Friday, July 30, 2010

Labor market in the Great Recession

Arguably the biggest immediate economic challenge facing developed economies is that of lowering their unemployment rates. In the US, despite the unexpectedly strong economic recovery, equity market rise, and surges in corporate profits, unemployment rate has been stubbornly sticky.

In fact, in his semi-annual monetary policy report testimony to the Congress, Fed Chairman Ben Bernanke even opined that it will take at least three or four years for employment to return to its long-run sustainable level and the economy to recover the 8.5 million jobs lost in 2008 and 2009.

Assuming that the recession in the US officially ended in the summer of 2009, after adjusting for inflation, the economy grew 2.2% in 2009 Q3, 5.6% in 2009 Q4, and 2.7% in 2010 Q1. However, despite three quarters of output growth, the unemployment rate has remained stubbornly high, rising from 9.5% in June 2009, to 10.1% in October 2009, before falling back to 9.5% in June 2010, identical to the rate when growth restarted 12 months earlier.

Admittedly, the well-established lag between output growth and rise in job-hirings is a partial explanation for this stickiness. In the circumstances, there has been an interesting debate about the surprising rigidities in the labor market that are delaying labor market adjustments in the aftermath of the Great Recession.

Conservatives had attributed the persistently high unemployment rates among many European economies through much of the last two decades to the moral hazard induced incentive distortions generated by the continental economies' generous social welfare nets. It was argued that those out of jobs found it more convenient to live-off benefits than search for jobs.

However, the same argument can be hardly applied to the US economy, with its bare-minimum of social security benefits and flexible labor markets. In fact, the US Congress even voted early this month to not continue unemployment insurance benefits to long-term unemployed beyond 52 weeks (originally, it was 26 weeks but was extended once). However, early this week the Senate approved the extension of unemployment benefits for those out of work for more than 26 weeks for another six months till November 2010.

In the US, the search for structural causes for the persistent high unemployment rate has received much attention. They include skills mismatch, workers geographically "stuck" in place by negative housing equity, and the reduced employability of the long-term unemployed.

Automation and off-shoring has meant that it becomes possible for employers to lay-off workers performing manufacturing and low-end mangerial jobs and certain high-skill service sector jobs. As Daron Acemoglu argues, this along with the reluctance of men previously performing relatively well-paying manual tasks, manufacturing jobs and lower-end managerial jobs being unwilling to take lower paying service jobs means that structural unemployment will remain high for some time, atleast till recovery takes firmer roots and/or expectations of the workforce about job propsects change.

In other words, as Gilles Saint Paul writes, unemployment will persist "because of the reduced employability and job search intensity of the long term unemployed". He also cautions against "political hysteresis", or the possibility of those who have kept their jobs (the insiders) supporting tighter labour regulations (once the recession is past) that would increase their bargaining power in order to take advantage of the recovery to get higher wages, at the expense of new hires.

However, the search for explanations for the high unemployment in structural causes may be a red-herring. For a start, structural unemployment refers to those long-term unemployed that would be without jobs even if aggregate demand was on target, an extremely dubious assumption now. As Scott Sumner points out, the decline in nominal GDP has been so large (in 2009 nominal GDP in the US fell at the fastest rate since 1938) that "there is nothing at all surprising about the steep rise in unemployment during 2008-09". And after the ARRA expired, since nothing really effective is being done to nullify the effects of the lingering nominal GDP shock, the high rate of unemployment persists. He writes that the focus on structural causes "takes the pressure off policymakers to deliver adequate nominal GDP growth".

Richard Koo tool feels that the search for structural causes is diverting attention from attending to the real problem of reviving depressed aggregate demand. He argues that in a balance sheet recession the private sector minimizes its debt instead of maximising profits so as to repair their battered balance sheets. In the circumstances, governments need to step in and fill the deficit or risk letting the economy dip into a deflationary spiral with persistent and rising unemployment rates. He writes,

"When the deficit hawks manage to remove the fiscal stimulus while the private sector is still deleveraging, the economy collapses and re-enters the deflationary spiral. That weakness, in turn, prompts another fiscal stimulus, only to see it removed again by the deficit hawks once the economy stabilises. This unfortunate cycle can go on for years if the experience of post-1990 Japan is any guide. The net result is that the economy remains in the doldrums for years, and many unemployed workers will never find jobs in what appears to be structural unemployment even though there is nothing structural about their predicament. Japan took 15 years to come out of its balance sheet recession because of this unfortunate cycle where the necessary medicine was applied only intermittently."

Paul Krugman points to the interesting differences in the labor market outcomes in the recessions before nineties, which were due to real economy shocks and resultant macroeconomic policy actions, and those in the nineties and beyond, which have been induced by financial market crashes. He points to the variations in productivity across recessions in these two periods and writes

"(B)efore 1990, recessions were generally accompanied by a fall in productivity, mainly because businesses would hang on to workers, so as to be prepared to ramp up production quickly once recovery got underway. After 1990, this 'labor hoarding' effect basically vanished; if anything, productivity growth seemed to accelerate in times of weak demand. Partly this may have reflected structural changes in the economy; it might also reflect the (correct) perception that recovery from financial-crisis-induced recessions is much slower than recovery from recessions created by tight money, imposed by the Fed to control inflation."

In a much discussed recent post, Tyler Cowen felt inclined towards the view that "many previously employed workers simply have a current marginal product pretty close to zero". His reasoning is based on the fact that the economy produces much the same amount of goods with the reduced number of labor, without any major new technological breakthroughs. In other words, since a lesser number of workers are now producing more than earlier, the marginal product of those now laid-off workers had to be zero. See also Arnold Kling here.

Further, examination of the US labor market reveals that the unemployment rate is very low for highly-educated workers but very high for those at the other end of the spectrum (in the US, 34.7% of the labor force is college graduate and above, whereas their share of the long-term unemployed, more than six months, was just 18.7%). Similarly, lower-income Americans are affected greatly by the current recession while people at the top of the pyramid have almost negligible unemployment rates.

However, among the long-term unemployed, those with more education tend to have longer length of unemployment, while those with less education are more likely to find another quickly. This is understandable given the fact that more educated are more likely to wait for an appropriate opening before they re-join the labor market, whereas those at the other end are most likley to join the first opportunity that comes their way.

Greg Mankiw points to the steep rise in the median duration of unemployment and feels that traditional unemployment-output-inflation relationships like Okun's law, Phillips Curve, NAIRU etc may be due for revision.

On the shift in NAIRU, he also points to recent research findings that the "long-term unemployed put less downward pressure on inflation" and argues that therefore "the increase in long-term unemployment may mean that we will see less deflationary pressure than we might have expected from the high rate of unemployment".

Interestingly, Paul Krugman draws attention to a paper by Laurence Ball who points to the strong tendency of high unemployment to become permanent. Those out of work for a long enough period become permanently unemployable (or atleast less productive and less competitive in the job market), both psychologically and in terms of their skill-erosion. Ball provides compelling evidence that weak policy responses to high unemployment tend to raise the level of structural unemployment, so that inflation tends to rise at much higher unemployment rates than before.

Krugman also points to the possibility that the Beveridge curve — the relationship between job vacancies and the unemployment rate — already seems to have shifted out dramatically, signaling a worsening of the NAIRU. David Altig writes that the "most tempting explanation for the seeming shift in the Beveridge curve relationship is a problem with the mismatch between skills required in the jobs that are available and skills possessed by the pool of workers available to take those jobs".

Heidi Shierholz explains away the falsehood in the five big unemployment myths - unemployment benefits make people less likely to find jobs (since there are now roughly five unemployed workers for every available job, simply searching for jobs won't help you find one!); unemployment insurance doesn't contribute to economic recovery (see this); we can't afford to do this right now (see this, primary causes of US long-run deficits are rising health-care costs and low revenues and the stimulus is responsible for no more than 1 to 2% of US's long-run fiscal gap); private sector can take care of unemployment on its own (to get down to the pre-recession unemployment rate within five years, the labor market would have to add an average of roughly 280,000 jobs every single month between now and then); and unemployment rate gives us a good sense of how many people are affected by the downturn (the underemployment rate which includes jobless people who have given up looking for work and people who are working part time but want full-time jobs stands at 16.5%).

Tim Duy has this excellent summary of the debate in the blogosphere about the possibility of NAIRU having risen.

Update 1 (1/8/2010)
Minneapolis Fed has a series of superb graphics that highlights the magnitude of the Great Recession in relation to previous ones in terms of changes is employment and output.

Update 2 (8/8/2010)
Mark Thoma argues that irrespective of whether the unemployment is structural or cyclical, governments have a major role to play in lowering it, especially in deep recessions like the current one. Mark Thoma also argues that governments should intervene with fiscal support even with structural unemployment.

Update 3 (1/9/2010)

During the 2007-09 recession, middle income jobs were the worst hit, while lower paying jobs have been recovering faster than higher paying ones.

David Autor of the MIT also writes that from 2007 to 2009, the paper said, there was relatively little net change in total employment for both high-skill and low-skill occupations, while employment plummeted in so-called middle-skill occupations.

Update 4 (20/9/2010)

Mike Konczal (also this) and Paul Krugman point to evidence that it is weak aggregate demand and not skills mismatch that is driving the high unemployment rates in the US.

Update 5 (11/11/2010)

A San Francisco Fed working paper questions the structural reasons are causing high unemployment hypothesis. They write that "analysis of data on employment growth and jobless rates across industries, occupations, and states suggests only a limited increase in structural unemployment, indicating that cyclical factors account for most of the rise in the unemployment rate".

Update 6 (5/12/2010)

The structural or skills-mismatch hypotheisis of Kocherlakota et al (with its attendant claim that government cannot do anything) is not borne out by the across-the-board nature of job losses. As Matt Yglesias points out, it looks like a "massive and largely avoidable plague of idless in which the country is producing far fewer goods and services than it is capable of producing".

Update 7 (12/3/2011)

Economix (from Cleveland Fed report) points to an excellent graphic that highlights how "America’s most educated workers have been blessed with significantly lower unemployment rates and higher wages gains than their less-educated counterparts".

No comments: