The Milken Institute Review

Long-Term Unemployment: Anatomy of the Scourge

When unemployment rose during the Great Recession, so did long-term unemployment – defined here as joblessness lasting at least six months. This is the norm in recessions, but the gravity of the problem after 2008 was unprecedented.

In the 11 recessions since World War II, unemployment reached 9 percent in just three (1974-75, 1981-82, 2008-9). Only in the most recent slump, though, did the rate of longterm unemployment exceed 3 percent. Indeed, it reached 4.5 percent in April 2010, almost two percentage points higher than the peak in any previous postwar business cycle. And the problem is worryingly persistent: by April 2012, the long-term rate had exceeded 3 percent of the labor force for 34 successive months.

Between 2007 and 2011, the fraction of the nation's unemployed who were unemployed six months or longer increased from 18 percent to 44 percent.

What’s going on here – and what can we do about it?

Just the Facts

The U.S. Bureau of Labor Statistics defines an unemployed person as a potential worker who is currently jobless, has actively sought work in the previous four weeks and is available for work. The BLS also counts workers as unemployed who have been temporarily laid off and anticipate being recalled, even if they’re not actively looking for another job. People who are neither employed unemployed by this definition are classified as “not in the labor force.”

The government does not have an official definition of “long term” unemployment; economists generally use six months or a year of involuntary joblessness as the benchmark. Because unemployment lasting more than a year has historically been rare in the United States, in many BLS reports the chronically unemployed are simply placed in the “27 weeks or longer” category.

In other industrialized countries, by contrast, the long-term unemployed have long represented a large fraction of total unemployment. In 2007, for example, more than a fifth of France’s unemployed and almost 40 percent of Germany’s had been out of work for two years or longer. (That same year, fewer than 18 percent of unemployed Americans reported being jobless longer than 6 months, and fewer than one in 10 had been unemployed for as much as a year.)

Very long periods of unemployment were only rarely reported in the United States before the Great Recession. Laid-off workers, along with those just entering the labor force, typically found jobs relatively quickly. From 1990 to 2007, for example, more than a quarter of all American workers who were classified as unemployed in a given month found jobs the following month.

The pace of exit from unemployment is sensitive to the business cycle – finding a job is, of course, easier in a boom than in a bust. At the end of the prosperous 1990s, more than one-third of the unemployed in a given month had found work one month later. Jobfinding success fell sharply in the 2001 recession, reaching a low point in early 2003 before recovering during the economic expansion that ended in 2007. The 2008-9 recession brought another plunge in the rate of worker exit from unemployment. The monthly job-finding rate of the unemployed was 28 percent in 2007; by the second half of 2009 it had fallen to just 16 percent. That is, at the nadir of the recession, less than one in six unemployed workers was successful in finding a job within a month.

Though jobless workers were less successful in finding jobs in the 2008-9 recession, they were also less likely to give up their search by dropping out of the work force altogether. The predictable result: average unemployment durations lengthened. Even though the job-finding rate slowly improved after the worst of the recession passed, at the beginning of 2012 it was still one-third lower than the average between 1990 and 2007.

The obvious reason jobless workers were less successful in finding jobs after 2007 is that employers had fewer vacancies to fill: the job-opening rate fell more than 40 percent from 2007 to 2009. The vacancy rate has grown since the recession ended, but in early 2012 it was still almost one-fifth below the 2007 level. In the last quarter of 2007, there were 1.6 unemployed workers for every job vacancy reported by the BLS. Two years later, this ratio had jumped to six to one. And while the ratio improved during the recovery, it was still 3.8 to one at the end of 2011.

Damaged Goods

A worker’s success at finding a job tends to decline with the length of unemployment. The percentage of all unemployed who found work within a month fell from 28 percent in 2007 to 17 percent in 2011. But in both years, workers were much more successful finding jobs in the first weeks of joblessness. In 2007, for example, workers who had been unemployed less than 5 weeks had a 37 percent chance of landing a job within a month. Workers reporting unemployment longer than six months had only a 16 percent chance of finding employment in the coming month.

The Great Recession has thus pushed jobless workers into unemployment-duration groups with poor odds of finding work, even after economic recovery. From 2007 to 2011, the fraction of the nation’s unemployed who were unemployed six months or longer increased from 18 percent to 44 percent.

A simple explanation for why workers’ job-finding success drops as the duration of their unemployment rises is that the workers whose skills are in shortest supply will tend to be re-employed fastest. These workers thus tend to leave the unemployment queue before they have been jobless for long. A disproportionate number of the chronically unemployed simply lack skills currently prized by employers.

A second common-sense explanation is that workers become discouraged. In a recent survey, the economists Alan Krueger (Princeton) and Andreas Mueller (Stockholm University)found that the longer unemployment lasts, the less time the unemployed spend looking for work.

A third reason: employers (legally) discriminate against job applicants who have been idle for a long time. Catherine Rampell of The New York Times reviewed job postings on Web sites like and Craigslist and found hundreds “that said employers would consider (or at least ‘strongly prefer’) only people currently employed or just recently laid off.”

Even employers who do not impose outright bans on hiring the long-term unemployed may nonetheless follow a de facto policy of discrimination. Faced with résumés from three qualified applicants – a fresh graduate, a worker dismissed one week earlier when his employer went bankrupt and a laidoff worker who has been unsuccessful in 18 months of job search – it is hardly surprising that employers often assume the third applicant has the most problematic job record. Indeed, in a buyer’s market, it may not be rational for employers to expend the effort to dig deeper.

Even employers who do not impose outright bans on hiring the long-term unemployed may nonetheless follow a de facto policy of discrimination.

Another explanation sometimes offered is that tangible work skills decline as workers remains idle. I am not persuaded. A job seeker with extensive experience using the general skills needed in a prospective job should, in theory, have an advantage over untested applicants. And this should be the case whether the experience was accumulated in a job that ended with a layoff a week ago or a year ago.

It seems more likely that a worker’s perceived readiness for work may be hurt by a long period of involuntary idleness. Long spells of unemployment are disheartening, and this can be reflected in a job applicants’ skills in selling themselves to interviewers. Krueger and Mueller found that workers exhibit more frequent episodes of sadness as the duration of their unemployment increases.

The Economics of Unemployment Benefits

Even in serious recessions, jobless Americans tend to be out of work for shorter periods than their European counterparts. This difference became evident in the 1980s, when for the first time in decades European unemployment rose above the U.S. rate and remained there. A popular explanation was the difference in social protection on the two sides of the Atlantic. By generously insuring laid-off workers against earnings losses for very long periods, the argument goes, European countries reduced unemployed workers’ incentive to search intensively for jobs. The United States provided less generous protection, especially for workers who remained jobless for very long.

So why didn’t these differences in social protection have a similar impact before the 1980s? One possibility is that before the big recessions in the mid-1970s and early 1980s relatively few European workers were exposed to the adverse incentives created bygenerous jobless benefits. And even those who were exposed didn’t have much time to develop a taste for living on the dole. Indeed, labor markets were so tight that many rich Western European countries were forced to import labor to make up for the shortfall of native workers.

Though unemployment insurance and other income protection programs have been scaled back in Europe, Europeans are still better sheltered than Americans. According to the OECD, the United States ranks near the bottom in generosity among affluent nations when income replacement rates were measured over a two-year unemployment spell. In 2007, just 14 percent of an American worker’s pre-layoff earnings were replaced. The median net replacement rate among the other 20 countries surveyed was 60 percent.

The special stimulus measures passed in 2008 and 2009 did narrow the gap. The twoyear replacement rate increased from 14 percent to 43 percent, mainly because workers were permitted to collect unemployment insurance benefits for greatly extended periods. By late fall of 2009, workers in most states with high unemployment rates could draw up to 99 weeks of benefits – almost four times longer than in 2007.

In addition to extending UI protection, the stimulus programs temporarily increased weekly payments, reduced income-tax liability on unemployment benefits, and provided generous subsidies for laid-off workers who chose to keep on buying health insurance through their ex-employers. Weekly benefits are now back to pre-recession levels, and the extensions of eligibility for benefits are slated to expire at the end of the year. But even if Congress once again extends the extensions, it will almost certainly let them lapse when unemployment dips below 7 percent.

More generally, there is no reason to believe that the country is becoming more generous to the unemployed. The percentage of unemployed eligible for UI programs fell substantially in the 1980s and has only partially recovered since then. At the peak of stimulus spending in early 2010, nearly 70 percent of unemployed Americans collected UI – a higher percentage than in the recessions of the last few decades, but a smaller percentage than in 1975.

Nor is there a trend toward higher income replacement rates for those who do collect UI. Benefits were not subject to the income tax before 1979, making them more generous in 1975 than they are today.

It’s likely that the above-average generosity of stimulus-related UI benefits delayed the re-employment of workers. But did increased generosity make a material difference to longterm unemployment?

Most estimates of the impact of benefit improvements have focused on the effect on overall unemployment rather than on longterm unemployment. The most recent assessments suggest they added 0.4 to 1.8 percentage points to the United States unemployment rate. I believe the actual impact was at the low end of this range.

There is no reason to believe that the country is becoming more generous to the unemployed.

In perhaps the most careful statistical analysis of the relationship between UI benefits and worker behavior in the Great Recession, Jesse Rothstein (University of California, Berkeley) concluded that the benefit extensions reduced workers’ monthly exit rate from unemployment by one to three percentage points (compared with a base exit rate of 22.4 percent per month). Only about half the reduction took the form of lower exit into employment; the other half was a consequence of the fact that fewer unemployed chose to leave the work force altogether.

Rothstein’s findings suggest that eligibility for longer-duration benefits did reduce workers’ re-employment rates. But not surprisingly, it also reduced the odds they would leave the work force. And while both these changes raise the measured rate of unemployment, only the first reduces the number of employed Americans.

Analyses of the effect of UI on unemployment duration usually miss two macroeconomic effects. As noted above, even with extended benefits in effect, far less than 100 percent of the unemployed qualify for UI. And those who do not qualify may find work faster because they face less competition for jobs from those who do qualify. At the peak of the emergency UI programs in early 2010, for example, less than three-quarters of the unemployed collected benefits. This implies that more than a quarter of the unemployed were not directly influenced by the adverse incentives of more generous UI – but indirectly benefited because jobs became a bit easier for them to find.

A second macro effect of generous UI is to increase the consumption of the unemployed, raising the total demand for goods and services – and labor. UI eligibility extensions and higher replacement rates roughly doubled outlays on UI compared with the benefits that would have been available without the stimulus measures. All told, emergency and extended UI benefits added an average of $56 billion a year to government transfer payments from July 2008 to March 2012, which almost certainly had a ripple effect on jobs. What’s more, the fiscal multiplier associated with extended UI benefits was probably higher than the multiplier for other kinds of stimulus spending because the recipients saved little or nothing from their checks.

In sum, the extensions to UI did increase the length of average unemployment spells suffered by UI recipients. But part of the increase was offset by shorter average unemployment durations among workers who did not qualify for UI. Remember, too, that government generosity created incentives for UI claimants to keep on looking for work, a socially desirable effect even though it increases measured unemployment. Finally, extra government outlays on UI increased household spending – and thus total employment – by improving the household finances of workers suffering long periods of joblessness.

Thinking Bigger

There are good reasons to worry about longterm unemployment. The main focus of concern is, of course, the welfare of those unemployed. The costs exacted by a recession are distributed unevenly, with the long-term unemployed suffering the most. UI provides very modest compensation for the income losses they experience.

Economists also worry about the enduring consequences of permitting unemployment to become chronic. Some of those affected eventually exit the work force, retiring long before their capacity and willingness to work end. Others may find employment, but in jobs that are a poor match for their skills and experience. These losses have consequences for the broader economy, shrinking potential output and increasing the burden on costly transfer programs, including Medicaid and Social Security Disability Insurance.

A secondary consequence is that the longterm unemployed eventually become invisible to both the labor market and to policymakers and thus cease to serve as a check on wage inflation as the economy nears full capacity. High unemployment leads workers to be more cautious in their wage demands and makes employers more reluctant to grant big pay increases. It’s unclear, though, whether the long-term unemployed have the same weight as the short-term unemployed in changing labor market expectations.

The short- and long-term unemployed would have equal weight if employers regarded them as equally eligible to fill job vacancies. If they don’t view the long-term unemployed as adequate substitutes for newly laid-off workers, however, an increase in the long-term unemployment rate should have a smaller restraining influence on wages and prices than an equivalent increase in shortterm unemployment.

Ricardo Llaudes, an economist at the European Central Bank, investigated the relative impact of short- and long-term unemployment rate, examining the experiences of 19 rich countries between the late 1960s and 2002. In most of them, the short-term unemployed exerted a considerably greater restraining influence than their long-term counterparts. In France, for example, their estimated impact was three times as great.

The difference was much less pronounced in the United States, where short-term unemployment has only a 16 percent greater weight than the long-term unemployment. Llaudes’s data do not, however, include the big run-up in long-term unemployment since 2007. And conceivably, the American experience is converging with the European, a process that effectively raises the “natural” rate of unemployment – the lowest rate sustainable without inflation.

The long-term unemployed eventually become invisible to both the labor market and to policymakers, and thus cease to serve as a check on wage inflation as the economy nears full capacity.

From a political perspective, familiarity with long-term unemployment may breed collective indifference; addressing the challenge of long-term unemployment becomes a lower priority for policymakers when the media (and voters) lose interest. If each American worker faced identical odds of losing a job, and each job loser then suffered an identical spell of unemployment, most workers would probably take keen interest in minimizing both the risk of unemployment and its likely duration. In fact, the risk of losing a job varies tremendously across regions, industries and job categories, and the duration of unemployment varies greatly among those who do lose their jobs. As a result, the risk of job loss is of little concern to many workers (especially when the economy is growing) and the severe problems of the long-term unemployed become a smaller concern.

The percentage of employed workers who became unemployed in a typical month averaged about 1.2 percent in 2007, the last year of the economic expansion. From October 2008 to September 2009, the monthly probability of losing one’s job surged to 1.8 percent. Most workers became acutely aware of the increased risk, providing policymakers with a base of support for measures to halt the economic slide and to aid the workers harmed by it.

But the monthly percentage of workers who become unemployed has fallen since the economy began to expand in late 2009. Though job-loss rates remain higher than they were at the end of the last expansion, they are well below their recession peaks. Rousing sympathy for the unemployed in general, and the long-term unemployed in particular, is thus becoming more difficult.

What Could Be Done

Almost any policy that increased the demand for labor would trickle down to the benefit of the long-term unemployed. But economists are divided on what’s needed to revive hiring. Many (including me) see this as a classic problem of inadequate aggregate demand, which can be traced to the loss of household wealth in the collapse of the housing bubble. Measured in terms of purchasing power, United States households were nearly one-fifth poorer at the end of 2011 than they were five years earlier.

The standard remedy for lagging aggregate demand (if interest rates are already near zero and thus can’t be lowered further with monetary policy) is to use fiscal policy to increase private and public spending. That was the idea behind the stimulus package – and there’s good reason to believe that it worked as promised. But its magnitude was modest compared with the drop in demand during the Great Recession. Nonetheless, many voters and virtually all Republican lawmakers are opposed to additional stimulus that would increase the government debt.

There are alternative ways to help the longterm unemployed. One is to replace more of their lost income; another is to make them cheaper to hire.

The former is apparently a political nonstarter. Rather than extending income protection for the long-term unemployed, the nation is likely to scale it back. UI extensions in a recession have always been temporary and will surely be trimmed back this time around, even if long-term unemployment remains high. That doesn’t mean workers idle for more than six months will be completely cut off from government help. Many will qualify for Medicaid, and others will collect Supplemental Nutritional Assistance (a k a food stamps). But there’s no getting around the reality that chronic unemployment is a ticket to poverty in America.

A more fundamental fix for long-term unemployment would require serious reforms in labor market rules, such as changes that spread the sacrifice associated with recessions more broadly across the work force.

Another option for middle-aged and older workers is Social Security Disability Insurance (SSDI). To become eligible, workers must demonstrate that they are temporarily or permanently disabled. But disability is a fuzzy legal concept. Larger numbers of jobless workers manage to meet the legal standard when work prospects are poor to nil. In every recession since the early 1990s, a slump in the labor market has been followed by a surge in applications – and benefit awards – for SSDI. This recession was no different: between 2007 and 2009 applications increased 21 percent. Sadly, once workers are enrolled in the SSDI program, few ever return to work.

Can training help? Education and training programs serve dual purposes in a severe slump. First, they can make the long-term unemployed more attractive to potential employers after the economy recovers. Some kinds of training and retraining have a solid track record: graduates go on to earn better wages or find more stable employment. Note, too, that a slump is an ideal time to invest in such training. One of the major societal costs – the earnings workers give up to be in a classroom – is low. What’s more, training programs reduce the queue for whatever jobs are available. People upgrading their skills in a training center or college classroom will not be competing for openings with other laidoff workers.

The catch, of course, is that training costs money. And neither Congress nor the public seems in the mood to make the investment.

Another way to make the long-term unemployed more attractive to employers is to provide them with a financial endowment of sorts – a narrowly targeted hiring subsidy, payable to employers. For example, the government might temporarily waive the employer portion of the payroll tax on the wages of a worker coming off a long bout of unemployment.

The strategy of selectively subsidizing the long-term unemployed has been tried in the past, though I don’t know of any rigorous evaluations of its impact. There is some troubling evidence, however, that subsidies narrowly targeted on disadvantaged populations can do more harm than good. Employers may be reluctant to hire applicants who are identified as disadvantaged, or they may fear the paperwork burden required to collect the subsidy.

Hiring subsidies that are not narrowly targeted are probably more effective in inducing employers to add to their payrolls. But unless the subsidy is targeted, employers would still be inclined to favor the newly unemployed over the long-term unemployed.

That said, I think a non-targeted hiring subsidy would be worth trying. At the very least, it would reduce the number of unemployed workers who remain idle long enough to become classified as long-term unemployed.

A more fundamental fix for the problem of long-term unemployment would require serious reforms in labor market rules – in particular, changes that spread the sacrifice associated with recessions more broadly across the work force. One way to accomplish this would be to give employers both the discretion and the incentive to cut hours rather than jobs during downturns.

German law and labor market practices have long favored this shared-sacrifice approach. Katherine Abraham (University of Maryland) and Susan Houseman (Upjohn Institute) found that the German system delivered as promised, with greater cyclical variability in hours per worker and less variability in total number employed than the American labor market.

Some perspective here: Germans who do lose their jobs in spite of the shared-sacrifice approach are likely to be out of work for a very long time. Nonetheless, by moving toward policies that reduce the number of permanent layoffs when there is weakness in aggregate demand, we could reduce the number of short-term unemployed who are at risk of becoming part of the long-term unemployed.

* * *

Over the postwar era, long-term unemployment has increased as a percentage of all unemployment, and labor market recoveries from recessions have gotten slower. Ragurum Rajan, the former chief economist of the IMF, and others argue that the nature of recessions is changing, and that employment is slower to recover than in the past. So long as actual output is far below potential output, workers who are idle for long periods will remain a large percentage of the unemployed.

As the cliché goes, there are no easy answers. Politics and ideology have made it unpalatable to use fiscal stimulus to help either the shortor the long-term unemployed. Micro fixes – training, hiring subsidies – can be expensive and may not be terribly effective. There’s no ignoring the reality, though, that the failure to come to grips with long-term unemployment has serious consequences both for individual welfare and potential economic growth. I’m convinced that, in the long run, the cost of sustained idleness in terms of both productivity and human dignity far exceeds the cost of addressing the problem.