Why economists disagree: An illustration of irreconcilability using the U.S. state level unemployment rate data

The literature suggests two theories regarding unemployment rate: the natural rate hypothesis vs. the hysteresis hypothesis. The former hypothesis refers to the rate of unemployment towards which the economy naturally gravitates in the long run. It is sometimes described as the equilibrium unemployment rate defined as the unemployment rate that would be observable when the economy is at its full employment level of output. According to the natural rate hypothesis, the equilibrium unemployment rate is determined by labor market structure, i.e. labor demand and labor supply, and it is generally unaffected by actual unemployment. The hysteresis hypothesis refers to the idea that the equilibrium unemployment rate depends on the history of the actual unemployment rate. According to this hypothesis, a random shock, such as recession, will have a permanent effect on the unemployment rate.

In the debate between the two hypotheses, there exists a deep schism as they have fundamentally opposing policy implications. The natural rate hypothesis supports the philosophy that the ‘market economy is self-correcting.’ This implies that adverse shocks (favorable shocks) leading the economy to recessions (economic expansions) are in fact market correction mechanism. The natural rate hypothesis, therefore, sees no need for government intervention to actively manage the economy. In contrast, the hysteresis hypothesis suggests that high unemployment that results from an adverse shock, if left by itself, may persist and continue to remain a serious social problem even in the long run. This possibility provides the rationale for government to play an active role in the economy and fight against unemployment.

Empirical evidence on U.S. unemployment rates is inconclusive. Studies, for example, by Mitchell (1993), Breitung (1994), and Hatanaka (1996) provide support for the US unemployment rate to be non-stationary. In contrast, studies by Nelson and Plosser (1982), Perron (1988), Xiao and Phillips (1998) and Song and Wu (1997) provide support for unemployment rate to be stationary.

In this paper, we analyze a panel of unemployment rate data for all 50 states over the period 1976 – 2016 and test these two competing hypotheses. Not surprisingly, we find support for both the hysteresis hypothesis and the natural rate hypothesis. In conclusion, using unemployment rate as an example, we report more anomalies about the unit root tests that are commonly in use. We contribute to the literature by explaining why reconciliation among economists that hold opposing views with regard to unemployment rate is not possible.