This chapter examines whether the recovery will be accompanied by significant job creation or will be “jobless” like the previous two U.S. recoveries. It compares the recent recession with previous episodes and employs panel and time-series regressions to pin down the fundamental factors underlying the relationship between growth and (un)employment. The recent crisis has destroyed more jobs than any other post-Depression episode. However, if economic uncertainty recedes significantly, employment should rebound more strongly when compared to past jobless recoveries; although probably not strong enough to prevent a slow decline in the unemployment rate.

Abstract

This chapter examines whether the recovery will be accompanied by significant job creation or will be “jobless” like the previous two U.S. recoveries. It compares the recent recession with previous episodes and employs panel and time-series regressions to pin down the fundamental factors underlying the relationship between growth and (un)employment. The recent crisis has destroyed more jobs than any other post-Depression episode. However, if economic uncertainty recedes significantly, employment should rebound more strongly when compared to past jobless recoveries; although probably not strong enough to prevent a slow decline in the unemployment rate.

This chapter examines whether the recovery will be accompanied by significant job creation or will be “jobless” like the previous two U.S. recoveries. It compares the recent recession with previous episodes and employs panel and time-series regressions to pin down the fundamental factors underlying the relationship between growth and (un)employment. The recent crisis has destroyed more jobs than any other post-Depression episode. However, if economic uncertainty recedes significantly, employment should rebound more strongly when compared to past jobless recoveries; although probably not strong enough to prevent a slow decline in the unemployment rate.

A. Introduction

1. The recent economic recession has had a severe impact on employment by historical standards and the recovery has not been “job rich” so far (Figure 1). The unemployment rate increased by over 5 percentage points since the onset of the crisis, reaching levels comparable to historic post-war records in the early-1980s recession. However, unlike past deep recessions and similarly to the previous two recessions, the unemployment rate has not improved much since the economic recovery started in the middle of 2009, and the duration of unemployment has actually increased (Figure 2).

Figure 1.
Figure 1.

Real GDP Growth and the Change in the Unemployment Rate, 1902-2009

Citation: IMF Staff Country Reports 2010, 248; 10.5089/9781455206759.002.A003

Sources: Bureau of Economic Analysis; Bureau of Labor Statistics; Historical Statistics of the United States; U.S. Census Bureau; Haver Analytics; and Fund staff estimates.
Figure 2.
Figure 2.

Long-Term Unemployment Across U.S. Postwar Recessions

Citation: IMF Staff Country Reports 2010, 248; 10.5089/9781455206759.002.A003

Sources: Bureau of Labor Statistics and Haver Analytics.

2. These developments raise the specter of a “jobless recovery” going forward. Labor markets have changed significantly since the 1980s, with the past two recoveries being characterized by high productivity growth and little job creation. In addition, the current recession originated in a deep financial turmoil; a shock that historically has tended to have persistent impacts on job flows. These two facts suggest that jobs and unemployment may be slow to recover this time around as well.

3. This chapter argues that this recovery will be richer in jobs than the past two episodes if economic uncertainty declines significantly, although unemployment will probably remain elevated for a while. A comparison with past recessions shows that the recent episode produced a deeper labor adjustment than usual, including very weak labor cost growth. Econometric estimates show that the growth/(un)employment relationship has changed over time in the United States, with unemployment being more responsive to output changes during the recent crisis than in the previous 20 years. Shocks in financial conditions, relative labor costs, and economic uncertainty—measured here by stock market volatility—explain a good share of changes in the growth/(un)employment relationship, and suggest that rapid future employment gains will depend on a decisive reduction in uncertainty.

B. How Does this Recession Compare to Previous Ones?

4. The 2007 downturn followed a general pattern that places it in between the postwar recessions and the Great Depression (Figures 36). In particular:

  • The current recession marked the largest postwar upswing in the unemployment rate and a similar upswing to that observed during the first phase (1930–1933) of the Great Depression. It also marked the largest postwar contraction in employment, but half the decline seen in the Great Depression.

  • In the recent recession, more hours were lost relative to the ‘mild’ 1991 and 2001 downturns, but fewer hours (half) were lost relative to the Great Depression. The decline in total labor input followed the postwar trend with a 70:30 heads/hours split, contrary to the Great Depression during which more hours than persons were lost (with roughly a 40:60 heads/hours split).

  • Unit labor costs have dropped more this time than in earlier downturns, although less than in the Great Depression, a sign that wages are responding to the large flows into unemployment and to reductions in hiring rates. That bodes well for future employment growth vis-à-vis past recessions that faced weaker downward adjustments in labor costs.

Figure 3.
Figure 3.

Rolling Okun's Law Coefficients and Growth Compatible with Stable Unemployment, 1915-2009

Citation: IMF Staff Country Reports 2010, 248; 10.5089/9781455206759.002.A003

Sources: Bureau of Economic Analysis; Bureau of Labor Statistics; Historical Statistics of the United States; U.S. Census Bureau; and Fund staff estimates.
Figure 4.
Figure 4.

Comparing the Steep Recessions and Job-Rich Recoveries Contractions

(Peak to Trough)

Citation: IMF Staff Country Reports 2010, 248; 10.5089/9781455206759.002.A003

Sources: Bureau of Economic Analysis, Bureau of Labor Statistics, and Fund staff estimates.
Figure 5.
Figure 5.

Comparing the Most Recent Recessions and Jobless Recoveries Contractions

(Peak to Trough)

Citation: IMF Staff Country Reports 2010, 248; 10.5089/9781455206759.002.A003

*Unemployment is scaled by a factor of 10.Sources: Bureau of Economic Analysis, Bureau of Labor Statistics, and Fund staff estimates.
Figure 6.
Figure 6.

Comparing the Great Depression and the Great Recession Contractions

(Peak to Trough)

Citation: IMF Staff Country Reports 2010, 248; 10.5089/9781455206759.002.A003

Source: Bureau of Economic Analysis; Bureau of Labor Statistics; Kendrick (1961); Haver Analytics; and Fund staff estimates.

5. As a result, the relationship between unemployment and growth was stronger during the recent recession than at any post-war recession (Figure 3). Increases in the rate of output growth have been associated consistently with a lower unemployment rate—with a ratio of 2½:1—and, statistically, this relationship has not shown structural breaks over 1900–2009 (in other words the ratio has not shifted permanently to higher or lower levels over time, hovering around its mean). However, this relationship has varied a lot in particular periods of time, and the current recession marks the second time in history following the Great Depression that the ratio shifted down drastically from its long-run average (implying a ratio of 1¾:1).2 A lower ratio suggests that—other things equal—each percentage point of growth above trend creates more jobs and vice versa.

C. How Will the Recovery be Like?

6. Turning to the recovery seen in the data, the initial evidence (2009 Q3-2010 Q1) points to similarities with the 1933–1934 recovery with respect to the behavior of unemployment and labor costs (Figures 36). In particular:

  • Unemployment has continued to deteriorate in a way similar to the 1933–1934 recovery. To a lesser extent this also characterized other postwar recoveries, as unemployment is a lagging indicator of the cycle. Employment and the labor force however are adjusting more sluggishly this time around than during the Great Depression, likely reflecting differences in the type of shock behind each downturn.3

  • Unit labor costs have declined sharply in 2009—also thanks to large increases in productivity—similarly to the Great Depression, but much less than in previous post-war recoveries.

7. To predict the path of recovery going forward, we focus on the relationship between employment and growth when economic uncertainty is high. In particular, we estimate a simple tri-variate SVAR relating changes in employment, real GDP growth, and stock market volatility (a proxy for economic uncertainty) using U.S. data from 1930 to 2009. As Figure 7 indicates, the highest-on-record spikes in volatility (in 1930 and 2008) coincide with lowest-on-record Okun’s ratios, suggesting that uncertainty about the economic environment is associated with more employment losses than during more certain times, given the same growth rate in real GDP. Impulse response functions from the SVAR (with shocks identified using the following Cholesky ordering: stock market volatility➜ real GDP growth➜changes in employment) reaffirm that an adverse shock to stock market volatility depresses employment growth, other things equal (Figure 8).

Figure 7.
Figure 7.

Stock Market Volatility versus Rolling Okun's Law Coefficients

Citation: IMF Staff Country Reports 2010, 248; 10.5089/9781455206759.002.A003

Sources: Bloomberg, LP and Fund staff estimates.
Figure 8.
Figure 8.

Impulse Responses from a Tri-Variate SVAR

Citation: IMF Staff Country Reports 2010, 248; 10.5089/9781455206759.002.A003

Sources: Bureau of Economic Analysis; Bureau of Labor Statistics; Historical Statistics of the United States; U.S. Census Bureau; Bloomberg, LP; Haver Analytics; and Fund staff estimates.

8. Our results show that the stronger-than-normal decline in employment per output lost between 2007 and 2008 reflected in great part mounting exceptional economic uncertainty at that time—as shown by the sharp increase in stock market volatility. With no recovery in sight and bad domestic and world macroeconomic news flowing in—an environment reminiscent of the 1930s, as acknowledged for example in Daly and Hobijn (2010) and Krueger (2010) 4—firms opted for more firing and less hiring this time relative to previous recessions.

9. Predictions using the same model suggest that uncertainty could be a key factor influencing the speed of employment growth going forward. Figure 9 plots alternative forecast scenarios for the unemployment rate implied by the model’s forecasts for employment growth conditioned on different assumptions for future stock market volatility. In the first scenario, volatility decreases steadily over the forecast horizon to its long-term average level, while the second scenario supposes volatility jumps in 2010 due to a financial shock before falling. The decreased volatility under the first scenario helps to support job creation, and employment is back to its pre-crisis level by 2012–13. By contrast, employment under the high-volatility scenario takes an additional year to recover to the pre-recession level. These two volatility paths help to highlight risks to the WEO baseline forecast (assuming a given path for labor force participation), with the steadily falling volatility implying a swifter (albeit still slow) fall in the unemployment rate, while a financial shock could slow the decline in the unemployment rate.

Figure 9.
Figure 9.

Unemployment Scenarios

Citation: IMF Staff Country Reports 2010, 248; 10.5089/9781455206759.002.A003

Sources: Bureau of Labor Statistics; Bloomberg, LP; Haver Analytics; and Fund staff estimates.

10. Taking into account labor force dynamics, implications for the unemployment rate are less clear, however, even within a scenario of reduced uncertainty. The relationship between unemployment, employment and labor force participation exhibits a break in 1990, as labor force participation has become more pro-cyclical than before.5 Hence the ultimate decline in unemployment, given improvements in employment, will be a function of how participation behaves during the recovery.

11. Cross-country evidence also shows that uncertainty linked to financial stress would reduce employment growth, while slow growth in labor costs vis-à-vis capital costs (due to abundant labor supply) would raise it. Table 1 shows panel data estimates for 16 OECD countries of a labor demand equation relating employment growth to GDP growth, relative price of labor vis-à-vis capital, and measures of financial stress used in previous chapters of IMF’s World Economic Outlook.6 The results corroborate the view that weak wage growth vis-à-vis the cost of capital and a reduction in financial stress would raise employment growth for a given GDP path. Estimates of a modified Okun’s Law equation—in which relative labor costs and measures of financial stress work as shifters in the relationship between GDP growth and the unemployment rate—show that unemployment would also decline faster for a given rate of GDP growth under weak wage growth and reduced financial stress.

Table 1.

Estimating Labor Demand and Okun's Law for a Panel of Countries

article image
Notes: Standard errors are shown in parentheses. ***, **, and * denote significance at the 1 percent, 5 percent, and 10 percent levels, respectively. The dependent variable in equation (1) is the annual percent change in the number of workers, and in equation (2), the dependent variable is the first difference of the unemployment rate. The regressors included the percent change in real GDP, the difference between the hourly wage inflation rate (hourly wages were computed as the aggregate wage bill divided by aggregate hours) minus the inflation rate on capital goods (capital goods prices were measured as the implicit deflator on gross fixed capital formation), the bank lending rate, and a financial stress index, which is a weighted average of banking-related financial indicators, securities-market indicators, and exchange rate volatility (a higher value for index indicates increased stress in financial markets). Each equation also included country dummy variables to capture country fixed effects as well as year dummies. The sample included 11 advanced economies over 1983-2004.

D. Conclusions and Policy Implications

12. Overall, the analysis in this chapter points to a faster recovery in employment than observed in the “jobless recoveries” of the early 1990s and early 2000s. The basic reason for this conclusion can be put simply: as opposed to the other two episodes, employment has reacted very strongly to declines in production during the recession—probably because of the sharp increase in economic uncertainty. Going ahead, and also because of the large supply of available labor and the resulting sluggish labor costs, there are equilibrating pressures to hire more people for a given unit of production than during past shallow recessions.

13. However, some specific factors may dampen the rebound in employment, justifying additional policy support to job creation. The current episode has seen a surge in involuntary part-time employment—implying hours may grow ahead of “bodies” in the upturn—and ongoing economic shocks (more recently from Europe) are keeping uncertainty high. Tax incentives for net hiring could nudge firms to contract more labor for each unit of output being produced. Indeed, as discussed in Katz (2010), evidence for the United States suggests that firms respond to short-run reduction in marginal wage costs by moderately expanding employment (e.g., Card, 1990). Other evidence suggests that a net job creation tax credit would be an effective way to raise employment (Bartik and Bishop, 2009, and Congressional Budget Office, 2010). Estevão (2007) shows that subsidies to direct hiring by the private sector have been the best alternative among a set of active labor market policies to raise employment rates sustainably across a panel of OECD countries. To minimize economic distortions, subsidies should be temporary, though, and be unwound once unemployment rates get closer to structural levels. In addition, given the fiscal situation in the United States, any increase in spending in this area should be offset by a reduction in outlays in other areas or an increase in fiscal revenue.

  • Bartik, Timothy, and John Bishop, 2009, “The Job Creation Tax Credit: Dismal Employment Projections Call for Quick, Efficient, and Effective Response,” EPI Briefing Paper #248.

    • Search Google Scholar
    • Export Citation
  • Card, David, 1990, “Unexpected Inflation, Real Wages, and Employment Determination in Union Contracts,” American Economic Review, 79, September.

    • Search Google Scholar
    • Export Citation
  • Congressional Budget Office, 2010, Policies for Increasing Economic Growth and Employment in 2010 and 2011, January.

  • Daly, Mary and Bart Hobijn (2010), “Okun’s Law and the Unemployment Surprise of 2009,” FRBSF Economic Letter, 2010-07, Federal Reserve Bank of San Francisco, March.

    • Search Google Scholar
    • Export Citation
  • Estevão, Marcello, 2007, “Labor Policies to Raise Employment,” Staff Papers, Vol. 54, No. 1, pp. 113138, March.

  • Katz, Lawrence, 2010, Unemployment in the Great Recession: Structural Problems and Policy Responses, Testimony for the Joint Economic Committee, U.S. Congress, February.

    • Search Google Scholar
    • Export Citation
  • Knotek, Edward S., II, 2007, “How Useful Is the Okun’s Law?” Economic Review, Federal Reserve Bank of Kansas City, fourth quarter, pp. 73103.

    • Search Google Scholar
    • Export Citation
  • Krueger, Alan, 2010, Statement before the Joint Economic Committee, U.S. Department of Treasury,May 5.

  • Okun, Arthur, 1962, “Potential GNP: Its Measurement and Significance,” Proceedings of the Business and Economics Statistics Section, American Statistical Association, pp. 98104.

    • Search Google Scholar
    • Export Citation
1

Prepared by Nicoletta Batini, Marcello Estevão, and Geoffrey Keim.

2

These estimates are similar to Okun’s original estimates for the entire postwar era in the United States and are also close to Knotek’s (2007) updates of those estimates using data between 1948 Q2 and 2007 Q2.

3

As argued by many (e.g., Katz, 2010), the shock triggering the recent turmoil was different to the one in the Great Depression: (1) the recent shock has originated in the market for housing mortgages and thus hit the market for dwellings, generating a geographic lock-in effect that reduced mobility and job creation; (2) the shock penalized fast-growing areas; and finally (3) due to the specific type of credit crunch it generated, the recent shock proved particularly pernicious for start-ups in job-creating sectors.

4

“Job losses in this period (2007–2009) exceeded what one would predict from the sharp concurrent contraction in GDP by about 25 percent. The sharp loss in jobs around the time of the financial crisis resulted because the seizure of credit markets caused a sharp drop in economic activity, and because the panic that took hold of financial markets likely spread to employers in other sectors, causing them to react more than normally to a contraction in demand for their goods and services by shedding workers” (Krueger,2010).

5

Running a Chow Breakpoint test, the null of no breaks at a 1990 breakpoint can be rejected at the 1 percent confidence level [F-statistic= 4.35, Prob. F(3, 44) = 0.009; Log-likelihood ratio= 12.99; Prob. Chi Square (3) =0.0046]. Breaks at other dates can be rejected within the period going from 1960 to 2009.

6

Measures of stock market volatility are quite correlated across countries, thus the choice of using a broader measure of financial stress here. Stock market volatility is available for a much longer sample period, which justifies its use in the SVAR estimations for the United States presented in the previous paragraphs.

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

    Real GDP Growth and the Change in the Unemployment Rate, 1902-2009

  • View in gallery

    Long-Term Unemployment Across U.S. Postwar Recessions

  • View in gallery

    Rolling Okun's Law Coefficients and Growth Compatible with Stable Unemployment, 1915-2009

  • View in gallery

    Comparing the Steep Recessions and Job-Rich Recoveries Contractions

    (Peak to Trough)

  • View in gallery

    Comparing the Most Recent Recessions and Jobless Recoveries Contractions

    (Peak to Trough)

  • View in gallery

    Comparing the Great Depression and the Great Recession Contractions

    (Peak to Trough)

  • View in gallery

    Stock Market Volatility versus Rolling Okun's Law Coefficients

  • View in gallery

    Impulse Responses from a Tri-Variate SVAR

  • View in gallery

    Unemployment Scenarios