The 3rd quarter 2009 GDP growth figure came in at 3.5% annually, but as both Dean Baker and Paul Krugman have pointed out, historically this growth rate does not correspond to significant employment-rate improvement.
Krugman gives a plot of GDP and unemployment change over the last 60 years, but going back further in time the relationship is even better. This shows the correlation of year-over-year change in GDP with change in unemployment rate since 1922.
This correlation is one of the best in economics over this time period (R = 0.89) and is especially good on the growth (right) end. The intercept on the x-axis is 3.37%, meaning that improvement in unemployment rate should not be expected if yearly GDP growth is not greater than this.
Also obvious from the color-coding is the moderation over time in swings of both GDP and unemployment - all the points above +10 and below -5 dGDP are before 1950. Ben Bernanke has taken credit for this moderation on behalf of the Fed, but this may be equivalent to the witch doctor taking credit for the sunrise. Unfortunately if we look at average rates of change during the recession-recovery cycles as defined by the NBER (red is "recession", black is "recovery") we see that what has been moderated most is unemployment recovery rate.
There has always been a pronounced asymmetry in unemployment change rate - unemployment goes up very fast in the down-cycle and recovers more slowly. This is not getting any better. Krugman has pointed out that in the previous two recessions unemployment has not even begun to turn around until well after GDP did (and after the NBR called the "recession" or down-cycle over). The current climb in unemployment is probably not over yet (end October 2009), and a rapid recovery should not be expected.
Looking at actual year-over-year change in unemployment rate, by month (pre-1948 values were obtained by interpolation from yearly averages), instead of averages, tells the same story with a some significant details:
In this diagram, anything above the zero line represents increasing unemployment rate - unemployment does not begin to decrease until the line is crossed going down. The small extent of the blue areas of overlap between increasing unemployment and NBER recessions show that the NBER has usually called the end of the recession at or before the point at which the job loss rate has turned around (the positive peaks in Fig. 3), not when unemployment rate has turned down (when the curve crosses zero).
GDP has almost always turned around before unemployment. Since the NBER apparently bases its cycle dates primarily on GDP this can be misleading for the great majority of people, who think of recessions in terms of unemployment.
After the recession of 1981-82 there was a pretty large spike in unemployment recovery, but this did not last long. What is not shown in Figure 3 is that unemployment was already high before 1981 because of incomplete recovery from the preceding recessions in 1970, 1974 and 1980. This is quite clear from the plot of unemployment itself:
Reduction from that level took a long time and the rate was slow - in fact a return to late-60's level is now many years in the future, if it ever occurs. A series of recessions from 1945 to 1960 also elevated the level of unemployment, which was only slowly reversed during the boom of the 60's
The peaks of unemployment in the recessions of 1990 and 2001 were not high but broad in time, which displaced the turnaround point (when the curve passed through zero in Fig. 3 or when the curve turned down in Fig. 4) even further from the GDP turnaround.
Over the last 30 years or so the difference in turnaround between GDP and employment seems to have become more pronounced. Why is this? One possible reason is that GDP has become increasing based on the financial industries, which thanks to government subsidies and bailouts (don't forget the banking and S&L crisis of the 80's-90's) have been able to get back into business easily. Reduction of short-term interest rates, which is what the Fed does in a recession, is a boon to all types of leveraged speculation. There has also been an overall trend of decreasing employment in domestic manufacturing and non-financial services, as these have been outsourced to countries where labor is cheaper. Domestic employment has been swimming upstream against this flow.
Although unemployment recovery rate was actually quite high after the bottom of the Depression in 1933, the rate in the early 40's (Figure 3) was obviously boosted by WW II. There was a short recession and climb in unemployment at the end of the war, but the unemployment rate just went up a little from the unrealistically low war-time rate. There was a small spike in the recession of 1949, but recovery was fast (Figure 4).
In other words the huge government "stimulus" of the war, which removed the last of the Depression unemployment, really caused no major problem when it ended. Not only did the private economy immediately pick up the war-industry jobs, but also the returning servicemen (although some women and other war-time workers retired from the work force after the war).
Private spending and investment had nothing to do with the employment recovery during the war (tax rates were very high), but when the war was over they surged back immediately and the economy kept growing at a high rate for many years. The huge debt of the war was made insignificant by this growth.
10/31/09