This Selected Issues paper for the United States discusses the microeconomics of the country—household wealth and savings. Households’ consumption-saving decisions have an important bearing on the U.S. economic outlook. This paper demonstrates how households with consistently lower income, which have shown growth in the years prior to the crisis, experienced larger declines in their saving rates and a larger rise in their indebtedness before the crisis, contributing significantly to the dynamics of the mean saving rate.

Abstract

This Selected Issues paper for the United States discusses the microeconomics of the country—household wealth and savings. Households’ consumption-saving decisions have an important bearing on the U.S. economic outlook. This paper demonstrates how households with consistently lower income, which have shown growth in the years prior to the crisis, experienced larger declines in their saving rates and a larger rise in their indebtedness before the crisis, contributing significantly to the dynamics of the mean saving rate.

IV. Is Long-Term Unemployment Pushing up Structural Unemployment? 1

A. Introduction

1. The recession has left deep scars on the U.S. labor market. The unemployment rate surged from 4.4 percent in May 2007 to 10 percent in October 2009 on the back of massive job destruction. Compared to the recoveries from previous postwar recessions, GDP growth and job creation have been particularly weak (the behavior of output per hour worked, on the other hand, has been similar to previous recoveries). Three years after the end of the recession, labor underutilization remains significant: the unemployment rate exceeds 8 percent, an additional 6 percent of the labor force is either marginally attached to the labor market or employed part-time for economic reasons, the employment-population ratio is barely above its post-recession trough, and the labor force participation rate has declined to its early 1980s levels (Figure 1).

Figure 1.
Figure 1.

The Emergence of a Long-Term Unemployment Problem?

Citation: IMF Staff Country Reports 2012, 214; 10.5089/9781475504910.002.A004

Sources: Bureau of Labor Statistics, JOLTS, staff calculations based on the CPS basic month, and Fund staff estimates.

2. A novel and worrisome feature of the post-recession U.S. labor market has been the emergence of a large and persistent pool of long-term unemployed. During the postwar period, the U.S. labor market had never experienced large and persistent long-term unemployment (LTU), in contrast to several European countries. 2 3 This changed with the 2008–09 recession. The LTU rate increased from 0.7 percent in early 2007 to 4.5 percent in early 2010—almost double the previous historical peak—and has been unusually persistent since then (Figure 1).4 In contrast, movements in short-term unemployment (STU) have been similar to those observed during previous U.S. labor market cycles. As a result, while in 2007 individuals unemployed for over one year accounted for 10 percent of overall U.S. unemployment, significantly below the OECD average of 30 percent, by 2011 their share in U.S. unemployment had surged to 31.3 (compared to a OECD average of 33.6 percent in 2011).

3. The increase in LTU has raised the possibility that the U.S. unemployment rate may remain higher than its level before the crisis, even as the economy recovers. Indeed, the experience from some European countries, albeit with higher unemployment rates, in the 1980s suggests that the longer people remain out of work, the greater is the risk they will fall out of the workforce altogether (OECD, 2012). Long spells of unemployment carry the risk that the long-term unemployed progressively lose their skills and attachment to the labor force. Thus, LTU could both raise the natural rate of unemployment and lower the labor force participation rate.

4. This paper investigates the extent to which the sharp increase in LTU in the U.S. may lead to an increase in the natural rate of unemployment. The current weakness in the U.S. labor market is widely estimated to be mostly cyclical though a number of studies suggest that the natural rate of unemployment has increased since the recession (e.g., Elsby et al., 2011; Guichard and Rusticelli, 2011; Valletta and Kuang, 2012). To assess the risk that LTU could morph into a structural unemployment problem, the paper first analyses the characteristics of the long-term unemployed (Section B), and then assesses whether they are becoming less employable (Section C). The assessment of Section C draws both from estimates of the probabilities that the long-term unemployed individuals transition into a job, and from an assessment of whether labor markets find relatively more difficult to match supply and demand for the LTU. Section D concludes and provides policy recommendations.

5. The paper’s main findings are twofold. First, long-term unemployed are significantly less likely to find a job now than before the crisis, suggesting there is now a greater risk that they will lose skills and abandon the labor force altogether. Second, the loss of efficiency in the matching process between job vacancies and the unemployed workers is mainly a phenomenon regarding the long-term unemployed. Together, these results point to the possibility that the equilibrium rate of unemployment might be greater now than before the crisis.

B. Who are the Long-Term Unemployed?

6. LTU is particularly high among less educated and younger individuals, and in lower-skill occupations, but broad-based across industries:

  • Age: LTU rates are higher among younger individuals—for example in 2011, the LTU rate for those aged 16–19 years was one and a half times larger than the rate for those 55 years and older. In terms of shares, the 16–19 year old group account for 5.3 percent of total LTU, while they represent only 3.7 percent of the labor force. This is in contrast with other age groups—the group of 35 to 44 year olds, for example, account for 18.9 percent of the LTU, while representing 21.3 percent of the labor force.

  • Education: the level of education has a large negative impact on LTU incidence. Individuals without a high school diploma, for example, accounted for 15.6 percent of total LTU, but only 8.7 percent of the labor force in 2011. The gap between the LTU share and the labor force share decreases as education levels rise, and eventually reverses. Those with a college degree, for example, account for 19.5 percent of the LTU but 35.3 percent of the labor force (and 36.6 percent of total employment).

  • Occupation: LTU rates are higher among lower-skill occupations, such as support services (e.g., cleaning and maintenance), construction, and production and transportation.5 Similar findings arise in terms of shares: for example, production and transportation occupations account for about 12 percent of the labor force but about 50 percent of total LTU in 2011. While high-skill occupations workers face a lower LTU rate, once they become unemployed, they face similar length of unemployment spells as other occupations.

  • Industry: LTU is broad-based across industries, as would be expected when unemployment is primarily due to a lack of aggregate demand. LTU rates are higher among unemployed individuals previously employed in some industries—e.g, leisure and hospitality, manufacturing and construction, professional and business services, information, trade, and transportation—but all the industries experienced a notable increase in their LTU rate since the recession. Industries that created jobs throughout the recession and recovery are have also experienced a large increase in their sectoral LTU rate (Figure 1).

C. Is there Evidence that the Long-Term Unemployed are Becoming Less Employable?

7. To assess whether the long-term unemployed are becoming less employable, we estimate the probability that someone unemployed may find a job or drop out of the labor force from one month to the next. To estimate these “transition probabilities,” we use individual-level data from the Current Population Survey (CPS), a monthly survey of a sample of about 60,000 households that is representative of the entire civilian non-institutional population of the United States, with information on the labor force participation, employment, and unemployment status of individuals. As the CPS does not follow individuals over long periods of time, we track unemployed individuals from one month to the next. In March 2012, some 3,800 individuals that were unemployed in February 2012 were tracked. We estimate the monthly transition probabilities of the unemployed from 1994 to 2012. This is done using a multinomial logit model, where the dependent variable is the current labor force status of individuals that were unemployed in the previous month (that is, either unemployed, employed, or out of the labor force), and the explanatory variables are unemployment duration, age, gender, and education.6

8. We find evidence that longer unemployment spells are associated with a lower probability of finding a job and with a higher probability of leaving the labor force, but only up to a period surrounding the end of unemployment-insurance benefits. In March 2012, the probability of finding a job was significantly lower for the long-term unemployed—about 5 percent, compared to about 17 percent for all other unemployed (Table 1, regression 1).7 This finding is robust across the sample period and across demographic groups. However, we find some evidence of nonlinearities in the impact of unemployment duration on the probability of finding a job. Specifically, unemployment duration lowers the exit rate into employment only up to 83 weeks, after which further duration slightly improves the exit rate (Table 1, regression 4). This could reflect an optimal decision related to the proximity of the expiration of unemployment benefits the maximum duration of which was 99 weeks (as of March 2012).8 We also find that the monthly probability of leaving the labor force is significantly higher for the long-term unemployed compared to all other unemployed (24 percent versus 21 percent—Table 1, regression 1). The probability of leaving the labor force increases by 7 percentage points once the duration of unemployment becomes larger than 99 weeks (Table 1, regression 5).

Table 1.

Labor Force Predicted Transitional Probabilities of Unemployed Workers and Marginal Effects1/

article image

Multinomial logit model. *** refers to significant at the 1 percent level, ** refers to significant at the 5 percent level, and * refers to significant at the 10 percent level.

U: unemployed; E: employed; NILF: not in labor force

9. We also find that the probability that the LTU transition into employment has fallen markedly after 2009, and has remained depressed since then (Figure 3). The impact of unemployment duration on the predicted transition probability into employment has been large, negative, and consistently statistically significant over the past two decades. On average over the sample period 1994–2007, this probability had been relatively stable at around 14 percentage points for the LTU (and about 13 percentage points lower than for the rest of the unemployed). However, this predicted probability has dropped to record lows with the 2007–09 recession, and has remained persistently low during the recovery: at 5 percent in 2012, it was two-thirds lower than during the 2001 recession. That the transition probability of the LTU into employment is still depressed three years after the end of the recession is a concern, as the longer the unemployed remain out of work, the more acute is their risk of deskilling and dropping out of the labor force.9 While the probability has not improved, we do not find evidence that it has trended down during the recovery—which would have been evidence of hysteresis effects: that is, evidence that longer-duration unemployment spell reduces the probability that an unemployed individual finds a job.10

10. Our results also show a weaker attachment to the labor force after the last recession, raising the risk that the recent decline in the participation rate may not fully reverse. While secular trends were already pushing the labor force participation rate down since early 2000s, the period following the 2008-09 recession saw an unprecedented drop in participation rates. Almost half of the drop can be explained by older workers permanently abandoning the labor force while the other half is likely to be more cyclical, reflecting a mix of increased school enrollment and discouraged individuals (Figure 2).11 Our results indicate that the probability that the unemployed leave the labor force has declined during the last recession and the early recovery phase (Figure 1). This could reflect the strong labor force attachment of those who became unemployed during the recession, as well as the large increase in extended unemployment benefits, which temporarily increases the unemployed workers’ marginal attachment to the labor force. However, since 2011, the probability that the unemployed leave the labor force has risen. Part of that increase arises towards the end of the extended unemployment benefits period (99 weeks), which points to the risk of a further decline in participation rates going forward as a growing number of individuals exhaust their emergency unemployment benefits. The aggregate impact of LTU on the labor force participation rate remains limited in the near term as only 2½ percent of the unemployed were within 6 months of the 99 weeks threshold in March 2012.12 By contrast, as of March 2012, about 40 percent of the unemployed had been out of a job for a duration between 27 and 72 weeks. There is a risk that the labor force participation rate declines if the unemployment-duration of these individuals becomes greater than 99 weeks.

Figure 2.
Figure 2.

A permanent Deterioration in Labor Force Participation and in Labor Market Efficiency?

Citation: IMF Staff Country Reports 2012, 214; 10.5089/9781475504910.002.A004

Sources: Bureau of Labor Statistics, JOLTS, Haver Anallytics, and Fund staff estimates.
Figure 3.
Figure 3.

Job Matching and Finding Rates, by Duration of Unemployment

Citation: IMF Staff Country Reports 2012, 214; 10.5089/9781475504910.002.A004

Sources: Bureau of Labor Statistics, JOLTS, Barichon (2010), staff calculations based on the CPS basic month.

11. Shifts in the Beveridge curves suggest rising mismatches in the labor market due to high and persistent long-term unemployment. The Beveridge curve plots the relationship between the vacancy rate and the unemployment rate through a downward sloping curve. The closest is the curve to the origin, the more efficient is the labor market in matching demand and supply, as for a given vacancy rate there will be fewer unemployed individuals. In ranking models of unemployment the LTU are at the end of the hiring queue—or the least likely to get employed (Blanchard and Diamond, 1994; Lockwood, 1991). If so, in an environment with cyclically weak labor demand, the demand for LTU is significantly depressed. In such a situation, plotting the overall vacancy rate against the STU rate, instead of the total unemployment rate, can reveal whether the last recession has caused a loss of efficiency in labor markets which is related to the duration of unemployment. While the Beveridge curve has moved outward since the last recession (Box 1), the modified Beveridge curve (with STU instead of total unemployment) does not exhibit any shift: as the economy entered recession, the vacancy rate fell and the STU rate rose but when the economy started to recover both the vacancy and the STU rate retraced their earlier path (Figure 2). This suggests that the matching efficiency of the labor market seems to have been preserved after the 2007–09 recession for the STU, while mismatches have worsened for the LTU. The same results hold at a sectoral level,13 in particular for manufacturing sector (Figure 3). Although it is too early to assess the extent to which these mismatches could become permanent, the more protracted is the recovery in the labor market, the more likely are these mismatches to become entrenched.

D. Conclusions and Policy Recommendations

12. Our analysis suggests that while high long-term unemployment has not yet morphed into a permanent structural problem, it does pose an upward risk to the structural rate of unemployment. We have found that long-term unemployed are significantly less likely to find a job now than before the crisis, and that the loss in labor market matching efficiency observed since the recession is entirely due to a worsening of the labor matching of the long-term unemployed. Together, these results point to a risk that the structural rate of unemployment might be greater now than before the crisis.

13. Hence, forceful measures should be introduced that reduce long-term unemployment and address the risks associated with long spells of unemployment, namely skills erosion and a weaker attachment to the labor force. These measures include policies to increase demand for the long-term unemployed in the short run (active labor market policies, ALMP). When appropriately designed, such policies have been shown to be effective in improving employment and earnings prospects of long-term unemployed workers (Card et al, 2010; Card and Levine, 2000; Heinrich et al., 2008; Hotz et al., 2006). In particular, as discussed in the Staff report, a significant increase in ALMP resources is warranted given the persistently large pool of long-term unemployed and the risk that, as duration lengthens, their skills and attachment to the workforce might erode. Indeed, in terms of resources per long-term unemployed, the United States spends relatively little on active labor market policies, both compared to other OECD countries, and relative to its own pre-recession levels. 14

The Beveridge Curve and Structural Unemployment

1. Evidence of a rightward shift in the Beveridge curve is consistent with an increase in the natural rate of unemployment. From the literature on search and matching models, a rightward shift in the Beveridge curve indicates that the labor market matching efficiency has deteriorated. Several authors have pointed to this shift in the Beveridge curve since the 2007–09 recession (Figure 2) as evidence of a rising natural rate of unemployment (e.g., Barnichon et al., 2010; Daly et al., 2011; Weidner and Williams, 2011). Potential causes of the shift have been proposed: skills mismatch, geographic mismatch due to a post-housing bubble housing-lock effect, and extended unemployment insurance (CBO, 2012). Sahin et al. (2011) find that skills mismatch have indeed been a key factor behind the recent shift in the Beveridge curve but they also find that mismatches across sectors and occupations have started to decrease and are therefore likely to be cyclical. The housing-lock effect has been found to be small or insignificant in recent studies while the extended unemployment insurance impact is expected to fade as extended benefits are phased out, starting in 2012 (Elsby et al., 2011; CBO, 2012).

2. The Beveridge curve has shifted rightward during severe recessions, with the shift ranging from three years to being “permanent”. Given the depth of the 2007–09 recession, it is instructive to analyze how the Beveridge curve behaved in previous severe recessions. Figure 2 reveals that in both the 1973 and the 1980 recessions a pronounced rightward shift of the Beveridge curve occurred. For the former the shift proved “permanent” while for the latter the curve eventually returned to its pre-recession locus but did so only three years after the unemployment peak of December 1982.

3. Drawing from Beveridge curve shifts during severe recessions, the current rightward shift is expected to persist for some time. Using as benchmarks the Beveridge curve movements during severe recessions, one would conjecture that, even if the current shift of the Beveridge curve were to be temporary, it is likely to persist for a few more years since (1) the unemployment peak was reached “only” 2.5 years ago (in October 2009), (2) the 2007–09 recession was deeper than any of the previous postwar recessions, (3) the Beveridge curve is not showing signs of looping back and, if anything, is shifting further rightward1—which at this stage of the recovery can indicate that structural unemployment is creeping up—and, importantly, (4) in contrast to previous severe recessions/recoveries, the U.S labor market is currently confronted with a large pool of long-term unemployed individuals which might not be as rapidly employable as individuals with shorter unemployment spells (as detailed in the previous section, LTUs have been facing depressed hiring rates since 2009).

1/ For a given vacancy rate, the gap between the unemployment rate during the recession and the recovery has slightly widened in early 2012 compared to the second half of 2011, from 2.4 to 2.2 percentage points, respectively.

References

  • Aaronson, D., J. Davis, and L. Hu (2012) “Explaining the Decline in the U.S. Labor Force Participation Rate” FRB Chicago Fed Letter, March, No. 296.

    • Search Google Scholar
    • Export Citation
  • Aaronson, D., B. Mazumder, and S. Schechter (2010) “What is Behind the Rise in Long-Term Unemployment?” FRB Chicago Economic Perspectives, Q2 2010, p. 2851.

    • Search Google Scholar
    • Export Citation
  • Barnichon, R. (2010) “Building a composite Help-Wanted Index,” Economics Letters, Vol. 109, pp. 1758.

  • Barnichon, R., M. Elsby, B. Hobijn, and A. Sahin (2010) “Which Industries are shifting the Beveridge Curve,” FRBSF Working Paper 2010–32.

    • Search Google Scholar
    • Export Citation
  • Blanchard, O.J and P. Diamond (1994) “Ranking, Unemployment Duration, and Wages,” Review of Economic Studies, Vol. 61, pp. 41734.

    • Search Google Scholar
    • Export Citation
  • Blanchard, O.J and L.H. Summers (1987) “Hysteresis in Unemployment,” European Economic Review, Vol. 31, pp. 288-95.

  • Card, D., J. Kluve and A. Weber (2010) “Active Labor market Policy Evaluation: A Meta-Analysis”, NBER Working Paper No. 16173.

  • Card, D. and P.B. Levine (2000) “Extended Benefits and the Duration of Unemployment Insurance Spells: Evidence From the New Jersey Extended Benefit Program,” Journal of Public Economics, 2000, v78, pp. 10738.

    • Search Google Scholar
    • Export Citation
  • Daly, M., B. Hobijn, and R. Valletta (2011) “The Recent Evolution of the Natural Rate of Unemployment” IZA Discussion Paper No. 5832.

    • Search Google Scholar
    • Export Citation
  • Elsby, M.W. L., B. Hobijn, A. Şahin, and R.G. Valletta (2011) “The Labor Market in the Great Recession: An Update,” Brooking Papers on Economic Activity, Fall, pp. 35371.

    • Search Google Scholar
    • Export Citation
  • Guichard, S. and E. Rusticelli (2011) “Reassessing the NAIRUs after the Crisis,” OECD Economics Department Working Paper No. 918, OECD, Paris.

    • Search Google Scholar
    • Export Citation
  • Heinrich, Carolyn J., P. Mueser, and K. Troske (2008) “Workforce Investment Act Non-Experimental Net Impact Evaluation,” IMPAQ International.

    • Search Google Scholar
    • Export Citation
  • Hotz, V. Joseph, G. W. Imbens, and J. A. Klerman (2006) “Evaluating the Differential Effects of Alternative Welfare-to-Work Training Components: A Reanalysis of the California GAIN Program.” Journal of Labor Economics, Vol. 24, pp. 521566.

    • Search Google Scholar
    • Export Citation
  • Krueger A., and A. Mueller (2011) “Job Search, Emotional Well Being, and Job Finding in a Period of Mass Unemployment: Evidence from High-Frequency Longitudinal Data,” Brooking Papers on Economic Activity, Spring 2011.

    • Search Google Scholar
    • Export Citation
  • Lockwood, B. (1991) “Information Externalities in the Labour Market and the Duration of Unemployment,” Review of Economic Studies, Vol. 59, pp. 51450.

    • Search Google Scholar
    • Export Citation
  • Machin, S., and A. Manning (1999) “The Causes and Consequences of Long-Term Unemployment in Europe,” in Handbook of Labor Economics, Vol. 3C, pp. 30853139, Elsevier Science, New York.

    • Search Google Scholar
    • Export Citation
  • OECD (2012) 2012 Economic Survey—United States, forthcoming, OECD Paris.

  • OECD (2011) OECD Employment Outlook 2011, OECD Paris.

  • Rothstein, J. (2011) “Unemployment Insurance and Job Search in the Great Recession,” Brooking Papers on Economic Activity, Fall, pp. 14396.

    • Search Google Scholar
    • Export Citation
  • Sahin, A., J. Song, G. Topa, and G. Violante (2011) “Measuring Mismatch in the U.S. Labor Market,” mimeo FRBNY and NYU.

  • Valletta, R. and K. Kuang (2012) “Why Is Unemployment Duration So Long?” FRBSF Economic Letter 201203.

  • Valletta, R. and K. Kuang (2010) “Extended Unemployment and UI Benefits,” FRBSF Economic Letter 201012.

  • Van Zandwedhe, W. (2012) “Interpreting the Recent Decline in Labor Force Participation,” FRB Kansa City Economic Review, Q1 2012.

  • Weidner, J. and J.C. Williams (2011) “What Is the New Normal Unemployment Rate?” FRBSF Economic Letter, 201105.

  • White House (2012a), Economic Report of the President, White House, Washington D.C.

  • White House (2012b) “President’s Plan to Provide Americans with Job Training and Employment Services,” White House and Department of the Treasury, Washington D.C.

    • Search Google Scholar
    • Export Citation
  • White House (2011), American Jobs Act, White House, Washington D.C.

1

Prepared by Eric Le Borgne.

2

Unless otherwise specified, the use of “long-term unemployment, (LTU)” refers to persons unemployed for 27 weeks or more, as currently used in the United States (internationally, LTU is associated with a duration of 52 weeks or more).

3

The LTU share was on a secular upward trend prior to the financial crisis but remained low for OECD countries. Both the rate and share of individuals unemployed for 52 weeks or more were low in the United States. compared to the OECD average (0.5 percent versus 1.7 percent, and 10 percent versus 29 percent, respectively, OECD, 2011).

4

Three years into the recovery the LTU rate has receded by ¼ of its pre-recession surge against ⅔ for the STU.

5

LTU rates by a given occupation and industry are calculated as the ratio of the long-term unemployed workers who were previously employed in a given occupation or industry, divided by the sum of those that are currently employed in that occupation or industry and the unemployed workers that were previously employed in that occupation or industry.

6

Unobserved heterogeneity (e.g., the quality of a worker not captured through education), could be an issue in analyzing the impact of unemployment duration on exit rates. The negative correlation between unemployment duration and the probability that a worker finds a job due to unobserved quality becomes more acute when the overall exit rate from unemployment is high (Machin and Manning, 1999). Thus, unobserved heterogeneity is less of a concern when unemployment is mostly cyclical, as is currently the case.

7

Krueger and Mueller (2011) find similar results.

8

That unemployment to employment exit rates increase as unemployment benefits near exhaustion has also been found in Aaronson et al. (2010) using older data.

9

Should the LTU exit probability remain at its March 2012 level, it would take an additional 15 months for half (of the current LTU to find a job. By contrast, it would take 2.5 years for ¾ of the current LTU to find a job.

10

Unemployment hysteresis embodies the idea that the equilibrium unemployment rate depends on the history of the actual unemployment rate (Blanchard and Summers, 1987).

11

Aaronson et al. (2012) and Van Zandweghe (2012) find similar magnitudes of the impact of cyclical versus structural factors in the decline of the labor force participation rate since 2007.

12

Rothstein (2011) also finds that the extended unemployment insurance significantly affects the probabilities that the long-term unemployed leave the labor force, but that the aggregate impact on labor force participation is minimal.

13

Sectoral Beveridge curves assume that the unemployed are searching for jobs in their previous industry of employment; such an assumption would likely not hold in sectors that experienced large and protracted shocks.

14

For example, in the President’s FY2013 budget proposal, the allocation to the Universal Displaced Worker Program can finance up to one million workers per year to receive job assistance and training. This can provide re-employment services to 19 percent of the long-term unemployed. The budget allocation in 2007 would have been able to provide these services to about 40 percent of the long-term unemployed.

United States: Selected Issues
Author: International Monetary Fund
  • View in gallery

    The Emergence of a Long-Term Unemployment Problem?

  • View in gallery

    A permanent Deterioration in Labor Force Participation and in Labor Market Efficiency?

  • View in gallery

    Job Matching and Finding Rates, by Duration of Unemployment