξgδn
hµ,,,, ,
idr
denote the optimal solution to Equation 11. Then, the efficiency score for the Sage 1 portion
of DMUo is given by
∏
∏
=
ξh
gn
e
ez
ex
ˆ.
o
do
d
io
i
1
d
i It then follows that the Stage 2 score is given by =
ee e
ˆ{ˆ/[ˆ]}
ooagg o
pp211/
12
.
Similar derivations apply in the case that Stage 2 is chosen as the leader, with Stage 1 being the follower.
Restricted Projection to the Frontier
It is useful to look at the dual envelopment version of Equation 8, which is Equation 12. It is noted that in
this conventional additive model, the sum of slacks in the envelopment problem arises from the imposition
of lower bounds of unity, as opposed to zero, on the multipliers ni, hd, µr in the primal multiplier problem
in Equation 8. The result is a projection to the frontier in a direction that accounts for those dimensions for
which the lower bound of unity was imposed. Clearly, if a different lower bound was imposed, like e, the same
projection would result. Furthermore, if one relaxes these lower bound constraints and executes such nonzero
102 JCC: The Business and Economics Research Journal
lower bounds on only a subset of the primal multipliers, projection to the frontier will only be in those direc-
tions. This represents an important feature of the model as a tool for competitive evaluation of a set of DMU
in that any given DMU can decide which dimensions or factors are the ones on which it wishes to compete.
Users of the model can then evaluate where it stands on those factors.
sssmin ;
r
r
i
i
d
d
123
∑∑∑
-++
pp;
j
j
j
j
12
12
∑∑
λλ-≥-
zzsppz d
ˆˆ ˆ,;
jdj
j
jdjd
j
do
123
12
∑∑
λλ
()
--≥- ∀
p;
j
j
1
1
∑
λ
-≥
(12)
p;
j
j
2
2
∑
λ≥
xs px i
ˆˆ
,;
jiji
j
io
12
1
∑
λ--≥- ∀
ys py r
ˆˆ
,;
jrjr
j
ro
21
2
∑
λ-≥
-∀
sssjidr,,,, 0, ,, ,.
jjri d
12123
λλ ≥∀
In the section to follow, we apply this model to data pertaining to the performance of 37 business schools.
Efciency Analysis of Business Schools
Table 1 provides the list of universities included in our sample. This is followed by Table 2 which displays
the data on 37 business schools relative to the variables discussed in Section 2. Table 3 presents the same data
as in Table 2 but in natural logarithmic form.
The first analysis of efficiency is presented in Table 4. Here, Column 2 presents the log efficiency of each
of the business programs arrived at by solving the log-additive Equation 8. A reminder that the dual form of
this model takes the form of the envelopment model shown as Equation 12. Hence, the log efficiency score
can be viewed in terms of the sum of all input and output slacks along all dimensions. Column 3 presents the
same results but as the actual efficiencies, the antilog of the values in Column 2. It is noted that in this first
analysis we have set p1 = p2 = 0.50, giving equal importance to performance in each of the two stages. In
this analysis, two schools, #37 and #24, are DEA efficient, showing scores of 100% in Column 3. These are
the only DMUs on the Cobb-Douglas frontier. All other schools are inefficient.
To arrive at efficiency scores for the two stages in each DMU, one of the two stages was chosen as the
leader and the appropriate log efficiency model was solved. In the case of Stage 1 being the leader, this would
follow Equation 11. The resulting efficiency score in the case of choosing Stage 1 as the leader is then given
by
∏
∏
=
ξh
gn
e
ez
ex
ˆ.
o
do
d
io
i
1
d
iWith this rating in place, the corresponding Stage 2 follower efficiency score is then given
by =
ee e
ˆ{ˆ/[ˆ]}
ooagg o
pp211/
12
. Note that with Stage 2 as the leader, Equation 11 would be modified by replacing
the objective function with
∑∑
δµ
ξh
=+ --
eyzlogˆmaxˆˆ
orro
r
dd
o
d
2
. Given that there is a choice of leader,
two sets of analyses were now conducted; first setting Stage 1 as the leader and deriving the Stage 1 and
Stage 2 ratings and then repeating this with Stage 2 as the leader. In this particular application, the values for
the Stage 1 and Stage 2 scores turned out to be invariant to the choice of leader. This phenomenon may be
the result of a unique optimal solution arising from Equation 8 for any given DMU. Columns 4 and 5 show
the Stage 1 log efficiencies and actual efficiencies, respectively; Columns 6 and 7 display the corresponding
Stage 2 scores. It is noted that only two DMUs for #37 and #24 are efficient in both stages, meaning that in
the aggregate these two DMUs are efficient as well.
103
Competitiveness among Higher Education Institutions
To examine the sensitivity of efficiency scores of DMUs to the choice of weights p1, p2 applied to the two
stages, two additional analyses were carried out, namely using p1 = 0.25, p2 = 0.75. The results of these analy-
ses are shown in Table 5. The results of p1 = 0.10, p2 = 0.90, appears in Table 6. No particular trend emerges
from these three scenarios other than the observation that significantly different scores can arise, depending
on the choice of these weights. What this means is that in any given situation a clear understanding of the
importance to be attached to the two stages should be made before embarking on an analysis of efficiency.
From a competitive standpoint, perhaps the choice of weights p1, p2 is secondary to choosing which input
and output factors to focus attention on. Specifically, one might criticize the Cobb-Douglas methodology
for its characterization of efficiency in terms of all slacks (Equation12). Efficiency scores as represented in
Table 4, for example, presume that projections to the frontier along all dimensions are allowable. However,
projections that call for an increase in a factor such as the percentage of students obtaining jobs are definitely
relevant and of interest to the organization; a slack implying a reduction in the percentage of students in the
top quartile is not relevant. This feature of the additive model is clearly undesirable. As discussed above,
the structure has the advantage of permitting one to select those factors on which the organization wishes to
compete and then defines efficiency in terms of only those factors. With this concept in mind, two additional
analyses were carried out. Table 7 shows the results when efficiency is defined only in terms of Jobs. That is,
a lower bound of unity was applied only to the multiplier for Jobs, while all other lower bounds were set to
zero, meaning that the objective function in Function 12 would be restricted to -
smin []
1
1
. A second analysis
was then carried out focusing on both Jobs and Internships, specifically using a restricted objective function
-+
ssmin []
1
1
3
3
. Table 7 displays the efficiency scores in this case, while Table 8 shows the actual slacks for
both Jobs and Internships. Table 9 provides useful information to the school in terms of its competitive posi-
tion on these two key factors.
Conclusions
This paper provides a methodology for evaluating the relative performance of a set of competing entities,
with specific reference to a set of business schools. We view the DMU, the school, from the perspective of
its efficiency in attracting top students. Subsequently, the success of the school is measured in the number
of students earning scholarships, internships, and, later on, employment. We view the process as involving
two stages: the first stage relates to the admission process in attracting student applications; the second stage
describes the success of students in the job market following graduation, relative to such factors as their abili-
ties to acquire scholarships and internships while still in their program.
Rather than selecting the conventional DEA model as commonly applied in such studies, we proceeded
with a Cobb-Douglas methodology. This approach is somewhat more amenable to the percentage data involved.
Unlike earlier two-stage models expressed in multiplicative form, like in Kao and Hwang (2008) which attaches
equal weights to the two stages, our methodology permits variable, user-specified weights on the stage-scores.
Furthermore, one of the useful and appealing features of our approach is its ability to focus on user-specified
competitive drivers. We illustrate this with two analyses, one with Jobs only as the driver and another where
both Jobs and Internships are the focus.
In the usual multi-stage models, the principal focus is on the outputs from the final stage. The model
used herein, permits one to select those outputs from whichever stages they choose and consider projection
to the frontier specifically along those dimensions. This is rather like designating certain inputs or outputs as
discretionary versus nondiscretionary.
References
Athanassopoulos, A. D., & Shale, E. (1997). Assessing the comparative efciency of higher education institutions in the UK
by means of data envelopment analysis. Education Economics, 5(2), 117-134. dx.doi.org/10.1080/09645299700000011
Avilés-Sacoto, S. V. (2012). A Restricted Multiplier DEA Model for identifying best practices for attracting students into
a Mexican university Case Study: ENCSH at ITESM. Instituto Tecnológico y de Estudios Superiores de Monterrey,
Mexico, and Schulich School of Business, York University, Canada.
Avkiran, N. (1999). Investigating technical and scale efciencies of Australian universities through data envelopment
analysis. Socio-Economic Planning Sciences, 35(1), 57-80. dx.doi.org/10.1016/S0038-0121(00)00010-0
104 JCC: The Business and Economics Research Journal
Banker, R. D., Charnes, A., & Cooper, W. W. (1984). Some models for estimating technical and scale efciencies in data
envelopment analysis. Management Science, 30(9), 1078-1092. dx.doi.org/10.1287/mnsc.30.9.1078
Beasley, J. E. (1995). Determining teaching and research efciencies. Journal of the Operational Research Society, 46(4),
441-452.
Beasley, J. E. (1990). Comparing university departments. Omega, 18(2), 171-183. dx.doi.org/10.1016/0305-0483(90)90064-G
Bessent, A., Bessent, W., Kennington, J., & Reagan, B. (1982). An application of mathematical programming to assess
productivity in the Houston independent school district. Management Science, 28(12), 1355-1367. dx.doi.org/10.1287/
mnsc.28.12.1355
Boyd, B. K., Bergh, D. D., & Ketchen, D. J. J. (2010). Reconsidering the reputation–performance relationship: A resource-
based view. Journal of Management, 36(3), 588-609. dx.doi.org/10.1177/0149206308328507
Caballero, R., Galache, T., Gomez, T., Molina, J., & Torrico, A. (2004). Budgetary allocations and efciency in the human
resources policy of a university following multiple criteria. Economics of Education Review, 23(1), 67-74. dx.doi.
org/10.1016/S0272-7757(03)00049-9
Caroline, M. C. N., Castano, N., & Cabanda, E. C. (2007). Performance evaluation of the efciency of Philippine private
higher educational institutions: Application of frontier approaches. International Transactions in Operational Research,
14(5), 431-444. dx.doi.org/10.1111/j.1475-3995.2007.00599.x
Chalos, P. (1997). An examination of budgetary inefciency in education using data envelopment analysis. Financial
Accountability and Management, 13(1), 55-69. dx.doi.org/10.1111/1468-0408.00026
Chalos, P., & Cherian, J. (1995). An application of data envelopment analysis to public sector performance measurement and
accountability. Journal of Accounting and Public Policy, 14(2), 143-160. dx.doi.org/10.1016/0278-4254(94)00015-S
Chang, T-Y., Chung, P-H., & Hsu, S-S. (2012). Two-stage performance model for evaluating the managerial efciency
of higher education: An application by the Taiwanese tourism and leisure department. Journal of Hospitality, Leisure,
Sport and Tourism Education, 11(2), 168-177. dx.doi.org/10.1016/j.jhlste.2012.04.003
Charnes, A., Cooper, W. W., Golany, B., Seiford, L. M., & Stutz, J. (1985). Foundations of data envelopment analy-
sis and Pareto-Koopmans empirical production functions. Journal of Econometrics, 30(1-2), 91-107. dx.doi.
org/10.1016/0304-4076(85)90133-2
Charnes, A., Cooper, W. W., & Rhodes, E. (1978). Measuring the efciency of decision making units. European Journal
of Operational Research, 2(6), 429-444. dx.doi.org/10.1016/0377-2217(78)90138-8
Charnes, A., Cooper, W. W., & Rhodes, E. (1981). Evaluating program and managerial efciency: An application of
data envelopment analysis to program follow through. Management Science, 27(6), 668-697. dx.doi.org/10.1287/
mnsc.27.6.668
Charnes, A., Cooper, W. W., Seiford, L., & Stutz, J. (1983). Invariant multiplicative efciency and piecewise Cobb-
Douglas envelopments. Operations Research Letters, 2(3), 101-103. dx.doi.org/10.1016/0167-6377(83)90014-7
Chen, Y., Cook, W. D., Kao, C., & Zhu, J. (2013). Network DEA pitfalls: Divisional efciency and frontier projection.
European Journal of Operational Research, 226(3), 507-515. dx.doi.org/10.1016/j.ejor.2012.11.021
Chen, Y., Cook, W. D., Li, N., & Zhu, J. (2009). Additive efciency decomposition in two-stage DEA. European Journal
of Operational Research, 196(3), 1170–1176. dx.doi.org/10.1016/j.ejor.2008.05.011
Chen, Y., Cook, W. D., & Zhu, J. (2010). Deriving the DEA frontier for two-stage processes. European Journal of
Operational Research, 202(1), 138–142. dx.doi.org/10.1016/j.ejor.2009.05.012
Cohn, E., Rhine, S. L. W., & Santos, M. C. (1989). Institutions of higher education as multi-product rms: Economies of
scale and scope. The Review of Economics and Statistics, 71(2), 284-290. dx.doi.org/10.2307/1926974
Cook, W. D., Green, R. H., & Zhu, J. (2006). Dual-role factors in data envelopment analysis. IIE Transactions, 38(2),
105-115. dx.doi.org/10.1080/07408170500245570
Cook, W. D., Liang, L., & Zhu, J. (2010). Measuring performance of two-stage network structures by DEA: A review and
future perspective. Omega, 38(6), 423–430. dx.doi.org/10.1016/j.omega.2009.12.001
Cook, W. D., & Zhu, J. (2005). Modeling performance measurement: Applications and implementation issues in DEA.
Boston, MA: Springer Science.
Cook, W. D., Zhu, J., Bi, G., & Yang, F. (2010). Network DEA: Additive efciency decomposition. European Journal of
Operational Research, 207(2), 1122–1129. dx.doi.org/10.1016/j.ejor.2010.05.006
Cook, W. D., Zhu, J., & Liang, L. (2011). DEA Efciency in Two-Stage Networks with Feedback. IIE Transactions, 43(5),
309-322. dx.doi.org/10.1080/0740817X.2010.509307
Cooper, W. W., Seiford, L. M., & Tone, K. (2006). Introduction to data envelopment analysis and its uses. New York, NY:
Springer-Verlag.
Delaney, A. M. (2008). Typical institutional research studies on students: perspective and examples. New Directions for
Higher Education, 2008(141), 57-67. dx.doi.org/10.1002/he.293
105
Competitiveness among Higher Education Institutions
Deller, S. C., & Rudnicki, E. R. (1993). Production efciency in elementary education: The case of Maine public schools.
Economics of Education Review, 12(1), 45-57. dx.doi.org/10.1016/0272-7757(93)90042-F
Diamond, A. M., & Medewitz, J. N. (1990). Use of data envelopment analysis in an evaluation of the efciency of the
DEEP program for economic education. Journal of Economic Education, 21(3), 337- 354. dx.doi.org/10.2307/1182252
Doyle, S., Gendall, P., Meyer, L. H., Hoek, J., Tait, C., McKenzie, L., & Loorparg, A. (2010). An Investigation of factors
associated with student participation in study abroad. Journal of Studies in International Education, 14(5), 471-490.
dx.doi.org/10.1177/1028315309336032
Elliott, K. (1992). The third wave breaks on the shores of accounting. Accounting Horizons, 6(2), 61-85.
Fandel, G. (2007). On the performance of universities in North Rhine-Westphalia, Germany: Government’s redistribution
of funds judged using DEA efciency measures. European Journal of Operational Research, 176(1), 521-533. dx.doi.
org/10.1016/j.ejor.2005.06.043
Farrell, M. J. (1957). The measurement of productive efciency. Journal of the Royal Statistical Society, Series A
(General), 120(3), 253-290. dx.doi.org/10.2307/2343100
Franek, R., Meltzer, T., Maier, C., Olson, E., Doherty, J., Owens, E., & Bakhshi, M. (2010). The Best 373 Colleges (2011
ed.). New York, NY: Random House.
Färe, R. S., Grosskopf, S., Norris, M., & Zhang, Z. (1994). Productivity Growth, Technical Progress, and Efciency
Change in Industrialized Countries. The American Economic Review, 84(1), 66-83.
Färe, R. S., & Grosskopf, S. (2000). Network DEA. Socio-Economic Journal, 5(1-2), 9-22.
Gansemer-Topf, A. M., & Schuh, J. H. (2006). Institutional sensitivity and institutional expenditures: Examining organiza-
tional factors. Research in Higher Education, 47(6), 613-642. dx.doi.org/10.1007/s11162-006-9009-4
Glass, J. C., McKillop, D. G., & Hyndman, N. (1995). Efciency in the provision of university teaching and research:
An empirical analysis of UK universities. Journal of Applied Econometrics, 10(1), 61-72. dx.doi.org/10.1002/
jae.3950100106
Johnes, G. (1988). Determinants of research output in economics departments in British universities. Research Policy,
17(3), 171-178. dx.doi.org/10.1016/0048-7333(88)90041-8
Johnes, G. (1990). Measures of research output: University departments of economics in the UK. The Economic Journal,
100(401), 556-560. dx.doi.org/10.2307/2234141
Johnes, G., & Johnes, J. (1993). Measuring the research performance of UK economics departments: An application of
data envelopment analysis. Oxford Economic Papers, 45(2), 332-347.
Johnes, J., & Johnes, G. (1995). Research funding and performance in U.K. university departments of economics: a fron-
tier analysis. Economics of Education Review, 14(3), 301-314. dx.doi.org/10.1016/0272-7757(95)00008-8
Johnes, J., & Yu, L. (2008). Measuring the research performance of Chinese higher education institutions using data envel-
opment analysis. China Economic Review, 19(4), 679-696. dx.doi.org/10.1016/j.chieco.2008.08.004
Kao, C., & Hung, H-T. (2008). Efciency analysis of university departments: An empirical study. OMEGA, 36(4),
653–664. dx.doi.org/10.1016/j.omega.2006.02.003
Kao, C., & Hwang, S-N. (2008). Efciency decomposition in two-stage data envelopment analysis: An application to non-
life insurance companies in Taiwan. European Journal of Operational Research, 185(1), 418-429. dx.doi.org/10.1016/j.
ejor.2006.11.041
Kaplan, R. S., & Norton, D. P. (1996). Translating Strategy into Action. Boston, MA: Harvard Business School Press.
Kingsley, G., & Malecki, E. J. (2004). Networking for Competitiveness. Small Business Economics, 23(1), 71-84. dx.doi.
org/10.1023/B:SBEJ.0000026022.08180.b7
Lamotte, G., & Carter, G. (2000). Are the balanced scorecard and the EFQM excellence model mutually exclusive or
do they work together to bring value added to a company? (Working Paper). Brussels, Belgium: Balanced Scorecard
Collaborative.
Lascelles, D. M., & Peacock, R. (1996). Self-assessment for business excellence. Maidenhead, UK: McGraw-Hill.
Lauer, C. (2002). Participation in higher education. The role of cost and return expectations. International Journal of
Manpower, 23(5), 443-457. dx.doi.org/10.1108/01437720210450897
Liang, L., Chen, Y., Cook, W. D., Du, J., & Zhu, J. (2011). A bargaining game model for measuring performance of
two-stage network structures. European Journal of Operational Research, 210(2), 390-397. dx.doi.org/10.1016/j.
ejor.2010.08.025
Liang, L., Cook, W. D., & Zhu, J. (2008). DEA models for two-stage processes: Game approach and efciency decomposi-
tion. Naval Research Logistics, 55(7), 643-653. dx.doi.org/10.1002/nav.20308
Liang, L., Yang, F., Cook, W. D., & Zhu, J. (2006). DEA models for supply chain efciency evaluation. Annals of
Operations Research, 145(1), 35-49. dx.doi.org/10.1007/s10479-006-0026-7
Lubitz, V. D., & Wickramasinghe, N. (2006). Dynamic leadership in unstable and unpredictable Environments.
International Journal of Management and Enterprise Development, 3(4), 339-350.
106 JCC: The Business and Economics Research Journal
Madden, G., Savage, S., & Kemp, S. (1997). Measuring public sector efciency: A study of economics departments at
Australian Universities. Education Economics, 5(2), 153-168. dx.doi.org/10.1080/09645299700000013
Martin, E. (2006). Efciency and quality in the current higher education context in Europe: An application of the data
envelopment analysis methodology to performance assessment of departments within the University of Zaragoza.
Quality in Higher Education, 12(1), 57-79. dx.doi.org/10.1080/13538320600685172
McCarty, T. A., & Yaisawarng, S. (1993). Technical efciency in New Jersey school districts. In H. O., Fried, C. A., Lovell,
& S. S., Schmidt (Eds.), The Measurement of Productive Efciency: Techniques and Applications (pp. 271-287). New
York, NY: Oxford University Press.
Moingeon, B., & Edmondson, A. (1996). Organizational Learning and Competitive Advantage. London, UK: Sage
Publications Ltd. dx.doi.org/10.4135/9781446250228
Podobnik, D., & Dolinsek, S. (2008). Competiveness and performance development: An integrated management model.
Journal of Organizational Change, 21(2), 213-229. dx.doi.org/10.1108/09534810810856444
Porter, M. E. (1998). On Competition. Boston, MA: Harvard Business School.
Ray, S. C. (1991). Resource-use efciency in public schools: A study of Connecticut data. Management Science, 37(12),
1620-1628. dx.doi.org/10.1287/mnsc.37.12.1620
Rinn, A. N., & Plucker, J. A. (2004). We recruit them, but then what? The educational and psychological experiences of
academically talented undergraduates. Gifted Child Quarterly, 48(1), 54-67. dx.doi.org/10.1177/001698620404800106
Sinuany-Stern, Z., Mehrez, A., & Barboy, A. (1994). Academic departments’ efciency via DEA. Computers & Operations
Research, 21(5), 543-556. dx.doi.org/10.1016/0305-0548(94)90103-1
Spender, J-C. (1996). Organizational knowledge, learning and memory: Three concepts in search of theory. Journal of
Organizational Change Management, 9(1), 63-78. dx.doi.org/10.1108/09534819610156813
Sum, N-L., & Jessop, B. (2013). Competitiveness, the knowledge-based economy and higher education. Journal of the
Knowledge Economy, 4(1), 24-44. dx.doi.org/10.1007/s13132-012-0121-8
Tomkins, C., & Green, R. (1988). An experiment in the use of data envelopment analysis for evaluating the efcien-
cy of UK university departments of accounting. Financial Accountability and Management, 4(2), 147-164. dx.doi.
org/10.1111/j.1468-0408.1988.tb00066.x
Tone, K. (2001). A slacks-based measure of efciency in data envelopment analysis. European Journal of Operational
Research, 130(3), 498-509. dx.doi.org/10.1016/S0377-2217(99)00407-5
Tyagi, P., Yadav, S. P., & Singh, S. P. (2009). Relative performance of academic departments using DEA with sensitivity
analysis. Evaluation and Program Planning, 32(2), 168-177. dx.doi.org/10.1016/j.evalprogplan.2008.10.002
Vidal-Salazar, M. D., Ferrón-Vílchez, V., & Cordón-Pozo, E. (2012). Coaching: An effective practice for busi-
ness competitiveness. Competitiveness Review: An International Business Journal, 22(5), 423-433. dx.doi.
org/10.1108/10595421211266302
Wireman, T. (2004). Benchmarking Best Practices in Maintenance Management. New York, NY: Industrial Press Inc.
Retrieved from http://common.books24x7.com.ezproxy.library.yorku.ca/toc.aspx?bookid=9012
Wongrassamee, S., Gardiner, P. D., & Simmons, J. E. L. (2003). Performance measurement tools: The balanced scorecard and
the EFQM Excellence Model. Measuring Business Excellence, 7(1), 14-29. dx.doi.org/10.1108/13683040310466690
Authors Note
Sonia Valeria Avilés-Sacoto, Quality and Manufacturing Chair Department, Instituto Tecnológico y de Estudios
Superiores de Monterrey (ITESM), Av. Eugenio Garza Sada 2501 Sur, Colonia Tecnológico, Monterrey, NL, Mexico.
Wade Douglas Cook, Operations Management and Information Systems, Schulich School of Business, York University,
Toronto, Canada.
David Güemes-Castorena, Quality and Manufacturing Chair Department, Instituto Tecnológico y de Estudios
Superiores de Monterrey (ITESM), Av. Eugenio Garza Sada 2501 Sur, Colonia Tecnológico, Monterrey, NL, Mexico.
Correspondence concerning this article should be addressed to Sonia Valeria Avilés-Sacoto, Email: sonia_v_a_s@
hotmail.com
The authors (S. Avilés-Sacoto and D. Güemes-Castorena) gratefully acknowledge CONACYT - Mexico, for their main-
tenance scholarship and for the support for the research and scientific development of teachers and students (national
and international). Wade D. Cook wishes to acknowledge the Natural Sciences and Engineering Research Council
(NSERC) of Canada for financial support. Thanks also to those professors at Instituto Tecnológico y de Estudios
Superiores Monterrey (ITESM) in Mexico and at Schulich School of Business at York University in Canada, who made
contributions toward this study.
107
Competitiveness among Higher Education Institutions
Appendix
Table 1
List of Universities (37 DMUs)
DMU Universities
1 University of Notre Dame (Mendoza)
2 University of Virginia (McIntire)
3 Massachusetts Institute of Technology (Sloan)
4 University of Pennsylvania (Wharton)
5 Cornell University
6 Emory University (Goizueta)
7 University of Michigan (Ross)
8 Boston College (Carroll)
9 University of Texas (McCombs)
10 New York University (Stern)
11 University of North Carolina (Kenan-Flagler)
12 University of Richmond (Robins)
13 Miami University (Farmer)
14 Babson College
15 Wake Forest University
16 Indiana University (Kelley)
17 Villanova University
18 Bentley University
19 Carnegie Mellon University (Tepper)
20 College of William and Mary (Mason)
21 University of Illinois
22 Pennsylvania State University (Smeal)
23 Southern Methodist University (Cox)
24 University of Washington (Foster)
25 Rensselaer Polytechnic Institute (Lally)
26 Boston University
27 Case Western Reserve University (Weatherhead)
28 Santa Clara University (Leavey)
29 DePaul University
30 University of Wisconsin
31 Michigan State University (Broad)
32 Texas A&M University (Mays)
33 Syracuse University (Whitman)
34 Fordham University
35 University of Georgia (Terry)
36 Georgia Institute of Technology
37 ITESM
108 JCC: The Business and Economics Research Journal
Table 2
Data for Schools of Business
Stage 1 Stage 2
Inputs Outputs Inputs Outputs
x1 x2 x3 z1 z2 I z1 z2 I J
DMU
%
students in
top 25%
of their
class
% of
applicants
rejected
Academic
rating
% of
accepted
applicants
enrolled
% receiving
institutional
scholarships
%
receiving
internships
% of
accepted
applicants
enrolled
% receiving
institutional
scholarships
% receiving
internships
%
students
who get
jobs
10.950 0.730 0.900 0.990 0.370 0.790 0.990 0.370 0.790 0.950
20.970 0.680 0.930 0.960 0.200 0.863 0.960 0.200 0.863 0.780
31.000 0.890 0.970 0.650 0.580 0.909 0.650 0.580 0.909 0.940
41.000 0.830 0.910 0.750 0.430 0.852 0.750 0.430 0.852 0.930
50.980 0.810 0.920 0.800 0.420 0.853 0.800 0.420 0.853 0.930
60.980 0.700 0.920 0.990 0.370 0.862 0.990 0.370 0.862 0.840
70.990 0.500 0.830 0.820 0.380 0.897 0.820 0.380 0.897 0.810
80.950 0.700 0.890 0.280 0.300 0.841 0.280 0.300 0.841 0.830
90.950 0.560 0.740 0.640 0.190 0.842 0.640 0.190 0.842 1.000
10 0.920 0.620 0.810 0.360 0.530 0.923 0.360 0.530 0.923 0.930
11 0.960 0.680 0.830 1.000 0.480 0.773 1.000 0.480 0.773 0.970
12 0.870 0.610 0.940 0.260 0.610 0.777 0.260 0.610 0.777 1.000
13 0.740 0.210 0.770 0.200 0.650 0.708 0.200 0.650 0.708 0.710
14 0.870 0.600 0.850 0.280 0.460 0.851 0.280 0.460 0.851 0.640
15 0.910 0.620 0.890 0.900 0.420 0.788 0.900 0.420 0.788 0.930
16 0.710 0.270 0.760 0.460 0.600 0.738 0.460 0.600 0.738 0.900
17 0.880 0.540 0.880 0.300 0.500 0.890 0.300 0.500 0.890 0.820
18 0.790 0.570 0.800 0.330 0.600 0.862 0.330 0.600 0.862 0.900
19 0.930 0.640 0.990 0.210 0.500 0.970 0.210 0.500 0.970 0.850
20 0.980 0.660 0.940 0.950 0.310 0.617 0.950 0.310 0.617 0.970
21 0.890 0.290 0.720 0.450 0.310 0.749 0.450 0.310 0.749 0.720
22 0.860 0.480 0.770 0.450 0.260 0.737 0.450 0.260 0.737 0.950
23 0.730 0.470 0.760 0.520 0.790 0.749 0.520 0.790 0.749 0.800
24 0.130 0.420 0.750 0.740 0.150 0.636 0.740 0.150 0.636 0.550
25 0.900 0.570 0.840 0.250 0.900 0.564 0.250 0.900 0.564 0.620
26 0.910 0.580 0.720 0.350 0.420 0.821 0.350 0.420 0.821 0.760
27 0.870 0.300 0.810 0.100 0.500 0.735 0.100 0.500 0.735 0.900
28 0.760 0.410 0.830 0.230 0.680 0.688 0.230 0.680 0.688 0.820
29 0.290 0.260 0.750 0.330 0.630 0.629 0.330 0.630 0.629 0.660
30 0.910 0.430 0.770 0.990 0.460 0.683 0.990 0.460 0.683 0.630
31 0.700 0.270 0.710 0.990 0.260 0.678 0.990 0.260 0.678 0.920
32 0.890 0.330 0.710 0.530 0.380 0.581 0.530 0.380 0.581 0.740
33 0.730 0.400 0.800 0.380 0.590 0.870 0.380 0.590 0.870 0.680
34 0.730 0.510 0.840 0.190 0.700 0.876 0.190 0.700 0.876 0.800
35 0.890 0.460 0.720 0.990 0.710 0.532 0.990 0.710 0.532 0.750
36 0.950 0.410 0.720 0.610 0.220 0.612 0.610 0.220 0.612 0.780
37 0.350 0.180 0.843 0.680 0.230 0.471 0.680 0.230 0.471 0.690
109
Competitiveness among Higher Education Institutions
Table 3
Log Data for Schools of Business
Stage 1 Stage 2
Inputs Outputs Inputs Outputs
x1 x2 x3 z1 z2 I z1 z2 I J
DMU
% students
in top 25%
of their
class
% of
applicants
rejected
Academic
rating
% of
accepted
applicants
enrolled
% receiving
institutional
scholarships
% receiving
internships
% of
accepted
applicants
enrolled
% receiving
institutional
scholarships
% receiving
internships
% students
who get
jobs
1 -0.051 -0.315 -0.105 -0.010 -0.994 -0.236 -0.010 -0.994 -0.236 -0.051
2 -0.030 -0.386 -0.073 -0.041 -1.609 -0.147 -0.041 -1.609 -0.147 -0.248
3 0.000 -0.117 -0.030 -0.431 -0.545 -0.095 -0.431 -0.545 -0.095 -0.062
4 0.000 -0.186 -0.094 -0.288 -0.844 -0.160 -0.288 -0.844 -0.160 -0.073
5 -0.020 -0.211 -0.083 -0.223 -0.868 -0.159 -0.223 -0.868 -0.159 -0.073
6 -0.020 -0.357 -0.083 -0.010 -0.994 -0.149 -0.010 -0.994 -0.149 -0.174
7 -0.010 -0.693 -0.186 -0.198 -0.968 -0.109 -0.198 -0.968 -0.109 -0.211
8 -0.051 -0.357 -0.117 -1.273 -1.204 -0.173 -1.273 -1.204 -0.173 -0.186
9 -0.051 -0.580 -0.301 -0.446 -1.661 -0.172 -0.446 -1.661 -0.172 0.000
10 -0.083 -0.478 -0.211 -1.022 -0.635 -0.080 -1.022 -0.635 -0.080 -0.073
11 -0.041 -0.386 -0.186 0.000 -0.734 -0.257 0.000 -0.734 -0.257 -0.030
12 -0.139 -0.494 -0.062 -1.347 -0.494 -0.252 -1.347 -0.494 -0.252 0.000
13 -0.301 -1.561 -0.261 -1.609 -0.431 -0.345 -1.609 -0.431 -0.345 -0.342
14 -0.139 -0.511 -0.163 -1.273 -0.777 -0.161 -1.273 -0.777 -0.161 -0.446
15 -0.094 -0.478 -0.117 -0.105 -0.868 -0.238 -0.105 -0.868 -0.238 -0.073
16 -0.342 -1.309 -0.274 -0.777 -0.511 -0.304 -0.777 -0.511 -0.304 -0.105
17 -0.128 -0.616 -0.128 -1.204 -0.693 -0.117 -1.204 -0.693 -0.117 -0.198
18 -0.236 -0.562 -0.223 -1.109 -0.511 -0.149 -1.109 -0.511 -0.149 -0.105
19 -0.073 -0.446 -0.010 -1.561 -0.693 -0.030 -1.561 -0.693 -0.030 -0.163
20 -0.020 -0.416 -0.062 -0.051 -1.171 -0.483 -0.051 -1.171 -0.483 -0.030
21 -0.117 -1.238 -0.329 -0.799 -1.171 -0.289 -0.799 -1.171 -0.289 -0.329
22 -0.151 -0.734 -0.261 -0.799 -1.347 -0.305 -0.799 -1.347 -0.305 -0.051
23 -0.315 -0.755 -0.274 -0.654 -0.236 -0.289 -0.654 -0.236 -0.289 -0.223
24 -2.040 -0.868 -0.288 -0.301 -1.897 -0.453 -0.301 -1.897 -0.453 -0.598
25 -0.105 -0.562 -0.174 -1.386 -0.105 -0.573 -1.386 -0.105 -0.573 -0.478
26 -0.094 -0.545 -0.329 -1.050 -0.868 -0.197 -1.050 -0.868 -0.197 -0.274
27 -0.139 -1.204 -0.211 -2.303 -0.693 -0.308 -2.303 -0.693 -0.308 -0.105
28 -0.274 -0.892 -0.186 -1.470 -0.386 -0.374 -1.470 -0.386 -0.374 -0.198
29 -1.238 -1.347 -0.288 -1.109 -0.462 -0.464 -1.109 -0.462 -0.464 -0.416
30 -0.094 -0.844 -0.261 -0.010 -0.777 -0.381 -0.010 -0.777 -0.381 -0.462
31 -0.357 -1.309 -0.342 -0.010 -1.347 -0.389 -0.010 -1.347 -0.389 -0.083
32 -0.117 -1.109 -0.342 -0.635 -0.968 -0.543 -0.635 -0.968 -0.543 -0.301
33 -0.315 -0.916 -0.223 -0.968 -0.528 -0.139 -0.968 -0.528 -0.139 -0.386
34 -0.315 -0.673 -0.174 -1.661 -0.357 -0.132 -1.661 -0.357 -0.132 -0.223
35 -0.117 -0.777 -0.329 -0.010 -0.342 -0.631 -0.010 -0.342 -0.631 -0.288
36 -0.051 -0.892 -0.329 -0.494 -1.514 -0.491 -0.494 -1.514 -0.491 -0.248
37 -1.050 -1.715 -0.171 -0.386 -1.470 -0.753 -0.386 -1.470 -0.753 -0.371
110 JCC: The Business and Economics Research Journal
Table 4
Efciency Scores (p1 = 0.50 and p2 = 0.50)
1234567
DMU Log eoagg e^ Log eoagg Log eo1 e^ Log eo1 Log eo2 e^ Log eo2
1 -1.236 0.291 -0.891 0.410 -1.580 0.206
2 -1.730 0.177 -1.464 0.231 -1.996 0.136
3 -1.467 0.231 -0.606 0.546 -2.329 0.097
4 -1.471 0.230 -0.897 0.408 -2.045 0.129
5 -1.454 0.234 -0.817 0.442 -2.091 0.124
6 -1.687 0.185 -0.541 0.582 -2.834 0.059
7 -1.511 0.221 -0.457 0.633 -2.564 0.077
8 -1.671 0.188 -2.052 0.129 -1.291 0.275
9 -0.363 0.696 -0.725 0.484 0.000 1.000
10 -1.225 0.294 -0.598 0.550 -1.851 0.157
11 -1.028 0.358 -0.570 0.566 -1.486 0.226
12 -0.786 0.456 -1.572 0.208 0.000 1.000
13 -0.537 0.584 0.000 1.000 -1.074 0.342
14 -1.667 0.189 -1.584 0.205 -1.750 0.174
15 -1.267 0.282 -0.653 0.521 -1.880 0.153
16 -0.863 0.422 0.000 1.000 -1.725 0.178
17 -1.513 0.220 -1.231 0.292 -1.796 0.166
18 -1.316 0.268 -0.596 0.551 -2.036 0.131
19 -1.637 0.195 -1.233 0.291 -2.042 0.130
20 -1.086 0.338 -2.107 0.122 -0.064 0.938
21 -0.935 0.392 -1.009 0.365 -0.862 0.423
22 -0.898 0.408 -1.645 0.193 -0.151 0.860
23 -1.290 0.275 -0.296 0.744 -2.283 0.102
24 0.000 1.000 0.000 1.000 0.000 1.000
25 -1.668 0.189 -2.296 0.101 -1.040 0.353
26 -1.279 0.278 -1.166 0.312 -1.391 0.249
27 -1.049 0.350 -2.098 0.123 0.000 1.000
28 -1.273 0.280 -1.522 0.218 -1.025 0.359
29 -0.622 0.537 0.000 1.000 -1.243 0.289
30 -1.481 0.227 -0.629 0.533 -2.334 0.097
31 -0.280 0.756 0.000 1.000 -0.560 0.571
32 -0.775 0.461 -0.552 0.576 -0.998 0.369
33 -1.316 0.268 -0.318 0.728 -2.314 0.099
34 -1.380 0.251 -1.110 0.329 -1.651 0.192
35 -1.158 0.314 0.000 1.000 -2.316 0.099
36 -1.114 0.328 -1.956 0.141 -0.271 0.763
37 0.000 1.000 0.000 1.000 0.000 1.000
111
Competitiveness among Higher Education Institutions
Table 5
Efciency Scores (p1 = 0.25 and p2 = 0.75)
1234567
DMU Log eoagg e^ Log eoagg Log eo1 e^ Log eo1 Log eo2 e^ Log eo2
1 -1.397 0.247 -0.920 0.399 -1.557 0.211
2 -1.393 0.248 -3.237 0.039 -0.779 0.459
3 -1.798 0.166 -1.762 0.172 -1.810 0.164
4 -1.732 0.177 -1.849 0.157 -1.693 0.184
5 -1.745 0.175 -1.774 0.170 -1.735 0.176
6 -1.959 0.141 -2.036 0.131 -1.934 0.145
7 -1.831 0.160 -1.704 0.182 -1.874 0.154
8 -1.104 0.332 -3.669 0.026 -0.249 0.779
9 -0.181 0.834 -0.725 0.484 0.000 1.000
10 -1.387 0.250 -1.803 0.165 -1.248 0.287
11 -1.253 0.286 -0.583 0.558 -1.476 0.229
12 -0.393 0.675 -1.572 0.208 0.000 1.000
13 -0.806 0.447 0.000 1.000 -1.074 0.342
14 -1.634 0.195 -2.237 0.107 -1.434 0.238
15 -1.569 0.208 -0.674 0.510 -1.867 0.155
16 -1.294 0.274 0.000 1.000 -1.725 0.178
17 -1.532 0.216 -1.980 0.138 -1.383 0.251
18 -1.548 0.213 -1.585 0.205 -1.536 0.215
19 -2.320 0.098 -6.052 0.002 -1.077 0.341
20 -0.547 0.579 -2.187 0.112 0.000 1.000
21 -0.890 0.411 -1.117 0.327 -0.814 0.443
22 -0.500 0.607 -1.703 0.182 -0.099 0.906
23 -1.787 0.168 -0.296 0.744 -2.283 0.102
24 0.000 1.000 0.000 1.000 0.000 1.000
25 -1.354 0.258 -2.296 0.101 -1.040 0.353
26 -1.335 0.263 -1.166 0.312 -1.391 0.249
27 -0.525 0.592 -2.098 0.123 0.000 1.000
28 -1.149 0.317 -1.522 0.218 -1.025 0.359
29 -0.932 0.394 0.000 1.000 -1.243 0.289
30 -1.908 0.148 -0.629 0.533 -2.334 0.097
31 -0.420 0.657 0.000 1.000 -0.560 0.571
32 -0.875 0.417 -0.606 0.546 -0.965 0.381
33 -1.717 0.180 -1.019 0.361 -1.949 0.142
34 -1.410 0.244 -1.826 0.161 -1.272 0.280
35 -1.517 0.219 -0.654 0.520 -1.804 0.165
36 -0.659 0.518 -2.455 0.086 -0.060 0.942
37 0.000 1.000 0.000 1.000 0.000 1.000
112 JCC: The Business and Economics Research Journal
Table 6
Efciency Scores (p1 = 0.10 and p2 = 0.90)
1234567
DMU Log eoagg e^ Log eoagg Log eo1 e^ Log eo1 Log eo2 e^ Log eo2
1 -1.493 0.225 -0.920 0.399 -1.557 0.211
2 -1.024 0.359 -3.237 0.039 -0.779 0.459
3 -1.805 0.164 -1.762 0.172 -1.810 0.164
4 -1.709 0.181 -1.849 0.157 -1.693 0.184
5 -1.739 0.176 -1.774 0.170 -1.735 0.176
6 -1.944 0.143 -2.036 0.131 -1.934 0.145
7 -1.857 0.156 -1.704 0.182 -1.874 0.154
8 -0.591 0.554 -3.669 0.026 -0.249 0.779
9 -0.073 0.930 -0.725 0.484 0.000 1.000
10 -1.304 0.272 -1.803 0.165 -1.248 0.287
11 -1.387 0.250 -0.583 0.558 -1.476 0.229
12 -0.157 0.855 -1.572 0.208 0.000 1.000
13 -0.967 0.380 0.000 1.000 -1.074 0.342
14 -1.473 0.229 -2.535 0.079 -1.354 0.258
15 -1.732 0.177 -1.361 0.257 -1.774 0.170
16 -1.553 0.212 0.000 1.000 -1.725 0.178
17 -1.443 0.236 -1.980 0.138 -1.383 0.251
18 -1.541 0.214 -1.585 0.205 -1.536 0.215
19 -1.228 0.293 -2.594 0.075 -1.077 0.341
20 -0.219 0.804 -2.187 0.112 0.000 1.000
21 -0.844 0.430 -1.117 0.327 -0.814 0.443
22 -0.214 0.808 -2.136 0.118 0.000 1.000
23 -2.085 0.124 -0.296 0.744 -2.283 0.102
24 0.000 1.000 0.000 1.000 0.000 1.000
25 -1.145 0.318 -2.511 0.081 -0.993 0.371
26 -1.262 0.283 -1.884 0.152 -1.192 0.304
27 -0.210 0.811 -2.098 0.123 0.000 1.000
28 -1.074 0.342 -1.522 0.218 -1.025 0.359
29 -1.119 0.327 0.000 1.000 -1.243 0.289
30 -2.163 0.115 -0.629 0.533 -2.334 0.097
31 -0.504 0.604 0.000 1.000 -0.560 0.571
32 -0.879 0.415 -1.174 0.309 -0.846 0.429
33 -1.856 0.156 -1.019 0.361 -1.949 0.142
34 -1.327 0.265 -1.826 0.161 -1.272 0.280
35 -1.689 0.185 -0.654 0.520 -1.804 0.165
36 -0.299 0.741 -2.455 0.086 -0.060 0.942
37 0.000 1.000 0.000 1.000 0.000 1.000
113
Competitiveness among Higher Education Institutions
Table 7
Efciency Scores (p1 = 0.25 & p2 = 0.75) Lower bound = 1 on Jobs Multiplier
1234567
DMU Log eoagg e^ Log eoagg Log eo1 e^ Log eo1 Log eo2 e^ Log eo2
1 -0.038 0.962 0.000 1.000 -0.051 0.950
2 -0.186 0.830 0.000 1.000 -0.248 0.780
3 -0.046 0.955 0.000 1.000 -0.062 0.940
4 -0.054 0.947 0.000 1.000 -0.073 0.930
5 -0.054 0.947 0.000 1.000 -0.073 0.930
6 -0.131 0.877 0.000 1.000 -0.174 0.840
7 -0.158 0.854 0.000 1.000 -0.211 0.810
8 -0.140 0.870 0.000 1.000 -0.186 0.830
9 0.000 1.000 0.000 1.000 0.000 1.000
10 -0.054 0.947 0.000 1.000 -0.073 0.930
11 -0.023 0.977 0.000 1.000 -0.030 0.970
12 0.000 1.000 0.000 1.000 0.000 1.000
13 -0.202 0.817 0.000 1.000 -0.270 0.764
14 -0.335 0.716 0.000 1.000 -0.446 0.640
15 -0.054 0.947 0.000 1.000 -0.073 0.930
16 -0.074 0.928 -0.001 0.999 -0.099 0.906
17 -0.149 0.862 0.000 1.000 -0.198 0.820
18 -0.079 0.924 0.000 1.000 -0.105 0.900
19 -0.122 0.885 0.000 1.000 -0.163 0.850
20 -0.013 0.987 -0.052 0.949 0.000 1.000
21 -0.234 0.791 -0.025 0.975 -0.304 0.738
22 -0.033 0.968 -0.042 0.959 -0.030 0.971
23 -0.167 0.846 -0.014 0.986 -0.218 0.804
24 0.000 1.000 0.000 1.000 0.000 1.000
25 -0.301 0.740 -0.640 0.527 -0.188 0.829
26 -0.206 0.814 0.000 1.000 -0.274 0.760
27 -0.036 0.965 -0.142 0.867 0.000 1.000
28 -0.137 0.872 -0.189 0.827 -0.120 0.887
29 -0.210 0.811 0.000 1.000 -0.280 0.756
30 -0.340 0.712 -0.024 0.976 -0.446 0.640
31 -0.038 0.963 0.000 1.000 -0.050 0.951
32 -0.162 0.850 -0.247 0.781 -0.134 0.875
33 -0.289 0.749 0.000 1.000 -0.386 0.680
34 -0.167 0.846 0.000 1.000 -0.223 0.800
35 -0.160 0.852 -0.347 0.707 -0.097 0.907
36 -0.110 0.896 -0.298 0.742 -0.048 0.953
37 0.000 1.000 0.000 1.000 0.000 1.000
114 JCC: The Business and Economics Research Journal
Table 8
Efciency Scores (p1 = 0.25 & p2 = 0.75) Lower bound = 1 on Jobs & Internships Multiplier
1234567
DMU Log eoagg e^ Log eoagg Log eo1 e^ Log eo1 Log eo2 e^ Log eo2
1 -0.229 0.795 -0.099 0.906 -0.273 0.761
2 -0.405 0.667 -0.155 0.856 -0.488 0.614
3 -0.286 0.752 -0.013 0.988 -0.377 0.686
4 -0.269 0.764 -0.061 0.940 -0.339 0.713
5 -0.273 0.761 -0.051 0.950 -0.347 0.707
6 -0.366 0.693 -0.017 0.983 -0.483 0.617
7 -0.404 0.668 0.000 1.000 -0.538 0.584
8 -0.284 0.753 -0.226 0.798 -0.304 0.738
9 -0.014 0.986 -0.056 0.946 0.000 1.000
10 -0.262 0.770 -0.035 0.966 -0.337 0.714
11 -0.203 0.816 -0.116 0.890 -0.232 0.793
12 -0.043 0.958 -0.172 0.842 0.000 1.000
13 -0.207 0.813 0.000 1.000 -0.277 0.758
14 -0.489 0.613 -0.149 0.861 -0.602 0.548
15 -0.237 0.789 -0.104 0.901 -0.281 0.755
16 -0.130 0.878 0.000 1.000 -0.173 0.841
17 -0.330 0.719 -0.099 0.906 -0.406 0.666
18 -0.240 0.787 -0.082 0.921 -0.292 0.747
19 -0.328 0.720 -0.062 0.940 -0.417 0.659
20 -0.088 0.916 -0.352 0.703 0.000 1.000
21 -0.300 0.741 -0.015 0.985 -0.395 0.674
22 -0.103 0.902 -0.252 0.777 -0.053 0.948
23 -0.267 0.766 -0.109 0.897 -0.320 0.726
24 0.000 1.000 0.000 1.000 0.000 1.000
25 -0.301 0.740 -0.640 0.527 -0.188 0.829
26 -0.314 0.730 0.000 1.000 -0.419 0.658
27 -0.054 0.947 -0.218 0.804 0.000 1.000
28 -0.166 0.847 -0.304 0.738 -0.119 0.887
29 -0.210 0.811 0.000 1.000 -0.280 0.756
30 -0.442 0.643 -0.153 0.858 -0.538 0.584
31 -0.048 0.953 0.000 1.000 -0.064 0.938
32 -0.162 0.850 -0.247 0.781 -0.134 0.875
33 -0.448 0.639 -0.133 0.876 -0.597 0.550
34 -0.307 0.736 -0.133 0.876 -0.365 0.695
35 -0.160 0.852 -0.347 0.707 -0.097 0.907
36 -0.142 0.867 -0.364 0.695 -0.069 0.934
37 0.000 1.000 0.000 1.000 0.000 1.000
115
Competitiveness among Higher Education Institutions
Table 9
Inefciencies in Jobs and Internships
12345
DMU Log eoagg e^ Log eoagg Internships Inefciency Job Inefciency
1 -0.229 0.795 0.214 0.015
2 -0.405 0.667 0.341 0.064
3 -0.286 0.752 0.269 0.016
4 -0.269 0.764 0.243 0.026
5 -0.273 0.761 0.245 0.027
6 -0.366 0.693 0.259 0.107
7 -0.404 0.668 0.272 0.131
8 -0.284 0.753 0.247 0.036
9 -0.014 0.986 0.014 0.000
10 -0.262 0.770 0.248 0.013
11 -0.203 0.816 0.203 0.000
12 -0.043 0.958 0.043 0.000
13 -0.207 0.813 0.012 0.195
14 -0.489 0.613 0.198 0.290
15 -0.237 0.789 0.207 0.029
16 -0.130 0.878 0.085 0.044
17 -0.330 0.719 0.224 0.105
18 -0.240 0.787 0.203 0.036
19 -0.328 0.720 0.255 0.072
20 -0.088 0.916 0.088 0.000
21 -0.300 0.741 0.188 0.111
22 -0.103 0.902 0.103 0.000
23 -0.267 0.766 0.134 0.132
24 0.000 1.000 0.000 0.000
25 -0.301 0.740 0.000 0.300
26 -0.314 0.730 0.156 0.158
27 -0.054 0.947 0.036 0.017
28 -0.166 0.847 0.069 0.095
29 -0.210 0.811 0.000 0.209
30 -0.442 0.643 0.118 0.323
31 -0.048 0.953 0.048 0.000
32 -0.162 0.850 0.000 0.162
33 -0.448 0.639 0.204 0.243
34 -0.307 0.736 0.190 0.116
35 -0.160 0.852 0.000 0.159
36 -0.142 0.867 0.142 0.000
37 0.000 1.000 0.000 0.000
116 JCC: The Business and Economics Research Journal
The mission of the Journal of CENTRUM Cathedra (JCC): The Business and Economics Research Journal is to disseminate knowledge
generated by academic research in the areas of management, finance, and economics.
Criteria for Publication
The Journal of CENTRUM Cathedra (JCC): The Business and Economics Research Journal is the official academic journal of CENTRUM
Católica Graduate Business School, the business school of the Pontificia Universidad Católica del Perú. JCC publishes scholarly
research articles offering new theoretical developments, and or empirical contributions. Articles cover such areas as leadership and
business strategy, human resources, organizational behavior, organizational theory, marketing, finance, and economics.
The JCC seeks to publish theoretical and empirical research that proves, expands, and builds global strategic management theory
and contributes to the practice of management.
Requirements for the submission of articles
Manuscripts should be written as simply and concisely as possible without sacrificing meaning and clarity of presentation. As guide-
line, contributions should be about 40 double-spaced pages.
Authors who submit their manuscripts to the journal accept the following:
1. Their manuscripts will not be reviewed by, or be sent to, any other publication during the period of review.
2. Their manuscripts report the elaboration of theory or empirical results that have not been published previously.
3. Their manuscripts are free from any form of plagiarism and display all credits due.
4. Their manuscripts have not been previously submitted to the journal for review.
5. Submit manuscripts in Word without columns.
The review process
The review process will include at least two stages. The first stage will be undertaken by the Editor-in-Chief to assess the relevance
of the article and its fit with the mission of the journal. The second stage will be conducted by anonymous reviewers, who will make
their comments and observations on content in writing, to be forwarded to the author. In the case of rejection, the decision will be
sent to the author. In the case of conditional approval, the comments and observations will be forwarded to the author for correction
or clarification and the author will be given a deadline for submission of his article. In the case of approval, the author will be notified
of the date of publication.
Format requirements for accepted articles
1. It is made explicit that authors submit their manuscripts for review within the existing stylistic criteria of the APA Publication
Manual, 6th edition.
2. The whole article should be submitted in Word format, Times New Roman 12, doubled-spaced. This includes tables, figures, charts,
and equations.
JOURNAL OF CENTRUM CATHEDRA
INFORMATION FOR CONTRIBUTORS
For more information or to order The Journal of CENTRUM Cathedra visit:
centrumwebs.pucp.edu.pe/jcc
Journal of
CENTRUM
Cathedra
JCC
