Abstract

During the Great Recession, reported unemployment duration has increased to much higher levels than during earlier severe recessions. Previous studies using data from non-recessionary periods have found that part of this increase was caused by a 1994 redesign in the way the Current Population Survey measures unemployment duration. This paper uses data from recessionary and non-recessionary periods to determine how the size of the redesign’s effect on the distribution of unemployment duration changes over the business cycle. I find that, for most measures of unemployment duration, the redesign effect is relatively constant across business cycle conditions, but the redesign effect on median unemployment duration tends to shrink during recessionary periods. This suggests a constant adjustment factor will adequately correct for the CPS redesign for most variables I consider, but for median duration the use of a constant adjustment factor will lead to underestimating the true median duration during high unemployment periods such as the Great Recession.