Changes to data collected on unemployment duration
Effective with data for January 2011, the Current Population Survey (CPS) was modified to allow respondents to report longer durations of unemployment. Prior to that time, the CPS accepted unemployment durations of up to 2 years; any response of unemployment duration greater than this was entered as 2 years. Starting with data for January 2011, respondents were able to report unemployment durations of up to 5 years. This change affected estimates of average (mean) duration of unemployment. The change did not affect the estimate of the number of unemployed persons and did not affect other data series on the duration of unemployment.
There was an unprecedented rise in the number of persons with very long durations of unemployment during the recent labor market downturn. Nearly 11 percent of unemployed persons had been looking for work for about 2 years or more in the fourth quarter of 2010. Because of this increase, BLS and the Census Bureau updated the CPS instrument to accept reported unemployment durations of up to 5 years. This upper bound was selected to allow reporting of considerably longer durations while limiting the effect of erroneous extreme values (outliers).
The new upper bound of 5 years for reported unemployment duration was phased in over the first 4 months of 2011, as the duration question is only asked of a portion of those unemployed in any given month. (The question is asked of unemployed persons who were not interviewed in the prior month and the newly unemployed. Duration is updated automatically for unemployed respondents who remain unemployed the following month.) By April 2011, all households were able to report the new duration upper limit.
Impacts on published estimates
Because of the previous upper limit on recorded values, BLS cannot determine the duration of unemployment for persons who had been unemployed for longer than 2 years for data prior to January 2011. Monthly estimates of average (mean) duration of unemployment for 2011 produced using the 5-year upper limit are higher than those using the 2-year upper limit. (See tables below.) Only the average (mean) duration of unemployment was affected by this change in data collection. The median duration of unemployment was not affected by this change, nor were distributions of unemployment by weeks unemployed.
BLS incorporated the new data into the existing official, published data on unemployment duration beginning with data for January 2011. Consequently, there is a break in series for average (mean) duration of unemployment effective in January 2011, though the full effects of the break in series were not evident until April 2011, when the entire CPS sample was able to report the new upper limit of 5 years.
BLS also tabulated, for research purposes, average (mean) duration of unemployment data that replicate the previous 2-year upper limits. These tables allow BLS and data users to gauge the effects of the new upper limit on the average (mean) duration of unemployment. The tables below compare average duration using the previous (2-year) and new (5-year) upper bounds. These tabulations were produced through June 2011.
A comparison of average (mean) unemployment duration using previous and new upper bounds, January–June 2011
Public use microdata
Through March 2011, the CPS public use microdata files (produced by the Census Bureau) contained variables on duration that were restricted by the upper bound of 2 years. With the completion of data collection for April 2011—at which point the 5-year upper bound was fully phased in—the variables for the public use files reflect the 5-year upper bound. (Public use files prior to April 2011 will not be revised to reflect the 5-year upper bound.) Duration values are subject to top-coding in the CPS public use microdata files.
Last Modified Date: July 8, 2011