| Abstract: | The stability of Okun's law coefficient in the United States from 1949 to 2015 is examined using a regression with GARCH errors in order to capture the volatility of the series. Rolling estimations suggest that taking the volatility of the series into account yields more stable results compared to the simple OLS estimation, irrespective of the specification (gap or growth model), the data frequency (monthly or quarterly), or the length of the rolling window. The results also suggest that the persistence of shocks became much more important in explaining contemporaneous volatility when data from the recent global financial crisis were incorporated. In contrast, the feedthrough of output shocks in next period's output volatility was more important in the past, and especially during the 1970s stagflation period, but has been declining since. |
