You are on page 1of 6

This article was downloaded by:[HEAL-Link Consortium]

On: 23 June 2008


Access Details: [subscription number 772811123]
Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954
Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Applied Economics
Publication details, including instructions for authors and subscription information:
http://www.informaworld.com/smpp/title~content=t713684000

An examination of Okun's law: evidence from regional


areas in Greece
Nicholas Apergis a; Anthony Rezitis b
a
Department of International Economics and Political Sciences, University of
Macedonia, Thessaloniki, Greece.
b
Department of Economics, University of Ioannina, Ioannina, Greece.
Online Publication Date: 10 January 2003
To cite this Article: Apergis, Nicholas and Rezitis, Anthony (2003) 'An examination
of Okun's law: evidence from regional areas in Greece', Applied Economics, 35:10,
1147 1151
To link to this article: DOI: 10.1080/0003684032000066787
URL: http://dx.doi.org/10.1080/0003684032000066787

PLEASE SCROLL DOWN FOR ARTICLE


Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf
This article maybe used for research, teaching and private study purposes. Any substantial or systematic reproduction,
re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expressly
forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents will be
complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should be
independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings,
demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or
arising out of the use of this material.

Electronic copy available at: http://ssrn.com/abstract=1150126

Downloaded By: [HEAL-Link Consortium] At: 11:25 23 June 2008

Applied Economics, 2003, 35, 11471151

An examination of Okuns law:


evidence from regional areas in Greece
N I C H O LA S A P E R G I S * and A N T H O N Y R EZ I T I S #
*Department of International Economics and Political Sciences,
University of Macedonia, Thessaloniki, Greece and
#
Department of Economics, University of Ioannina, Ioannina, Greece

This paper estimates Okuns coefficient for certain regional areas in Greece over the
period 19601997. Through the HodrickPrescott filtering and the band-pass
filtering the empirical analysis shows that that the coefficients do not exhibit
substantial interregional differences, except for the cases of Epirus and North
Aegean Islands. In these two cases, the estimates are larger than the regional average
under both detrending methodologies. The empirical findings also show that Okuns
relationship undergoes a structural change in 1981. After this break, unemployment
becomes less responsive to output changes in all regional areas.

I. INTRODUCTION
Okuns famous law (Okun, 1962) stated that excess
capacity or a lack of demand is the major source for the
presence of unemployment. More specifically, the law
postulated a simple relationship between unemployment
and output growth. According to this, an increase of 1% in
the unemployment rate leads to a fall of about 3% in the
growth of real GNP. The law has been frequently used in
conjunction with the Phillips curve for economists as well
as policy makers to understand the implications of certain
economic policy measures for labour as well as commodity
markets (Knoester, 1986).
The statement of the Okuns law yields:
Y  yT  u  uT

where y represents output, expressed in natural logarithms,


u is the unemployment rate, expressed in percentage terms,
and the exponent T in the variable refers to corresponding equilibrium values, i.e. potential output and natural
unemployment, respectively. The two dierences in
Equation 1 dene the output and the unemployment gap,
respectively. Finally, is the Okun coecient.
Relevant studies have investigated the properties of
Okuns law for certain economies (Motley, 1984;
Knoester, 1986; Holloway, 1989; Watts and Mitchell,
1991; Prachowny, 1993; Moosa, 1997). While the majority

of these studies investigated the validity of the law for the


USA, the relationship has also been examined for other
countries as well (Knoester, 1986; Kaufman, 1988; Moosa,
1997). Knoester (1986) reported that the law was not
enough to explain the high unemployment rates for Europe,
especially after the 1970s. Prachowny (1993) and Attfield
and Silverstone (1995) provided evidence that a 1%
reduction in unemployment leads to only a 2/3% increase
in output. Attfield and Silverstone (1995) estimated the
coefficient of Okuns law for the UK by applying recent
developments in time series analysis, i.e. cointegration
techniques.
This study attempts to test Okuns law using data for
Greeces regional economies as they are defined by the
National Statistical Service over the period 19601997
in order to get evidence on regional differences in the
responsiveness of labour markets to output changes. The
results will motivate policy makers to adopt strategies
that will rescue the economy from the problem of
unemployment.
The methodology used to decompose output and
unemployment from their trend, i.e. to get the output and
the unemployment gap, is that of the HodrickPrescott
filter (Hodrick and Prescott, 1997) and the band-pass filter
suggested by Baxter and King (1995). Although the former
methodology has frequently been used to obtain the cyclical
component of a series, the latter exhibits the advantage
that it filters out both the high frequency fluctuations,

Applied Economics ISSN 00036846 print/ISSN 14664283 online 2003 Taylor & Francis Ltd
http://www.tandf.co.uk/journals
DOI: 10.1080/0003684032000066787

Electronic copy available at: http://ssrn.com/abstract=1150126

1147

Downloaded By: [HEAL-Link Consortium] At: 11:25 23 June 2008

1148
e.g. irregular components, and the low frequency fluctuations, e.g. autocorellation. These high and low frequency
fluctuations are not eliminated after taking the first
differences, and applying the HodrickPrescott filter,
respectively, of the relevant series.
The paper is organized as follows. Section II presents a
brief economic historical background for the economy of
Greece related to Okuns relationship, while Section III
provides the empirical analysis and discusses certain policy
implications related to the unemploymentgrowth issue.
Finally, Section IV provides some concluding remarks.

II. A BRIEF HISTORY BACKGROUND


In the 1960s and 1970s, the Greek economy was characterized by low unemployment rates. By contrast, beginning in
1981, the economy experienced a disappointing growth
performance that led to a substantial increase in unemployment. More specifically, unemployment jumped from 2.7%
in 1980 to 8.1% by 1984. Since then, unemployment has
continued to persist at high levels and in 2000 reached a
record high at 11.3%, despite the economy having
experienced high growth rates since 1993. In the light of
Greeces participation in the European Monetary Union
(EMU) from 1 January 2001, Greek policy makers had to
pursue economic policies characterized by conflicting objectives. In particular, the desirability of the Governments
objective of achieving a very low inflation rate was in
conflict with the objective of maintaining economic growth
at sufficiently high rates to reduce unemployment.
It is also highly likely that certain structural changes,
mainly after 1981, tended to have had a strong impact on
the growth rate of potential GNP as well as on the
sensitivity of unemployment to current real GNP changes.
More specifically:


After 1980 the services sector experienced high growth


rates, while the contribution of manufacturing to total
output was signicantly lower. In particular, employment in the services sector increased by 55%, while its
counterpart in the industrial sector fell by 11.6%. Since
the demand for services diers from the demand for
industrial output in terms of sensitivity to economic
uctuations, it is highly likely that Okuns coecient
must have changed over the post-1980 period.
After 1980, the productivity of labour critically slowed
down and it remained stagnant over the entire period. In
particular, over the period 19611980 the hourly labour
productivity increased, on average, by 5% on an annual
basis. By contrast, over the period 19811997 it
experienced only a 0.6% increase. These facts suggest a
changed relationship between unemployment and
growth, i.e. a dierent Okun coecient.

N. Apergis and A. Rezitis




Finally, in 1980, percentage unemployment gures were


around 2.8%, while those gures reached 8% in 1988, a
gure which represented the average OECD and
European Union (EU) unemployment rates. Today,
unemployment in Greece lies at a gure above 11%
(11.3% at the end of 2000). After 1980, the legislation
that protects massive layos was critically improved in
favour of the workers. Moreover, the most painful
characteristic of unemployment gures in Greece is that
a substantial portion of total unemployment is characterized as long-term unemployment. Over the period
19811988 the percentage of those who were unemployed for more than 12 months rose from 21% to
42.2%. Today, over 50% of unemployed can be
characterized as long-term unemployed. Meghir et al.
(1989) provided empirical support to the prolonged
duration of unemployment in Greece. The reported data
on unemployment in Greece suggest the presence of
persistence and hysteresis phenomena. Elmeskov and
MacFarlan (1993) and Apergis (1997) have provided
evidence in favour of such phenomena in Greece. The
observed hysteresis phenomena in the labour market
prevent employment from falling during recession times;
however, in the case of boom conditions, corporations
are reluctant to hire unemployed individuals, resulting in
a slow decrease in unemployment. Thus, the presence of
hysteresis phenomena is expected to have changed
Okuns coecient over the post-1980 period.

To date there has been no specic study that has


investigated relevant issues for specic regions in Greece.
Only the study by Paraskevaides (1993) has demonstrated
that the stagation conditions that Greece has experienced
over the last twenty years have hurt more painfully, in
terms of higher unemployment, the regions of Epirus, West
Macedonia, and the Islands of North Aegean Sea. The
income of these regions originated primarily through the
primary sector of production (over 80%). In this manner, it
is expected that the changes mentioned above, which
occurred in the production side of the Greek economy after
1981, are expected to have dramatically changed, apparently for the worse, the employment conditions in those
regions.
I I I . E M P I R IC A L A N A L Y S I S
Data
The empirical analysis is carried out using annual data on
output measured by real GDP (Y) at 1990 prices and the
unemployment rate (U) covering the period 19601997 for
eight regional areas of Greece. These regions are defined
as: East Macedonia, Central and West Macedonia, Attiki

1149

Downloaded By: [HEAL-Link Consortium] At: 11:25 23 June 2008

An examination of Okuns law: evidence from regional areas in Greece


plus the Rest of Sterea Hellas, Epirus, Thessalia, West
Hellas plus The Ioanian Islands plus Pelloponisos, The
Islands of the North Aegean Sea, and The Islands of the
South Aegean Sea plus Crete. Data were obtained from
the Statistical Service of Greece. For the empirical purposes
of this study, gaps in output and unemployment data had
to be estimated. These gaps are viewed as the cyclical
component of the output and unemployment series. As was
mentioned in the introductory section, two alternative
methodologies were employed to estimate the cyclical
components of the two series under investigation, i.e. the
HodrickPrescott filter decomposition and the band-pass
decomposition. For the empirical purposes of the paper,
the band-pass analysis follows the suggestions of Baxter
and King (1995), who recommend for annual data a highfrequency cut-off point of 2 years and a low-frequency cutoff point of 8 years. The empirical analysis was carried out
with the assistance of RATS 4.2 and MATLAB software.

Table 1. Estimates of Okuns coefficient for the Greek regions from 1960
to 1997

Regions
East Macedonia
LM
Central and West Macedonia
LM
Thessalia
LM
Epirus
LM
Attiki and Sterea Hellas

Regional estimates of Okuns coefficient


Table 1 reports estimates of Okuns coefficients for the
regional data in Greece under the two alternative
methodologies and for the entire period under study. The
results show that the Okuns coefficients for all the regions
under study are statistically significant at 5% across the
detrending methodology followed. Serial correlation (LM)
tests, performed for each equation, show the absence of
any serial correlation problems. Likelihood Ratio tests,
reported in Table 2, indicate that the coefficients do not
exhibit substantial interregional differences, except for the
cases of Epirus and the North Aegean Islands, where they
are larger than the regional average under both detrending
methodologies.
Furthermore, ZivotAndrews (1992) endogenous break
unit root tests (available upon request) performed on the
original income and unemployment data indicated the
presence of a break in 1981. Based on the endogenous
break tests, the filtered series were re-estimated for each of
the separate subperiods under study. Table 3 presents
estimates of Okuns coefficients for the regional data in
Greece, but this time the sample has been split into two
alternative time periods, prior and after 1981. The results
point out that the estimates of Okuns coefficients differ
over the two sample periods. In particular, all regional
coefficients experienced a decrease under both alternative
estimation methodologies. In order to test the statistical
significance of the equality of coefficients across the
estimated Okun equations we proceed with the following
analysis:
Let:
y  yT t 0 1t u  uT t ut

y  yT t 00 11 u  uT t ut

LM
West Hellas and Ioanian
Islands and Pelloponisos
LM
North Aegean Islands
LM
South Aegean Islands and Crete
LM

HodrickPrecott
filter

Band-pass
filter

1.15
(2.54)*
3.79[0.43]

1.22
(3.01)*
4.08[0.38]

1.54
(2.88)*
4.29[0.23]

1.66
(2.97)*
4.17[0.27]

1.63
(2.87)*
3.13[0.37]

1.87
(4.11)*
3.48[0.34]

2.97
(3.24)*
3.21[0.36]

3.19
(2.87)*
2.78[0.52]

1.17
(3.36)*
4.53[0.23]

1.42
(3.13)*
3.39[0.36]

1.56
(3.61)*
4.22[0.24]

1.77
(4.38)*
4.51[0.19]

3.56
(3.14)*
5.18[0.17]

3.69
(4.09)*
4.64[0.26]

1.35
(4.51)*
2.96[0.41]

1.44
(3.76)*
2.55[0.59]

Notes: Figures in parentheses denote t-statistics. For the estimation of


regional average the ratio of income in the region to total income was
used as weighting. LM is a serial correlation test. * Denotes statistically
significant at 5%.

be the two regressions corresponding to the pre- and post1981 period. Dene a dummy variable Dt, which takes
the value 0 over the pre-1981 period and 1 otherwise.
Introducing the dummy variable into Equation 2 yields a
modied regression of the form:
y  yT t 0 1 u  uT t Dt u  uT t  vt

Over the pre-1981 period Equation 4 turns out to be:


y  yT t 0 1 u  uT t vt

while over the post-1981 period Equation 4 yields:


y  yT t 0 1 u  uT t vt

Downloaded By: [HEAL-Link Consortium] At: 11:25 23 June 2008

1150

N. Apergis and A. Rezitis

Table 2. Likelihood ratio tests (testing restrictions across Okun equations)

a1

a2

a3

a1

a2

a3

a4

a5

a6

0.72
[0.39]

0.03
0.87]

4.89
[0.02]

0.30 1.58 3.93 2.31


[0.58] [0.21] [0.03] [0.13]

2.11
[0.17]

5.17
[0.00]

1.87 1.96 17.26 0.05


[0.26] [0.28] [0.00] [0.83]

7.15
[0.00]

0.62 1.79 6.44 2.03


[0.43] [0.18] [0.00] [0.19]

Table 3. Estimates of Okuns coefficient for the Greek regions from 1960 to
1980 and from 1981 to 1997

a7
Regions

HodrickPrecott filter

Band-pass filter

1.03
(3.22)*
2.79[0.37]
0.97
(2.69)*
2.68[0.39]

1.14
(2.87)*
2.55[0.43]
1.01
(3.44)*
2.42[0.47]

Central and West Macedonia


19601980
1.29
(3.26)*
LM
3.09[0.31]
19811997
0.88
(3.47)*
LM
2.91[0.34]

1.50
(4.05)*
2.86[0.35]
1.29
(4.71)*
2.18[0.52]

East Macedonia
19601980
LM
19811997
LM

a4

4.92 4.15 5.46 6.26


[0.03] [0.04] [0.01] [0.00]

a5

0.85 0.02 2.16


[0.36] [0.89] [0.15]

a6

7.97 0.78
[0.00] [0.38]

Thessalia
19601980

4.32
[0.04]

LM
19811997

a7

RATIO

7.73[0.005] 8.94[0.002]

Notes: Figures denote Likelihood Ratio tests. The coefficients a1, a2, a3,
a4, a5, a6, and a7 correspond to the combination between East Macedonia
and Central and West Macedonia, between East Macedonia and
Thessalia, and so on. RATIO is a Likelihood Ratio test (under the two
alternative detrending methods) that inspects the restriction that the
estimated Okun coefficients are all equal across equations. Figures in
parentheses denote p-values.

The testable hypothesis is H0: 0 against H1: 6 0. The


results reported in Table 4 demonstrate that in all regions
the gamma ( ) coecient is shown to be dierent from
zero, implying that Okuns coecients have changed
(declined) over the post-1981 period for reasons mentioned
in Section II.

IV. CONCLUDING REMARKS AND


POLICY IMPLICATIONS
The goal of this paper has been to provide estimates for
Okuns relationship for regional areas in Greece over the
period 19601997. By making use of two alternative
detrending techniques, namely, the HodrickPrescott
filter and the band-pass filter, the results showed that the
coefficients did not exhibit substantial interregional differences, except for the cases of Epirus and North Aegean
Islands. In these two cases, the estimates were larger than
the regional average under both detrending methodologies.
Moreover, the empirical findings showed that Okuns
relationship underwent a structural change in 1981. After
this year, unemployment was shown to be more responsive
to output changes, i.e. economic growth. These findings

LM
Epirus
19601980
LM
19811997
LM
Attiki and Sterea Hellas
19601980
LM
19811997
LM

1.41
(2.66)*
4.64[0.14]
1.12
(2.81)*
2.33[0.56]

1.55
(3.91)*
3.95[0.20]
1.28
(3.55)*
2.13[0.59]

2.31
(3.12)*
3.69[0.25]
2.16
(3.42)*
4.21[0.19]

2.44
(3.68)*
3.44[0.29]
2.38
(3.17)*
3.73[0.23]

1.06
(3.65)*
3.65[0.26]
0.72
(4.09)*
2.08[0.62]

1.33
(3.79)*
3.38[0.35]
0.97
(3.28)*
1.76[0.68]

West Hellas and Ioanian Islands and Pelloponisos


19601980
1.14
(3.12)*
LM
3.49[0.34]
19811997
0.84
(2.48)*
LM
3.11[0.40]
North Aegean Islands
19601980

1.36
(4.79)*
3.27[0.39]
1.11
(3.73)*
2.71[0.45]

2.79
(3.41)*
2.65[0.38]
2.56
(3.44)*
2.31[0.40]

2.81
(4.52)*
2.25[0.43]
2.58
(4.59)*
1.89[0.57]

South Aegean Islands and Crete


19601980
1.04
(3.67)*
LM
3.04[0.34]
19811997
0.78
(2.77)*
LM
2.41[0.49]

1.17
(3.91)*
2.58[0.46]
0.89
(3.44)*
2.07[0.53]

LM
19811997
LM

Notes: As for Table 1.

Downloaded By: [HEAL-Link Consortium] At: 11:25 23 June 2008

An examination of Okuns law: evidence from regional areas in Greece


Table 4. Testing restrictions across Okun equations over the pre- and post1981 period

Region

t-statistic (of )

East Macedonia
3.76[0.0]
Central and West Macedonia
3.49[0.0]
Thessalia
4.03[0.0]
Epirus
4.56[0.0]
Attiki and Sterea Hellas
3.61[0.0]
West Hellas and Ioanian Islands and Pelloponisos 3.45[0.0]
North Aegean Islands
4.83[0.0]
South Aegean Islands and Crete
3.03[0.0]
Notes: Figures in brackets denote p-values.

have some serious policy implications regarding the


strategy against unemployment the policy makers should
follow. More specifically, policy makers must place more
emphasis on deregulating certain sectors in the economy,
particularly, the labour market. This will lead to higher
labour productivity and competition, which in turn, will
increase the productivity of the overall economy and
decrease unemployment in all regions, especially in those
that need it most, e.g. Epirus and North Aegean Islands.

REFERENCES
Apergis, N. (1997) Hysteresis in unemployment and business
cycles: the case of Greece. Working Paper, University of
Macedonia.
Attfield, C. L. F. and Silverstone, B. (1998) Okuns law,
cointegration and gap variables, Journal of Macroeconomics,
20, 62537.

1151

Baxter, M. and King, R. G. (1995) Measuring business cycles:


approximate band-pass filters for economic time series.
Working Paper 5022, National Bureau of Economic
Research.
Elmeskov, J. and MacFarlan, M. (1993) Unemployment persistence. OECD Economic Studies, 21, 6088.
Hodrick, R. J. and Prescott, E. C. (1997) Postwar US business
cycles: an empirical investigation, Journal of Money, Credit,
and Banking, 29, 116.
Holloway, T. M. (1989) An updated look at Okuns law, Social
Science Quarterly, 70, 497504.
Kaufman, R. T. (1988) An international comparison of Okuns
law, Journal of Comparative Economics, 17, 182203.
Knoester, A. (1986) Okuns law revisited, Weltwirtschaftliches
Archiv, 122, 65765.
Meghir, C., Ioannides, Y., and Pissarides, C. (1989) Female
participation and male unemployment duration in Greece,
European Economic Review, 33, 395406.
Moosa, I. A. (1997) A cross-country comparison of Okuns
coefficient, Journal of Comparative Economics, 24, 33556.
Motley, B. (1994) How soon will the US reach full employment?
an assessment based on Okuns law, Federal Reserve Bank of
San Francisco Economic Review, 3, 2639.
Okun, A. M. (1962) Potential GNP: its measurement and
significance,
In
American
Statistical
Association.
Proceedings of the Business and Economic Statistics Section.
Paraskevaides, P. (1993) Income Inequalities and Regional
Distribution of the Labour Force, Center of Planning and
Economic Research, Athens.
Prachowny, M. F. J. (1993) Okuns law: theoretical foundations
and revised estimates, The Review of Economics and
Statistics, 75, 3316.
Watts, M. and Mitchell, W. (1991) Alleged instability of the
Okuns law relationship in Australia: an empirical analysis,
Applied Economics, 23, 182938.
Zivot, E. and Andrews, D. W. K. (1992) Further evidence on the
great crash, the oil-price shock, and the unit root hypothesis,
Journal of Business and Economic Statistics, 10, 25170.

You might also like