Income Inequality and Economic Convergence in Turkey - Springer Link

2 downloads 0 Views 284KB Size Report
Income Inequality and Economic Convergence in Turkey. Julide Yıldırım1 and Nadir Öcal2. 1Department of Econometrics, Gazi University, Ankara, Turkey.
Transition Studies Review (2006) 13 (3): 559–568 DOI 10.1007/s11300-006-0124-x

Transition Studies Review Ó Springer-Verlag 2006 Printed in Austria

Income Inequality and Economic Convergence in Turkey Julide Yıldırım1 and Nadir Öcal2 1

Department of Econometrics, Gazi University, Ankara, Turkey (E-mail: [email protected]) 2 Department of Economics, Middle East Technical University, Ankara, Turkey (E-mail: [email protected])

Abstract. Even though the convergence of regional per capita incomes has been a highly debated issue internationally, empirical evidence regarding Turkey is both limited and contradictory. This paper is an attempt to investigate regional income inequality and convergence dynamics in Turkish gross domestic product. First, the Theil coefficient of concentration index has been employed in order to analyze the dispersion aspects of the convergence process which shows a procyclical character. Then, we investigate the convergence dynamics, taking regional interdependencies into account. Empirical results indicate that there is convergence at the national level. Moreover, the spatial error model is preferred by the model selection criteria, indicating that the typical least-squares regional convergence model is misspecified. Keywords: income inequality; convergence; spatial analysis; Turkey.

There have been traditionally two opposing views about the expected longrun trajectories of regional development. First, it has been argued that interregional mobility of capital and labour and sufficient time for tuning would eventually self-correct the regional inequalities. However, the existence of significant adjustment costs of flowing inputs among spatially distinct regions contributes to the second view that regional divergence is more likely. In particular, economies of scale, agglomeration of human capital, institutional framework, and geographical structures of certain regions accrue economic rents to be more local (Martin and Sunley 1998). Recent studies have revealed that there are economic disparities within each country which are generally higher than those observed at the country level (Barro and Sala-i-Martin 1991, Neven and Gouyette 1995, Fagerberg and Verspagen 1996, Quah 1996, Pekkala 1999, Terrasi 1999, Azzoni 2001, Akita 2003). Empirical studies provide evidence concerning the convergence of regional economies, which provides assistance in planning and evaluating regional policy

560

Yıldırım and Öcal

measures. So, the challenge for the national governments would be to provide sufficient incentives to amend unequal regional development. The persistent disparities in aggregate growth and the large differences in wealth of Eastern and Western regions has been the main concern of policy makers in Turkey. Since 1963, eight five-year development plans have been launched to achieve regional convergence. Although wealth disparities across Turkish regions and provinces have been a debated issue, there is limited empirical evidence concerning regional economic convergence in Turkey. Filiztekin (1999) investigated convergence across Turkish provinces between 1975–1995 applying the single cross-section methodology and found divergence of per capital output in all periods except for 1990–1995. He reported that the dominant sector in Turkey is still the low-productive agricultural sector, even though there has been massive flow of labor from that sector to others, especially services. However, Tansel and Gungor (1998) repeated the single cross-section studies for the same time period but came up with results contradictory to those of Filiztekin (1999). The difference between the two studies can be due to the fact that Filiztekin (1999) was concerned with per capita incomes, whereas Tansel and Gungor (1998) dealt with convergence in labor productivity. Using again the labor productivities at province level, Temel et al. (1999) provided evidence of polarization around highly industrialized regions. In their nonparametric regression analysis, they showed that besides a significant labor mobility from eastern agricultural provinces to western industrialized provinces, there is significant concentration around mainly the three western provinces of Turkey. Yet, the previous studies presented a regional-inequality analysis for Turkey at a disaggregated level and ignored the spatial dimension to the pattern of growth across regions. In a more recent paper, Gezici and Hewings (2003) explored the regional inequalities considering the spatial patterns and found similar indication of disparities between the east and the west of Turkey during the period of 1980–1997. Although the existing intraregional inequalities were found to be declining, they argue that spatial dependence to few wealthier provinces would be persistent in Turkey. The aim of this study is to provide a new look at the existing regional economic differences in Turkey and emphasize the fact that regional convergence needs to be properly spatialized. In this paper, the regional inequality issue has been investigated by the Theil coefficient of concentration using spatially disaggregated data for Turkey for the period of 1979–2001. Then, a convergence analysis has been implemented incorporating the spillover effects between the provinces. Empirical analysis using disaggregated gross domestic product (GDP) data of NUTS (Nomenclature of Units for Territorial Statistics) level 3 areas gives evidence in favor of regional convergence at the national level. However, there is a significant income inequality between the regions; because inequality within regions is relatively small, the regional classification that has been 4

Transition Governance Working Papers

Yıldırım and Öcal

561

made is justified. It appears that the Theil coefficient exhibits a procyclical character, such that it has a tendency to increase in periods of economic expansion and to decrease in periods of recession. In the second part of the paper, the beta convergence analysis for 67 provinces of Turkey taking regional interdependencies into account is presented. The empirical analysis reveals that there is beta convergence. However, when spatial autocorrelations are taken into account, the convergence rate increases where the spatial error model outperforms the other models, indicating that there are interdependencies in growth rates of provinces.

Methodology and empirical results Regional inequalities This section presents an inequality analysis where inequality indicators are calculated and their evolution over time is investigated. The regional convergence studies claim that the growth process of regions is similar to that of national states, mainly due to free capital and labour mobility compared to international level. However, Terassi (1999) for Italy, Petrakos and Saratsis (2000) for Greece, and Azzoni (2001) for Brazil indicate that there are serious income inequalities among regions which may show oscillations in time. Whereas Fagerberg and Verspagen (1996), Funke (1995), and Chatterji and Dewhurst (1996) report the existence of selective tendencies, convergence clubs, and asymmetric shocks within economies which result in spatial inequalities. In this study the Theil coefficient of concentration index (Theil 1967) has been employed in order to analyze the dispersion aspects of the convergence process, using NUTS level 3 data, relating to 67 provinces of Turkey1 for the time period of 1979–2001, by forming 4 subregions – (1) Marmara; (2) Ege, Akdeniz, and West Anatolia; (3) East and South East Anatolia; and (4) Karadeniz and Mid-Central Anatolia –, considering the spatial dimension of the post-1980 development.2 The population data was obtained from the State Institute of Statistics. Population data that are based on the 1980, 1985, 1990, and 1997 official censuses were interpolated for the years that do not coincide with the census. Real GDP data on the other hand come from two sources. The State Institute of Statistics has published provincial real GDP data since 1987. GDP data for earlier years are obtained from Özötün (1988). Even though price deflators for all provinces do not exist, price deflators for one major province in each region are available, which are used to construct price indices for each province. Then they have been used to deflate the GDP data from Özötün (1988). The Theil coefficient of concentration index is a very popular index for analyzing spatial distributions, for it is independent of the number of regions Transition Governance Working Papers

5

562

Yıldırım and Öcal

and thus compares inequalities of different regional systems. Additionally it is decomposable into between- and within-group inequalities. The following formulas are used to calculate the index: X T¼ yi logðyi =xi Þ ¼ Tbr þ Twr ; ð1Þ i

X

Yr logðYr =Xr Þ;

ð2Þ

" # X Yr ðyi =Yr Þ logðyi =Yr =xi =Xr Þ ;

ð3Þ

Tbr ¼

r

Twr ¼

X r

i

where T denotes the total inequality, Tbr between-region inequality and Twr within-region inequality; yi and xi are regional shares of national income and population, respectively, and Yr and Xr are the same shares for regions. The results of the income inequality analysis are presented in Table 1 and Figs. 1 and 2. The analysis indicates a convergence throughout the period under consideration. However, it has a procyclical character, such that it has a tendency to increase in periods of economic expansion and to decrease in periods of recession. Turkey has experienced one of the most severe economic crises in the Republic era, which contributed to the military intervention, in 1980. In that year, inequality index shrank almost by 45%. During the expansion period, from 1983 till 1988, the Theil index exhibited an increasing trend. However, from 1989 onwards the Theil index decreased, hitting bottom in 1994 after another major economic crisis. A similar behaviour can be observed for the 1999 and 2001 crisis years, supporting the hypothesis that in expansion periods richer regions receive more benefits than poorer regions, thus increasing the inequality. However, in recession periods, the richer areas would be affected more quickly and seriously compared with the poorer regions. This could be due to the fact that recessions are generally more severe than expansions. This finding is in line with Petrakos and Saratsis (2000) and Gezici and Hewings (2003), who report that regional inequalities have a procyclical nature in Greece and Turkey, respectively. Moreover, within-region inequality indices for regions 2 and 3 indicate an increase in inequality within the respective regions, implying divergence, whereas for regions 1 and 4 the opposite is observed. Additionally, region 1, which is the richest region, has the highest level of within-region inequality, especially in expansion years, among all regions, thus contributing to nationwide inequality considerably. Decomposition of the overall regional inequality into between-region and within-region components reveals the procyclical nature of the inequality once again. Even though until 1987 withinregion inequality accounts for almost a quarter of total inequality, after 1987 it 6

Transition Governance Working Papers

Yıldırım and Öcal

563

Table 1. Theil coefficients for the different aspects investigated a Year

T

T1

T2

T3

T4

Twr

%Twr

Tbr

%Tbr

1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

0.043 0.024 0.041 0.033 0.050 0.052 0.046 0.047 0.049 0.043 0.044 0.034 0.033 0.034 0.036 0.033 0.036 0.036 0.037 0.036 0.034 0.036 0.031

0.078 0.053 0.074 0.064 0.082 0.082 0.083 0.085 0.073 0.070 0.074 0.066 0.067 0.067 0.069 0.060 0.065 0.067 0.071 0.068 0.066 0.067 0.062

0.014 0.006 0.018 0.012 0.020 0.022 0.015 0.014 0.028 0.025 0.022 0.020 0.017 0.018 0.019 0.023 0.022 0.019 0.017 0.018 0.017 0.018 0.016

)0.027 )0.008 )0.027 )0.016 )0.027 )0.027 )0.027 )0.028 )0.026 )0.026 )0.027 )0.027 )0.027 )0.027 )0.027 )0.028 )0.028 )0.028 )0.028 )0.028 )0.029 )0.029 )0.028

)0.022 )0.027 )0.024 )0.027 )0.025 )0.024 )0.025 )0.025 )0.026 )0.026 )0.025 )0.025 )0.024 )0.024 )0.024 )0.023 )0.023 )0.021 )0.022 )0.021 )0.020 )0.020 )0.019

0.012 0.009 0.010 0.011 0.011 0.011 0.012 0.011 0.013 0.006 0.005 0.003 0.003 0.003 0.004 0.003 0.003 0.004 0.004 0.004 0.003 0.003 0.003

28.099 39.304 24.930 33.940 21.953 21.282 25.758 22.879 27.270 14.975 11.055 8.287 9.545 9.395 10.020 8.557 8.517 9.968 10.061 9.929 8.585 9.535 9.991

0.031 0.014 0.031 0.022 0.039 0.041 0.034 0.036 0.036 0.037 0.039 0.031 0.030 0.031 0.032 0.030 0.033 0.033 0.034 0.032 0.031 0.032 0.028

71.901 60.696 75.070 66.060 78.047 78.718 74.242 77.121 72.730 85.025 88.945 91.713 90.455 90.605 89.980 91.443 91.483 90.032 89.939 90.071 91.415 90.465 90.009

a

Subscripts 1 to 4 refer to regions considered. %Tbr and %Twr refer to percentage shares of Tbr and Twr in total T

0.1 0.08

Theil T

0.06 0.04 0.02 0

Theil

T1

T2

T3

T4

−0.02 −0.04 ’79 ’80 ’81 ’82 ’83 ’84 ’85 ’86 ’87 ’88 ’89 ’90 ’91 ’92 ’93 ’94 ’95 ’96 ’97 ’98 ’99 ’00 ’01

Fig. 1. Theil coefficients

Transition Governance Working Papers

7

564

Yıldırım and Öcal

0.06 TBR

0.05

TWR

Theil T

0.04 0.03 0.02 0.01 0 ’79 ’80 ’81 ’82 ’83 ’84 ’85 ’86 ’87 ’88 ’89 ’90 ’91 ’92 ’93 ’94 ’95 ’96 ’97 ’98 ’99 ’00 ’01

Fig. 2. Inequality decomposition by Theil index

is relatively stable around 10% of total inequality. Thus, the decomposition analysis indicates a meaningful partition of 67 provinces into four regions, in that income disparity within the regions is smaller than the between region inequality, so that these four regions can be assumed to be homogeneous in nature. Overall the inequality analysis indicated that income inequality has a procyclical nature in Turkey in the time period under consideration, which raises an important question concerning the relationship between regional inequalities and economic performance. Moreover, even though the overall income inequality decreased, regional disparities are observed.

Spatial analysis The issue of economic convergence at subnational level has attracted a lot of attention in recent years. With the seminal work of Romer (1986) and Barro and Sala-i-Martin (1991), a large number of studies has been devoted to investigating variations in economic performance of countries. These studies reported huge economic disparities within each country. Beta convergence analysis has generally been employed in order to investigate convergence across economies or regions using cross-sectional data, implementing the following equation:

logðyit =yi0 Þ ¼ a þ d log yi0 þ ui ;

ð4Þ

where yit denotes real income or real GDP of region i at some time t ¼ 1; 2; . . . ; N; yi0 denotes income or GDP per capita at some initial time 0; a is the intercept term which may incorporate any rate of technological progress; u is a random error term independently and identically distributed ð0; r2 Þ, 8

Transition Governance Working Papers

Yıldırım and Öcal

565

which may represent random shocks to technology or tastes. A negative value of d signifies the beta convergence and convergence rate is calculated using the following formula:3

b ¼ ½ln½1  d=N:

ð5Þ

However, this approach assumes that all regions or economies under consideration have the same steady-state income path. But this is a highly restrictive assumption and may induce a significant heterogeneity bias in estimates of the convergence coefficient. Moreover, as Quah (1993) points out, the traditional cross-section approach does not reveal the dynamics of the growth processes. In empirical literature, two alternative approaches have been introduced to correct the heterogeneity bias associated with the traditional cross-section analysis. The first is to employ time series analysis to investigate the rates of convergence by looking for common stochastic trends in the individual regional time series data. But this approach is applicable only if long time series data are available at both the regional level and the national level. Alternatively, control variables that can proxy or capture the differences in the paths of steady-state incomes of regions, such as rates of accumulation of physical capital, net migration rates, differences in industrial structure, can be included in the traditional cross-section estimates. However, to obtain long time series data as well as reliable proxy data is a difficult task especially for a developing country such as Turkey. Another dimension of the convergence analysis is that the regional economic growth may follow a spatial pattern. It is important to investigate the spatial patterns that may indicate the spillover effects among regions. Gezici and Hewings (2003) point out that if the growth rates of the poor regions are higher than the growth rates of the rich regions, the spatial inequality may decrease over time, which may result in convergence. Even though the neoclassical model assumes perfect mobility of factors of production between economies, there may be significant adjustment costs or barriers to mobility for labour and possibly also for capital. In cases where regions pursue their own growth-promoting policies, there may be spillover effects from those regions to the adjacent regions. Cheshire and Gordon (1998) point out that economic rents from research and development and other sources may be more likely to accrue locally, where regions are more self-contained. Moreover, Fagerberg et al. (1996) claim that rates of technological diffusion may follow a spatial pattern as regions may have different capacities to create or absorb new technologies. Thus, incorporating spatial effects into the analysis may impact significantly on any estimated convergence effects. Spatial dependence can be handled in beta convergence in alternative ways.4 The first approach, spatial error model, assumes that the spatial Transition Governance Working Papers

9

566

Yıldırım and Öcal

dependence operates through the error process, where any random shocks follow a spatial pattern, so that shocks are correlated across adjacent regional economies, such that the error term in Eq. (4) may reveal a significant degree of spatial covariance, which can be represented as follows:

logðyit =yi0 Þ ¼ a þ d log yi0 þ ui

with

ui ¼ qWui þ ei ;

ð6Þ

where q is the spatial error coefficient, ei is a white-noise error component, and W is a spatial weighting matrix. W may be constructed using information on physical distance between pairwise combinations of economies in the sample or may be defined such that element wij = 1 if i and j are physically adjacent, and 0 otherwise. In this paper the latter approach is preferred. Alternatively, the spatial lag model examines the extent to which regional growth rates depend on the growth rates of adjacent regions, conditioning on the level of initial income:

logðyit =yi0 Þ ¼ a þ d log yi0 þ qW logðyit =yi0 Þ þ ui ;

ð7Þ

where q denotes the spatial autoregressive parameter. Moreover, the spatial cross-regressive model allows any spatial spillovers to be reflected in the initial levels of income as follows:

logðyit =yi0 Þ ¼ a þ d log yi0 þ sW log yi0 þ ui ;

ð8Þ

where s represents the spatial spillovers. In this section, beta convergence analysis is performed, taking spatial dimension into account, in order to overcome the heterogeneity bias associated with the traditional convergence analysis. In addition to estimating the degree of beta convergence using Eq. (4), estimates of each of the three spatial spillover models described in Eqs. (6), (7), and (8) are also provided. Table 2 reports estimates of convergence equations for real GDP growth between 1979 and 2001 using the province level data. Conventional least-squares estimates of the speed of convergence suggest a speed of convergence of close to 0.02% per year. Model selection criteria suggest the selection of the spatial error model, which has the highest convergence rate, implying a convergence rate of 0.8% per year. Additionally, the null hypothesis of zero spatial autocorrelation in the least-squares regression residuals is rejected for spatial lag and spatial error models, indicating that the typical least-squares regional convergence model is misspecified.

Conclusion The existence of wealth disparities across Turkish regions and provinces is a well-known and debated issue. However, the limited empirical evidence concerning regional economic convergence in Turkey cannot support an 10

Transition Governance Working Papers

Yıldırım and Öcal

567

Table 2. Beta convergence regressions a

Constant log y0

OLS

Spatial lag

Spatial error

Spatial crossregressive

0.5732 (0.685) )0.0047 (0.964)

2.5197** (0.043) )0.1539*** (0.104) 0.2330* (0.000)

2.0911* (0.087) )0.1743*** (0.078) 0.2401* (0.000)

0.5676 (0.698) )0.0054 (0.956)

q s 2

R Loglikelihood AIC Schwartz b

0.34 22.885 106.4923 106.1444 0.0002

0.61 41.018 93.08844 92.56667 0.0072

0.61 41.115 92.97926 92.45748 0.0083

0.0002 (0.951) 0.34 22.887 108.4912 107.9694 0.0002

a

Values in parentheses are the p values; single, double, and triple asterisks indicate significance at 1%, 5%, and 10%, respectively

agreement on that issue. The aim of this study is twofold: First the regional inequality issue was investigated employing the Theil coefficient of concentration using spatially disaggregated data for Turkey for the time period of 1979–2001. Then, convergence analysis was performed incorporating spillover effects between the provinces. The Theil coefficient of correlation analysis indicated that within-region inequality accounts for a small portion of the overall inequality. Thus, these four regions can be assumed to be homogeneous in nature. Moreover, the Theil coefficient exhibits a procyclical character, such that it has a tendency to increase in periods of economic expansion and to decrease in periods of recession. Next, the convergence dynamics are investigated taking spatial dimension into account. Empirical results indicate that there is convergence at the national level. Moreover, the spatial error model is preferred by the model selection criteria, indicating that the typical least-squares regional convergence model is misspecified.

Notes 1 From 1990 onwards the number of provinces has increased from 67 to 81. But the original 67 provinces have been included in our analysis, as the data relating to new provinces do not cover the time period under consideration. 2 A second and traditional west–east partition has also been investigated. But the analysis indicates a greater within-region inequality than between-region inequality, thus the inequality aspects relating to regions cannot be clarified. 3 See, for example, Sala-i-Martin (1996) for a detailed description of estimation methods. 4 For a detailed analysis of spatial econometric techniques and methods, see Anselin (1988) and Henley (2005). Transition Governance Working Papers

11

568

Yıldırım and Öcal

References Akita T (2003) Decomposing regional income inequality in China and Indonesia using two-stage nested Theil decomposition method. Annals of Regional Science 37: 55–77 Anselin L (1988) Spatial econometrics: methods and models. Kluwer, Dordrecht Azzoni CR (2001) Economic growth and regional income inequality in Brazil. Annals of Regional Science 35: 133–152 Barro R, Sala-i-Martin X (1991) Convergence across states and regions. Brookings Papers on Economic Activity 1 1991: 107–182 Chatterji M, Dewhurst JHLL (1996) Convergence clubs and relative economic performance in Great Britain, 1977–1991. Regional Studies 30: 31–40 Cheshire PC, Gordon IR (1998) Territorial competition: some lessons for policy. Annals of Regional Science 32: 321–346 Fagerberg J, Verspagen B (1996) Heading for divergence? Regional growth in Europe considered. Journal of Common Market Studies 34: 431–448 Fagerberg J, Verspagen B, Marjolein C (1996) Technology, growth and unemployment across European regions. Regional Studies 31: 457–466 Filiztekin A (1999) Convergence across Turkish provinces and sectoral dynamics. Background paper for the report ‘‘Turkey: economic reforms, living standards and social welfare study’’, World Bank Report 20029-TU, Poverty Reduction and Economic Management Unit, World Bank Funke M (1995) Europe’s monetary future: one market, one money. Discussion Paper DP 1-95. Centre for Economic Forecasting, London Business School, London Gezici F, Hewings GJD (2003) Spatial analysis of regional inequalities in Turkey. Paper presented at the 43rd Congress of European Regional Science, 27–30 August 2003, Jyvaskyla, Finland Henley A (2005) On regional growth convergence in Great Britain. Regional Studies 39: 1245–1260 Martin R, Sunley P (1988) Slow convergence? The new endogenous growth theory and regional development. Economic Geography 74: 201–227 Neven D, Gouyette C (1995) Regional convergence in the European Community. Journal of Common Market Studies 33: 47–65 Özötün E (1988) Provincial distribution of Turkish GDP 1979–1986 (in Turkish). Istanbul Chamber of Industry Publication 1988.8. Istanbul Chamber of Industry, Istanbul Pekkala S (1999) Regional convergence across the Finnish provinces and subregions, 1960–94. Finnish Economic Papers 12: 28–40 Petrakos G, Saratsis Y (2000) Regional inequalities in Greece. Papers in Regional Science 79: 57–74 Quah D (1993) Empirical cross-section dynamics in economic growth. European Economic Review 37: 426–434 Quah D (1996) Regional convergence clusters across Europe. European Economic Review 40: 951–958 Romer P (1986) Increasing returns and long-run growth. Journal of Political Economy 94: 1002–1037 Sala-i-Martin X (1996) Regional cohesion: evidence and theories of regional growth and convergence. European Economic Review 40: 1325–1352 Tansel A, Gungor ND (1998) Economic growth and convergence: an application to the provinces of Turkey 1975–1995. ERC Working Paper 98/9. Middle East Technical University, Ankara Temel T, Tansel A, Albersen P (1999) Convergence and spatial patterns in labour productivity: nonparametric estimates for Turkey. Journal of Regional Analysis and Policy 29: 3–19 Terrasi M (1999) Convergence and divergence across Italian regions. Annals of Regional Science 33: 491–510 Theil H (1967) Economics and information theory. Rand McNally, Chicago

12

Transition Governance Working Papers