United States: Selected Issues
Author: International Monetary Fund
Publication Date:
25 Jul 2011
eISBN:
9781462317356
Language:
English
Keywords:
ISCR; CR; price; mortgage; foreclosure; liquid asset; work skill; job reallocation; work-skill loss; job-search effort; readiness assistance; Employment; Unemployment; Labor markets; Manufacturing; skills training
Download PDF (639 KB)

The note delves on the U.S. housing market outlook, the potential benefits of mitigating distressed sales household deleveraging, and the recovery. Policies to facilitate labor market adjustment to reduce the large employment volatility without affecting efficient labor allocation could prevent problems. U.S. firms are hoarding money but it is likely to be spent to boost firms’ capital expenditure, rather than kept as precautionary balances. The note discusses commodity price shocks affecting Treasury inflation protected securities (TIPS), budget institutions for federal fiscal consolidation, and mortgage delinquencies in the United States.

Abstract

The note delves on the U.S. housing market outlook, the potential benefits of mitigating distressed sales household deleveraging, and the recovery. Policies to facilitate labor market adjustment to reduce the large employment volatility without affecting efficient labor allocation could prevent problems. U.S. firms are hoarding money but it is likely to be spent to boost firms’ capital expenditure, rather than kept as precautionary balances. The note discusses commodity price shocks affecting Treasury inflation protected securities (TIPS), budget institutions for federal fiscal consolidation, and mortgage delinquencies in the United States.

III. Policies to Facilitate Labor Market Adjustment1

Following massive layoffs during the Great Recession, the recovery has produced relatively few jobs so far, with some sectors doing significantly better than others. These factors contribute to record-high long-term unemployment and sectoral reallocation pressures, which would affect work skills among the unemployed and, thus, structural unemployment. With this backdrop, public policy has an important role in supporting labor market adjustment, in particular by boosting training programs. Measures to reduce the large employment volatility in the United States without affecting efficient labor allocation could prevent similar problems in the future.

A. introduction

1. The Great Recession has caused major dislocations in the U.S. labor market. After reaching 10.1 percent in late 2009, the unemployment rate has declined only modestly and a record-high number of people have been out of work for six months or more. Joblessness broadly defined (which includes discouraged and involuntary part-time workers) remains near the peak 17 percent of the labor force reached last year while labor force participation dropped and employment as a share of working-age population is at its lowest level in 25 years. This shock had regional and sectoral dimensions, with unemployment rates varying from 3.9 percent in North Dakota to 14.9 percent in Nevada in 2010. Construction remains very depressed while production of durable goods has turned around and demand for health care workers continued broadly unabated by the crisis.

2. Recent research has shown that these large dislocations could have raised structural unemployment. Estevão and Tsounta (2011) show that, historically, severe housing shocks and increased skill mismatches are associated with higher unemployment rates at the state level even after controlling for common cyclical effects, with compounded effects if mismatches and bad housing market conditions interact. Given the rise in mismatch and unequal housing market performance during the crisis, the paper estimates that the structural unemployment rate could have reached 6¾ percent in 2010. Research by the Federal Reserve Bank of San Francisco has estimated that extended unemployment benefits have restrained job-search effort and displaced workers in construction could have a hard time finding jobs in other sectors, both factors raising structural unemployment somewhat.2

3. This chapter focuses on the need for large sectoral reallocation of workers and the “jobless” aspect of the current recovery; both potential structural issues. It shows that (1) the recent cycle was characterized by significant sectoral heterogeneity, with many industries actually posting employment losses so far in the recovery; and that (2) the recovery has been “jobless” so far, even after accounting for weak output growth. The chapter recognizes that long-term joblessness is also a key pressure on structural unemployment, as work skills deteriorate with idleness. Thus, the observed slower job creation in this recovery is a possible structural problem as it helps to perpetuate long unemployment spells. The chapter concludes that there is a role for public policy to facilitate job reallocation, and either boost or maintain worker skills through training programs while they search for jobs.

B. Structural Job Losses During the Great Recession

4. To diagnose employment flows in each industry as being either structural or cyclical, we first track the direction of job flows during and after the recession. Job losses (or gains) during the recession that are reversed during the recovery would be classified as cyclical.3 If, instead, employment declines (or increases) continued during the recovery, the adjustments would be classified as structural, at least so far in the recovery. (Figure 1 and Tables 1–4) Even if the latter category recovers at a certain point, the changes triggered by the crisis would still be “structural” unless sectoral employment shares return to pre-crisis levels.

Figure 1.
Figure 1.

Cyclical and Structural Changes in Employment across U.S. Sectors

Citation: IMF Staff Country Reports 2011, 202; 10.5089/9781462317356.002.A003

Sources: Bureau of Labor Statistics; Haver Analytics; and Fund staff estimates.* The dates of peaks and troughs are those defined by the NBER’s Business Cycle Dating Committee, the recovery periods are for the 23 months following the NBER’s trough.
Table 1.

Industries with Structural Employment Losses, December 2007-May 2011

(aligning with green triangles in figure 1)

article image
Sources: Bureau of Labor Statistics; Haver Analytics; and Fund staff calculations.
Table 2.

Industries with Structural Employment Gains, December 2007 - May 2011

(aligning with orange circles in figure 1)

article image
Sources: Bureau of Labor Statistics; Haver Analytics; and Fund staff calculations.
Table 3.

Industries with Procyclical Employment Flows, December 2007 - May 2011

(aligning with blue diamonds in figure 1)

article image
Sources: Bureau of Labor Statistics; Haver Analytics; and Fund staff calculations.
Table 4.

Industries with Countercyclical Employment Flows, December 2007 - May 2011

(aligning with red x’s in figure 1)

article image
Sources: Bureau of Labor Statistics; Haver Analytics; and Fund staff calculations.

5. A second method—based on deviations from usual cyclical changes in employment—can also help to identify structural labor market changes. If a recovery’s job content changes significantly, flows out of the unemployment pool could be either larger or smaller than usual, thus affecting the length of unemployment spells and, possibly, labor force attachment and work skills. To gauge the sensitivity of sectoral employment to the cycle, we regressed changes in industry’s employment on real GDP change (our measure of the cycle), dummy variables denoting recession periods and recoveries, interactions of real GDP growth with both dummies, and additional controls for wars, strikes, and other employment disruptions. To generate a comparison of employment behavior during the current episode with that of the other postwar cycles since 1957, we re-estimated the equation eight times, stopping the sample in the quarter corresponding to a business cycle peak.4 After that, we calculated deviations between sectoral employment growth and simulated out-of-sample forecasts seven quarters ahead to match the latest cycle.

Either procedure suggests that significant structural changes might be at play:

  • The first measure (Figure 1) demonstrates the substantial heterogeneity in employment performance across industries. It suggests that the housing bust and healthcare needs have triggered structural changes in U.S. employment. Construction sector industries (and related ones, like furniture producers) are among the top structural losers (Table 1 and southwestern quadrant of Figure 1), while health care providers and related industries feature prominently within the list of structural gainers (Table 2 and northeastern quadrant of Figure 1). Interestingly, federal civilian employment has grown during the crisis and during the recovery, thus supporting the labor market throughout the whole cycle. The durable goods manufacturing sector, which has also benefitted from the depreciation of the U.S. dollar, is among the largest procyclical industries (Table 3 and northwestern quadrant of Figure 1). Only a few industries behaved countercyclically, including state and local employment (Table 4 and southeastern quadrant of Figure 1).

  • Both measures indicate that employment reductions in the latest episode were particularly bleak. Figure 1 shows that the job losses in the industries with structural losses during the late-2000s recession were outsized compared with both the early-2000s and early-1990s episodes, even though structural losses during early-2000s recovery (gray circles in the southwest quadrant of the chart on the left) were in some cases larger. Likewise, labor cutbacks during the 2007–09 recession were much larger than historical employment-output relationships would lead to expect.5 (Figure 2)

Figure 2.
Figure 2.

Sectoral Employment Changes Above and Beyond Usual Patterns During Downturns

Citation: IMF Staff Country Reports 2011, 202; 10.5089/9781462317356.002.A003

Source: Fund staff estimates.

  • Virtually all sectors have produced so far fewer jobs than implied by usual rebound patterns, suggesting this recovery has been “jobless”. Model-predicted employment in the last three cycles has almost always been too optimistic when compared with actual payrolls. (Figure 3) In the 2001 cycle every sectoral employment outcome performed below expectations, in the 2007 episode only mining and logging payrolls and durable goods manufacturing were stronger than the model’s forecast, and in the early 1990s only government outperformed expectations. By comparison, employment growth was stronger than expected in 10 industries in the 1974 recession, in 11 industries in the 1982 recession, and in 9 industries during the 1957 recession. This result is consistent with other research pointing to the “jobless” character of the recoveries in the early 1990s and early 2000s,6 and shows that the current recovery has followed a similar pattern.

Figure 3.
Figure 3.

Sectoral Employment Changes Above and Beyond Usual Patterns During Recoveries

Citation: IMF Staff Country Reports 2011, 202; 10.5089/9781462317356.002.A003

Source: Fund staff estimates.

C. Policy Implications

6. Public policies could facilitate the resolution of potentially large reallocation needs in the job market and other emerging structural problems. In terms of so-called active labor market programs, when well designed:7

  • Classroom and on-the-job training programs yield relatively positive impacts in the medium term, even if these programs often have insignificant (or even negative) short-term impacts on employment.8 Thus, training programs would be ideal for situations where cyclical unemployment is large but long unemployment spells and potential future job reallocation needs threaten future job matches.

  • Job-search assistance programs usually yield positive results in the short term.9 Thus, they can be used to help unemployed workers to find jobs in the current situation. Even though job availability is still the key issue in the current cyclical situation, job-search programs would be more effective if they help workers to find jobs across state boundaries, which would ease regional unemployment disparities.

  • There is some cross-country evidence that direct subsidies to private sector jobs are effective in raising employment rates in the short term.10

7. However, more research on the effectiveness of U.S. active labor market policies is needed before new programs are created or existing ones are better calibrated. The U.S. system is characterized by a decentralized provision of government services for the unemployed, which may be more attuned to local needs, but may lack a consolidated, national strategy to address current labor market problems. In particular:

  • The federal government has nearly 50 different programs dispersed across nine federal departments geared toward helping the unemployed, both through training and job-search assistance programs.11

  • At face value, such proliferation of programs may cause duplications and inefficient spending, although the one-stop shop career centers maintained by the Labor Department are an effort to channel these programs efficiently.12

  • States have their own training and job-search programs, which add to the focus on decentralized service provision.

8. Existing active labor market programs cost little to federal and state budgets, and could be expanded after an evaluation of how the overall system performs. Although, many people are affected by the federal programs, the cumulative budget dedicated to them amounted to only about $18 billion in 2009 or a bit more than 0.1 percent of GDP.13 State-level programs are funded in many different ways, including through transfers from the federal government (already included in the total amount of resources dedicated to these policies at the federal level). There are no readily available, up-to-date consolidated figures of how much states spend on these policies, but indirect evidence suggest that the amounts are very small as a share of GDP.14

9. The U.S. education system can be further leveraged to provide training for the unemployed, in particular through community colleges. Recently, the U.S. government has partnered with the private sector to increase training opportunities for manufacturing workers in community colleges across 30 U.S. states.15 Further changes to the unemployment insurance system by making the provision of benefits conditional not only on active job search, but also on enrollment in certified training courses, could help sustain labor force attachment and skill acquisition until the cyclical situation improves.

10. At a deeper level, excessive layoffs exacerbated the persistent unemployment problem during the current cycle. There are many incentives for U.S. firms to fire workers, instead of changing the workweek, as a way to adjust labor inputs in response to cyclical shocks.16 Importantly, the U.S. unemployment insurance system creates a bias toward this type of adjustment. Under the current system, benefits are available for workers who are laid off, but not, in general, for workers whose hours have been cut back for economic reasons. Moreover, the system is financed with revenues from employers’ payroll taxes, but that tax liability is based only partially on the employer’s layoffs history. After layoffs reach a certain threshold, marginal tax rate increases are set to zero, limiting the cost of firing additional workers. However, if the cycle turns out to be more prolonged and uncertainty on future economic prospects does not clear, what started as a cyclical issue could become a structural one, as firms chose productivity-raising methods or increased working hours from incumbent employees to meet production targets, instead of hiring the unemployed.

11. Looking ahead, institutional reforms to incentivize employee retention could limit the volatility of U.S. employment in future cycles, and reduce the risks of larger cyclical shocks being transformed into structural problems. Changing the incentives to layoffs in the current system would help to limit the economic fallout from possible long-term unemployment in future business cycles. For instance, broader introduction of short-time compensation programs (STC) in the United States, which would provide benefits to employees working less than the usual workweek, would better align incentives across different ways to adjust labor input to cyclical shocks.17 That would strengthen job stability, and avoid work-skill losses and exits from the labor force.

12. The international experience shows that more workweek flexibility could improve labor market performance through the cycles, without affecting labor reallocation following structural changes. Canada and Germany are good study cases, with broader and stronger STC programs (especially in Germany). As a result, fewer jobs were lost during the crisis in both countries than in the United States (even though output declines were of the same order of magnitude), while the workweek was reduced more drastically. (Figure 4) The unemployment rate has also increased less in Canada and has actually declined in Germany during the crisis. (Figure 5) Moreover, at least one in-depth study of the German institutional setup has shown that its higher reliance on short-term flexibility in working hours has not curbed efficient work allocation across economic sectors, as STC arrangements are temporary. Structural changes in the economy are still followed by firings, hirings, and worker reallocations.18

Figure 4.
Figure 4.

Comparison of Labor Cutbacks in the United States, Canada, and Germany

Citation: IMF Staff Country Reports 2011, 202; 10.5089/9781462317356.002.A003

Sources: Statistiches Bundesamt; Statistics Canada; Bureau of Labor Statistics; Haver Analytics; and Fund staff calculations.
Figure 5.
Figure 5.

Cumulative Change in Unemployment Rates in the United States, Canada, and Germany

Citation: IMF Staff Country Reports 2011, 202; 10.5089/9781462317356.002.A003

Sources: IMF, World Economic Outlook.
Table 5.

Estimates of the sensitivity of industrial employment growth to output growth, 1947–2007

article image
Source: Fund staff estimates.*, ** denotes significance at the 5 percent and 1 percent levels, respectively.
Table 6.

Seven Largest Programs: Estimated Amount Spent on Employment and Training Activities in Fiscal Year 2009 and Estimated Number of Participants Served

article image
Source: GAO survey of agency officials.

Estimates may include funds provided by the American Recovery and Reinvestment Act of 2009 (Recovery Act).

Officials provided the estimated number of participants for the most recent year for which data were available.

This number represents the monthly average number of individuals receiving TANF cash assistance who were engaged in work activities such as subsidized employment, work experience, on-the-job training, job search and job readiness assistance, community service, vocational educational training, job skills training, and in certain circumstances education directly related to employment. It does not include the number of individuals engaged in unsubsidized employment. Officials were unable to provide an annual estimate.

Appendix 1. Data Definitions and Sources

Our measures of output growth and employment growth are those compiled by the U.S. authorities. The dependent variable, Δeit, is the seasonally adjusted annualized percent change in quarterly average nonfarm payroll employment in each of 15 broad-based industries. These data come from the Bureau of Labor Statistics’ Current Employment Statistics (CES) program and are available for most of the industries back to the late-1930s. The output variable is the percent change in real growth from the previous quarter, also seasonally adjusted and at an annual rate. This series is published by the Bureau of Economic Analysis in their National Income and Product Accounts release. Quarterly output data are available beginning in 1947, which is when our estimation period also begins.

Definitions of additional variables:

In equations (2) and (3), the recession periods included in CONt are those defined by the NBER’s Business Cycle Dating Committee. The Dating Committee does not provide dates for recoveries from recessions, so we coded the variable RECt as equal to 1 from the end of the recession to the closing of the gap between the actual unemployment rate and its Hodrick-Prescott filtered trend. We added variables to control for major factors influencing employment growth:

  • a) A dummy variable for two major War periods—1950–1953 (Korean War) and 1965–1973 (Vietnam War).

  • b) Major strike activity distorted employment growth in mining, information, utilities, and leisure and hospitality industries, due to BLS’ convention of leaving striking employees off their counts of payroll employment. When these major work stoppages occurred, employment suddenly dropped, depressing the quarterly average of employment. Then in the quarter following the strike, employment growth jumped, as growth was being calculated from an artificially low base. To capture these differential effects, we coded two variables for each strike event. The first provides a measure of both timing and severity and is coded as the number of months in the quarter during which a strike took place. To capture the artificial employment surge as employees returned to work, we coded a dummy variable, equal to 1 in the quarter following the strike.

  • c) We gathered the dates of strike activity from several sources. For months starting in January 1973, archived Employment Situation news releases available on the Federal Reserve Bank of St. Louis’ FRASER archive (Federal Reserve Archival System for Economic Research). Some strike data were also gathered from FRASER’s archived Economic Report of the President for various years. For mining over 1948–50, we obtained months of mining strike activity from Bernstein and Lovell (1953).

References

  • Abraham, Katharine and Susan Houseman, 1993, Job Security in America, The Brookings Institution, Washington, DC.

  • Batini, Nicoletta, Marcello Estevão, and Geoffrey Keim, 2010, “Production and Jobs: Can We Have One Without the Other?” Selected Issues Paper, Chapter 3, United States—Article IV Consultation, July.

    • Search Google Scholar
    • Export Citation

  • Bernstein, Irving and Hugh G. Lovell, 1953, “Are Coal Strikes National Emergencies?” Industrial and Labor Relations Review, Vol. 6, No. 3, pp. 352–367, April.

    • Search Google Scholar
    • Export Citation

  • Card, David, Jochen Kluve, and Andrea Weber, 2011, “Active Labor Market Policy Evaluations: A Meta-Analysis,” NBER Working Paper 16173, National Bureau of Economic Research, Cambridge, MA.

    • Search Google Scholar
    • Export Citation

  • Duscha, Steve and Wanda Lee Graves, “The Employer as the Client: State-Financed Customized Training” Department of Labor, http://wdr.doleta.gov/research/eta_default.cfm?fuseaction=dsp_resultDetails&pub_id=2345&bas_option=Title&start=1&usrt=4&stype=basic&sv=1&criteria=customized

    • Search Google Scholar
    • Export Citation

  • Estevão, Marcello, 2007, “Labor Policies to Raise Employment,” IMF Staff Papers, Vol. 54, No. 1, pp. 113–138, March.

  • Estevão, Marcello and Evridiki Tsounta, 2011, “Has the Great Recession Raised U.S. Structural Unemployment?” IMF Working Paper, WP/11/105, May.

    • Search Google Scholar
    • Export Citation

  • Groshen, Erica and Simon Potter, 2003, “Has Structural Change Contributed to a Jobless Recovery?” Current Issues in Economics and Finance, Vol. 9, No. 8, Federal Reserve Bank of New York, August.

    • Search Google Scholar
    • Export Citation

  • Schweitzer, Mark, 2003, “Another Jobless Recovery?” Economic Commentary, Federal Reserve Bank of Cleveland, March.

  • Valletta, Rob and Katherine Kuang, 2010, “Is Structural Unemployment on the Rise?” FRBSF Economic Letter, 2010-34, Federal Reserve Bank of San Francisco, November.

    • Search Google Scholar
    • Export Citation
1

Prepared by Marcello Estevão and Geoffrey Keim.

4

The equation is Δeit = β0+β1Δyt+β2CONt+β3RECt+β4CONt·Δyt+β5RECt·Δyt+β4Zit, where the subscripts i and t represent industry and time, respectively; Δeit refers to the change in average quarterly employment; Δyt refers to the change in real aggregate GDP; CONt is a dummy variable denoting NBER recession periods; RECt is a dummy variable denoting recoveries; and Zit refers to other factors unrelated to aggregate growth that could be influencing employment growth. For a fuller discussion of the data sources, please consult the appendix. The cycles for which we produced forecasts were the ones with peaks in 1957Q3, 1960Q2, 1969Q4, 1973Q4, 1981Q3, 1990Q3, 2001Q1, and 2007Q4. The 1980 recession was excluded from this exercise, as the forecast period would end in the 1982 recession.

5

Batini, Estevão, and Keim (2010) present evidence that the high degree of uncertainty during the 2007–09 financial crisis was a key factor behind the breakdown of the employment/output relationship.

7

Active labor market policies consist mainly of training, targeted subsidies for job creation, public employment services, and other expenditures aimed at promoting employment. The definition excludes general macroeconomic policies and nontargeted policies to lower labor costs.

8

As shown in the comprehensive metadata evaluation presented in Card et al (2011).

11

U.S. Government Accountability Office, “Employment and Training Programs: Opportunities Exist for Improving Efficiency,” Testimony to Congress, April 7, 2011.

12

One-stop shop career centers are designed to provide a full range of assistance to job seekers under one roof. Established under the Workforce Investment Act of 1998, the centers offer training referrals, career counseling, job listings, and similar employment-related services. There are many centers in all U.S. states.

13

Table 6 reports the amount spent and the number of beneficiaries affected by the seven largest federal programs, which account for about 75 percent of the total federal budget for active labor market programs.

14

Duscha and Graves (2006) report that states spent an aggregate amount of only $560 million in training programs in 2006. This said, the figure may have increased during the crisis and do not account for expenditures in other programs, including job-search assistance.

16

For a more complete list of factors biasing labor adjustment actions towards laying off workers, see Abraham and Houseman (1993).

17

Several U.S. states have some form of STC program, but they tend to be small, loaded with pre-conditions, and not well publicized. In Canada, many of the pre-conditions for qualification for STC were waived during the recent recession, which served to increase participation.

Same Series

Other IMF Content

Other Publishers

Cover IMF Staff Country Reports
United States: Selected Issues
Author: International Monetary Fund
Volume/Issue:
Volume 2011: Issue 202
Publisher:
International Monetary Fund
ISBN:
9781462317356
ISSN:
1934-7685
Pages:
87
DOI:
https://doi.org/10.5089/9781462317356.002
  • View in gallery

    Cyclical and Structural Changes in Employment across U.S. Sectors

  • View in gallery

    Sectoral Employment Changes Above and Beyond Usual Patterns During Downturns

  • View in gallery

    Sectoral Employment Changes Above and Beyond Usual Patterns During Recoveries

  • View in gallery

    Comparison of Labor Cutbacks in the United States, Canada, and Germany

  • View in gallery

    Cumulative Change in Unemployment Rates in the United States, Canada, and Germany

Cyclical and Structural Changes in Employment across U.S. Sectors

Sectoral Employment Changes Above and Beyond Usual Patterns During Downturns

Sectoral Employment Changes Above and Beyond Usual Patterns During Recoveries

Comparison of Labor Cutbacks in the United States, Canada, and Germany

Cumulative Change in Unemployment Rates in the United States, Canada, and Germany