Time-varying okun coefficient and its determinants

Okun (1962) Law shows the trade-off between the unemployment rate and economic growth, which views the lack of aggregate demand and the presence of idle capacity as the main causes of unemployment. Some studies in the related literature document that the unemployment-growth relationship (Okun's coefficient) has changed over time as a result of various economic and socio-cultural factors, unexpected events, and the effects of crises and shocks. Okun's coefficient may vary across countries and over time for many reasons such as the changes in the production techniques, productivity shocks and labor market sturucture. There are some studies in the literature that focus on the time-varying structure of the Okun coefficient, employing rolling regression, smooth time-varying parameter, Bayesian analysis and Markov-switching methods. In this dissertation, we propose a distinct empirical methodology, multivariate autoregressive conditional variance-dynamic conditional correlation (DCC-GARCH) model to examine the time-varying structure of the Okun coefficient. There are several advantages of using the DCC-GARCH method for this purpose. First, this method takes into account of the fact that the shocks to both unemployment and growth series might be persistent and thus their variances might change over time. Second, this method takes into account of the time-dependent volatility in the linkage between unemployment and growth as well as the deviations resulting from such volatility. Using quarterly data on economic growth and unemployment over the periods of 1990:1-2017:4 from 45 countries, a DCC-GARCH(1,1) model is estimated for each country in the sample. Thus, the conditional variances of growth and unemployment rate and the conditional correlation coefficient between the two are estimated, from which the time-varying Okun coefficients are obtained for each country. Further, this study goes on to examine the possible factors affecting Okun's coefficient by second generation panel regression models. To this end, common correlated effect mean group (CCEMG) model is estimated. The findings show that the time-varying Okun coefficients are associated mainly with productivity shocks. More specifically, productivity shocks seem to reduce the magnitude of the linkage between unemployment and growth.