Regional economic convergence in Turkey: Does the government really matter for?

Gömleksiz, Mustafa; Şahbaz, Ahmet; Mercan, Birol Article Regional economic convergence in Turkey: Does the government really matter for? Economies Provided in Cooperation with: MDPI – Multidisciplinary Digital Publishing Institute, Basel Suggested Citation: Gömleksiz, Mustafa; Şahbaz, Ahmet; Mercan, Birol (2017) : Regional economic convergence in Turkey: Does the government really matter for?, Economies, ISSN 2227-7099, MDPI, Basel, Vol. 5, Iss. 3, pp. 1-16, https://doi.org/10.3390/economies5030027 This Version is available at: https://hdl.handle.net/10419/197029 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ economies Article Regional Economic Convergence in Turkey: Does the Government Really Matter for? † Mustafa Gömleksiz, Ahmet ¸Sahbaz * and Birol Mercan Faculty of Social Sciences and Humanities, Necmettin Erbakan University, Selçuklu 42060, Turkey; [email protected] (M.G.); [email protected] (B.M.) *Correspondence: [email protected] † Authors’ Note: This paper is an extended and modified version of an earlier study titled “Regional Economic Convergence and Role of Government: A Case Study on NUTS 2 Regions in Turkey” which was presented in WEI 2016 Academic Conference held in Rome, Italy. Academic Editor: Vassilis Tselios Received: 21 March 2017; Accepted: 18 July 2017; Published: 24 July 2017 Abstract: Solow (1956) has made an essential contribution to the Neo-classical growth approach through the economic convergence hypothesis. It assumes that poorer countries’ or regions’ per capita incomes tend to grow at faster rates than the richer ones. Convergence could occur either among a group of economies with the same steady states or within regions in which their fundamental dynamics differ, and thus they exhibit multiple steady states. This study aims to investigate convergence with respect to GDP per capita across NUTS 2 regions in Turkey for the time period 2004–2014. In the convergence process, we also inquire into role of government in terms of regional government investments and fixed investment incentives. All the empirical results confirm the validity of the convergence hypothesis at a regional level. Also, in the context of the convergence process, it is possible to conclude that the role of government is likely to be decisive in solving regional economic disparities. Keywords: regional economic convergence; regional incentives; government investments; regional economic disparities; panel data analysis JEL Classification: R11; R50; C23 1. Introduction In regard to Neo-classical growth theory, Solow (1956) asserts that poorer countries’ or regions’ per capita incomes tend to grow at faster rates than richer ones. Therefore, all countries or regions should converge in terms of per capita income eventually. In recent years, such a catch-up effect has been often discussed in many of the growth studies. Sala-i-Martin (1996a) states that the growing attention to the convergence concept has been attributed to a number of reasons. Firstly, the convergence approach allows testing validity of the modern growth theories and provides information about the share of capital in production. Secondly, along with emerging datasets covering a large number of countries, evolution of the convergence process has begun to be re-examined. In this regard, most researchers have focused on the question of whether per capita income tends to converge over time across countries or regions (Barro and Sala-i-Martin 1991,1992;Mankiw et al. 1992;Sala-i-Martin 1996b;Lall and Yilmaz 2000;Michelis et al. 2004;Varblane and Vahter 2005;Lopez-Rodriguez 2008; Bonnefond 2014). Regional disparities are one of the most important phenomena that can be encountered in almost every economy, especially in underdeveloped and developing countries where industrial activities are relatively low and infrastructure investments are lacking (Chaudhuri 2001). In the context of Economies 2017,5, 27; doi:10.3390/economies5030027 www.mdpi.com/journal/economies Economies 2017,5, 27 2 of 16 regional disparities, the convergence concept has also dealt widely with different aspects. At a regional level, convergence can be mentioned only if growth takes place via reducing inter-regional income disparities. Otherwise, an increase in income could cause a ‘divergence’ effect across regions. In the latter case, the role of both investments and incentives to relatively less developed regions provided by the public sector is assumed to be important to the effort to close the gap. Myrdal (1957) and Hirschman (1958) state that government affects both national and regional economies positively in terms of infrastructure, education, and health investments. Thus, government investments can be considered as external shocks to stimulate lagging regions and give rise to convergence across regions (Button 1998). Moreover, governments often design some measures including financial and non-financial support and incentives in order to ensure more rapid development in certain regions or economic activities. In this respect, investment incentives could influence investment decisions or give rise to increases in profits at a regional level (Gineviˇcius and Šimelyt ˙ e 2011). According to the Republic of Turkey Ministry of Economy (2014), investment incentives basically aim to reduce regional development disparities, steer savings into high value added investments, boost production and employment, increase foreign direct investments, and promote investments for clustering and environmental protection. In this respect, governments may use some instruments such as value added tax (VAT) exemption, customs duty exemption, tax deduction, income tax withholding support, and land allocation to further accelerate investment decisions. Regional economic disparities reveal many economic, social, and political problems. First of all, the fact that there are so many serious differences between the regions is a symptom which induces inefficient economic structure. Especially since the 1960s, regional development has always been a main part of the development policies which have been implemented in Turkey. In recent years, differences between regions stand as an important controversial issue within both planned development efforts and the process for joining the European Union. However, despite all the policy proposals, debates and struggles, regional disparities in Turkey are still quite problematic ( Filiztekin 2009 ). In empirical literature, it can be said that various studies which mainly focused on regional disparities and the process of convergence revealed different results for Turkey. Thus, a group of studies conclude a significant convergence between regions (e.g., Tansel and Gungor 1998;Sa˘gba¸s 2002;Erlat 2005; Yildirim et al. 2009;Ersungur and Polat 2010;Önder et al. 2010;Zeren and Yilanci 2011;Aslan and Kula 2011;Karaalp and Erdal 2012;Gerni et al. 2015;Özgül and Karada˘g 2015), while others indicate that there is no tendency to converge across regions in Turkey (e.g., Filiztekin 1999;Berber et al. 2000; Erk et al. 2000;Gezici and Hewings 2004). This study aims to investigate the existence and degree of convergence among 26 NUTS 1 2 Regions in Turkey for the time period between the years 2004 and 2014. We also examine the role of government in terms of regional investments and fixed investment incentives in the convergence process. The dataset used in this analysis is obtained from the Turkish Statistical Institute (TurkStat), the Republic of Turkey Ministry of Development, and the Ministry of Economy databases. In this respect, the originality of this paper is said to be two-fold. First, the study discusses the midterm regional reflections of the economic reforms which have been put into effect in 2001 in Turkey. In this context, the Investment Incentive Program which is first introduced by Incentive Law No. 5084 in 2004 provides a number of advantages to the priority provinces and regions in development 2 . Some of the objectives of the program are to reduce regional development disparities, increase foreign direct investments, and encourage regional and large scale strategic investments. Accordingly, regional incentive 1 The NUTS classification (Nomenclature of territorial units for statistics) is a hierarchical system for dividing up the economic territory of the EU. 2 The implementation period of the Incentive Law No. 5084 was finalized on 31.12.2009. However, the period of benefiting from the incentives was extended until 2012 with the Law No. 5568 and the amendment made in Article 7 of Law No. 5084. Finally, the last investment incentive program, which put into effect in June 2012, is still ongoing (Republic of Turkey Ministry of Economy 2014). Economies 2017,5, 27 3 of 16 applications include: (i) customs tax exemption; (ii) VAT exception; (iii) tax reduction; (iv) insurance premium support for employers; (v) investment grants; (vi) interest support; (vii) support for income tax withholding; and (viii) insurance premium support ( Republic of Turkey Official Gazette 2012 ). Secondly, as a primary indicator of the convergence process, we use an up-to-date data of per capita GDP which has just been released at a regional level, instead of proxy variables. Thus, we might well take into account income convergence in both the cross-sectional and panel data of the regions. The rest of the paper is organized as follows. Section 2presents the theoretical background of convergence hypothesis. Empirical literature is reviewed in Section 3. Dataset and methodology used in analysis is given in Section 4. Section 5presents results from econometric investigation of convergence process in NUTS 2 Regions and the last section discusses the results and concludes the paper. 2. Theoretical Background of Convergence Approach In recent years, there has been much concern about convergence analysis in growth studies investigating the spillover effect of growth components across countries or regions over time. In this regard, the phenomenon of the convergence process is referred to as beta ( β ) and sigma ( σ ) (convergence in the literature. In the context of β -convergence, a pioneering study by Sala-i-Martin (1996b) , which extends the empirical evidence on regional growth and convergence across the United States, Japan, and five European countries, indicates that regions similarly tend to converge at a speed of approximately 2% annually. Another type of convergence is σ -convergence which emerges only if there are declines in the dispersion of real per capita income across economic units over time (Sala-i-Martin 1996b, p. 1327). In other words, σ -convergence is concerned with the behavior of the cross-sectional standard deviation, or of the coefficient of variation of per capita output over time. As a seminal paper, Baumol (1986) conducts a long-run convergence analysis within sixteen industrialized countries and introduces a basic method for testing the convergence in the context of the Neo-classical growth model in a cross-sectional regression. A modified and expanded version of this approach by Barro and Sala-i-Martin (1991,1992) assumes that there is an inverse relation between the growth rate of an economy and the distance from its steady state only if all the economies share same steady state. Hence, convergence to same steady state is also called as ‘absolute’ or ‘unconditional’ β -convergence. However, if the economies have different steady states in terms of their technological level or saving rates, then such an inverse relation will no longer be in question ( Magrini 2004 ). If ignorance of such cross-sectional differences is reasonable, convergence occurs as ‘conditional’ β -convergence. Thus, each economy conditionally approaches its own unique equilibrium. In this case, some specific explanatory variables which may potentially affect the convergence process and represent proxies for the different steady states get involved in analysis. Furthermore, it can be said that the convergence analysis evolves with panel data techniques in recent studies (e.g., Islam 1995 ; Lee et al. 1998;Gaulier et al. 1999;Michelis and Neaime 2004;Piras and Arbia 2007;Ranjpour and Takanlou 2008;Shen et al. 2008;Cuaresma et al. 2008;Bonnefond 2014). In the context of convergence analysis, σ and β -convergence have been subject to debate in some of the initial studies. Friedman (1992) and Quah (1996) argue that β -convergence in a cross-section regression only demonstrates the average behavior of the units and so it is irrelevant and uninformative for a distribution ' s dynamics. Also, regression fallacies arising from the tendency of a unit of variable such as income to move toward the mean over time are addressed to β -convergence analysis. Thus, they state that σ -convergence should be of primary interest in examining whether the distribution of income across economies is becoming more equitable (Friedman 1992;Quah 1996). However, Sala-i-Martin (1996b) suggests that these two concepts deal with the convergence process in different ways. Accordingly, σ -convergence discusses how the distribution of income evolves over time, while β -convergence focuses on the mobility of income within the same distribution (Sala-i-Martin 1996b, p. 1328). Moreover, Temple (1999) discusses potential problems such as endogeneity, the correlation in Economies 2017,5, 27 4 of 16 error terms, and measurement errors in estimating and interpreting growth regressions. In the case of β -convergence, Baumol (1986) asserts that conditional convergence takes place between a group of countries which are identical in initial conditions and some other factors. So, there are actually more than one ‘convergence club’, which belongs to industrial countries, middle income countries, and the poorer, less developed countries respectively (Baumol 1986, p. 1080). In respect to the convergence process, Romer (1996) concludes that countries are headed towards their balanced growth paths and differences in paths are due to differences in their initial capital endowments. Also, less developed countries with lower capital intensity have higher marginal capital efficiency; this causes the flow of capital from the developed countries to the underdeveloped countries (Romer 1996, p. 28). 3. Convergence Studies in Turkey In the empirical literature, there is a vast amount of studies investigating economic convergence at regional and provincial levels in Turkey. In this context, it is possible to evaluate these studies in chronological order, as given in Table 1. As one of the preliminary studies that emerged in the related period, Tansel and Gungor (1998) examine convergence in productivity measures across 67 Turkish provinces between the years 1975 and 1995. Results of the analysis show the presence of σ -convergence for the 1980–1995 period and absolute β -convergence in productivity measure for the whole period across the provinces. In the conditional analysis, they conclude that both saving rates and human capital increase the convergence rate among the provinces. Findings also imply a faster rate of convergence between homogenous groups of poorer and richer provinces in comparison with the total set of provinces. Notwithstanding, using provincial-level data on GDP and GDP per capita for the 1975–1995 period, Filiztekin (1999) is concerned with the same period considering single cross-sectional analysis. The results provide evidence that provinces diverge in an absolute sense in all periods except 1990–1995, though the rate of divergence is low. However, he finds evidence of conditional convergence of about 1.7% per year when region-specific dummies and the share of agriculture in total provincial output are included in the analysis. In another study, Berber et al. (2000) test the convergence hypothesis in seven geographical regions in 1975–1997 period. In this study, it is concluded that the regions do not converge in terms of per capita income and also diverge from each other. This result also partially overlaps with the research findings of Erk et al. (2000) which show no evidence for σ and absolute β -convergence across the 67 provinces in the 1979–1997 period. Sa˘gba¸s (2002) claims that the growth takes place in a path and it reduces the income disparities between the provinces in the 1986–1997 period. In context of conditional convergence, Sa˘gba¸s (2002) also examines the effects of public expenditures on the convergence process. However, he does not find any significant relationship between the public expenditures and the growth rates of the provinces in the second part of analysis. On the other hand, Gezici and Hewings (2004) empirically examine convergence across provinces and 16 functional regions in Turkey for the period 1980–1997. It is concluded that convergence between regions or provinces has not been observed in the related period. Within the 1975–2001 period, Erlat (2005) also does not find a clear evidence of convergence across provinces or regions. Another study by Yildirim et al. (2009) attempts to explore the causes of regional income inequalities and the convergence process across NUTS 1 regions for the 1987–2001 period. The results support the β -convergence hypothesis in terms of the higher speed of convergence in the relatively poorer Eastern and Southeastern regions. In the context of conditional convergence, they also find that the impact of real per capita government expenditures is relatively more outstanding in developed Western regions. Also, these results were partially confirmed in another study conducted by Ersungur and Polat (2010) for the same period. In another study dealing with control variables in the convergence process, Önder et al. (2010) examine the relationship between public capital, transportation capital stocks, and output per capita at NUTS 2 level. The findings of the conditional model based on the panel dataset for the 1980–2001 period indicate that per capita public capital stock has a positive and significant effect on output per capita and somewhat on regional Economies 2017,5, 27 5 of 16 convergence. Contrary to this, the effect of the transportation component of public capital stock on regional convergence is found as negative in all the models employed. Table 1. Summary of Convergence Studies in Turkey. Author/s Period/Sample Dependent Variable Method 1Some of the Findings Tansel and Gungor (1998)1975–1995 Labor productivity level and productivity growth NLS, OLS and panel FE i. σ-convergence for the 1980–1995 period. ii. absolute convergence in productivities across the provinces. iii. inclusion of both savings and human capital positively affect convergence rate. 67 provinces Filiztekin (1999)1975–1995 GDP and GDP per capita NLS and panel FE i. provinces diverge in absolute sense in all periods except 1990–1995. ii. provinces conditionally converge on each other about 1.7% per year. 65 provinces Berber et al. (2000)1975–1997 GDP per capita OLS i. no evidence for σand absolute β-convergence across the provinces. 7 geographical regions Erk et al. (2000)1979–1997 Real GDP per capita OLS and NLS i. no evidence for σand absolute β-convergence across the provinces. 67 provinces Sa˘gba¸s (2002)1986–1997 Real GDP per capita OLS i. absolute β-convergence across the provinces. ii. no relationship between government expenditures and growth in the context of convergence process. 67 provinces Gezici and Hewings (2004) 1980–1997 GDP per capita OLS i. no clear evidence for σ-convergence across the provinces and functional regions. ii. no evidence for absolute and conditional β-convergence across both provinces and the functional regions. 16 functional regions and 67 provinces Erlat (2005) 1975–2001 Real GDP per capita IPS, ADF and CADF panel unit root tests i. no clear evidence for absolute convergence across the provinces and regions. 7 geographical regions and 65 provinces Yildirim et al. (2009)1987–2001 Real GDP per capita OLS, SEM, SAR, and GWR i. absolute and conditional β-convergence across provinces. ii. eastern and southeastern provinces show higher speeds of convergence. 67 provinces Önder et al. (2010) 1980–2001 Real GDP per capita Pooled panel, FE, GMM-DIF, GMM-SYS i. σand conditional β-convergence across the regions. ii. per capita public capital stock has a positive effect on GDP per capita. 26 NUTS 2 regions Economies 2017,5, 27 6 of 16 Table 1. Cont. Author/s Period/Sample Dependent Variable Method 1Some of the Findings Ersungur and Polat (2010)1987–2000 GDP per capita OLS i. no evidence for σand absolute β-convergence across the regions. 12 NUTS 1 regions Aslan and Kula (2011)1975–2001 GDP per capita Univariate and panel LM i. absolute β-convergence across provinces, except two provinces. 67 provinces Zeren and Yilanci (2011)1991–2000 GDP per capita Panel FE and RE i. evidence for absolute and conditional convergence for the average of the regions. ii. regional bank deposits have a positive effect on GDP per capita. 26 NUTS 2 regions Karaalp and Erdal (2012)1993–2001 GDP per capita Panel FE and GMM i. evidence for absolute and conditional convergence across the provinces. ii. agglomeration effects slow down the convergence process. 73 provinces Gerni et al. (2015) 2004–2012 Regional tax revenue per capita OLS i. absolute β-convergence across the regions. ii. no evidence for absolute β-convergence across the provinces. iii. investment incentives have a negative impact on income per capita. 81 provinces 26 NUTS 2 regions Özgül and Karada˘g (2015) 1990–2001 GDP per capita and per person employed OLS i. σ-convergence across the regions. ii. some evidence for absolute β-convergence. iii. socio-economic indicators have no effect on regional growth. 26 NUTS 2 regions 1 ADF: augmented Dickey-Fuller test; CADF: cross-sectionally augmented DF test; GMM-DIF: generalized method of moments estimator in difference; GMM-SYS: system GMM estimator; NLS: non-linear least squares; GWR: geographically weighted regression; IPS: Im, Pesaran and Shin unit root test; LM: Lagrange multiplier unit root test; OLS: ordinary least squares; SAR: spatial autoregressive model; SEM: spatial error model. In the context of convergence analysis, Aslan and Kula (2011) consider structural breaks and thus a stochastic convergence process for the 67 provinces in Turkey. Using data on per capita income over the period 1975–2001, their analysis offers strong evidence for convergence except for the provinces of Bitlis and Erzurum. However, panel data techniques are also used for convergence analysis in some of the studies. For instance, Zeren and Yilanci (2011) empirically investigate regional convergence among NUTS 2 regions for the 1991–2000 period. Their analysis, which is based on the panel data with the random effects estimator, points out the validity of both absolute and conditional convergence across NUT 2 regions. At a region-specific level, they also conclude the validity of absolute convergence for 17 regions and conditional convergence for 25 regions. The results indicate that deposits have a positive impact on per capita income. In another study, Karaalp and Erdal (2012) examine the effects of agglomeration economies and the growth of neighboring cities on the convergence process at a provincial level. In the context of analysis, agglomeration coefficients in the Turkish manufacturing industry are calculated using the Herfindahl Index, the Gini Coefficient, and the Location Quotient for the period between 1993 and 2001. Results from the panel data estimation of β -convergence model indicate that income differences between provinces decrease over time. In another empirical study, Gerni et al. (2015) examine convergence at both the provincial and regional levels in Turkey over the 2004–2012 period. The results from the analysis indicate that per capita income increases lead to an absolute convergence between regions. Lastly, a recent study by Özgül and Karada˘g (2015) consider welfare measures and socio-economic indicators to investigate convergence across NUTS 2 regions in Turkey for the 1990–2001 period. In this context, the cross-sectional analysis of convergence reveals Economies 2017,5, 27 7 of 16 some evidence of unconditional convergence in Turkey. The results also show that socio-economic indicators have no effect on regional growth in the convergence process. 4. Data and Model In this study, we analyze income convergence along with a dataset which includes both cross-sectional and time series data for 26 NUTS 2 regions 3 for the time period between 2004 and 2014. We prefer real GDP per capita in order to measure income as the dependent variable. Also, concerning the purpose of investigating the role of government in the convergence process, we employ regional per capita government investments and per capita fixed investment incentives as control variables in this analysis. In generating per capita values of control variables, we use regional population estimations for the 2004–2006 period and the Address-Based Population Registration System (ABPRS) for the 2007–2014 period. The dataset is obtained from Turkish Statistical Institute (2017a,2017b), Republic of Turkey Ministry of Development (2015), and the Republic of Turkey Ministry of Economy (2016) . Also, all the data are expressed in Turkish Lira (TL) and in real prices. For this purpose, we use the consumer price index at a national level. As already mentioned in Section 2, σ -convergence occurs when income differentiation between economies decreases over time. In this respect, it is possible to say that dispersion of income levels can be measured by standard deviation of income per capita among economies. Also, the coefficient of variation (CV) can be used instead of standard deviation. In the analysis process, we use the coefficient of variation of GDP per capita which is formulated in Equation (1). CV =Standart Deviation Mean (1) Additionally, we conduct a regression of trend line of CV for GDP per capita to verify decreases in dispersion over time. In Equation (2), the dependent variable is the coefficient of variation of GDP per capita levels across regions while the independent variable is the time variable ( t= 1 . . . 11) for the period between 2004 and 2014. CVyt=γ0+γ1t+ut(2) For the purpose of investigating absolute or unconditional β -convergence, we regress a logarithmic equation based on cross-sectional data. In Equation (3) the left-hand-side of the equation form represents the average growth rate of region i in the time period T ( T=t0. . . t ). Also, Yi,t0 is initial year of period Tand γ0is a constant. T−1logYi,t Yi,t0=γ0+α1logYi,t0+εi(3) Additionally, we prefer a modified version of Equation (3) in order to test conditional β -convergence considering the specific characteristics of each region. In Equation (4) the Gi,t0 and Vi,t0 represent per capita regional government investments and fixed investment incentives, respectively. T−1logYi,t Yi,t0=γ0+α1logYi,t0+α2log Gi,t0+α3logVi,t0+εi(4) In panel data models, the choice of the estimation model is often discussed due to several reasons such as controlling for individual heterogeneity, informativeness, variability, and multicollinearity (Baltagi 2005, pp. 4–7). In general, one can choose between pooled ordinary least squares (OLS), fixed effects (FE), and random effects (RE) in estimation, considering different assumptions. In this respect, 3Also see the Appendix Afor regional context and region codes. Economies 2017,5, 27 8 of 16 a pooled OLS approach assumes that the error term is independent of the cross-sectional units and individually and identically distributed (I.I.D.). So, it does not take time-invariant specific effects into account. Instead, the FE or RE consider object-specific time-invariant effects (Wooldridge 2010). Also, the FE model is often applied when the differences between regions can be viewed as parametric shifts of the regression (Greene 2003, p. 293). We estimate the regression form given in Equation (5), which is based on panel data, in order to investigate convergence hypothesis. logYi,t−logYi,t−1=γ0+α1logYi,t−1+α2log Gi,t+α3logVi,t+εi,t(5) In Equation (11), where Yi,t Yi,t represents GDP per capita in region iwhile Gi,t and Vi,t are regional government investments and investment incentives, respectively. Also, following Barro and Sala-i-Martin (2004), we calculate the speed of convergence4as given in Equation (6). β=−T−1ln(1+α1T)(6) According to the equation, if convergence occurs ( α< 0) then higher initial income levels have a negative effect on final growth. Thus, β measures the annual convergence rate of an economy towards its steady state income level. 5. Empirical Findings In this section, we conduct an analysis based on the CV of regions, cross-sectional OLS, and panel data in order to investigate and verify σ and β convergence approaches for 26 NUTS 2 Regions in Turkey. The models that we employed are estimated for different specifications, including pooled OLS and FE estimators, that is, depending on different assumptions about the error term. The results from the analysis are presented in the sub-sections. 5.1. σ-Convergence of NUTS 2 Regions Table 2reports the results obtained from the analysis of σ -convergence for NUTS 2 Regions. The results indicate that coefficients of variation and standard deviation tend to decrease when real GDP per capita decreases across regions. In other words, during the process of growth, real GDP per capita levels of the regions become more equal and the variation between their real GDP per capita levels decreases. Thus, one might say that σ -convergence exists across NUTS 2 regions in Turkey for the time period 2004–2014. In Table 2, data show that the standard deviation of the per capita real GDP of the regions is 0.4148 in 2004 (it decreases to 0.3647 in 2012). However, the NUTS 2 regions tend to diverge relatively in the last two years and differentiation increases up to 0.3704 among regions in 2014, as well as a slight increase in the year 2006 and 2011. Thus, it causes a divergence effect across regions, as can be seen in Figure 1. However, during 2004–2014, it is possible to conclude that sustainable growth rates in per capita GDP enable lower income dispersion at a regional level. Figure 1shows the coefficients of variation of NUTS 2 regions along with the trend line for the whole period. Thus, it can be said that the NUTS 2 regions reveal s-convergence during 2004–2014. The CV of the regions decreases over the period except in the year 2006. In 2004, the CV of the regions is 0.0473, and it decreases to 0.0373 in 2014. In this period, average GDP per capita of the NUTS 2 regions increases about 13%. Also, Figure 1reports the trend line regression of the regions where the dependent 4 Barro and Sala-i-Martin (2004, pp. 466–67) estimate the speed of convergence for the U.S. states by an univariate regression model, 1 /T . log(yiT/yi0) = α−[( 1 −e−βT)/T] . log(yi0) + wi0t , where yiT and yi0 are final and initial year per capita GDP, α is a constant, T is the length of the period, and β is the coefficient of convergence speed. In this equation, the coefficient of initial income, [( 1 −e−βT)/T] , also equals to α1 in Equations (4) and (5), and then α1=−[( 1 −e−βT)/T] . In order to obtain β equation, we take the log of both sides of this equation [logee−βT=loge( 1 +α1T)] −βT/−T= [ln( 1 +α1T)/−T] . Thus, we reach the βcoefficient as given in Equation (6). Economies 2017,5, 27 15 of 16 Cuaresma, Jesus Crespo, Doris Ritzberger-Grünwald, and Maria Antoinette Silgoner. 2008. Growth, convergence and EU membership. Applied Economics 40: 643–56. [CrossRef] Durlauf, Steven, and Danny T. Quah. 1999. The new Empirics of Economic Growth. In Handbooks of Macroeconomics. Edited by John B. Taylor and Michael Woodford. Cornwall: North-Holland, vol. 1A, pp. 235–308. Erk, Nejat, Sanli Ate¸s, and Tuba Direkçi. 2000. Convergence and growth within GAP region (South Eastern Anatolia Project) and overall Turkey’s regions. In Proceedings of the METU International Conference in Economics IV, Ankara, Turkey, 13–16 September 2000. Erlat, Haluk. 2005. Türkiye’de Bölgesel Yakınsama Sorununa Zaman Dizisi Yakla¸sımı. In Bölgesel Geli¸sme Stratejileri ve Akdeniz Ekonomisi. Edited by Haluk Erlat. Ankara: Türkiye Ekonomi Kurumu, pp. 251–76. Ersungur, ¸S. Mustafa, and Özgür Polat. 2010. Türkiye’de Bölgeler Arasında Yakınsama Analizi. Atatürk Üniversitesi Sosyal Bilimler Enstitüsü Dergisi 8: 335–43. Filiztekin, Alpay. 1999. Convergence across Turkish Provinces and Sectoral Dynamics. Available online: http: //myweb.sabanciuniv.edu/alpayf/files/2010/04/provconv.pdf (accessed on 13 February 2017). Filiztekin, Alpay. 2009. Türkiye’de Bölgesel Farklar. Available online: http://research.sabanciuniv.edu/11643/1/ BolgeselFarklar_YKY.pdf (accessed on 13 February 2017). Friedman, Milton. 1992. Do Old Fallacies Ever Die? Journal of Economic Literature 30: 2129–32. Gaulier, Guillaume, Christophe Hurlin, and Philippe Jean-Pierre. 1999. Testing Convergence: A Panel Data Approach. Annales d’Économie et de Statistique 55/56: 411–27. [CrossRef] Gerni, Cevat, Selahattin Sarı, Haktan Sevinç, and Ömer Selçuk Emsen. 2015. Bölgesel Dengesizliklerin Giderilmesinde Yatırım Te¸sviklerinin Rolü ve Ba¸sarı Kriteri Olarak Yakınsama Analizleri: Türkiye Örne˘gi. In Proceedings of the International Conference on Eurasian Economies, Kazan, Russia, 9–11 September 2015. Gezici, Ferhan, and Geoffrey J. D. Hewings. 2004. Regional Convergence and Economic Performance of Peripheral Areas in Turkey. Review of Urban and Regional Development Studies 16: 113–32. [CrossRef] Gineviˇcius, Romualdas, and Agn ˙ e Šimelyt ˙ e. 2011. Government incentives directed towards foreign direct investment: A case of central and eastern Europe. Journal of Business Economics and Management 12: 435–50. [CrossRef] Greene, William H. 2003. Econometric Analysis, 5th ed. New Jersey: Prentice Hall. Hirschman, Albert O. 1958. The Strategy of Economic Development. New Haven: Yale University. Islam, Nazrul. 1995. Growth empirics: A panel data approach. Quarterly Journal of Economics, 110: 1127–70. [CrossRef] Karaalp, Hacer Simay, and Fuat Erdal. 2012. Sanayile¸smenin Bölgesel Yı˘gılması ve Kom¸su ˙ Illerin Büyümesi Gelir Farklılıklarını Artırır mı? Türkiye için Bir Beta Yakınsama Analizi. Ege Akademik Bakı¸s 12: 475–86. Lall, Somik, and Serdar Yilmaz. 2000. Regional Economic Convergence: Do Policy Instruments Make a Difference? Available online: http://siteresources.worldbank.org/WBI/Resources/wbi37161.pdf (accessed on 10 January 2017). Lee, Michael, Ritchard Longmire, Laszlo Matyas, and Mark Harris. 1998. Growth convergence: Some panel data evidence. Applied Economics 30: 907–12. [CrossRef] Lopez-Rodriguez, Jesus. 2008. Regional Convergence in the European Union: Results from a Panel Data Model. Economics Bulletin 18: 1–7. Magrini, Stefano. 2004. Regional (di)convergence. In Handbook of Regional and Urban Economics. Edited by J. Vernon Henderson and Jacques-François Thisse. Amsterdam: Elsevier B. V., vol. 4, pp. 2741–96. Mankiw, N. Gregory, David Romer, and David N. Weil. 1992. A Contribution to the Empirics of Economic Growth. The Quarterly Journal of Economics 107: 407–37. [CrossRef] Michelis, Leo, and Simon Neaime. 2004. Income Convergence in the Asia-Pacific Region. Journal of Economic Integration 19: 470–98. [CrossRef] Michelis, Leo, Athanasios P. Papadopoulos, and Gregory T. Papanikos. 2004. Regional convergence in Greece in the 1980s: an econometric investigation. Applied Economics 36: 881–88. [CrossRef] Myrdal, Gunnar. 1957. Economic Theory and Underdeveloped Regions. London: University Paperbacks, Methuen & Co Ltd. Önder, A. Özlem, Ertu˘grul Delikta¸s, and Metin Karada˘g. 2010. The Impact of Public Capital Stock on Regional Convergence in Turkey. European Planning Studies 18: 1041–55. [CrossRef] Özgül, Seda, and Metin Karada˘g. 2015. Regional Convergence in Turkey Regarding Welfare Indicators. Sosyoekonomi 23: 38–50. Economies 2017,5, 27 16 of 16 Piras, Gianfranco, and Giuseppe Arbia. 2007. Convergence in per-capita GDP across EU-NUTS 2 regions using panel data models extended to spatial Autocorrelations Effects. Statistica 67: 157–72. Quah, Danny T. 1996. Empirics for economic growth and convergence. European Economic Review 40: 1353–75. [CrossRef] Ranjpour, Reza, and Zahra Karimi Takanlou. 2008. Evaluation of the income convergence hypothesis in ten new members of the European union: A panel unit root approach. Panoeconomicus 55: 157–66. [CrossRef] Republic of Turkey Ministry of Development. 2015. Available online: http://www2.kalkinma.gov.tr/kamuyat/il. html?2015_0 (accessed on 20 December 2016). Republic of Turkey Ministry of Economy. 2014. Investment Incentives. Available online: http://www.economy. gov.tr/portal/faces/home/invest-incent/invest-incentives? (accessed on 17 February 2017). Republic of Turkey Ministry of Economy. 2016. ˙ Istatistikler ve Yayınlar. Available online: http://www.ekonomi. gov.tr/portal/content/conn/UCM/uuid/dDocName:EK-229074 (accessed on 10 October 2016). Republic of Turkey Official Gazette. 2012. Yatirimlarda Devlet Yardimi Hakkinda Karar. Available online: http://www.resmigazete.gov.tr/main.aspx?home=http://www.resmigazete.gov.tr/eskiler/2012/ 06/20120619.htm&main=http://www.resmigazete.gov.tr/eskiler/2012/06/20120619.htm (accessed on 9 January 2017). Romer, David. 1996. Advanced Macroeconomics. New York: McGraw-Hill. Sa˘gba¸s, ˙ Isa. 2002. Türkiye’de Kamu Harcamalarının Yakınsama Üzerindeki Etkisi. Afyon Kocatepe Üniversitesi ˙ I.˙ I.B.F. Dergisi 4: 137–48. Sala-i-Martin, Xavier X. 1996a. The Classical Approach to Convergence Analysis. The Economic Journal 106: 1019–36. [CrossRef] Sala-i-Martin, Xavier X. 1996b. Regional Cohesion: Evidence and Theories of Regional Growth and Convergence. European Economic Review 40: 1325–52. [CrossRef] Shen, Daqing, Ruiqiang Wang, and Nan Yi. 2008. Analysis of the Convergence of Regional Economic Growth in Beijing. International Journal of Business and Management 3: 52–58. [CrossRef] Solow, Robert M. 1956. A contribution to the theory of economic growth. Quarterly Journal of Economics 70: 65–94. [CrossRef] Studenmund, A. H. 1992. Using Eonometrics: A Practical Guide, 2nd ed. New York: Harper Collins Publishers. Tansel, Aysit, and Nil Demet Gungor. 1998. Economic growth and convergence: An application to the provinces of Turkey, 1975–1995. Available online: https://www.academia.edu/7142026/Economic_Growth_and_ Convergence_An_Application_to_the_Provinces_of_Turkey_1975-1995 (accessed on 10 January 2017). Temple, Jonathan. 1999. The New Growth Evidence. Journal of Economic Literature 37: 112–56. [CrossRef] Turkish Statistical Institute. 2017a. Regional Statistics. Available online: https://biruni.tuik.gov.tr/ bolgeselistatistik/anaSayfa.do?dil=en (accessed on 12 February 2017). Turkish Statistical Institute. 2017b. Population Estimates, 2000–2006, Address Based Population Registration System (ABPRS), 2007–2016. Available online: www.tuik.gov.tr/PreIstatistikTablo.do?istab_id=1590 (accessed on 20 February 2017). Varblane, Urmas, and Priit Vahter. 2005. An Analysis of the Economic Convergence Process in the Transition Countries. Available online: https://ssrn.com/abstract=757204 (accessed on 21 December 2016). Wallace, T. Dudley, and J. Lew Silver. 1988. Econometrics: An Introduction. Reading: Addison-Wesley Publishing Company. Wooldridge, Jeffrey M. 2010. Econometric Analysis of Cross Section and Panel Data, 2nd ed. Cambridge: MIT Press. Yildirim, Julide, Nadir Öcal, and Süheyla Özyildirim. 2009. Income Inequality and Economic Convergence in Turkey: A Spatial Effect Analysis. International Regional Science Review 32: 221–54. [CrossRef] Zeren, Fatma, and Veli Yilanci. 2011. Türkiye’de Bölgeler Arası Gelir Yakınsaması: Rassal Katsayılı Panel Veri Analizi Uygulaması. Business and Economics Research Journal 2: 143–51. © 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).