- Email: [email protected]
Budget Deficits Sustainable? An Empirical Analysis for OECD Countries
Available online at www.sciencedirect.com
ScienceDirect Procedia - Social and Behavioral Sciences 131 (2014) 258 – 263
WCETR 2013
Budget Deficits Sustainable? An Empirical Analysis For OECD Countries Mehmet Mercana* a
Assist. Prof. Dr., Hakkari University, FEAS, Department of Economics,Hakkari, TURKEY
Abstract In this study, budget deficit sustainability in OECD countries, are investigated via intertemporal budget constraint approach in 1980-2012 periods and quarterly data used. In analysis firstly, Cross Sectionally Dependency (CD) in country was examined CDLM test (Cross Sectionally Dependency Lagrange Multiplier) developed by Pesaran (2004). Stationary of series was searched CADF test (Cross-Sectionally Augmented Dickey-Fuller) developed by Pesaran (2006) considering CD. Cointegration relationship among series was examined panel cointegration test with multiple structural breaks developed by Basher and Westerlund (2009). As a result of analysis, cointegration relationship between the series was determined. According to the longrun analysis, budget deficits of these countries are sustainable in weak form.
© 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license © 2014 The Authors. Published by Elsevier Ltd. (http://creativecommons.org/licenses/by-nc-nd/3.0/). Selection and peer-review under responsibility of the Organizing Committee of WCETR 2013. Selection and peer-review under responsibility of the Organizing Committee of WCETR 2013.
Keywords: Budget Deficit, Sustainability, Cross-Sectional Dependency, Panel Cointegration. Jel Codes: H61, H62, H63 Introduction Implementation of Keynesian policies predicting the intervention of the government to the system especially in crisis period caused negative effects on budget balance of many countries. Countries chose to go into debt in order to meet the budget deficit and the insolvable debts in maturity periods generally were financed by new debts. This became a vicious cycle of debt-interest and it affected several macro economical variables. Despite Maastrich contract anticipating that budget deficit should be 3% most, the budget deficits in many European Union and OECD countries having the crisis recently exceeded 3% limit. For instance, in 2010 budget deficits in Ireland is 32.4%, in the USA 10.7%, in Greece 10.4%, in England 10.3%, in Spain 9.3%, in Portugal
* Corresponding Author: Mehmet Mercan Tel.: +90-505-386-6949 E-mail address:[email protected], [email protected]
1877-0428 © 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/3.0/). Selection and peer-review under responsibility of the Organizing Committee of WCETR 2013. doi:10.1016/j.sbspro.2014.04.114
Mehmet Mercan / Procedia - Social and Behavioral Sciences 131 (2014) 258 – 263
9.2%, in Japan 8.2% and in most of OECD countries 7.7% (OECD, 2013). In Europe, indebtedness reached a very high state; it passed over acceptable limits and became serious problem and threat to global economy (Staněk, 2004). When a budget deficit continues to grow over time, governments must pay attention to its potential effects on the economy and they should take necessary steps (Yoon, 2012). Hakkio and Rush (1991), for the US economy, relationship between government expenditures and government revenues examined by using 1950:Q2-1988:Q4 period data. The cointegration relationship between the series tested whether the cointegration coefficient is equal to one. If the parameter is equal to one, the budget deficits are sustainable, while smaller than one is considered to be unsustainable budget deficits in the long run. Quintos (1995) expanded these conditions. If the coefficient of the budget expenditure is equal to one, the budget deficits sustainability is considered strong. If it is between zero and one, sustainability is in a weak form. 2. Analysis In this study, 18OECD member countries†, whose ratio of budget deficit to GDP has been greater than 3% in 2012 year, quarterly data of INC (General government revenue Percent of GDP) and EXP (General government total expenditure Percent of GDP) for 1996-2012 period annual data have been used. Data was taken from the IFS (International Financial Statistics) data base. Gauss 9.0 package program was used in this study. 2.1. Testing the cross-sectional dependency Before proceeding with further steps, cross-section dependence must be tested. Otherwise, results may be biased and inconsistent (Breusch and Pagan, 1980; Pesaran, 2004). Therefore, prior to further analyses, the existence of cross-section dependency in the series and the cointegration equation should be tested. The existence of a cross-section dependency among countries is tested via the Breusch-Pagan (1980) LM test when time dimension is greater than the cross-section dimension. Pesaran (2004) improved this test for time dimension is smaller than the cross-section dimension and time dimension is greater than the cross-section dimension. This test is biased when the average group is zero, but the average individual is different from zero. Pesaran et al. (2008) adjusted this deviation by adding the variance and the average to the test statistics. Therefore, it is called the bias-adjusted LM test (LMadj). The first form of LMadjtest statistics is as the following:
Where represents the avarage, represents the variance. The test statistics to be obtained here show a standard normal distribution as asymptotic (Pesaran, et al. 2008). The null hypothesis of the LMadj test is no crosssection dependency. The LMadj test was used and obtained results are presented in Table 1. Table 1: Crosssectional Dependency (LMadj) Test Results Test Prob. Statistics Value INC 4.016 0.000 EXP 2.051 0.020 Cointegration Equation 44.56 0.000 Note:p-values were computed 1000 bootstrap replications. As can be seen from Table 1, since the probability values of series and cointegration equation are smaller than Variables
†
Belgium, Canada, Czech Republic, Denmark, France, Greece, Ireland, Israel, Japan, Netherlands, New Zeland, Poland, Portugal, Slovak Republic, Slovenia, Spain, United Kingdom and United States.
259
260
Mehmet Mercan / Procedia - Social and Behavioral Sciences 131 (2014) 258 – 263
0.05, H0 hypotheses are strongly rejected and it has been decided that there is cross-sectional dependency among these countries. This reveals to a significant change in the series in one of the countries also affects the others. Therefore, while the decision makers in these countries set their policies, should take into consideration to policies of the other countries and the other external factors. Furthermore, since cross-section dependency determined, while choosing the unit root and cointegration tests method, this situation should be taken into account. Therefore, panel unit root tests and cointegration analysis considering the cross-section dependency have been also used in the analysis. 2.2. Panel unit root test The panel unit root tests considering the information about both the time and the cross-section dimension of the data are accepted to be statistically stronger than the time series unit root tests considering the information only about the time dimension (Maddala and Wu, 1999; Levin, Lin and Chu, 2002; Beyaert and Camacho, 2008) because the variability in the data increases with the addition of the cross-section dimension to the analysis. The first problem in the panel unit root test is whether or not the cross-sections forming the panel are independent to each other. Panel unit root tests here are divided into two as first and second generation tests. First generation tests are Levin, Lin and Chu (2002), Breitung (2005), Hadri (2000), Im, Pesaran and Shin (2003), Maddala and Wu (1999) and Choi (2001). First generation unit root tests are based on the hypothesis that the cross-section units forming the panel are independent and all the cross-section units are equally affected by the impact to one of the units in the panel. However, it is a more realistic approach that units are differently affected from the impact to one of the crosssectional units in the panel if it is thought that national economies are related to each other today. In order to overcome this deficiency, second generation unit root tests carrying out the unit root analysis considering the crosssection dependence between the cross-section units have been developed. Main second generation unit root tests are MADF (Taylor and Sarno, 1998), SUDARF (Breuer, Mcknown and Wallace, 2002) and CADF (Pesaran, 2006). In this study, since it has not been identified any cross-section dependency between the countries in the panel for the INC and EXP variables used in the study, stationary of the series has been analyzed with one of the second generation unit root test CADF test developed by Peseran (2006). Through CADF unit root test can be performed in each crossection unit in the series forming the panel. So the stationary of the series can also be estimated one by one for the panel’s overall and each cross-section. CADF test hypothesing that every country is affected differently from time effects and considering the spatial autocorrelation is used in T>N and N>T situations. Stationary for each country is tested by comparing the statistics values of this test with Peseran’s CADF critical table values. If CADF critical table value is greater than CADF statistics value, the null hypothesis is rejected and it is found out that the series of only that country is stationary. CADF test statistics is estimated as the following:
(2)
(3) Here shows unobservable common effects of each country, shows individual-specific error. Equations (2), (3) and unit root hypotheses can be written as the following: H0: for all i H1: i=1, 2, . . . . , N1, i=N1+1, N1+2,…, N.
(4)
(Series is non stationary.) (Series is stationary.)
Test statistics and critical values have been computed for each country and panel (overall). Results are presented in Table 2. Table 2: CADF Unit Root Test Results Variables
Level
INC
-0.831
First Difference -1.903
Critical Value -1.840
Mehmet Mercan / Procedia - Social and Behavioral Sciences 131 (2014) 258 – 263
EXP -1.282 -1.841 -1.840 Note: Without trend model for INC and EXP series and %10 significant level have been selected as a test model. Gauss 9.0 package program was used in the analysis. Results in Table 2 show that series are non-stationary at levels but become stationary at first differences; they are said to be integrated of first order, I(1). In this case, it has been concluded that the existence of cointegration relationship between these series can be tested since series under consideration are integrated of the same order. 2.3. Cointegration analysis The existence of the cointegration relationship between the series has been investigated through the panel cointegration test developed by Westerlund (2008), which considers the cross-section dependency. The null hypothesis denotes no cointegrating relationship. When the probability value of the calculated test is smaller than 0.05, H0is rejected and it is decided that there are cointegration relationship between the series. Cointegration test results were presented in Table 3. Table 3: Cointegration analysis results
Durbin-H Group
Cointegtarionstat 4.331
Durbin-H Panel
2.437
Prob. 0.000 0.007
Decision Cointegratio n Cointegratio n
2.4. Estimation of long term coefficients In this part of the study, the long run individual cointegration coefficients will be estimated with the Common Correlated Effects (CCE) method which is developed by Pesaran (2006). CCE is an estimator that can generate results providing consistent and asymptotic normal distribution when the time dimension is both greater and smaller than the cross-section dimension and that can separately calculate the long term cointegration coefficients for the cross-section units (Pesaran, 2006). Long term cointegration coefficients of panel were calculated with the Common Correlated Effects Mean Group (CCEMG) method. CCE and CCEMG estimations that have been carried out using equation (5) and results are presented in Table 4. Table 4: Long term coefficients Countries Belgium Canada Czech R. Denmark France Greece Ireland Israel Japan Panel
Coefficient -0.342 [-1.15] -0.259 [-1.26] 0.11[1.46]*** 0.32[2.60]* 0.19[0.64] -0.35[1.72]** 0.50[3.17]* 0.57[6.61]* 1.91 [4.47]* 0.28[2.99]*
Countries Netherland New Zeland Poland Portugal Slovak Rep.
Coefficient 0.46[1.62]** 0.28 [2.48]* -0.07[-0.53] 0.89[5.44]* 0.64[1.93]**
Slovenia
0.48[6.70]*
Spain U. Kingdom U. States
0.43[1.38]*** 0.30[1.20] 0.26 [1.15]
261
262
Mehmet Mercan / Procedia - Social and Behavioral Sciences 131 (2014) 258 – 263
Note: Autocorrelation and heteroscedasticity problems were adjusted with the Newey-West.[ ]; shows t statistics. *, **, ***; Indicates significance level in 1%, 5% and 10% respectively. According to Table 4, the long-run cointegration coefficients are smaller than one. According to Hakkio and Rush (1991) and Quintos (1995), the budget deficits in these countries are sustainable in weak form. 3. Results and policy implications In this study, the sustainability of budget deficit has been examined for 18 OECD countries which their budget deficit exceed 3% of the GDP in the year 2012 by means of panel cointegration method under cross-section dependence for the period of 1996-2012 annual data. It is found that in these countries budget deficits are weakly sustainable according to Hakio and Rush (1991) and Quintos (1995) for the cointegration coefficient is smaller than one. According to macroeconomic theory, in order to sustain high budget deficits some policies should be implemented in undesirable occasions such as scarcity of investment, lack of consumption, foreign high borrowing and low private savings. These economic precautions should be implemented even they are not welcomed in some parts of the society as in the sample of Greece. Since it is found that the sustainability of budget deficits is low as a result of the analysis, OECD and EU economy management should take the necessary precautions, alert the related countries and implement the necessary sanctions. In this context, it will be helpful to implement the fiscal rule in countries and to establish the independent supervisory and regulatory institutions. References Basher, S. A. & Westerlund, J. (2009). Panel cointegration and the monetary exchange rate model, Economic Modelling, 26, 506-513. Breuer, B., Mcnown, R. & Wallace M. (2002). Series-Specific Unit Root Test with Panel Data, Oxford Bulletin of Economics and Statistics 64(5): 527-546. Breusch, T.S & Pagan A.R. (1980). The Lagrange Multiplier Test and Its Applications to Model Specification Tests in Econometrics, Review of Economic Studies, 47: 239-53. Breitung, J. (2005). A Parametric Approach to the Estimation of Cointegrating Vectors in Panel Data, Econometric Reviews, 24(2): 151 173. Choi, I. (2001). Unit Roots Tests for Panel Data, Journal of International Money and Finance 20: 229-272. Hadri, K. (2000). Testing for Stationarity in Heterogenous Panels, Econometrics Journal, 3: 148-161. Hakkio, C. S. & Rush, M. (1991). Is the Budget Deficit Too Large? Economic Inquiry, pp. 429-445. Im, K., Pesaran H. & Shin Y. (2003). Testing for Unit Roots in Heterogenous Panels, Journal of Econometrics, 115(1): 53-74. Levin, A., Lin C.F. & Chu C.S.J. (2002). Unit Root Tests in Panel Data: Asymptotic and Finite Sample Properties, Journal of Econometrics. Maddala, G.S. & Wu, S. (1999). A Comparative Study of Unit Root Tests with Panel Data and a New Simple Test, Oxford Bulletin of Economics and Statistics, 61: 631-652. Pesaran, M. H. (2004). General Diagnostic Tests for Cross Section Dependence in Panels, Cambridge Working Papers in Economics, 435. Pesaran, H. (2006). A Simple Panel Unit Root Test in the Presence of Cross Section Dependency, Cambridge Working Papers in Economics, 0346. Quintos, C. E. (1995). Sustainability of the Deficit Process with Structural Shifts, Journal of Business and Economic Statistics, 13: 409–417. Staněk, P. (2004). Indebtedness as Immanent Feature of Current Economies, Ekonomicky Casopis, 52(3): 255-269. Taylor, M. & Sarno, L. (1998). The Behaviour of Real Exchange Rates during the Post-Bretton Woods Period, Journal of International Economics, 46: 281-312. Yoon, G. (2012). Explosive U.S. Budget Deficit, Economic Modelling, Pesaran, M. H., Ullah, A. & Yamagata, T., (2008). A Bias-Adjusted LM Test of Error Cross-Section Independence, Econometrics Journal, 11(1), 105-127. Westerlund, J. (2008). Panel Cointegration Tests of the Fisher Effect, Journal of Applied Econometrics, 23, 193 233.
Mehmet Mercan / Procedia - Social and Behavioral Sciences 131 (2014) 258 – 263
Beyaert A. & Camacho M. (2008). TAR Panel Unit Root Tests And Real Convergence: an Application to the EU Enlargement Process, Review of Development Economics, 12(3), 668-681.
263
ScienceDirect Procedia - Social and Behavioral Sciences 131 (2014) 258 – 263
WCETR 2013
Budget Deficits Sustainable? An Empirical Analysis For OECD Countries Mehmet Mercana* a
Assist. Prof. Dr., Hakkari University, FEAS, Department of Economics,Hakkari, TURKEY
Abstract In this study, budget deficit sustainability in OECD countries, are investigated via intertemporal budget constraint approach in 1980-2012 periods and quarterly data used. In analysis firstly, Cross Sectionally Dependency (CD) in country was examined CDLM test (Cross Sectionally Dependency Lagrange Multiplier) developed by Pesaran (2004). Stationary of series was searched CADF test (Cross-Sectionally Augmented Dickey-Fuller) developed by Pesaran (2006) considering CD. Cointegration relationship among series was examined panel cointegration test with multiple structural breaks developed by Basher and Westerlund (2009). As a result of analysis, cointegration relationship between the series was determined. According to the longrun analysis, budget deficits of these countries are sustainable in weak form.
© 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license © 2014 The Authors. Published by Elsevier Ltd. (http://creativecommons.org/licenses/by-nc-nd/3.0/). Selection and peer-review under responsibility of the Organizing Committee of WCETR 2013. Selection and peer-review under responsibility of the Organizing Committee of WCETR 2013.
Keywords: Budget Deficit, Sustainability, Cross-Sectional Dependency, Panel Cointegration. Jel Codes: H61, H62, H63 Introduction Implementation of Keynesian policies predicting the intervention of the government to the system especially in crisis period caused negative effects on budget balance of many countries. Countries chose to go into debt in order to meet the budget deficit and the insolvable debts in maturity periods generally were financed by new debts. This became a vicious cycle of debt-interest and it affected several macro economical variables. Despite Maastrich contract anticipating that budget deficit should be 3% most, the budget deficits in many European Union and OECD countries having the crisis recently exceeded 3% limit. For instance, in 2010 budget deficits in Ireland is 32.4%, in the USA 10.7%, in Greece 10.4%, in England 10.3%, in Spain 9.3%, in Portugal
* Corresponding Author: Mehmet Mercan Tel.: +90-505-386-6949 E-mail address:[email protected], [email protected]
1877-0428 © 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/3.0/). Selection and peer-review under responsibility of the Organizing Committee of WCETR 2013. doi:10.1016/j.sbspro.2014.04.114
Mehmet Mercan / Procedia - Social and Behavioral Sciences 131 (2014) 258 – 263
9.2%, in Japan 8.2% and in most of OECD countries 7.7% (OECD, 2013). In Europe, indebtedness reached a very high state; it passed over acceptable limits and became serious problem and threat to global economy (Staněk, 2004). When a budget deficit continues to grow over time, governments must pay attention to its potential effects on the economy and they should take necessary steps (Yoon, 2012). Hakkio and Rush (1991), for the US economy, relationship between government expenditures and government revenues examined by using 1950:Q2-1988:Q4 period data. The cointegration relationship between the series tested whether the cointegration coefficient is equal to one. If the parameter is equal to one, the budget deficits are sustainable, while smaller than one is considered to be unsustainable budget deficits in the long run. Quintos (1995) expanded these conditions. If the coefficient of the budget expenditure is equal to one, the budget deficits sustainability is considered strong. If it is between zero and one, sustainability is in a weak form. 2. Analysis In this study, 18OECD member countries†, whose ratio of budget deficit to GDP has been greater than 3% in 2012 year, quarterly data of INC (General government revenue Percent of GDP) and EXP (General government total expenditure Percent of GDP) for 1996-2012 period annual data have been used. Data was taken from the IFS (International Financial Statistics) data base. Gauss 9.0 package program was used in this study. 2.1. Testing the cross-sectional dependency Before proceeding with further steps, cross-section dependence must be tested. Otherwise, results may be biased and inconsistent (Breusch and Pagan, 1980; Pesaran, 2004). Therefore, prior to further analyses, the existence of cross-section dependency in the series and the cointegration equation should be tested. The existence of a cross-section dependency among countries is tested via the Breusch-Pagan (1980) LM test when time dimension is greater than the cross-section dimension. Pesaran (2004) improved this test for time dimension is smaller than the cross-section dimension and time dimension is greater than the cross-section dimension. This test is biased when the average group is zero, but the average individual is different from zero. Pesaran et al. (2008) adjusted this deviation by adding the variance and the average to the test statistics. Therefore, it is called the bias-adjusted LM test (LMadj). The first form of LMadjtest statistics is as the following:
Where represents the avarage, represents the variance. The test statistics to be obtained here show a standard normal distribution as asymptotic (Pesaran, et al. 2008). The null hypothesis of the LMadj test is no crosssection dependency. The LMadj test was used and obtained results are presented in Table 1. Table 1: Crosssectional Dependency (LMadj) Test Results Test Prob. Statistics Value INC 4.016 0.000 EXP 2.051 0.020 Cointegration Equation 44.56 0.000 Note:p-values were computed 1000 bootstrap replications. As can be seen from Table 1, since the probability values of series and cointegration equation are smaller than Variables
†
Belgium, Canada, Czech Republic, Denmark, France, Greece, Ireland, Israel, Japan, Netherlands, New Zeland, Poland, Portugal, Slovak Republic, Slovenia, Spain, United Kingdom and United States.
259
260
Mehmet Mercan / Procedia - Social and Behavioral Sciences 131 (2014) 258 – 263
0.05, H0 hypotheses are strongly rejected and it has been decided that there is cross-sectional dependency among these countries. This reveals to a significant change in the series in one of the countries also affects the others. Therefore, while the decision makers in these countries set their policies, should take into consideration to policies of the other countries and the other external factors. Furthermore, since cross-section dependency determined, while choosing the unit root and cointegration tests method, this situation should be taken into account. Therefore, panel unit root tests and cointegration analysis considering the cross-section dependency have been also used in the analysis. 2.2. Panel unit root test The panel unit root tests considering the information about both the time and the cross-section dimension of the data are accepted to be statistically stronger than the time series unit root tests considering the information only about the time dimension (Maddala and Wu, 1999; Levin, Lin and Chu, 2002; Beyaert and Camacho, 2008) because the variability in the data increases with the addition of the cross-section dimension to the analysis. The first problem in the panel unit root test is whether or not the cross-sections forming the panel are independent to each other. Panel unit root tests here are divided into two as first and second generation tests. First generation tests are Levin, Lin and Chu (2002), Breitung (2005), Hadri (2000), Im, Pesaran and Shin (2003), Maddala and Wu (1999) and Choi (2001). First generation unit root tests are based on the hypothesis that the cross-section units forming the panel are independent and all the cross-section units are equally affected by the impact to one of the units in the panel. However, it is a more realistic approach that units are differently affected from the impact to one of the crosssectional units in the panel if it is thought that national economies are related to each other today. In order to overcome this deficiency, second generation unit root tests carrying out the unit root analysis considering the crosssection dependence between the cross-section units have been developed. Main second generation unit root tests are MADF (Taylor and Sarno, 1998), SUDARF (Breuer, Mcknown and Wallace, 2002) and CADF (Pesaran, 2006). In this study, since it has not been identified any cross-section dependency between the countries in the panel for the INC and EXP variables used in the study, stationary of the series has been analyzed with one of the second generation unit root test CADF test developed by Peseran (2006). Through CADF unit root test can be performed in each crossection unit in the series forming the panel. So the stationary of the series can also be estimated one by one for the panel’s overall and each cross-section. CADF test hypothesing that every country is affected differently from time effects and considering the spatial autocorrelation is used in T>N and N>T situations. Stationary for each country is tested by comparing the statistics values of this test with Peseran’s CADF critical table values. If CADF critical table value is greater than CADF statistics value, the null hypothesis is rejected and it is found out that the series of only that country is stationary. CADF test statistics is estimated as the following:
(2)
(3) Here shows unobservable common effects of each country, shows individual-specific error. Equations (2), (3) and unit root hypotheses can be written as the following: H0: for all i H1: i=1, 2, . . . . , N1, i=N1+1, N1+2,…, N.
(4)
(Series is non stationary.) (Series is stationary.)
Test statistics and critical values have been computed for each country and panel (overall). Results are presented in Table 2. Table 2: CADF Unit Root Test Results Variables
Level
INC
-0.831
First Difference -1.903
Critical Value -1.840
Mehmet Mercan / Procedia - Social and Behavioral Sciences 131 (2014) 258 – 263
EXP -1.282 -1.841 -1.840 Note: Without trend model for INC and EXP series and %10 significant level have been selected as a test model. Gauss 9.0 package program was used in the analysis. Results in Table 2 show that series are non-stationary at levels but become stationary at first differences; they are said to be integrated of first order, I(1). In this case, it has been concluded that the existence of cointegration relationship between these series can be tested since series under consideration are integrated of the same order. 2.3. Cointegration analysis The existence of the cointegration relationship between the series has been investigated through the panel cointegration test developed by Westerlund (2008), which considers the cross-section dependency. The null hypothesis denotes no cointegrating relationship. When the probability value of the calculated test is smaller than 0.05, H0is rejected and it is decided that there are cointegration relationship between the series. Cointegration test results were presented in Table 3. Table 3: Cointegration analysis results
Durbin-H Group
Cointegtarionstat 4.331
Durbin-H Panel
2.437
Prob. 0.000 0.007
Decision Cointegratio n Cointegratio n
2.4. Estimation of long term coefficients In this part of the study, the long run individual cointegration coefficients will be estimated with the Common Correlated Effects (CCE) method which is developed by Pesaran (2006). CCE is an estimator that can generate results providing consistent and asymptotic normal distribution when the time dimension is both greater and smaller than the cross-section dimension and that can separately calculate the long term cointegration coefficients for the cross-section units (Pesaran, 2006). Long term cointegration coefficients of panel were calculated with the Common Correlated Effects Mean Group (CCEMG) method. CCE and CCEMG estimations that have been carried out using equation (5) and results are presented in Table 4. Table 4: Long term coefficients Countries Belgium Canada Czech R. Denmark France Greece Ireland Israel Japan Panel
Coefficient -0.342 [-1.15] -0.259 [-1.26] 0.11[1.46]*** 0.32[2.60]* 0.19[0.64] -0.35[1.72]** 0.50[3.17]* 0.57[6.61]* 1.91 [4.47]* 0.28[2.99]*
Countries Netherland New Zeland Poland Portugal Slovak Rep.
Coefficient 0.46[1.62]** 0.28 [2.48]* -0.07[-0.53] 0.89[5.44]* 0.64[1.93]**
Slovenia
0.48[6.70]*
Spain U. Kingdom U. States
0.43[1.38]*** 0.30[1.20] 0.26 [1.15]
261
262
Mehmet Mercan / Procedia - Social and Behavioral Sciences 131 (2014) 258 – 263
Note: Autocorrelation and heteroscedasticity problems were adjusted with the Newey-West.[ ]; shows t statistics. *, **, ***; Indicates significance level in 1%, 5% and 10% respectively. According to Table 4, the long-run cointegration coefficients are smaller than one. According to Hakkio and Rush (1991) and Quintos (1995), the budget deficits in these countries are sustainable in weak form. 3. Results and policy implications In this study, the sustainability of budget deficit has been examined for 18 OECD countries which their budget deficit exceed 3% of the GDP in the year 2012 by means of panel cointegration method under cross-section dependence for the period of 1996-2012 annual data. It is found that in these countries budget deficits are weakly sustainable according to Hakio and Rush (1991) and Quintos (1995) for the cointegration coefficient is smaller than one. According to macroeconomic theory, in order to sustain high budget deficits some policies should be implemented in undesirable occasions such as scarcity of investment, lack of consumption, foreign high borrowing and low private savings. These economic precautions should be implemented even they are not welcomed in some parts of the society as in the sample of Greece. Since it is found that the sustainability of budget deficits is low as a result of the analysis, OECD and EU economy management should take the necessary precautions, alert the related countries and implement the necessary sanctions. In this context, it will be helpful to implement the fiscal rule in countries and to establish the independent supervisory and regulatory institutions. References Basher, S. A. & Westerlund, J. (2009). Panel cointegration and the monetary exchange rate model, Economic Modelling, 26, 506-513. Breuer, B., Mcnown, R. & Wallace M. (2002). Series-Specific Unit Root Test with Panel Data, Oxford Bulletin of Economics and Statistics 64(5): 527-546. Breusch, T.S & Pagan A.R. (1980). The Lagrange Multiplier Test and Its Applications to Model Specification Tests in Econometrics, Review of Economic Studies, 47: 239-53. Breitung, J. (2005). A Parametric Approach to the Estimation of Cointegrating Vectors in Panel Data, Econometric Reviews, 24(2): 151 173. Choi, I. (2001). Unit Roots Tests for Panel Data, Journal of International Money and Finance 20: 229-272. Hadri, K. (2000). Testing for Stationarity in Heterogenous Panels, Econometrics Journal, 3: 148-161. Hakkio, C. S. & Rush, M. (1991). Is the Budget Deficit Too Large? Economic Inquiry, pp. 429-445. Im, K., Pesaran H. & Shin Y. (2003). Testing for Unit Roots in Heterogenous Panels, Journal of Econometrics, 115(1): 53-74. Levin, A., Lin C.F. & Chu C.S.J. (2002). Unit Root Tests in Panel Data: Asymptotic and Finite Sample Properties, Journal of Econometrics. Maddala, G.S. & Wu, S. (1999). A Comparative Study of Unit Root Tests with Panel Data and a New Simple Test, Oxford Bulletin of Economics and Statistics, 61: 631-652. Pesaran, M. H. (2004). General Diagnostic Tests for Cross Section Dependence in Panels, Cambridge Working Papers in Economics, 435. Pesaran, H. (2006). A Simple Panel Unit Root Test in the Presence of Cross Section Dependency, Cambridge Working Papers in Economics, 0346. Quintos, C. E. (1995). Sustainability of the Deficit Process with Structural Shifts, Journal of Business and Economic Statistics, 13: 409–417. Staněk, P. (2004). Indebtedness as Immanent Feature of Current Economies, Ekonomicky Casopis, 52(3): 255-269. Taylor, M. & Sarno, L. (1998). The Behaviour of Real Exchange Rates during the Post-Bretton Woods Period, Journal of International Economics, 46: 281-312. Yoon, G. (2012). Explosive U.S. Budget Deficit, Economic Modelling, Pesaran, M. H., Ullah, A. & Yamagata, T., (2008). A Bias-Adjusted LM Test of Error Cross-Section Independence, Econometrics Journal, 11(1), 105-127. Westerlund, J. (2008). Panel Cointegration Tests of the Fisher Effect, Journal of Applied Econometrics, 23, 193 233.
Mehmet Mercan / Procedia - Social and Behavioral Sciences 131 (2014) 258 – 263
Beyaert A. & Camacho M. (2008). TAR Panel Unit Root Tests And Real Convergence: an Application to the EU Enlargement Process, Review of Development Economics, 12(3), 668-681.
263