The Life and Strange Death of Okun's Law...
The President's Council of Economic Advisers in the Economic Report of the President, 2010:
Since the recession began in December 2007, 7.2 million jobs have been lost.... This jobs deficit is much larger than the vast majority of observers anticipated at the end of 2008. this is not the result of a slow economic turnaround. On the contrary, as described above, the change in the economy’s direction has been remarkably rapid given the economy’s condition in the first quarter of 2009. Rather, the jobs deficit reflects two developments.... [F]irst... the unanticipated severity of the downturn in the real economy in 2008 and early 2009.... [T]he Blue Chip Consensus released in mid-December 2008 projected fourth quarter growth would be -4.1 percent and first quarter growth would be -2.4 percent. the actual values turned out to be -5.4 percent and -6.4 percent.... [T]he Administration’s economic forecast made in January 2009... like the private forecasts, underestimated the speed of GDP decline in the first quarter. It also underestimated average growth over the remaining three quarters of 2009. For the four quarters of 2009, the Administration forecast overall growth of 0.3 percent; the actual value, according to the latest available data, is 0.1 percent....
[S]econd... the behavior of the labor market given the behavior of GDP... consensus forecasts for the unemployment rate... as of December 2008, unemployment in the fourth quarter of 2009 was forecast to be 8.1 percent, dramatically less than the actual value of 10.0 percent....
Some of the unanticipated rise in unemployment was the result of the worse-than-expected GDP growth in 2008 and the beginning of 2009 [i.e., about 0.4 percentage points]. CEA analysis, however, also suggests that the normal relationship between GDP and unemployment has fit poorly in the current recession.... Okun’s law... suggests that a fall in GDP of 1 percent relative to its normal trend path is associated with a rise in the unemployment rate of about 0.5 percentage point after four quarters.... [A]lthough the fit of Okun’s law is usually good, the relationship has broken down somewhat during this recession. the error was concentrated in 2009.... CEA calculations suggest that as of the fourth quarter of 2009, the unemployment rate was approximately 1.7 percentage points higher than would have been expected given the behavior of real GDP since the business cycle peak in the fourth quarter of 2007...
Erica Groshen and Simon Potter (2003), "Has Structural Change Contributed to a Jobless Recovery?" Current Issues in Economics and Finance (August) 9:8 (New York: Federal Reserve Bank of New York) http://www.newyorkfed.org/research/current_issues/ci9-8/ci9-8.html:
The current recovery has seen steady growth in output but no corresponding rise in employment. A look at layoff trends and industry job gains and losses in 2001-03 suggests that structural change—the permanent relocation of workers from some industries to others—may help explain the stalled growth in jobs.
A surge in payroll jobs used to be a reliable sign of the end of a recession—but not any longer.... [T]he National Bureau of Economic Research (NBER)... designated November 2001 as the end of the nation’s latest recession.... By the end of June 2003, GDP had risen 4.5 percent from its low in the third quarter of 2001 and significantly exceeded its pre-recession peak.... What troubled the committee was that payroll employment, which would normally rise in tandem with output, had shown no sign of recovery. Indeed, the payroll numbers fell almost 0.4 percent in 2002 and another 0.3 percent through July 2003....
We advance the hypothesis that structural changes—permanent shifts in the distribution of workers throughout the economy—have contributed significantly to the sluggishness in the job market. We find evidence of structural change in two features of the 2001 recession: the predominance of permanent job losses over temporary layoffs and the relocation of jobs from one industry to another. The data suggest that most of the jobs added during the recovery have been new positions in different firms and industries, not rehires. In our view, this shift to new jobs largely explains why the payroll numbers have been so slow to rise: Creating jobs takes longer than recalling workers to their old positions and is riskier in the current uncertain environment.... Although the weak performance of the labor market during the current recovery has been surprising, it is not without precedent. The period following the 1990-91 recession was dubbed the “jobless recovery” because the economy added so few jobs during the first year and a half after the expansion began. The current recovery parallels this earlier recovery in important respects. In 1991-92, output growth rose fairly steadily, but job growth remained near zero for more than a year. In 2002-03, real (inflation-adjusted) GDP has grown each quarter at annualized rates between 1.3 and 5.0 percent, while payroll growth averaged -0.4 percent at an annualized rate through July.2
The sluggishness of payroll growth during the 1991-92 and current recoveries stands in sharp contrast to the vigorous rebound in employment during earlier recoveries.... The divergent paths of output and employment in 1991-92 and 2002-03 suggest the emergence of a new kind of recovery, one driven mostly by productivity increases rather than payroll gains. The fact that no influx of new workers occurred in the two most recent recoveries means that output grew because workers were producing more. Although one might speculate that output increased because workers were putting in longer days, average hours worked by employees actually changed little during this and the previous jobless recovery....
Recessions mix cyclical and structural adjustments. Cyclical adjustments are reversible responses to lulls in demand, while structural adjustments transform a firm or industry by relocating workers and capital. The job losses associated with cyclical shocks are temporary: at the end of the recession, industries rebound and laid-off workers are recalled to their old firms or readily find comparable employment with another firm. Job losses that stem from structural changes, however, are permanent: as industries decline, jobs are eliminated, compelling workers to switch industries, sectors, locations, or skills in order to find a new job. A preponderance of structural—as opposed to cyclical—adjustments during the most recent recession would help to explain why employment has languished during the recovery.... [W]e look for evidence that structural change played a dominant role in the 2001 recession.... Did temporary layoffs decline relative to permanent job losses in the recession? Were many of these lost jobs relocated to different industries?...
[...]
The period after the 2001 recession will be remembered as the second jobless recovery. Our inquiry into the reasons for the current labor market slump suggests that structural change has played an important role. Industries that lost jobs during the recession have continued to shrink during the recovery, and permanent job losses have eclipsed temporary layoffs.... Although our analysis has focused on the recent past rather than the future, our findings suggest that a return to job growth may require a mix of two ingredients: improved financing options for riskier ventures and resolution of current uncertainties, including time for the dust to settle from all the recent structural changes...
I can't say "I told you so'--Erica and Simon told us so: they did the heavy lifting. But I can claim to say that I noticed that they were telling us so:
DeLong: Econ 101b: Fall 2003: Lecture Topic: The Erosion of Okun's Law (September 3, 2003): We used to have considerable confidence in Okun's law: that an extra one percentage point rise (or fall) in the unemployment rate over a year would reduce (or boost) that real GDP growth by an extra 2.5 percent over that year because a rising (or falling) unemployment rate would also be accompanied by a falling (rising) share of the population in the labor force and by falling (rapidly rising) productivity. Productivity would fall when the unemployment rate rose for two reasons: first, even when factories are not running at full capacity they still incur substantial setup and maintenance costs; second, even when there isn't enough work for them to do firms would rather hold onto skilled workers than watch them drift away and have to pay to train their replacements the next time the wheel of the business cycle turns.
Things have been different, however, in this recession (and to a lesser extent in the preceding early-1990s recession. The standard relationship between output growth and hours worked has gone substantially awry. See that branch poking out of the scatter diagram on the left side? That's the most recent data. (The smaller twig pointing out below and to the left of the branch is from the early-1990s recession and recovery.)
The fact that falling hours have been accompanied by rapidly-rising productivity is what has given us not a jobless recovery but a massive job-loss recovery. The normal pattern we would expect from the past two years' output growth would be that employment and hours would have been nearly flat. Why the different pattern this time? We think that it is because firms are no longer "hoarding labor" when times are slack because the industries losing jobs no longer expect employment to bounce back.
This means that we no longer have any confidence that we understand the cyclical pattern of productivity growth--which means that we have little ability to translate the (high) productivity growth numbers we see into information about what the underlying long-run trend growth rate of the economy is.
Why is this? Why have firms changed their behavior? Let me turn the mike over to Erica Groshen and Simon Potter of the New York Federal Reserve Bank...
2004 Economic Report of the President: Indeed, the performance of employment over the past couple of years has been appreciably weaker than in past business cycles (Chart 1-10). Employment was slow to pick up in the average previous recovery, perhaps because employers delayed hiring until they became confident that the increases in demand were sustainable. However, such sluggishness typically has been short-lived (a quarter or two) and followed by vigorous expansion. In contrast, in the current business cycle, employment did not begin its recovery until nearly two years after the upturn in real GDP. The performance of employment in this cycle has lagged even that of the so-called “jobless recovery” from the 1990-1991 recession...
2003 Economic Report of the President: In other ways, however, the recent behavior of the labor market has been different from that in past business cycles. One difference is the high fraction of job losers who reported a permanent rather than temporary separation in 2001. In the government’s monthly Current Population Survey, each respon- dent who reports a job loss is asked whether he or she expects to return to work with the same employer. (Those who expect to return are typically on an explicitly temporary layoff, although this need not be the case.) Research from the Bureau of Labor Statistics found that, in the initial quarters of the four recessions before 1990, slightly more than half of job losers were perma- nently separated from their previous employers, with the rest on temporary layoff. In the three quarters after the business cycle peak of 1990, however, the share of permanent job losers rose to almost three quarters, and the comparable proportion for the March 2001 peak is nearly 90 percent. The rising proportion of job losers facing a permanent separation in recessions may reflect structural changes in the labor market during the past two decades, including the rise in temporary help employment...
