Half a century ago, when business cycle research was blossoming, Arthur Okun explored the relation between GDP and unemployment during periods of recession and recovery. As a first approximation, one might have assumed that a one percent decline in output would be associated with roughly a one percent decline in employment and a one point rise in the unemployment rate. However Okun showed that this approximation was very wide of the mark. In what came to be known as Okun’s Law, he estimated that a one percentage point higher unemployment rate was associated with three percent less GDP. Thus if unemployment was 2 points above its full employment level, GDP would be 6 percent below its potential level-defined as the level of GDP at full employment.
The building blocks of Okun’s Law were several cyclical features of the economy and its job market. In a weak job market, many discouraged workers stop looking for jobs and thereby leave the labor force. So a cyclical decline in employment does not produce a corresponding increase in measured unemployment. Cyclical movements in aggregate output per worker also reduce the impact of GDP fluctuations on the unemployment rate. In part this comes because a disproportionate part of cyclical job loss come in high productivity industries like manufacturing. And in part it comes because, in all industries, firms experience procyclical variations in output per worker because many jobs have the characteristics of “overhead” labor—they are needed if the firm is in business at all rather than in proportion to the firm’s level of output. And even for production workers, whose jobs are closely tied to the level of current output, most firms vary hours worked per week as a buffer against layoffs.
The basic logic that underlies Okun’s law has proven to be robust over the decades. But the declining share of manufacturing in total employment and other changes in the structure of the economy have gradually altered the size of the cyclical effects. In recent decades, the Okun’s Law parameter, which had been 3 in the early postwar decades, has more recently been around 2.
So how big is the gap between actual and potential GDP in today’s economy? The math could not be simpler. The 4.6 percent unemployment rate in the two years preceding the Great Recession are a reasonable estimate of full employment. Unemployment in the second quarter of this year was 3 points higher than that, so the output gap was about 6 percent of GDP, or roughly $800 billion.
This measure of today’s output gap is quite consistent with estimates of today’s employment shortfall that my colleague Gary Burtless recently made using a different methodology. He estimated that employment today is about 5 percent, or 7.4 million jobs, below what it would be at full employment.
With the economy having been so far from full employment for so long, the error bands around both the output gap and the employment shortfall estimates are larger than they would normally be. But the broad message that there is a lot of slack in today’s economy is well supported. There is a lot of room for expansion and stimulating aggregate demand should be the policy imperative for an extended period.