TESTING FOR UNIT ROOTS AND PERSISTENCE IN OECD UNEMPLOYMENT RATES

The natural rate hypothesis in economics is represented as a trend-stationary (TS) process while the hysteresis hypotheses in economics are represented as difference-stationary (DS) processes. Persistence, a notion often confused with hysteresis, is shown to be long-memory TS process. The presence of persistence means that shocks to unemployment rates have long durations. While formally TS processes, highly persistent (near unit root) processes, in practice provide aggregate demand policy with scope to attentuate the effects of a shock before subsequent shocks occur. Extreme natural rate conceptions do not permit such policy effectiveness. Unit root tests aimed at distinguishing between TS and DS hypotheses are conducted. In most of the OECD countries examined, the null of the unit root (DS) is not rejected. An examination is made of the Perron argument, that the widespread acceptance of the unit root hypothesis within recent literature arises because of mis-specification in the testing regression, specifically the failure to account for segmented trends. This contention is not found to be crucial in our study which confirms the widespread acceptability of the unit root hypothesis in OECD unemployment rates even after accounting for possible breaks in the trend function. Recognising the low power of the unit root tests against near unit root alternatives, the paper also derives and calculates the degree of persistence in OECD unemployment rates, to assess whether there is scope for policy initiatives aimed at permanently lowering the unemployment rate. It is concluded that the rates exhibit large degrees of persistence which allows scope for policy intervention.