Are Labor Force Participation Rates Really Non-Stationary? Evidence from Three OECD Countries

In this paper we first examine the labor force participation rates (LFPR henceforth) for Australia, Canada and the USA and endogenously determine several structural break points in the series and discuss their possible causes. We employ a class of generalized univariate processes, called fractionally integrated processes (Granger and Joyeux, 1980; Hosking, 1981) with structural breaks. They are flexible enough to capture the mean-reverting dynamics in the series. Therefore, they allow modeling the Labor force participation rate movements over time better than the standard time series models. In order to examine the possibility of mean reversion, we use a test developed by Robinson (1994) which permits testing I(d) hypothesis allowing for breaks at known times. The results of our analysis indicate that the LFPRs that we consider are stationary. As a result we can conclude that the informational value of the unemployment rates about the behavior of labor markets and the causes of joblessness are useful in Australia, Canada and the USA. In these countries we can talk about one-to-one correspondence between the long-term changes in unemployment rates and the long-term changes in employment rates and that unemployment rate is a useful indicator of joblessness.