The Jobless Rate

This article from the Cleveland Fed compares the jobless rate for men and women to the unemployment rate. The jobless rate entire working-age population in the estimate labor market weakness, not just the labor force as in the unemployment rate calculation (the difference includes factors such as discouraged workers). There are interesting movements in the jobless rate over long periods of time as well as in response to economic conditions.

The one part that I have a bit of trouble with is the estimates of the trend jobless rate using the Hodrick-Prescott (HP) filter. What is the HP filter? As Paul Krugman would say, don't ask. You want to know anyway? Well, OK. It is a means of separating a time series into trend (long-run) and cycle (short-run movements around the trend) components. However, one of the parameters in the filter must be chosen by the econometrician and, depending upon the choice, it will appear that movements in the series are mainly caused by movements in the trend, or by movements around the trend. For example, using output data for the U.S., it's possible to make the Great Depression look like a trend event rather than a deviation from the long-run, full employment path for the economy. There are conventions about which values of this parameter to use for GDP, but it is still ad hoc and the answers you get depend upon the choice you make (see here and here). That is one of the big shortcomings of the HP filter.

Let me say this another way. The HP filter is a mechanical, statistical means of decomposing a series into trend and cycle components, and the decomposition is ad hoc. For example, look at the very end of the last graph below. With a more sluggishly evolving trend component, the jobless rate would be above trend rather than cycling around it, and has a significant impact on the interpretation of recent movements in the jobless rate - things would look much worse. Without some structural based method of estimating the trend that captures its key movements, there is no way to know which view is correct:

Gender Differences in Employment Statistics, by Yoonsoo Lee and Beth Mowry, Cleveland Fed: One key measure of the condition of an economy’s labor market is the unemployment rate. Against the backdrop of slowing economic growth and the net loss of 260,000 jobs since the beginning of this year, some see recent unemployment data as further evidence of an economy in distress. The unemployment rate has inched slightly higher in the past few months (4.9 percent in January to 5.0 percent in April) and over the course of the past year (it was 4.5 percent last April).

Historically, though, the rate is actually somewhat low. At 5.0 percent, it is lower than the average for most years since the 1970s. Still, there is some concern that the low rate does not necessarily imply a strong labor market because of how the unemployment measure is calculated. For instance, the unemployment rate measures only a subset of employable people—those in a country’s workforce who are over the age of 16, do not currently have a job, and have been actively seeking work in the past month. It does not account for discouraged workers, workers who want a job but are not currently looking due to adverse job market conditions, or part-timers who would like full-time work if it were available. A sudden increase in the number of discouraged workers, for instance, could artificially lower the unemployment rate by reducing what the government calls the “labor force,” or the pool of people who either have jobs or are actively searching for one. By looking at the labor force instead of the entire work-eligible population, the unemployment rate is intended to keep people who are not interested in working from affecting the calculations.

But some observers are concerned that the proportion of discouraged workers among these unaccounted-for workers may be growing, making the unemployment rate a less accurate measure of true labor market conditions. One alternative measure of employment conditions is the jobless rate, defined as the percentage of the population without a job. Unlike the unemployment rate, its denominator is the entire working-age population, not just the labor force, so it does not have the problem of a denominator that fluctuates over the business cycle. Recently some observers have noted that the jobless rate for prime-age men is historically high.

The unemployment rate and the jobless rate for men followed each other closely up to the early 1980s. Both rates went up in recessions and down in recoveries. Both were low in the 1950s and 1960s and high in the 1970s and early 1980s. However, since the 1980s, the unemployment rate has followed a downward trend, while the jobless rate for men has continued its upward trend. The jobless rate for men ages 25-54 is currently 13.4 percent, whereas in the late 1940s it was just 5.6 percent.

It is important to note, though, that there is a clear long-run rise in the trend due to certain structural changes in the labor market, such as the increasing role of female workers. It therefore may not be appropriate for purposes of assessing labor market conditions to make simple comparisons of the levels without adjusting for the trend. In other words, one needs to distinguish changes in the jobless rate that are due to temporary fluctuations from those due to permanent changes. A comparison of the actual jobless rate for prime-age men to the trend (estimated here with the Hodrick-Prescott filter), shows that the rate has risen above the trend recently, but the deviation appears to be small when compared to those that have occurred during recessions.

The jobless rate for women shows quite different patterns than the men’s. The women’s rate has declined from 67 percent in 1948 to just 27 percent today. Additionally, it has fluctuated less than men’s over that time span.

For women, the unemployment rate has trended slightly downward over the long run, while the jobless rate has fallen quickly. This suggests that more women have entered the labor force during this period, and the proportion able to find jobs to those unable has held fairly constant, as shown by the unemployment rate.

The difference in jobless rate patterns between men and women can be explained in part by differences in their labor force participation. Women entered the labor force rapidly after World War II. Their entrance accelerated in the 1970s and stalled in the 1990s. This prominent change in the female workforce has led the overall jobless rate to decline from 37.4 percent to 20.4 percent, despite the increase in the male jobless rate alone. Male labor force participation, by contrast, has declined slightly, from 96.7 percent to 90.4 percent.

Despite these differing patterns in indicators between genders, the total jobless rate and total unemployment rate are quite low, historically (around 20 percent). At a time when the economy is losing jobs, as it has been since January, we should wonder whether the low rate of unemployment is concealing a growing demographic of discouraged workers. Given the shortcomings associated with any single measure, it would be more informative to look at multiple indicators when assessing employment conditions.

The jobless rate is a useful indicator because it overcomes this problem of movement in and out of the labor force. However, making assessments of the health of labor markets with the jobless rate should be done with caution because of existing structural changes and long-run trends. For that reason, short-run changes in the rate may be more important than levels. There seems to be an increase (deviation from the trend) in the total jobless rate, but the deviation appears to be small when compared to those that have occurred during recessions. The cyclical component, in other words, does not show substantial deviation from the trend line.