Here's the child's way to figure out the problems we have when measuring unemployment rates in the US right now:
Imagine that there's a fantastic event of some kind happening. A great concert or the sale of the century where you can buy a new car for ten dollars. This event is advertised everywhere and eager people start arriving at the location for it the previous day. They line up hours before, with packed lunches and blankets and even tents. Because the event is very popular, the line becomes immensely long.
Then the doors open and the first people are let in. Now think of this event as getting hired for a job. When the labor markets are bad the line is very long, very few people are allowed through the door at any one point in time, and more people arrive in the line than are let in during any particular length of time.
The people in the line are job-seekers, getting through the door is getting a job. Our most common measure of unemployment tries to capture the number of people in the line who are not getting in: add up the numbers of those already in and the people still in the line, then divide the number of people in the line by that sum and express the result as a percentage. That's how we get the usual measure of unemployment. Here's the summary information for some groups as it applies to December of 2013:*
Among the major worker groups, the unemployment rates for adult men (6.3
percent) and whites (5.9 percent) declined in December. The rates for adult
women (6.0 percent), teenagers (20.2 percent), blacks (11.9 percent), and
Hispanics (8.3 percent) showed little change. The jobless rate for Asians
was 4.1 percent (not seasonally adjusted), down by 2.5 percentage points
over the year. (See tables A-1, A-2, and A-3.
A shorter line looks better.
But when the job situation is very bad, there are other ways the line can shrink than getting through the door into the event (getting a job offer). People can get discouraged and leave. Some people might leave, whatever the job situation is, because they are not that keen on getting a job. But when few jobs are on offer, many more people get discouraged and end up leaving. The wait is too long, it is too cold, the door only opens once a day if that and only a few people get in. Some people stay in the line until they are moved by some external force. In the unemployment example these would be the long-term unemployed.
Others go and join a line for some other event, such as deciding to go back to school, getting into the military, trying to figure out if they could be in line for disability pension etc. Some try to make a living selling things on eBay. The long-term unemployed are the ones who hang on, staying in the line but not being admitted, for days, months and years.
In the past recessions the idea of measuring unemployment by looking at the number of people in the line divided by the total number inside the place of the event plus those in the line worked fairly well. That's because the rate at which people left the line didn't seem to be vastly influenced by the rate at which the doors opened for the event but by more longer-term events than temporary recessions.
That tendency to leave can be captured by the labor market participation rate (LMPR)**: the percentage of working-age population which is in the labor force (either looking for work or working). That percentage is slipping:
The unemployment rate in December fell to 6.7 percent, the lowest since October 2008, as more people dropped out of the labor force. Economists were looking for the jobless rate to hold steady at 7 percent.
The labor force participation rate, which gauges the proportion of the working-age population in the labor force, slipped to 62.8 percent, down 0.8 percent from a year ago, and the lowest since February 1978.
It's important to try to figure out who it is, exactly, that is leaving. If the people leaving are young people who decide to get more education instead of an immediate job (which might not be available), the consequences are different than if the people who leave are seeking early retirement. The consequences are probably the worst if the people who leave are most likely to be in the middle of their working lives (say, between 25 and 54 years of age). That's because rejoining the labor market is much harder to do than many think and thus those individuals could be permanently out of the marketplace. The obvious question then is how they will survive.
I haven't found clear expert agreement on which groups are the largest among the labor market dropouts. This agreement may exist, but so far I've been unable to get a good reference for it. Some economists argue that the problem seems to be changing as we watch: A few years ago individual workers near retirement chose early retirement if they could, but now it looks like people in that position hold on to their jobs for dear life.
What does all this mean and why should we, the readers, wade through a boring post like this one?
Not sure. Perhaps you are a bit masochistic? Or perhaps this matters a lot, because if the commonly used measurement of unemployment underestimates the actual level of joblessness, then it also underestimates the suffering and by using it we think we are doing much better than we actually are.
There's also the possibility that something rather foundational may have changed in the last recession, that the recovery hasn't really been a recovery for workers but mostly for those who invest in the stock market or own firms. Profits have recovered rather beautifully but wages are essentially not budging and too many people are giving up on that place in the line. We need to know where they are going and we desperately need to know how to create more events (more jobs which can actually support a person) so that the recovery is real for all people.
Gender, Racial And Ethnic Differences
Here the individual Bureau of Labor Statistics tables can be useful. For instance, this table shows us the unemployment rates and the LMPRs for adult (over 20) white women (5.3% in December 2013) and men (5.6% in December 2013) and black women (10.4% in December 2013) and men (11.5% in December of 2013) of various age groups and also for Asian-Americans (4.1% unadjusted in December 2013). That last data is not yet available separately for men and women or for the group of adult workers (over 20) and not yet available in the adjusted form. The actual unemployment rates for Asian-American men and women over 20 are likely to be lower than that 4.1% figure, because the overall figure includes the usually much higher unemployment of the workers between sixteen and twenty years of age.
The LMPR has dropped in all the groups of adult workers the table shows, on average about one percent. But the drop for black adult men is roughly two percent, leading to the lowest LMPR for this groups in several decades. Keeping those drops in mind affects how we view the clear drops in unemployment rates over 2013. For example, the rate of unemployment for black adult men was 13.9% in December 2012, dropping more than two percent over 2013. But the LMPR for that group also dropped about two percent.
The data on the Hispanic population can be found in this table. It's not yet available in adjusted form. Note that Hispanics (or Latinos) can here be of any race. The unadjusted unemployment rate for adult women in this group was 8.1% in December 2013 and the same rate for adult men was 7.5%. The LMPR dropped for women over 2013 (by approximately one percent) but did not drop for men.
What's the point of looking at those detailed statistics?
First, there are clear ethnic and racial differences between groups of workers, and most of those differences are of longer standing than just the aftermath of the most recent recession. The reasons for them probably include both non-discriminatory and discriminatory factors (I haven't been able to find a good recent aggregate-level (not just one industry or job) study which would look at both of these at the same time).
Among the latter are geographic factors and average educational differences*** (in December 2013 the unemployment rate for people without high school diplomas was 9.8% whereas the unemployment rate for people with at least a Bachelor's degree was 3.3%; the LMPR for people without high school diplomas was 43.7 (!) whereas it was 75.3% for those who had at least a Bachelor's degree.).
But the main reason for the low unemployment rates of Asian-Americans is, I believe, the higher average education level of that group.
Second, note that the comparison of male and female rates of unemployment and the LMPRs shows that not all ethnic and racial groups follow exactly the same pattern. For example, the female rates of unemployment in December 2013 were usually lower than the male rates (though approaching each other, as is the case at the end of recessions), but this is not the case for Hispanics, where the female rate is higher than the male rate of unemployment.
Sometimes disaggregated data is more useful than aggregate data, to understand what it is that creates this difference among the Hispanics. My guess is that it might be a consequence of the lower levels of education among adult Latinas, which would reduce the range of jobs they can apply for. Those lower levels of education, in turn, could be based on traditional ideas of the past about the proper roles of men and women and what those require.
Third, it's useful to keep in mind that the unemployment rates of men and women, overall, tend to be roughly the same over longer time periods. They diverge in recessions, because men are more likely to work in the bellwether industries (construction, manufacturing, shipping), and those industries are the first to lose jobs when a recession begins (though also the first to regain them when the recession is over). The traditionally female industries are less sensitive to recessions, which is why we get the greater male unemployment rates when economic times are bad. This, too, may be changing in the latest recession, partly because of the conservative "belt-tightening" in the public sector jobs, given that jobs such as teaching have a lot of women. But overall the male and female rates appear to be converging.
This is it for my unemployment post. I have left so much uncovered, such as the question whether the new jobs now created are as good as the old jobs that were lost, the question of what is happening to earnings for all the different groups mentioned here, and also some deeper analysis for the reasons of the gender, ethnic and racial differences I rapidly summarized as well as some I did not (median earnings, say).
*More about those percentages below.
**The LMPR does change and has changed in the past, but the reasons have mostly been unrelated to recessions. For example, the rate of women's labor market participation rose rapidly in the 1970s and continued to rise until 1990s when it stabilized. This reflected a larger social change from the few preceding decades. Men's rate of labor market participation has declined slightly over roughly the same time period, having to do with the tendency towards earlier retirement and more time spent in education.
***Location matters, because some US states have been hit more by the effects of the recession than others, and some of those states have a high percentage of blacks or Hispanics. Note, also, that educational differences could also be discriminatory if the reason for having less education is in past discrimination in the education system. The case of the Hispanic workers is complicated by the larger number of first generation immigrants. Because they may have been obtained their education before they immigrated, the skills that provided might not be as good a match for the US labor market as educational skills obtained here. Because each country teaches its children for the kinds of jobs it has.
Globalization and outsourcing are factors which also may affect certain ethnic and racial and educational groups more. Good blue-collar (and traditionally male) jobs have packed their bags and moved elsewhere. This is not yet quite the case with the jobs which require more education.