Will the recovery be "jobless"?

A lot of economists are worried that the recovery from the 2007-09 recession, like those following the 1990-91 and 2001 recessions, will be "jobless." Mark Thoma, for example, writes

Historically, a good rule of thumb has been that the peak in unemployment lags the trough in economic output by a quarter. Thus, if output begins recovering in the second quarter of the year, unemployment would not begin to decline until the third quarter... But as shown in these graphs of the unemployment rate and NBER-dated recessions, this changed with the 1990-91 and 2001 recessions... in the most recent recessions, the peak in the unemployment rate came a year or more after the trough in GDP... An important question is whether this new relationship between unemployment and output observed in the last two recessions will also be present in this recession. All indications are that it will.

I don't think there's any mystery here. The reason that unemployment was slower to fall after the last two recessions than in previous recoveries was that GDP growth was anemic. The average growth rate in the first four quarters of the recovery following the 1974-75 recession was 6 percent and after 1981-82 it was 5.5 percent. By contrast, in the first year of recovery following the 1990-91 recession GDP growth averaged 2.6 percent, and after 2001 it was 2.2 percent. Okun's Law says that the unemployment rate falls when growth exceeds its average rate, which from 1973-2009 was about 2.7%. If you think growth will be higher than this for the next year, you should expect to see unemployment begin to fall; if you think unemployment will continue to rise, you must believe GDP growth will be lower than 2.7%.

The graphs below demonstrate that the recoveries from the 1990-91 and 2001 recessions were not unusual. Suppose that in the recoveries following each of the last four major recession we wanted to predict the path of the unemployment rate based on GDP growth. A version of Okun's Law should do the trick: the change in the unemployment rate this quarter should equal one half the difference between last quarter's GDP growth rate and the post-1973 average growth rate (2.7%). (Note: because we're dealing with quarterly data here, we need to divide the annual growth rates by four before applying Okun's Law).

The graphs below show that Okun's Law does a great job predicting the path of the unemployment rate following the 1974-75 and 1981-82 recessions. After four quarters of growth beginning in 1975Q2, Okun's Law predicts an unemployment rate of 8.0%; the actual rate was 7.7%. After 12 quarters, Okun's Law predicts 6.2%, the actual rate was 6.3%. Same thing for the recovery beginning in 1982Q4. Four quarters in Okun's Law predicted an unemployment rate of 9.9%, actual was 9.4%; 12 quarters in Okun's Law predicted 7.0%, actual was 7.2%.



Was the pattern any different following the most recent two recessions? Not really. When output began to grow in 1991Q2, the unemployment rate was 6.8%. For the next four quarters GDP grew by 2.6%. Okun's Law would predict a slight increase in the unemployment rate, and indeed the rate rose to 7.4%. Twelve quarters into the recovery, a period during which GDP grew at an average rate of 3.1%, Okun's Law predicts an unemployment rate of 6.4%; the actual rate was 6.6%. (There was a divergence in the second year of recovery, but it disappeared in the third.)

The same pattern held following the 2001 recession. When GDP began to grow again in 2001Q4, the unemployment rate was 5.5%. After four quarters during which GDP grew at an average rate of 2.2%, Okun's Law predicts the unemployment rate should have increased to 5.6%; the actual figure was 5.7%.
In the first 12 months of recovery GDP growth was 2.7%. Okun's Law predicts that the unemployment rate should have ended up exactly where it had started, at 5.5%; the actual rate was 5.4%.

Despite all the structural changes that have occurred in the economy in the last few decades, Okun's Law remains a remarkably stable relationship. Until I hear convincing stories explaining why Okun's Law has suddenly broken down after all these years, I'm going to assume that the pace of decline in the unemployment rate for the next few years will be dictated by the strength of growth in GDP.

For reasons I have given in a number of posts below, I believe GDP growth is going to be about 4 percent over the next year. We've already had one quarter of 3.5% growth. Okun's Law predicts that 2009Q4 unemployment will be 0.5*(3.5-2.7)/4 = 0.1% below the level of 2009Q3, or 9.5%. If growth is 4% from 2009Q3-2010Q2, by 2010Q3 the unemployment rate should be down to 8.9% and down to 8.7% by Q4. Now these numbers are probably optimistic given that unemployment rose to 10.2% in October. If we assume unemployment peaks at 10.2% in 2009Q4, then four quarters of 4% growth from 2009Q4 to 2010Q3 brings us down to 9.5% by 2010Q4. Let's split the difference between optimistic and pessimistic scenarios: unemployment will be 9.1% by 2010Q4.

That's a grim unemployment forecast however you slice it. It calls for some more aid to states, extension of unemployment insurance, and so on. But it doesn't mean that there's been a dramatic change in the laws of economics.