On the first Friday of every month, the U.S. Department of Labor announces the unemployment rate for the previous month. Probably the most reported piece of economic data, this number, 4.1 percent for October 1999, is widely considered to be the single best indicator of labor market strength.

But behind this single statistic are several million individuals. Their reasons for unemployment and the length of time they spend unemployed vary greatly. There are young people looking for their first job and older workers who may have been downsized or quit voluntarily. Some will be recalled after a temporary layoff, others will end up in a new job, and still others will eventually leave the labor force.

What can details of this sort add to our ability to assess how tight the labor market has been and how fast the economy can grow? A spell of unemployment that ends with a discouraged worker leaving the labor force may, for example, have different significance than one that ends with a job. An increase in the unemployment rate caused by people voluntarily quitting may signify something different for labor market tightness than one caused by involuntary terminations. We have begun to look at these differences, in hopes of getting a better gauge of labor market conditions. We may also be able to shed light on some puzzling patterns in the recent unemployment data.

WHY HAS THE LENGTH OF TIME THAT PEOPLE ARE UNEMPLOYED REMAINED SO HIGH?

The median length of a spell of unemployment in 1990 was five weeks. This figure jumped with the onset of the 1990 recession and rose to nine weeks by the end of 1992, one year into the current expansion. By early 1998, in the midst of what otherwise seemed to be a booming economy, the median length of unemployment was still seven weeks.

This is somewhat surprising With the unemployment rate at 4.1 percent, we would expect the amount of time spent looking for work to have dropped back to its pre-recession level. Instead, the median length of a spell of unemployment (compared to the unemployment rate) has been at historically high levels for most of the 1990s.

One obvious cause of the increased duration would be a lack of jobs, but clearly this has not been the case. Looking at unemployment by groups based on their eventual labor market outcome, however, yields some clues. Unemployed workers who eventually stop searching for work and leave the labor force spend a longer time unemployed - about two weeks more - than unemployed workers who eventually find a job. And since 1991 this gap has widened, as unemployment duration increased for workers who ultimately left the labor force. At the same time, their share of all unemployed workers grew. Both trends help explain why, even in this strong labor market, people are spending a longer time in spells of unemployment.

Still, we are not sure why these changes have occurred. It could be that a greater share of the unemployed workers in the 1990s have outmoded skills, are older, or are hoping for jobs that offer dramatically higher wages, and they are therefore less likely to be successful in their job search. Or it may be that some workers feel more optimistic and are willing to spend more time searching for work, because of the strength of the economy. It will take additional research to sort out the possible explanations. And once we know more, we may also be in a better position to assess labor market tightness than with the overall unemployment rate alone.

WHY IS THE UNEMPLOYMENT RATE SO HIGH?

This may seem like an odd question, considering the recent historically low rates of unemployment in the United States. But given our recent high rates of growth in GDP - rates that by most accounts are in excess of our long-run potential growth rate - most economists would have predicted even lower unemployment rates than we currently enjoy. Some have suggested that productivity (which is hard to measure accurately) is growing faster than we think. If productivity rose, it would mean that the recent hefty rates of GDP growth could have occurred without exhausting the supply of available workers, and this could explain why the unemployment rate is still relatively high.

An alternative explanation focuses on differences in how people become unemployed in the first place. Grouped by reason for unemployment, individual unemployment spells vary in duration. Workers who are terminated by their firms (as opposed to those who quit or are temporarily laid off) have the longest spells of unemployment - more than 16 weeks following the 1982 and 1990 recessions. In contrast, those who have been temporarily laid off spend only three to six weeks unemployed, regardless of whether they ultimately get a job or leave the labor force.

Our research shows that after a recession begins, the median duration of unemployment first increases and then eventually declines, for most groups. But workers who've been terminated continue to experience increases in the length of time they spend unemployed - the last indicator of labor market weakness - until just before GDP growth surges and drives down the unemployment rate in the subsequent expansion. After the 1990 recession, their median duration reached a peak of 17 weeks as late into the recovery as 1993, had only dropped to 11 weeks in 1997, and was still around eight weeks in 1998.

Thus, the unusually long spells of unemployment in the 1990s, particularly for workers who were terminated by their firms and eventually stopped searching for work, may account for the higher-than-expected unemployment rate. This detailed information about why and how long people spend unemployed can improve our ability to gauge labor market conditions.

Hoyt Bleakley is a PhD candidate at MIT and Visiting Fellow at the Boston Fed. Ann Ferris is Research Associate and Jeffrey Fuhrer is Vice President and Economist at the Boston Fed. Their article, "New Data on Worker Flows During Business Cycles,” appeared in the July/August 1999 issue of The New England Economic Review.