Harvie, Charles and Pahlavani, M., Sources of Economic Growth in South Korea: An Application of the ARDL Analysis in the Presence of Structural Breaks – 1980-2005, Department of Economics, University of Wollongong, 2006.
The primary objective of this paper is to examine the major determinants of GDP growth in South Korea emphasizing the importance of investment, trade and human capital, using quarterly time series data covering the period 1980Q1 to 2005Q3. The time series properties of the data are, first, analyzed using the Zivot-Andrews (1992) model. The empirical results derived indicate that there is insufficient evidence against the null hypothesis of unit roots for all of the variables under investigation. Second, the Gregory-Hansen (1996) cointegration technique, allowing for the presence of potential structural breaks in the data, is applied, and is found to reject the null hypothesis of no cointegration relationship in favour of the existence of at least one cointegration relation in the presence of single structural breaks in the system. By applying these methodologies we find that most of the endogenously determined structural breaks coincide with the gradual effects of the Asian crisis on the Korean economy. Taking into account the resulting endogenously determined structural breaks the error correction version of the ARDL procedure is then employed, to specify the short- and long-term determinants of economic growth in the presence of structural breaks. Based on the preliminary empirical findings obtained we conclude that, in the longterm, policies aimed at promoting various types of physical and human capital, and trade openness, have improved Korea’s economic growth. More specifically, the empirical results show that while the effects of physical and human capital as well as exports are highly significant, as expected, total imports were found to be non significant, and this could be due to compositional changes away from the importation of capital goods to consumer goods as Korean standards of living have improved. It was also found that the speed of adjustment in the estimated models is relatively high and had the expected significant and negative sign.