Column: Five years after recession, we still can't agree on what causes joblessness

Reuters News
Posted: Aug 28, 2013 10:27 AM

By Allison Schrager

Although the Fed announced months ago it is considering pulling back its purchase of assets, unemployment remains historically high. What, if anything, can the government do to get people back to work?

In order to determine the right policy prescription, first we must diagnose what's causing unemployment. Is the high unemployment due to low demand from the recession, known as cyclical unemployment, or has the world changed and jobs are not coming back, known as structural unemployment? Most likely it's both. It is impossible to know precisely how much new unemployment is structural and how much is cyclical. This uncertainty has sparked a contentious debate about the nature of unemployment that has been raging since the start of the recession, and lately seems to be hardening into absolutism. The cyclical camp fears that acknowledging an increase in structural unemployment will be used as an excuse to support tightening monetary policy, and gives the government a pass on fixing unemployment. But actually, saying all unemployment is cyclical is what lets the government off too easy. Structural unemployment can be helped with policy, but the solutions take more political will, creativity and leadership. We can lower the structural rate by changing tax incentives to encourage mobility, both job and location, and building a wealth cushion to finance productive job transitions.

If unemployment is cyclical then all the government has to do is reignite demand. For example, if people aren't buying much then firms don't produce or sell as many goods, and therefore employ fewer people. Typically to compensate for recessions, monetary policy makers lower interest rates. This encourages people to buy more or firms to expand. That creates more demand and firms hire again. If the Fed creates more inflation, which lowers real wages, then firms can hire more cheaply.

Or the problem might be structural. Structural unemployment is often poorly understood because economists use several different definitions for it. It is often called the natural or permanent rate of unemployment — though there is nothing permanent about it; it varies over time. It is also known as the non-accelerating inflation rate of unemployment, meaning rates can be lowered and inflation created, but there will be little impact on unemployment. If all unemployment is structural and the Fed tries more expansion, all you get is high unemployment and more inflation.

An increase in the structural unemployment rate may be caused by several different factors. Employers may want to hire people, but can't find people with the necessary skills. Or if unemployment benefits are higher than what people can earn working, some people will prefer not to work. That is probably not a large factor in America today, but it explains why some European countries like France and Germany had a higher structural rate than the U.S., more than 8 percent before the crisis. Labor market reforms in Germany, which enhanced flexibility and encouraged work, are credited with lowering its structural rate. Another factor could be if employers must pay higher compensation, including salary and benefits, than they'd like to their workers. Geographic immobility, if job seekers can't move to where the jobs are, is another contributor. The definition can also be broadened to account for economic well-being. Suppose a laid-off manufacturing worker can only find a job in a fast food restaurant or part-time contract work. He no longer counts as unemployed, but a structural change undermined his economic security and an otherwise productive worker is underemployed.

There exists very compelling evidence that cyclical unemployment is present. Inflation is low, yet there's still high unemployment. That suggests there is more that monetary policy can do. Last summer at the Federal Reserve's Jackson Hole conference economist Edward Lazear presented strong evidence that a large component of unemployment was demand driven. Unemployment during the Great Recession was mostly in industries associated with structural change, manufacturing and construction. But since 2009, many people in these industries have found new work. Alas, we don't know what industry the new jobs were in and why unemployment remains high, especially when you count discouraged people who dropped out of the labor force. Lazear also found that industrial mismatch, the ratio of vacancies to unemployment across industries, increased at the start of the recession but fell as the recession dragged on. He claims the changing ratios show that, since the recession, jobs didn't disappear in some industries while new jobs in other industries couldn't be filled. One former structuralist champion, Minnesota Fed President Narayana Kocherlakota, conceded that he had underestimated the extent of cyclical unemployment, but with an important caveat.

"I should be clear — it's not that I used to think structural factors were the sole source of elevated unemployment, and now I think they don't matter. It's more nuanced — the evolution of the data has led me to put less weight on structural factors than I did earlier in 2012."

This is an important point: the existence of cyclical unemployment doesn't mean there's no change in the structural rate, too. The economy has been experiencing a structural change for decades. New technology and more globalization have changed the nature of work and the kind of skills American employers demand. Economist Josh Lerner believes that in order to understand structural shifts you need to look at employment by occupation, defined by on-the-job skills, not industry. He observed that manual jobs, such as traditional factory work, disappear during recessions and new job growth has been concentrated in high or low-skill jobs; this is often described as a "hollowing-out" of middle-skill occupations.

The economic factors that eliminated many middle-skill jobs have been unfolding for decades. But they didn't cause a gradual increase in the structural unemployment rate, as economists would expect. What happened was large discrete change following each recession. That's because cyclical and structural forces can interact. During recessions already vulnerable firms are more likely to close or have layoffs due to weak demand. The layoff may have been caused by a cyclical factor, the worker is structurally unemployed because his skills are harder to sell and his old job isn't coming back.

Research by economists Nir Jaimovich and Henry Siu found that 92 percent of job loss in middle-skill, routine jobs occurred during recessions, and many of these jobs never came back. Middle-skill job loss can account for most of the unemployment in each recession. Middle skill jobs accounted for 87 percent of job loss in the 1991 recession, 91 percent of job loss in the 2001 recession, and 95 percent in the Great Recession. The concentration of job loss in middle-skill occupations explains why, starting in the 1990s, jobs growth was slower during economic recoveries. Economic growth rebounded, but the unemployment rate took longer to come down.

The debate about the nature of unemployment has gotten contentious, in part from slippery definitions of structural unemployment. But if it can be defined as unemployment that won't be lowered by monetary policy and reflecting an underlying structural change, it's reasonable to believe that there's both deficient demand and a structural rate increase. That leaves plenty of work for policy makers.

There's more scope for the Fed to keep rates low and ensure capital is available to worthy, job-creating, small firms. To fix structural unemployment, the prescriptions are even harder to implement. Thoughtful and well-designed fiscal expansion may provide work and training to laid-off middle-skill workers. Or if health and retirement benefits were not tied employers, American firms could better compete with foreign counterparts. Changing the compensation structure, perhaps by eliminating the tax deductibility of health and retirement benefits, would also ease job-lock that keeps people stuck in less productive jobs simply to keep their benefits. That would make the labor market more fluid and able to absorb laid-off workers.

There also could be better policies to encourage geographic mobility. Right now many people have economic incentives to buy a home, but few to maintain a stock of liquid saving. If you suffer a setback and must move or retrain, that requires cash. According to 2009 survey data, half of Americans believe they probably couldn't come up with $2000 in a month if they had to. If we changed tax incentives there would be less concentration in a levered, illiquid, single asset — housing, which discourages mobility — and more in liquid saving assets, which finance productive transitions. Home ownership may be a worthy objective, but it's hard to justify encouraging people to keep all their wealth in a single, illiquid asset.

These are harder policy prescriptions than expansionary monetary policy and will take more time to work. But unless we accept that post-recession unemployment is a complicated, nuanced problem that requires innovative and thoughtful solutions, on both the monetary and fiscal side, it will continue and the structural part will get worse.

( Allison Schrager is a Reuters columnist. Opinions are her own)