Macroeconomic policy in recessions and unemployment hysteresis

Abstract

I adopt Ball’s (1999) cross-sectional approach to test for unemployment hysteresis to panel data. Long-run unemployment is explained with standard institutional controls, and proxies for monetary and fiscal policy reactions in recessions. The sample consists of 20 OECD countries for the period 1985 to 2008. The results indicate that fiscal consolidation in recessions has long-lasting effects on unemployment. No significant impact of monetary policy is found. However, tentative evidence suggests that the effects of fiscal spending are stronger when accommodated by expansionary monetary policy.

Keywords:

• monetary policy
• fiscal policy
• unemployment
• hysteresis
• recessions

JEL Classification:

• E520
• E620
• E630
• J080

Notes

¹ I also estimated dynamic versions of these models (the results are available from the author). The Arellano–Bond test, however, indicates serial correlation in the residuals. Because serial correlation in static specifications is a matter of efficiency, not consistency, I focus on the static results.

² Bertrand et al. (2004) illustrate the pitfalls of ignoring serial correlation in panel data, especially for long time series. They suggest the use of a clustered covariance matrix estimator proposed by Arellano (1987) as a possible resolution. The simulation results of Cameron et al. (2008) and Hansen (2007) show that with a cross-sectional dimension of 20 clustering SEs at the panel level is an appropriate procedure to account for serial correlation.

³ The countries are Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Ireland, Italy, Japan, Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, UK and USA.

⁴ To allow for a more precise timing in the spirit of Ball (1999), I include a dummy variable which is only equal to one at the first year of a recession period, and interact it with the monetary policy proxy. However, changes in real short-term interest rates are again not found to enter the specifications significantly.

⁵ As robustness check, I re-estimated all specifications until 2006 only to see if they are strongly influenced by the years of the Great Recession. The results remain very similar.