Convergence hypothesis of regional income in Korea

Abstract

Using data on per capita income among 13 regions in Korea over the period 1985–2002, regional convergence is evaluated. This study uses panel cointegration tests and a random coefficient model that allows both regional differences and similarities to estimate the Solow growth model. The model also corrects for heteroscedasticity and serial correlation. Evidence is found in favour of regional convergence in Korea, with a rate of convergence of around 8% a year. The results also indicate that the investment rate in physical capital has a significantly positive effect and the population growth has a significantly negative effect on the growth rate of per capita income, holding its initial level constant.

Notes

Taylor and Sarno (1998) mention that panel unit root rests need to be interpreted with caution, since rejection of the null hypothesis H ₀: ‘All series are I(1)’ does not necessarily imply H ₁: ‘All series are I(0)’, but rather H ₁: ‘At least one series is I(0)’.

In all studies, three regions (Gwangju, Daejeon and Ulsan) are excluded due to lack of data.

All but four of the estimated coefficients on lagged per capita income among 13 regions are now significantly negative as is the average. This means that four regions may not converge or only slowly converge to their steady-state levels.