Interregi̇onal inflation and income convergence in Turkey

In this study, existence of inflation convergence among 26 regions included in Level 2 of Nomenclature of Territorial Units for Statistics (NUTS) at 2004:M01-2016:M12 period is analyzed by 6 different models. Furthermore, impacts of inflation on income convergence among these regions and the validity of inter-regional income convergence hypothesis at 2004-2016 period is investigated by 22 different models. In analysis, absolute β convergence, conditional β convergence, σ convergence and variation coefficient convergence methods are employed. Stationarity of the series is tested by Levin, Lin, Chu (2002), Im, Pesaran and Shin (2003) and Hadri (2000) panel unit root tests and the series are found to be I(1). Existence of causality relationships between the series are examined with Granger (1969) and Dumitrescu and Hurlin (2012) panel causality tests. Existence of the cointegration relationship between the series in the models is tested with Kao (1999) panel cointegration test. The cointegration coefficients in the models are estimated by PDOLS method. It is found that the absolute convergence and conditional convergence hypothesis about inflation is not valid for the analysis period 2004-2016 in Turkey. At the general level of prices, there is a low speed divergence between regions. According to the sigma convergence and variation coefficient approaches for the 2004-2016 period, there is not inflation convergence among regions in Turkey, but there exists a low speed inflation divergence situation. In addition, whether the region is located in the western part of the country, average temperature, rain amount, time of snow on the ground and oil prices are the determinants of inflation and these variables have an effect on the inflation divergence. In the analysis on the existence of income convergence; it is seen that absolute β convergence hypothesis is not valid and the conditional β convergence is valid in some models and in some other models invalid. In the convergence models, the convergence speed and in the divergence models, the divergence speed is found to be low. Looking at the details of the models; it is observed that the East-West regional dummy variable is statistically insignificant, but the speed of divergence of income decreased slightly. The coefficient of the general level of prices is found to be statistically significant and for this model, conditional β convergence hypothesis is found to be valid. The convergence rate is also quite high. Moving from this results, it is evaluated the divergence of inflation in Turkey have positive impact on the income convergence among the regions from 2004 to 2016. For the period of 2004-2016 in Turkey, there arises a convergence in per capita national income. Looking at sub-periods; From the beginning of 2004 to August 2005, there was a low-speed income convergence, from that time to September 2007 the income diverges. The income convergence accelerated from September 2008 when the global economic crisis started. In 2010-2011, when the economy grew by 8.9% and 11.1% respectively, the income divergence started. In 2012, when the contractionary monetary policy applied to reduce the current account deficit and the economic growth declines to 4.8%, income convergence takes place. Income divergence started in 2013-2014, which are the years of relatively accelerated economic growth. In 2015-2016, by the national developments and the fact that FED started to implement the tightening monetary policy, income convergence was observed. As a general summary, the income diverges in the periods of accelerated economic growth in Turkey and income converges when the economic growth is in fall. Key Words: Regional Inflation Convergence, Regional Income Convergence, Determinants of Regional Inflation, Determinants of Regional Income.