Non-linear regional income divergence and policies: Turkey case

Abstract

The literature on economic convergence is strongly influenced by Neo-Classical Growth model. It describes a monotone saddle path along which each economy converges towards a unique steady state. Commonly employed method in convergence analysis is the linear cross-sectional regressions which links the annual growth rate of regions to their initial income level. Ignoring the non-linearities is important from a policy perspective that implications obtained from a linear regression can be very different to the policies learned from a nonlinear case. Aim of the present study is to analyze regional income convergence in Turkey by using nonparametric convergence regressions. We implement our study for 67 provinces and a period 1975-2000. We find that the relationship between initial income and growth takes a inverted-U shape which means that the very low-income and high-income group of provinces experince a slow growth pattern compared to middle-income group. This has several implications for regional economic policies. First, middle-income provinces are able to stimulate their economies and fulfill their potential for convergence by market forces. Second, however, the very low-income provinces need a substantial help and assistance.It, therefore, becomes a natural necessity to direct policy instruments such as subsidies, direct and indirect income transfers, tax exemptions and other resources to these areas. In this way, nonparametric estimations provide a very useful guide to the way how the resources should be allocated across provinces. © 2015 Hellenic Association of Regional Scientists. All rights reserved.