Assessing the Natural Rate of Unemployment

It has long been argued (Blanchard et al., 1992; Blanchard and Katz, 1997) that shifts in local labor market conditions, geographic variation in labor force demographics (labor supply), and sectoral shifts in industrial patterns (labor demand) might influence the natural rate of unemployment. On the supply side of the labor market this could occur through divergent trends in differential attachment of age, gender, racial, or ethnic groups to the labor force. On the demand side this could occur through differential job creation and destruction associated with changing industry technologies. As interesting as these possibilities are, they are exceedingly difficult to assess with official local labor market data because the methods used to construct official local area unemployment statistics (LAUS) ensure that at every level of aggregation the common trend and cycle components are identical to the components found in the national series. These statistics are designed, therefore, to measure how the local area deviates from the national series and not to provide a basis for assembling a national series built up from local data. By contrast, studies of the euro area (Berger, 2011) build up the aggregate labor market from the component country national statistics. In those studies the national trend and cycle components can be properly interpreted as aggregates of local (euro zone country) labor markets. Neither studies based on U.S. official unemployment data nor studies based on Eurostat official unemployment data for the euro zone can compare the trend and cycle components of a national series, estimated from national data, and the same components aggregated from independently measured local labor market ∗Abowd: Labor Dynamics Institute, Cornell University, Ithaca, NY ([email protected]). Goel: Labor Dynamics Institute, Cornell University, Ithaca, NY. Vilhuber: Labor Dynamics Institute, Cornell University, Ithaca, NY ([email protected]). Financial support from the U.S. Census Bureau and the National Science Foundation (Grants: SES 0922005, SES 1042181 and SES 1131848) is gratefully acknowledged. No confidential data were used in this paper; see the Data Appendix.