The heterogeneity of Okun's law : a metaregression analysis

Okun's law is an extremely influential parameter in empirical research and policy analysis, based on the sizable number of estimates from this perspective. Nevertheless, it is also subject to considerable heterogeneity. We first show graphical and statistical evidence on the existence of a high level of heterogeneity among Okun's law estimates in existing research, then analyze potential sources of heterogeneity. Using 1,213 estimates of Okun's law for various countries, regions, and time periods, separate metaregressions are estimated; one using estimates with the unemployment rate as the dependent variable, and the other with output as the dependent variable. Our findings indicate that the specification of the underlying model of the relationship has an effect on the magnitude of Okun's parameter. Differential labor market characteristics may also explain part of the observed heterogeneity. Finally, the results are also found to be influenced by methodological issues, such as the type of data (time series or panel data), the frequency of the data (annual or quarterly), the spatial coverage of the estimates (country, region, or group of countries), whether more variables are included in estimations, and whether a dynamic or static, symmetric or asymmetric model is estimated. This paper contributes to highlight the heterogeneity affecting the estimates of Okun's law and that needs to be taken into account. In order to know the "true" relationship between unemployment and economic growth, researchers should bear in mind that there are a number of methodological choices that have consequences for the results.