See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/332625967

The Relationship between Real Output (Real GDP) and Unemployment Rate:
An Analysis of Okun's Law for Eurozone

Article  in  Sosyoekonomi · April 2019

DOI: 10.17233/sosyoekonomi.2019.02.12

CITATIONS READS

2 1,240

1 author:

Utku Altunöz
Sinop Üniversitesi
85 PUBLICATIONS   103 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

ominal Faiz Oranı-Genel Fiyat Düzeyi İlişkisi İle Türkiye’de Gibson Paradoksunun Geçerliligi Analizi Article The Application Of The Laffer Curve In The Economy Of Turkey
Enter the title of your project View project

Türkiye’de Asgari Ücret, Işsizlik Ve Enflasyon Ilişkisinin: Var Analizi ile Tahmini View project

All content following this page was uploaded by Utku Altunöz on 24 April 2019.

The user has requested enhancement of the downloaded file.

Sosyoekonomi
ISSN: 1305-5577
DOI: 10.17233/sosyoekonomi.2019.02.12
RESEARCH
Date Submitted: 07.10.2018
ARTICLE
Date Revised: 05.03.2019
2019, Vol. 27(40), 197-210 Date Accepted: 15.03.2019

The Relationship between Real Output (Real GDP) and

Unemployment Rate: An Analysis of Okun’s Law for Eurozone

Utku ALTUNÖZ (http://orcid.org/0000-0002-0232-3108), Department of Business Administration, Sinop

University, Turkey; e-mail: [email protected]

Reel Çıktı (Reel GSYİH) ve İşsizlik Oranı İlişkisi: Euro Bölgesi İçin Okun
Yasası Analizi

Abstract
In this study, the relationship between economic growth and unemployment in the euro area
(eurozone) is investigated in the light of Okun’s Law by using annual panel data set covering the period
2000-2012. Besides panel error correction and panel integration methods are used to test the link
between unemployment and growth rate, unemployment hysteresis was examined by using several
non-stationary panel unit root techniques for in all countries included to the study. Findings of this
study propose that the Okun’s Law is valid. However, the cointegration coefficient is lower than the
Okun’s coefficient calculated for the United States and other studies which was done for developed
countries. The US is included in the analysis as it has the strongest economy in the world. In addition,
since growth and unemployment are more volatile in developing countries, they are included in the
analysis in order to compare with developed countries.
Keywords : Okun’s Law, Unemployment, Economic Growth, Panel
Cointegration, Eurozone.
JEL Classification Codes : R11, R15.
Öz
Bu çalışmada Euro bölgesi için ekonomik büyüme ile işsizlik arasındaki ilişi Okun Yasası
doğrultusunda 2000-2012 yılları için yıllık panel veri analizi ile test edilecektir. Bu amaçla işsizlik
histerisi panel hata düzeltme ve panel bütünleşme ve çeşitli panel birim kök testleri yardımıyla testleri
kullanılmıştır. Bu çalışmanın bulguları, Okun Yasasının geçerli olduğunu önermektedir. Bununla
birlikte, eş bütünleşme katsayısı, Okun’un ABD ve diğer ülkeler için yapılan diğer çalışmalar için
hesaplanan katsayısından düşüktür. ABD dünyadaki en güçlü ekonomiye sahip olduğu için analize
dâhil edilmiştir. Ayrıca, büyüme ve işsizlik gelişmekte olan ülkelerde daha değişken olduğundan,
gelişmiş ülkelerle karşılaştırmak amacıyla analize dâhil edilmiştir.
Anahtar Sözcükler : Okun Yasası, İşsizlik, Ekonomik Büyüme, Panel Eşbütünleşme, Euro
Alanı.
 Altunöz, U. (2019), “The Relationship between Real Output (Real GDP) and Unemployment
Rate: An Analysis of Okun’s Law for Eurozone”, Sosyoekonomi, Vol. 27(40), 197-210.

1. Introduction

Arthur Okun the study of the relationship between real growth rate and
unemployment is known as Okun’s Law in the economic literature. According to Okun, real
GDP growth by 1% above the trend value of 2.25%, reduces the unemployment rate by
0.5%. In the relevant study it was stated that the values are for the US economy and around
1% of annual population growth rate. Okun (1962) made use of three different models to
prove this claim. These models are called first differences, trial exploits, and appropriate
tendency and flexibility. In the first difference method, changes in the unemployment rate
(Y) every 3 months in percentage and 3 months in GNP (X) were associated. It was estimated
from 55 observations between 1947:2Q - 1960:4Q. The result is the equation (1) below:

𝑌 = 0.30 − 0.30𝑋 (1)

As can be understood from the equation, an increase of 1% in GDP reduces the

unemployment level by 0.3 points. It comes into play when selecting and testing certain
exponential values of the potential adversary in the test deficit section. The corresponding
regression equation is established with the arrow. Equation of unemployment as equation
(2) below;

𝑈 =𝑎+𝑏 (2)

The results of models made by Okun were presented to the unified economic
committee in 1961. The presented model is like equation (3).

𝑈 = 3.72_0.36 (3)

Equation (3) means that unemployment is 3.72% in the absence of a current deficit.

Last section was named as the appropriate tendency and flexibility by Okun. While
the first part is based on GNP and the change in unemployment, the second part is based on
the level values of these variables on the assumption that the growth trend is constant
unemployment rate in the second method (Muratoğlu, 2011: 26).

One of the most urgent problems in Europe is unemployment that rose from 4% in
the 1960s to 10% in the 1990s. However, despite the social and economic consequences of
high unemployment rates, some additional problems remain unresolved such as reason that
which one causes the existence of unemployment rates, does the size of unemployment
follow an autonomous route independent from other macroeconomic variables such as
inflation? (Christopoulos, 2004: 2). The answers are relevant to the empirical examination
of the relationship between unemployment and output. Mentioned variables, that is,
unemployment and output are affected to by the business cycle, we expect a long-term
relationship between these variables. This context, high output growth rates will lead to a
decrease in unemployment. But we find ourselves: when output expansion is limited,
unemployment moves in the opposite direction. Theoretical proposals on output and
unemployment are called the “Okun’s Law” (Okun, 1962). On an empirical basis, the

198
 Altunöz, U. (2019), “The Relationship between Real Output (Real GDP) and Unemployment
Rate: An Analysis of Okun’s Law for Eurozone”, Sosyoekonomi, Vol. 27(40), 197-210.

“Okun’s Law” allows the implementation of a more appropriate economic policy to reduce
unemployment.

Most of the empirical literature examining the validity of Okun’s Law is presupposed
to be a symmetrical relationship between unemployment and real output growth. Okun’s
(1962) original work states that the economy experienced a 1% unit increase in
unemployment for every 3% unit decrease gross domestic product from its long-run level.
Similarly, a 3% unit increase in gross domestic product from its long-run level is associated
with a 1% unit decrease in unemployment. It is important to keep in mind that changes in
output are related to changes in the number of changes in labor force participation, to ensure
that the relationship between changes in production and change in unemployment is not
individual. Changes in per capita working hours and labor productivity. Okun estimated that
3% unit increase in gross domestic product from its long-run level corresponded to a 0.5%
point increase in the labor force participation rate, a 0.5% unit increase in the hours worked
per employee, and a 1% unit increase in labor productivity, leaving the remaining 1% unit
to be the change in the unemployment rate. During the Okun’s period, the relationship
between changes in unemployment and changes in output has been predicted to be about 2
to 1 rather than the 3 to 1 that Okun originally proposed. In addition, economists have
pointed out that the relationship between changes in unemployment and changes in output
is not perfect, and that the Okun’s Law will be considered as a rule in general, rather than
an absolute management principle.

2. Real Output (Real GDP) and Unemployment in Eurozone

At the beginning of 2000, more than 20.5 million people were unemployed in the
EU-28, accounting for 9.2% of the total. At that time the trend of unemployment was down.
In the second quarter of 2001, the unemployment rate decreased to 19.6 million and the
unemployment rate to 8.7% kforce.

The increase in unemployment followed a long time.

At the end of 2004, the number of job seekers reached 21.2 billion, while the
unemployment rate was 9.2%. At the beginning of 2005, a period of unemployment started
to take shape until the first quarter of 2008. In this period, after the economic crisis, the EU-
28 unemployment rate exceeded 16.2 million. The unemployment rate between the second
quarter of 2008 and the middle of 2010 has increased by more than 6.7 million and it has
increased to the highest value recorded since the beginning of the year in 2000 by 9.7%. The
decline in unemployment in the following three quarters was a deceptive sign of a steady
recovery in the end of the crisis and labor market conditions in the EU-28. As a matter of
fact, until the second quarter of 2011 and the second quarter of 2013, unemployment has
risen steadily, reaching a record high of 26.5 million, reaching a record high of 11%. Since
then, the rate has fallen to 8.2% in 2016. As can be seen in Chart 1, the eurozone
unemployment rate rose to 7.2% in 2008, when the euro entered the market until spring
2008. With the onset of financial turmoil and the sharp economic slowdown associated with

199
 Altunöz, U. (2019), “The Relationship between Real Output (Real GDP) and Unemployment
Rate: An Analysis of Okun’s Law for Eurozone”, Sosyoekonomi, Vol. 27(40), 197-210.

it, unemployment has begun. Markedly increased. As of 2009, the eurozone unemployment
rate was 9.6% - the highest rate recorded in ten years.

Chart: 1
Unemployment Rate of Eurozone (2008-2018)

Source: Eurostat.

As of September of 2017, unemployment in the Eurozone has hit its lowest level
since January 2009, new data have revealed. The proportion of workers who were
unemployed in the Eurozone fell to 8.9 in September 2017, according to the European Union
data bank Eurostat. This unemployment rate is better than the 9-percentage point recorded
in August and a percentage point improvement in September 2016. According to Eurostat,
17.000 million men and women in 28 territories in the EU region and 14.000 million in the
euro area (EA-19) were unemployed in January 2011. According to December 2017, the
number of unemployed fell by 19.000 in the EU-28 and 10.000 in the euro area. As of
January 2017, the unemployment rate fell by 1.867 million in the EU-28 and 1.429 million
in the euro area. While real output (real domestic product) is widely used to measure the
overall size of an economy. And, it is commonly used to compare living standards derived
from indicators such as per capita gross domestic product those adjusted for differences in
dollar, dollar, or price levels. As to be seen in Chart 2 below, the global economic and
financial crisis caused a serious stagnation in Japan, The United States and European Union
in 2009. This was followed by a recovery in 2010. The crisis was already apparent in 2008,
when the rate of increase declined significantly. GDP in EU-28 and this was followed by a
4.4% decline in real GDP in 2009. The recovery in the EU-28 increased the volume index
of GDP (based on chain-linked volumes) by 2.1% in 2010 and increased by 1.7% in 2011.
Subsequently, GDP contracted by 0.5% in real terms in 2012, reaching d rates in 2013
(0.2%), 2014 (1.6%) and 2015 (2.2%). -19), the rates of change up to 2010 in EU-28 are
very similar. With growth in 2011 (-1.5%) and in 2012 (-0.9%) and in 2013 (-0.3%). Real
GDP growth in the euro area in 2014 and 2015 was weaker than in the EU-28.

200
 Altunöz, U. (2019), “The Relationship between Real Output (Real GDP) and Unemployment
Rate: An Analysis of Okun’s Law for Eurozone”, Sosyoekonomi, Vol. 27(40), 197-210.

Chart: 2
Gross Domestic Product of Eurozone (2008-2017)

Source: Eurostat.

The Gross Domestic Product in the Euro Area increased by 2.7% year-on-year in the
fourth quarter. In 2017, the GDP growth was 2.5%, well below the previous 2.5% estimate
and same level with 1.86% in 2016. The GDP Annual Growth Rate in the Eurozone reached
1.68% from 1995 to 2017, the highest level ever.

201
 Altunöz, U. (2019), “The Relationship between Real Output (Real GDP) and Unemployment
Rate: An Analysis of Okun’s Law for Eurozone”, Sosyoekonomi, Vol. 27(40), 197-210.

Chart: 3
Real GDP, Inflation, and Unemployment of Eurozone (2000-2016)

Source: IMF, October 2017.

Chart 3 shows together the real and linear GDP annual change in %, inflation and
linear inflation rate, unemployment rate of Euro area. According to Chart 2, tendency of the
unemployment and real GDP don’t support the coefficients claimed by Okun’s Law. This
situation can be observed in detail in Table 2.

3. Literature

In this part of the study, domestic and foreign studies on the relationship between
unemployment and economic growth will be examined.

Zagler (2003) used France, Germany, Italy, and England data for his study of Okun’s
Law with the vector error correction model. In the long run, he found that there is a
cointegration relationship between economic growth and unemployment and that the
relationship is positive. Christopoulos (2004) tested the validity of Okun’s Law for Greece
during the period 1971-1993. Applied on Greece economy, panel analysis showed that
Okun’s law can be confirmed for six out of the 13 regions. In the study that Holmes and
Silverstone (2006) tested the Okun’s Law for US economy on the Markov model, it couldn’t
reach the presence of a symmetrical relationship between growth and unemployment among
variables. Malley and Molana (2008) have considered the dynamics of labor markets such
as “effective wage”, “unionization”, “wage contracts”, “unemployment insurance” in their

202
 Altunöz, U. (2019), “The Relationship between Real Output (Real GDP) and Unemployment
Rate: An Analysis of Okun’s Law for Eurozone”, Sosyoekonomi, Vol. 27(40), 197-210.

studies examining the relationship between growth and unemployment for the 7 developed
countries. As a result, they did not see that the relationship between the variables in Germany
was more pronounced and stable. Pierdzioch et al. (2009) tested the Okun’s law for G7
countries used data covering the period 1989-2007. Result of study supported the
consistency between Okun’s law which means the real output growth rate and professional
economist’s forecasts of changes in unemployment rate. Zonzilos (2000) investigated
Okun’s Law for Greece, over the period 1965-1999. His findings recommend that, if
aggregate output increases by 1%, then unemployment decreases by 0.28% under the ceteris
paribus assumption. Based on these conclusions, the writer claims that the unemployment
rate in Greece cannot be further reduced by adopting policies aimed at strengthening total
demand. Villaverde ve Maza (2009) in a study that examines the relationship between
growth and unemployment for the Spanish economy, they found that due to increase in
unemployment, growth slows down. Demirgül (2010) aimed to contribute to the empirical
literature about Okun’s law. For this purpose, Demirgül used the periods between 1989 and
2007 with quarterly data for Turkey. The results show that there is stronger evidence of the
ineffectiveness of the law of Okun, particularly during periods of abnormal productivity
growth. Barış et al. (2010), examined the validity of Okun’s Law in Turkey and asymmetric
relationship between unemployment and economic growth, with data 1988-2008 period
using Markov-Switching approach. In the results of working; they reached the existence of
the asymmetric structure for the relation between the variables depending on the expansion
and contraction periods of the economy. In the study that Demirgil (2010) has been identified
as Okun’s relationship cannot be achieved over the periods 1987-2007 in Turkey when
productivity increases. Lal et al. (2010), predicted the Okun’s coefficient for the Asian
countries by using Engle Granger (1987) co integration technique over the period 1980-
2006. According to empirical evidences of study showed that Okun’s law interpretation may
not be applicable and the principle of non-accelerating inflation rate of unemployment
(NAIRU) does not hold its validity in some Asian developing countries. Ball, Leigh, and
Loungani (2012) examined the fits short-run unemployment movement of Okun’s Law in in
twenty advanced economies since 1980 and in the USA since 1948. At the end of the work,
it is understood that Okun’s law is stable and strong in many countries which have not
changed greatly during the Great Recession. In the end of study, demonstrated that Okun’s
law is a stable and strong in the most countries, one that did not change substantially during
the Great Recession. Bankole and Fatai (2013) predicted the coefficient of Okun and
examined the validity of Okun’s law for Nigeria Economy over the periods 1980-2008. For
this purpose, Bankole and Fatai applied the Engle granger co-integration test and Fully
Modified OLS. The empirical evidences demonstrated that there is positive coefficient in
the Regression which means Okun’s law interpretation is not applicable for Nigeria
economy. Altunöz (2014) analyzed the validity of Okun’s Law for Turkish economy
covering the periods 2000-2014. In the end of study, Altunöz revealed that output effects
unemployment while unemployment is not effective on output.

4. Methodology

According to Okun (1970) there should be two different type of model specification
one of which is a gap model. The other one is the first difference model. The first difference

203
 Altunöz, U. (2019), “The Relationship between Real Output (Real GDP) and Unemployment
Rate: An Analysis of Okun’s Law for Eurozone”, Sosyoekonomi, Vol. 27(40), 197-210.

model is described as, the relationship between the log of unemployment rate ( 𝑢𝑛𝑡 ) and log
of output (𝑔𝑡 ) is analyzed in the following function frame (Villaverdve & Maza, 2009: 291):

𝑔𝑡 − 𝑔𝑡−1 = 𝛼 + 𝛽 (𝑢𝑛𝑡 − 𝑢𝑛𝑡−1 ) + 𝜇𝑡 (5)

In equation (5), 𝛼 denotes the intercept and 𝛽 denotes the Okun’s coefficient that
measuring the change effect in the unemployment rate on real gross domestic product. Also,
random walking term denotes by 𝜇𝑡 .

The second specification model called gap model is obtained by using equation (5)
as below:

𝑔𝑡 − 𝑔𝑡∗ = 𝛼 + 𝛽 (𝑢𝑛𝑡 − 𝑢𝑛𝑡∗ ) + 𝜀𝑡 (6)

In equation (6), 𝑔𝑡∗ shows the natural rate of unemployment and the logarithm of
potential gross domestic product (output). Except the two of mentioned, all variables in
equation (6) are the same. The first term of the left-hand side in equation (6) denote the gap
of employment which is equal the difference between observed and potential unemployment
rates. However, the output gap which is equal to the difference between the observed and
potential output (real gross domestic product) is captured by 𝑔𝑡 − 𝑔𝑡∗ . For this analysis,
stationarity and cointegration conditions should be provided. In this paper, the first
difference model out of two specifications for Okun’s Law will be estimated.

On the other hand, the hypothesis of unemployment hysteresis shows that cyclical
fluctuations will permanently affect the rate of unemployment due to labor market rigidities.
In case of unemployment rate series has unit rate, we conclude that it carries a permanent
effect in it.

On this paper hysteresis hypothesis which is support the unemployment rates are non-
stationary, will be tested by focusing on ADF Panel unit root test as equation (7) below:

∆𝑢𝑛𝑖𝑡 = 𝛽0𝑡 + 𝛽1𝑡 𝑢𝑛𝑖𝑡−1 + 𝛽2𝑡 𝑡 + ∑𝑚

𝑗=1 𝛽3𝑡 ∆𝑢𝑛𝑖𝑡−𝑗 + 𝜔𝑖𝑡 (7)

By using different null and alternative hypothesis, several panel unit root tests
constructed. For analyzing the stationary, two of them will be consider. These are LLC
(Levin, et al., 2002), IPS (Im, et al., 2003). The alternative and null hypothesis of these unit
root tests for equation (7) are as follow:

Table: 1
Null and Alternative Hypothesis of Applied Tests
Unit Root Test Null Hypothesis Alternative Hypothesis
LLC 𝛽1𝑖 = 0 𝛽1𝑖 < 0 𝑇ü𝑚 𝑖 𝑙𝑒𝑟 𝑖ç𝑖𝑛
IPS 𝛽1𝑖 = 0 𝛽1𝑖 < 0 𝐵𝑎𝑧𝚤 𝑖 𝑙𝑒𝑟 𝑖ç𝑖𝑛

In this analysis, Doğru’s Study (2013) will be considered as a reference. co-

integration analyses and vector error correction (VEC) techniques will be preferred in this
study to determine the relationship between unemployment and real gross domestic product.

204
 Altunöz, U. (2019), “The Relationship between Real Output (Real GDP) and Unemployment
Rate: An Analysis of Okun’s Law for Eurozone”, Sosyoekonomi, Vol. 27(40), 197-210.

Many recent studies indicate that a more contemporary approach to I(1) series are
establishing a VEC model.

In vector error correction analysis, the first difference of each variables means as a
function of its lagged values, the lagged values of the other explanatory variables, and
cointegrating equation.

According to Doğru (2013) a simple vector error correction model with two variables
using this paper 𝑔𝑡 and 𝑢𝑛𝑡 besides lag length of 1 can be described as equation (8) below:

∆𝑔𝑡 = ∑𝑘𝑖=1 𝛼1𝑖 ∆𝑔𝑡−1 + ∑𝑘𝑖=1 𝛽1𝑖 ∆𝑢𝑛𝑡−1 + 𝜃11 ( ~𝜖1𝑡−1 ) + 𝜀𝑖𝑡 (8)

∆𝑢𝑛𝑡 = ∑𝑘𝑖=1 𝛼1𝑖 ∆𝑢𝑛𝑡−1 + ∑𝑘𝑖=1 𝛽2𝑖 ∆𝑢𝑛𝑡−1 + 𝜃21 ( ~𝜖1𝑡−1 ) + 𝜀2𝑡 (9)

In equation (8) and (9), t represents the time, the first difference of the logarithm of
the real output and the first difference of the logarithm of unemployment rate are respectively
~
represented by ∆𝑔𝑡 and ∆𝑢𝑛𝑡 . 𝜀1𝑡−1 = 𝑔𝑡−1 − 𝑦1 − ∅1 𝑢𝑛𝑡 means the 1 year lagged
disequilibrium residual. Estimated parameters are showed α, Ø, ƴ, ϕ and error correction
term is showed with 𝜀.

In this model, it is expected that ∅1 should be negative which means an increase in

unemployment related to decrease in real output.

In this part of the study, panel data analysis will be applied to the variables over the
period 2001-2017. For this purpose, 17 countries which are use Euro as a common currency
including the Netherlands, Spain, Slovakia, France, Slovenia, Portugal, Greece, Cyprus,
Luxembourg, Finland, Germany, Ireland, Italy, Estonia, Austria, Malta and Belgium, will
be included to the analysis. All data were obtained from IMF and World Bank.

Table: 2
Result of Panel Unit Root Test Results
At Level At First Difference
Variables Test LLC IPS LLC IPS
intercept -0.772 -0.312 -3288* -8041*
𝑢𝑛𝑖𝑡
intercept and trend 1.213 0.312 -7212** -9354*
intercept 1.615 -4.311 8131* -6423**
𝑔𝑖𝑡
intercept and trend 16.218 1.311 -12511* -31890**
Note: *and ** show the significance at the 10% level and at the 5% level respectively.

Before the individual, all variables were applied the panel unit root test as seen in
Table 2. According to panel unit root test, 𝐻0 cannot be rejected at 0.01%, 0.05% and 0.10%
significance in level for both unemployment and GDP panel data sets; nevertheless, it is
rejected for the first differences. Result means that both series are non-stationary; they are
I(1). Following the panel unit, individual intercept and trend and individual intercept and for
pooled un and g variables, Philips Peron (PP) and Augmented Dickey Fuller (ADF) unit root
tests are implemented for each country as individually. Akaike information criterion (AIC)
was considered for choosing the lag orders.

205
 Altunöz, U. (2019), “The Relationship between Real Output (Real GDP) and Unemployment
Rate: An Analysis of Okun’s Law for Eurozone”, Sosyoekonomi, Vol. 27(40), 197-210.

Table: 3
ADF and PP Individual Unit Root Test Results
ADF Unit Root Test Philips Perron Unit Root Test
Countries Unemployment Real Gross Domestic Product Unemployment Real Gross Domestic Product
At level First Difference At level First Difference At level First Difference At level First Difference
Netherlands 1.211 0.890** 1.312 0.087 1.451 0.899 2.901 0.011
Spain 1.720 0.097** 1.233 0.081 1.980 0.798 2.108 0.790
Slovakia 1.088 0.415** 1.532 0.012 2.767 0.556 1.987 0.209
France 0.981** - 1.089 0.413 3.019 0.376 1.078 0.014
Slovenia 0.977** - 0.076** - 0.091** - 0.780** -
Portugal 1.111 0.989** 0.776** - 0.057** - 0.060** -
Greece 1.431 0.652** 1.342 0.415 0.118 0.409 1.312 0.056
Cyprus 1.113 0.246** 1.665 0.331 2.109 0.899 1.111 0.866
Luxembourg 1.561 0.902** 0.071** - 2.190 0.312 0.067** -
Finland 1.123 0.211** 1.690 0.120 1.709 0.119 3.907 0.412
Germany 1.087 0.245** 1.809 0.149 1.590 0.856 1.120 0.221
Ireland 1.111 0.145** 1.765 0.217 1.760 0.908 2.901 0.512
Italy 1.098 0.456** 0.989 0.165 1.778 0.164 1.176 0.019
Estonia 1.761 0.667** 0.456** - 0.870** - 1.412 0.710
Austria 1.831 0.136** 1.443 0.133 1.809 0.243 1.117 0.030
Malta 0.897** - 1.338 0.190 1.567 0.768 1.087 0.121
Belgium 1.109 0.280** 1.578 0.114 1.412 0.990 4.901 0.118
Note: ** denotes the significance at the 5% level.

According to result of individual ADF unit root analysis seen in Table 3 the
unemployment rate variable for all countries except France, Slovenia and Malta are
stationary at first difference at least 10% significance level, while real output variable is
stationary at first differences for all countries but at level for Slovenia, Portugal,
Luxembourg and Estonia. Similarly, unemployment variable results of PP unit root test show
that all variables except Slovenia, Portugal and Estonia are stationary at first difference for
10% significance level. According to gross domestic product variable result of PP unit root
test, countries are stationary at first difference, but Slovenia, Portugal and Luxembourg are
at level. One of the most important and known cointegration tests in econometric literature
are these: McCoskey and Kao (1998), Kao (1999), Pedroni (1999), Larsson, Mark and Sul
(2003), Westerlund and Edgerton (2005) and Gutierrez (2005).

In this study, Pedroni (1999) tests are used to determine the validity of long run
relationship between unemployment rate and output in the pooled panel. Results is seen from
Table 4 below:

Table: 4
Result of Panel Cointegration Test
Within Dimension Tests Intercept Intercept and Trend
𝜐 statistic of panel 0.33 -0.87
rho(p) statistic of panel 1.57 -0.09**
PP Statistic of panel -5.41** -9.145**
ADF Statistic of panel -2.67* -5.89*
Between-dimension test
rho (p) statistic of Group 2.98 2.78
PP statistic of Group 3.12* -7.46**
ADF Statistic of Group 2.18* -3.12**
Note: *and ** show the significance at the 10% level and at the 5% level respectively.

Result of the pooled Panel Cointegration Test except group p-test that is in order to
determine of the validity of long run relationship between output and unemployment rate,
provides the evidence on the steady-state equilibrium which means there is a causal

206
 Altunöz, U. (2019), “The Relationship between Real Output (Real GDP) and Unemployment
Rate: An Analysis of Okun’s Law for Eurozone”, Sosyoekonomi, Vol. 27(40), 197-210.

interaction between unemployment and output. In the long-run for considered variables.
Following the panel cointegration test, panel vector error correction model should be
estimated to determine the causal interaction among variables. For this purpose, vector error
correction (VEC) model will be estimated by considering equation (8) and equation (9).

Table: 5
VECM estimation result of Panel Long and Short Run Estimation
Short Run Causality Lon-run Causality
∆𝒖𝒏 ∆𝝈 ERM (Error Correction Term)
∆𝒖𝒏 - 0.841(0.561) -0.65*
∆𝝈 0.238(0.066) - -4.12
Cointegration equation for panel: 𝑔𝑖𝑡 = −0.718𝑢𝑛𝑖𝑡 + 𝜇1𝑡
Note: *and ** show the significance at the 10% level.

The coefficient of the error correction term shows the speed of adjustment to the long-
term equilibrium. As can be seen in Table 5, the coefficient of the error correction term is -
0.65 which means that previous period disequilibrium is corrected at a speed of 65% by
system. Also, according to results in Table 5, in case of 1% increase in unemployment
expected reduction in real gross domestic product (realoutput) will reach to the 0.71% in the
long run term. In addition, results in Table 5 shows that there is a short run causality from
real gross domestic products to unemployment rate and it lasts to the long run. For long run
term, unemployment is the cause of real gross domestic products but there are no
relationships from unemployment to output. to make the results of the study more reliable,
coefficient of Okun’s Law will be estimated using fully modified ordinary least square and
Dynamic Ordinary Least Square methods which were developed by Pedroni (2001).
Coefficient of Okun’s Law by fully modified ordinary least square and Dynamic Ordinary
Least Square methods is presented Table 6 below:

Table: 6
Result of Panel Cointegration
𝑢𝑛 𝑔
DOLS (Panel) -0.038** -0.09**
FMOLS (Panel) -0.042** -0.032

Table 6 shows the coefficient of cointegration relationship for long run. According
to result of Panel cointegration in Table 6, both Fully modified least square and dynamic
ordinary least square, unemployment rate is negative related to real gross domestic product
(real output). Results in Table 6 means that 1% decrease in the real output (real gross
domestic product) causes to rise 0.03 unemployment rate besides, 1% increase in real gross
domestic product causes to decrease unemployment rate 0.09%.

According to Fully modified least square estimations, 1% decrease in real gross

domestic product causes to rise unemployment rate 0.03% and 1% rise in unemployment
rate causes to fall 0.042% in real gross domestic product.

207
 Altunöz, U. (2019), “The Relationship between Real Output (Real GDP) and Unemployment
Rate: An Analysis of Okun’s Law for Eurozone”, Sosyoekonomi, Vol. 27(40), 197-210.

5. Conclusion

In this study, validity of the Okun’s Law, which means the relationship between real
output (real gross domestic product and unemployment rate, was investigated for Eurozone
economy. For this purpose, Netherlands, Spain, Slovakia, France, Slovenia, Portugal,
Greece, Cyprus, Luxembourg, Finland, Germany, Ireland, Italy, Estonia, Austria, Malta and
Belgium were included to the analysis. Before the econometric analysis, the theoretical
infrastructure of Okun’s Law has been mentioned. Immediately after, unemployment and
real growth figures for the European Union have been analyzed comparatively. After the
literature survey, Okun’s Law was investigated for Eurozone by 2 different equations.

By using 2 different equations, panel integration methods and panel error correction
were used to test the link between growth and unemployment rate. And, it was investigated
the unemployment hysteresis using several non-stationary panel unit roots techniques for
unemployment hysteresis in the whole euro zone. The results of the study suggest that there
is evidence supporting the unemployment hysteria in many European countries and that
panel cointegration tests support the relationship between the actual output and the
unemployment rate in the long term. Also, according to results, in case of 1% increase in
unemployment expected decrement in real output will reach to the 0.71% in the long run
term which means that Okun’s Law is equal to -0. 71 pooled countries. In the other words,
in case of 1% increase in unemployment expected decrement in real output will reach to the
0.71% in the long run term. Results of Coefficient of Okun’s Law by fully modified ordinary
least square and Dynamic Ordinary Least Square methods are as a parallel of VECM
estimation result of Panel Long and Short Run Estimation that unemployment rate is
negative related to real gross domestic product (real output). In this study, the long-term
coefficient calculated from the panel data sets is lower than the coefficient obtained by Okun
for America.

References
Altunöz, U. (2015), “Reel Büyüme ve İşsizlik Bağlamında Türkiye İçin Okun Yasası”, Kamu-İş İş
Hukuku ve İktisat Dergisi, 1(14), 47-67.
Bankole, A.S. & B.O. Fatai (2013), “Empirical Test of Okun’s Law in Nigeria”, International
Journal of Economic Practices and Theories, 3(3), 227-231.
Barışık, S. & E.İ. Çevik & N. Kırcı-Çevik (2010), “Türkiye’de Okun Yasası, Asimetri İlişkisi ve
İstihdam Yaratmayan Büyüme: Markov-Switching Yaklaşımı”, Maliye Dergisi, 159, 88-
102.
Breitung, J. (2001), “The local power of some unit root tests for panel data”, in: B.H. Baltagi & T.B.
Fomby & R.C. Hill (eds.), Non-stationary Panels, Panel Cointegration, and Dynamic
Panels (Advances in Econometrics), Vol. 15, United Kingdom: Emerald Group
Publishing Limited, 161-177.
Christopoulos, K.D. (2004), “The relationship between output and unemployment: Evidence from
Greek regions”, Papers in Regional Science, 83, 611-620.
Demirgil, H. (2010), Okun Yasasının Türkiye için Geçerliliğine dair Ampirik bir Çalışma”, Alanya
İşletme Fakültesi Dergisi, 2(2), 139-151.

208
 Altunöz, U. (2019), “The Relationship between Real Output (Real GDP) and Unemployment
Rate: An Analysis of Okun’s Law for Eurozone”, Sosyoekonomi, Vol. 27(40), 197-210.

Doğru, B. (2013), “The link between unemployment rate and real output in Eurozone: A panel error
correction approach”, 9th International Strategic Management Conference, 99, 94-103.
EUROSTAT (online), Unemployment Statistics, <http://ec.europa.eu/eurostat/statistics-
explained/index.php/Unemployment_statistics>, 12.3.2018.
Holmes M.J. & B. Silverstone (2006), “Okun’s Law, Asymmetries and Jobless Recoveries in United
States: A Markov-Switching Approach”, Economics Letters, 92(2), 293-299.
Im, K.S. & M.H. Pesaran & Y. Shin (2003), “Testing for Unit Roots in Heterogeneous Panels”,
(revise version of 1997’s work), Journal of Econometrics, 115(1), 53-74.
Irfan, L. et al. (2010), “Test of Okun’s Law in some Asian Countries: Cointegration Approach”,
European Journal of Scientific Research, 40(1), 73-80.
Kao, C. (1999), “Spurious Regression and Residual-Based Tests for Cointegration in Panel Data”,
Journal of Econometrics, 90(1), 1-44.
Larsson, R. & L. Johan & L. Mickael (2001), “Likelihood-Based Cointegration Tests in
Heterogeneous Panels”, Econometrics Journal, 4, 109-142.
Laurence, B. & L. Daniel & L. Prakash (2012), “Okun’s Law: Fit at 50?”, Economics Working Paper
Archive 606, The Johns Hopkins University, Department of Economics.
Levin, A. & C.F. Lin & C.S.J. Chu (2002), “Unit Root Test in Panel Data: Asymptotic and Finite
Sample Propertiesî”, Journal of Econometrics, 108, 1-24.
Luciano, G. (2005), “Tests for cointegration in panels with regime shifts”, Econometrics 0505007,
University Library of Munich, Germany.
Malley, J. & H. Molana (2008), “Output, Unemployment and Okun’s Law: Some Evidence from
G7”, Economics Letters, 101, 113-115.
Mc Coskey, S.K. & K. Chihwa (1998), “A Residual-Based Test of the Null of Cointegration in Panel
Data”, Econometric Reviews, 17(1), 57-84.
Muratoğlu, Y. (2011), “Ekonomik Büyüme ve İşsizlik Arasındaki Asimetrik İlişki ve Türkiye’de
Okun Yasasının Sınanması”, Yayımlanmamış Yüksek Lisans Tezi, Hitit Üniversitesi,
Sosyal Bilimler Enstitüsü.
Okun, A.M. (1962), “Potential GNP: Its Measurement and Significance”, Cowles Foundation Paper
190, 1-7.
Pedroni, P. (1999), “Critical Values for Cointegration Tests in Heterogeneous Panels with Multiple
Regressors”, Oxford Bulletin of Economics and Statistics, 61, 653-670.
Pedroni, P. (2001), “Fully modified OLS for heterogeneous cointegrated panels”, Advances in
Econometrics, 93-130.
Pierdzioch, C. & J-C Rulke & G. Stadtmann (2009), “Do professional economists’ forecasts reflect
Okun’s law? Some evidence for the G7 countries”, Applied Economics, 43(11), 1365-
1373.
Villaverde, J. & A. Maza (2009), “The robustness of Okun’s law in Spain, 1980-2004 Regional
evidence”, Journal of Policy Modeling, 31(2), 289-297.
Westerlund, J. & E. David (2005), “Panel Cointegration Tests with Deterministic Trends and
Structural Breaks”, Lund University Working Papers, No. 42.
Zagler, M. (2003), “A Vector Error Correction Model of Economic Growth and Unemployment in
Major European Countries and An Analysis of Okun’s Law”, Applied Econometrics and
International Development, 3(3), 93-118.

209
 Altunöz, U. (2019), “The Relationship between Real Output (Real GDP) and Unemployment
Rate: An Analysis of Okun’s Law for Eurozone”, Sosyoekonomi, Vol. 27(40), 197-210.

Zonzilos, N.G. (2000), “The Phillips curve of the Greek economy and the time varying NAIRU”,
Bank of Greece Economic Bulletin, Vol. 15.

210

View publication stats