Notes: URAP Project 3: Fall 2005: Analyzing the Cyclical State of the Labor Market

Looking at the unemployment rate, today's U.S. labor market looks relatively strong:

Looking at the fraction of people at work, today's U.S. labor market looks relatively weak:

If the U.S. labor market is relatively strong, it is appropriate for the Federal Reserve to keep raising interest rates in order to prevent the emergence of an inflationary spiral and the resulting creation of fear that the Federal Reserve is not an effective guardian of price stability. If the U.S. market is relatively weak, there is no need to fear wage-push inflation, and it is appropriate for the Federal Reserve to cut interest rates in order to stimulate demand and get the economy closer to full employment. But our two main indicators disagree.

One way to gain more information about what is "trend" and what is "cycle" is to take a look at other time series indicators that we believe have a similar cyclical component. When the labor market is cyclically strong, we believe that the unemployment rate will be lower than trend, that the employment-to-population ratio will be higher than trend, that average hours worked will be relatively high (since firms are likely to increase both bodies and hours when their demand for labor is strong), and that the average duration of unemployment will be relatively low (because more of the fluctuations in quits, firings, and hires that drive the employment side of the business cycle are on the hires side). We know what unemployment and the employment-to-population ratio look like. What do these other two series look like? Here they are:

The issue at stake is essentially whether the past five years have seen a stable value for the trend unemployment rate and a fall in the underlying trend for the employment-to-population ratio (meaning that our current employment-to-population ratio is actually close to, not far below, the long-run trend) or whether the past five years have seen a stable value for the trend employment-to-population ratio and a fall in the underlying trend value for the unemployment rate. The weekly hours series and the unemployment spell duration series seem to vote with the employment-to-population ratio: three series seem to say that the current cycle component is large, that there has been only a smell recovery from the business cycle trough levels, and that we are still pretty far away from full employment. Only one indicator--the unemployment rate--seems to say that recovery is well advanced and that the cyclical component has substantially shrunk.

This, however, doesn't resolve the mystery: why is the indicator that is the unemployment rate giving a different signal? What has happened to keep workers whom we would have expected--given the behavior of unemployment spell duration, hours, and the employment-to-population ratio--to say that they are unemployed from saying so when the CPS interviewers come to call?

Some serious quantitative work on this would be genuinely useful--and could easily be done within the framework of a URAP

Comments

[pgl]

That the unemployment rate and the employment to population rate are giving mixed signals is something I noted over at Angrybear yesterday. During the prior two recession & recovery periods, they gave the same signals (roughly). So this period is different with the key being why has the labor force participation rate declined. David Altig says real business cycle shocks. I say - your attention to real wage behavior is one key to unraveling this, and then I note that David's RBC explanation does not hold water.

[JohnDL]

Two posibilities:

1: The BLS publishes several unemployment statistics; this one, which is the best-case scenario, subtracts nearly everyone not actively searching for work yet leaves in temps and PT'ers, and for current cultural/economic reasons these artificialities may mean that it does strictly align with what it measured in past decades. Temps have grown as a percentage of the workforce from a minor figure to a pretty sizeable one, and that alone could fox the resuts a bit, as they are considered employed even if they haven't worked in several days. So have PT'ers. Several large retailers, K-mart for example, do not hire full-time people (when possible) in order to avoid legally mandated costs associated with full-time employees. Ditto restaurants, etc; these service jobs make up a majority of the new jobs added to the economy since ~1990.

It would be interesting to take some of the other unemployment indicies the BLS publishes and see if they track the other macro data any better. It might pay also to look at income volatility, which has also risen sharply, as it is in some respects a measure of the "style" unemployment.

2: The second possibility is that the statistic is a lie. This administration has politicised damn near everything else; why should we trust this, especially as the admin has a very good political reason to issue a low number?

[knzn]

The most promising explanation, I think, is that low liquidity in the labor market has reduced the incentive to participate in the labor force. (In other words, in a quiescent labor market, new jobs are hard to find, so why bother to look?) Take a look at data on the BLS’s “Business Employment Dynamics” data on job creation and destruction (which I have presented graphically at http://www.flickr.com/photos/67919949@N00/36255919/ ) The last few years have been characterized by very low rates of both job creation and job destruction. If you already have a job, you’re likely to keep it, but if you don’t have one, there aren’t many new jobs available. There may even be a positive feedback mechanism (a “vicious cycle” if you like) here: fewer new jobs reduces the incentive to search, and fewer job seekers makes it harder to fill new jobs, thus reducing the incentive to create them.

[kharris]

The jobless rate and the participation rate (obviously) interact very strongly. I have seen several efforts to combine the two, to calculate a steady-participation-rate jobless rate, but never over several cycles. That might not distinguish cycle from trend, but it sure would take a lot of noise out of the jobless rate series, and give us a better single-number handle on how many people have a job.

I, by the way, and too lazy. This is a suggestion for others.

[wkwillis]

Possibly the real estate boom has enabled some people to devalue employment, especially if they are older and finding a well paying job is harder. My father is seventy three and unable to work since 1968. He took a sales job at Sears just to get back into work habits and increase his sales skills for interviews.

[spencer]

I wonder if there is a long term demographic at work here. The women having children now are the generation that the baby boomers had some 20-30 years ago. They are also the first children that were raised in an environment where it was the norm for mothers to work. Are we getting a negative feedback from these women in that they are deciding it is more important for them to stay at home and raise their children then go to work?
20 to 30 years ago womens liberation was in its early stages and partially because of this women largely made the decision in favor of work. But are their daughters now making the decision in favor of the children?

[Sebastian Holsclaw]

I'm curious about a partial demographic explanation. If you take the "employment to population rate" by age, what does it look like in narrow age bands over 30 years? I'm fairly certain that the employment to population rate of 80 year olds is pretty low compared to people in their 40s. But how is it compared to itself over 30 years?

[cm]

FWIW, here my collection of theses and speculations:

* The BLS statistics are sound, and provide correct estimates based on their definitions. The perception that their indicators don't square with reality is mostly caused by their definitions of terms being subtly or significantly different from "common language" definitions. U-6, as opposed to U-3, is closer to *my* definition of unemployment. "Self (un)employment" masks a good amount of people unable to secure formal jobs.

* All jobs are not the same. E.g. payrate, work schedule, benefits. ("Would you like fries with that?")

* I know people and of people who have adjusted to and found partial comfort in the idea of long-term or even life-long unemployment, and doing things they enjoy or that they could never get around to while on the rat race. Having had a lucky hand in past investments and house finance definitely helps. Even with subsistence & modest comfort taken care of, a big, if not the biggest, problem is adequate or reasonable healthcare insurance. Even a good amount of money in the bank will not buy that. I suspect that's a strong motivator for people 40+ or 50+ to seek employment (not just "a job").

* In a number of gray/white-collar professions, job growth abroad is rampant, and some domestic companies have instituted policies prescribing hiring quotas, or domestic hiring freezes. Yes, knzn, this would make for a quiescent labor market in those areas.

* Anecdotal evidence suggests high turnover at local retail chains, and generally "lower-rung" positions. Not clear what this means. Is the churn employer of employee driven?

* spencer -- People make choices, and take the path of least resistance. Given progressive tax rates, lower wages/salaries/career discrimination for women, daycare cost & troubles, and the "cost of doing business" (second car/commute), the marginal income from wives/SOs may indeed become "marginal".

* Sebastian -- BLS by-age employment shows a very well defined peak around birth year 1960. That corresponds to a current age of about 45. July YOY non-adjusted job growth was 80% at ages 45+.

[Sebastian Holsclaw]

"Sebastian -- BLS by-age employment shows a very well defined peak around birth year 1960. That corresponds to a current age of about 45. July YOY non-adjusted job growth was 80% at ages 45+."

Right, that is what I would expect. But that is only the first part of my question. I'm curious to see how

employment % age 45 in year 2005

compares to

employment % age 45 in year 1995

compares to

employment % age 45 in year 1985

compares to

employment % age 45 in year 1975

Ideally you would conduct this analysis for at least 4 or 5 different age groups. (My guess would be that 18, 30, 45, 50, and 60 would be revealing).

(Technically to be really accurate you might want to adjust for changes in life expectancy, but my guess is that they haven't been dramatic enough in the past 30 years to go through the work to control for them.)

Not being an economist, I have no idea where to get such data, but my intuition is that analyzing such data would help to answer the kind of questions that Brad is asking in that it might reveal if there is a societal shift going on, and if it is, where the shift is located.

[Sandwichman]

Unemployment is a dummy variable.

My brother-in-law works a few days a week some months of the year for health insurance. I'm currently copy editing a book a few hours a day for a handful of Euros. The dummy just gets dumber.

Years ago, when I was living with my ex, I could never succeed in convincing her that based on our income bracket, our coop housing situation and actual spreadsheet calculations we would be financially better off if she would just turn down a few more shifts a month. She kept thinking: more work = more money, period.

Leopard Geckos and Gerbils are both small pets. Just for fun this morning I put the Gerbil cage over by the Gecko's terrarium when the Geck was out of burrow. The gecko pulled himself up on the glass and stared in amazement at the furry creatures just about his own size. The rodents didn't seem to take any notice of the herp. Two gerbils + one gecko = three small pets.

[knzn]

Sebastian, You can get most of the data from the “Customized Tables” application on the BLS Web site:
http://data.bls.gov/PDQ/outside.jsp?survey=ln (Or go to the BLS home page at www.bls.gov and click on “Unemployment Rate” under “Employment and Unemployment”, and then scroll down the next page to “GET DETAILED CPS STATISTICS”.)

[cm]

knzn: Excellent link.

Sebastian: In addition to knzn's link, you may use these as well (they won't break down ages 55+ though).

http://www.bls.gov/webapps/legacy/cpsatab6.htm (CPS table A-6 statistics)

Select what you want, and on the next page go to "More formatting options" and fine-tune. Don't forget to include graphs, they are quite nifty.

http://www.bls.gov/data/home.htm (CPS statistics main page)

http://www.bls.gov/cps (CPS main page)