This article presents a simple framework for the systematic investigation of the relationship between net (and gross) flows among different labor market states and movements in the unemployment rate. The framework is then used to investigate the behavior of net flows of persons among employment, unemployment, and departure from the labor force (not in the labor force) in the United States over the 1990-2010 period. Understanding this behavior increases economists' understanding of the progression of unemployment over the business cycle and aids in identifying the characteristics that make the most recent recession different from previous ones. (1) The article contributes to the literature on gross flows (2) and flow probabilities among various labor market states by investigating net flows between states over long periods.
Stock-consistent worker flow data
The data that follow on worker flows are derived from the Current Population Survey (cps), a monthly sample survey of approximately 60,000 households that is carried out by the U.S. Census Bureau for the U.S. Bureau of Labor Statistics (BLS, the Bureau). Each month, the cps is administered to about three-quarters of the households that also were in the survey during the previous month. This month-to-month overlap allows the Bureau to track individuals who change their labor force status from one month to the next. In any given month, a person is in one of three labor force states: employed (E), unemployed (U), or not in the labor force (N). The next month, the person either remains in the same state or changes to one of the other two states. Changes (flows) are denoted by pairs of letters; the first letter indicates the labor force status of an individual in the previous month, and the second letter indicates the state of the same individual in the current month. Thus, there are six possible flows associated with changing states: EU, EN, UE, UN, NE, and NU. The Bureau makes available seasonally adjusted monthly estimates of these flows (also known as "gross flows") back to 1990. Although data on the six flows have been available from the cps for some time, discrepancies existed between the labor force stock changes implied by the flows and the net changes derived from the reported monthly stock estimates. Recently however, BLS researchers developed methods for reconciling the flows and the stock data; consequently, it is these stock-consistent data that are used in this article. (3)
The unemployment rate from 1990 to 2010
The unemployment rate is defined as the ratio of the number of unemployed to the total labor force. Chart 1 shows the trend in the unemployment rate from February 1990 to June 2010. The three recessions which occurred during that period are clearly visible, as are the recoveries from the first two recessions and the beginning of the recovery from the most recent recession. The analysis that follows examines the similarities and differences among selected subperiods, with an eye toward determining whether any systematic patterns are associated with periods of rising unemployment. Because the raw data on flows are extremely "noisy," averages of (monthly) seasonally adjusted data are presented for meaningful comparisons.
Several subperiods can be identified in the chart, based on the turning points in the unemployment rate. First, the aforementioned three recessions are clearly identifiable, defined for the purposes of this article as periods during which the unemployment rate was rising in a sustained fashion. (4) These recessions may
be dated as having occurred over the periods June 1990 to June 1992, January 2001 to June 2003, and April 2007 to October 2009.5 The periods between the recessions (July 1992 to December 2000 and July 2003 to March 2007) and after October 2009 can be thought of as economic recovery periods, although the most recent one should be regarded as not yet completed (indicated in note 1 in the tables that follow). An inspection of chart 1 suggests that the first recovery period can be usefully broken up into two subperiods, with unemployment falling at a faster rate in the first subperiod (July 1992 to March 1995) than in the second (April 1995 to December 2000).
[GRAPHIC 1 OMITTED]
Changes in the unemployment rate
This section discusses features that are common across all subperiods, features common to recessions, changes in net flows over the business cycle, and particular characteristics of the most recent recession. The discussion begins with the presentation of a simple, but general, framework that relates movements in the unemployment rate to the sizes of flows into and out of the unemployment pool. Gradually, the model is expanded to incorporate more details of the flows. As previously noted, the focus is on flows of persons and on net flows between three states: employed, unemployed, and not in the labor force. Clearly, this is but a first step toward a more disaggregated and detailed analysis, but the model can readily be generalized to explore the relationship between changes in any ratio and net or gross flows and to disaggregate data by gender, age, and other categories.
The change in the unemployment rate is defined as
[DELTA] (U / LF) = [U.sub.+1] / [LF.sub.+1] - U / LF, (1)…