Thursday, April 3, 2008

Employment Recessions.

Awhile back, there was a lot of controversy about whether or not the U.S. is in a "recession" at present. Officially, we're not (as far as we know) while most economists now seem to think that the current period will be likely to be dubbed a recession when the NBER gets around to it. Most of the non-economists seem to thin we're in one.

One generally-ignored dimension concerns the actual definition of a "recession." The "official" definition of a "recession" comes from the very-mainstream National Bureau of Economic Research's Business Cycle Dating committee. Journalists summarize their (relatively complex) definition by saying that having real GDP fall during two or more quarters in a row defines a recession.

The problem with these definitions -- especially the journalistic one -- is that the emphasis is totally on production sold through the market. That's what GDP is all about, and no more. This might be thought of as a totally capitalist definition of a recession.

Most workers, on the other hand, instead care about the growth of paid employment, i.e., the availability of jobs. So I decided to find "employment recessions" in the U.S. since World War II. As I define these animals, these are quarters where employment figures fall for two or more quarters in a row.

This is like the standard journalistic definition of a recession in terms of ease of calculation. The the NBER one is so hard to calculate. That version also reflects employment measures, I believe. My version also coincides with the common or "oral tradition" idea of a growth recession, in which GDP growth slows but does not turn negative, hurting employment.

I added one adjustment: the fall of employment is measured relative to the trend growth in employment in order to get a sense of the cycle, not the trend. (The trend rate has been falling over the decades, but that's another topic.)

Below, I listed "employment recessions." But let's jump to my conclusions:

1) There are several employment recessions that do not coincide with NBER recessions: in 1951/2, late 1959, late 1962, and early 1986. Somehow these receive much less press than the official ones do.

2) Employment recessions often begin before NBER recessions: in early 1957, late 1979, early 1981, early 1990, late 2000.

3) Employment recessions almost always end after NBER ones. The exception was in 1980. (It's possible that President Carter, who was running for re-election, begged Volcker to cease and desist.) This problem has gotten worse, with the "jobless recoveries" that followed the two Bush recessions that have occurred so far.

4) We're already in an employment recession, starting in the fourth quarter of 2006.

Here's The Complete List. Since 1951, we get the following "employment recessions" listed by year/quarter: dated by start and finish. In each of these quarters employment fell relative to trend growth, either preceded by or followed by another fall in employment relative to the trend. Note: the measures do not indicate the _depth_ of the employment recession, only its length.

1951/3 - 1952/3 -- does not coincide with an NBER recession.

1953/2 - 1954/4 -- coincides with an NBER recession, but ends two quarters after it.

1957/1 - 1958/3 -- begins 2 quarters before the NBER recession, while ending one quarter after it.

1959/3 - 1959/4 -- does not coincide with an NBER recession.

1960/2 - 1961/2 -- coincides with an NBER recession, but ends one quarter after it.

1962/3 - 1963/1 -- does not coincide with an NBER recession.

1969/4 - 1971/3 -- coincides with an NBER recession, but ends three quarters after it.

1973/4 - 1975/2 -- coincides with an NBER recession but starts one quarter after it and ends one quarter after it.

1979/4 - 1980/3 - coincides with an NBER recession but starts one quarter before it.

1981/1 - 1983/1 - starts 2 quarters before the NBER recession and ends 1 quarter after it.

1986/1 - 1986/2 -- does not coincide with an NBER recession.

1990/2 - 1992/4 -- starts 1 quarter before the NBER recession and ends 1 3/4 year after it. This is Bush the Father's famous "jobless recovery."

2000/3 - 2004/1 -- this one starts 2 quarters before the NBER one and ends 2 1/4 years after it. This is Bush the Son's repeat of the "jobless recovery." He outdid his father's example.

2006/4 - present. So far, it does not coincide with an official NBER recession.

Jim Devine


Econoclast said...

I'm commenting on my own note so that I can fix it so that later comments will be sent to my e-mail. (Blogger should be fixed so I don't have to do this.)

Bruce Webb said...

My Blogspot blog alerts me to every comment via e-mail.It may depend on whether you updated templates how you accomplish this but it can be done.

On my blog there is a 'Customize' tab between 'New Post' and 'Sign Out' at the upper right. If I click on that and then on 'Settings' and then on 'Comments' and then scroll to the bottom of the screen there is a box to enter an e-mail address to alert you to comments.

Otherwise you might need to click on 'Dashboard' and then select 'Settings' and proceed in the same way.

Bruce Webb said...

As to the post itself. I don't track employment numbers but I do track Social Security receipts which Treasury tracks monthly to the penny, so in principle giving a really good measure of how real wage earnings are performing. Through October it looked like we would smash through year end projections sometime in early November. When I finally checked actual year end numbers it turned out that while the DI component did come out ahead of projections that OAS lagged, not by much but surprisingly so. I didn't bother to try to track this month by month to see where the relative weakness started, but clearly Real Wages for people earning under the cap weakened in Q4 2007, no matter what the survey numbers show.

Social Security receipts for February should have come out on Tuesday. If you want an actual hard number proxy for employment wages you can get it with just a one month lag.

Chiasmus said...

Excellent stuff! Even better would be a graph showing NBER and employment recessions on a timeline--that might get some attention from other bloggers.

Michael Perelman said...

Very good work, Jim. I thought that employment was supposed to be a lagging indicator.

Also, if the quality of employment were taken into account, the picture would be grimmer.

Eleanor said...

So the 1950s had three recessions lasting 4 years in total. The 1960s had three recessions lasting 3 years. The 1970s had 3 recessions lasting 4 years, the 1980s 2 recessions lasting 2 years, the 1990s 1 recession lasting 2 years, and the 2000s thus far have had 2 recessions lasting 4 years.

The 1980s look better than I remember them, and the 1990s look great. What this does not track is the decline in good jobs and the deterioration of benefits such as unemployment. Way back, benefits were not taxed and you could collect even if you quit, though with a penalty period; and I could live quite comfortably on unemployment, even though I was working crappy office jobs. I used to think of unemployment as the working person's sabatical.