United States
Selected Issues

This Selected Issues paper focuses on U.S. potential growth in the aftermath of the crisis. It discusses recent productivity developments in the nonfarm business sector. The paper uses back-of-the envelope calculations to gauge the effect of diminished financial sector activity on GDP growth in coming years. A simple production function framework is used to give a long view of key factors explaining potential GDP growth in the United States in the last 30 years and explore possible developments in the next few years.

Abstract

This Selected Issues paper focuses on U.S. potential growth in the aftermath of the crisis. It discusses recent productivity developments in the nonfarm business sector. The paper uses back-of-the envelope calculations to gauge the effect of diminished financial sector activity on GDP growth in coming years. A simple production function framework is used to give a long view of key factors explaining potential GDP growth in the United States in the last 30 years and explore possible developments in the next few years.

I. U.S. Potential Growth in the Aftermath of the Crisis

Natalia Barrera, Marcello Estevão, and Geoffrey Keim

A. Introduction

1. Financial intermediation facilitates economic activity by allocating capital among different users; thus, financial activity is positively correlated with economic growth. Indeed, shocks to financial conditions seem to cause variations in real activity. That was likely the case in the United States between the mid-1990s and the mid-2000s, when a surge in securitization seems to have contributed to the simultaneous economic boom. Similarly, economic activity first slowed, and then cratered after the Lehman bankruptcy in September 2008 amid the ensuing overall tightening in financial conditions.

2. Looking forward, tighter financial conditions, together with economic restructuring, will slow U.S. potential growth.1 The protracted recession and tighter financial conditions will hurt investment and, thus capital accumulation, and the resulting high and more-persistent-than-usual unemployment rates will affect equilibrium rates of unemployment—both lowering potential growth. These factors will add to the downward pressure on potential growth from trends in labor force participation (driven mostly by demographics) and average hours worked per employee (a secular trend). Taking all into consideration, this chapter predicts that potential growth rates in the United States will fall below 2 percent in the coming years before rising slowly to about 2 percent by 2014.

3. The remainder of the chapter goes over the main arguments underlying staff’s potential growth projections. Noting that potential growth can be decomposed into trend labor productivity and labor input growth, the chapter first discusses recent productivity developments in the nonfarm business sector, an aggregation level preferred by productivity researchers and the Bureau of Labor Statistics. Then, back-of-the envelope calculations are used to gauge the effect of diminished financial sector activity on GDP growth in coming years. Finally, a simple production function framework is used to give a long view of key factors explaining potential GDP growth in the United States in the last thirty years and explore possible developments in the next few years.

B. Productivity Developments in the Nonfarm Business Sector

4. Labor productivity growth in the nonfarm business sector boomed between the mid-1990s and the mid-2000s. Growth in labor productivity (output divided by hours of work) in the U.S. nonfarm business sector—the most visible category tracked by the Bureau of Labor Statistics—rose from 1½ percent a year between 1973 and 1995 to 2½ percent between 1995 and 2000, and a whopping 3¼ percent between 2000 and 2004 (Table 1). Oliner and Sichel (2008) broke down productivity growth into different factors (as in formula (1) in footnote 2) and concluded that the first surge in labor productivity was driven by faster capital deepening in IT (information technology) equipment and machinery, and total factor productivity (TFP) in sectors producing this type of capital.2 The second surge occurred in capital deepening of other capital goods and in TFP growth outside sectors producing IT equipment. The non-IT related faster capital deepening in the period could be seen as the flip image of the jobless recovery after the 2001/2002 recession. Indeed, Oliner, Sichel, and Stiroh (2007) present evidence that the productivity boost during the first half of the 2000s was caused by a one-off industry restructuring in response to profit pressures and by a reallocation of material and labor inputs across industries.

Table 1.

United States: Contributions to Labor Productivity Growth in the Nonfarm Business Sector

(in percent)

article image
Sources: U.S. Bureau of Labor Statistics and Oliner and Sichel (2008).

Adjustments account for factor adjustment costs and utilization.

5. After the boom, productivity growth from 2005 to 2008 returned to rates similar to pre-1995 readings. Since 2004, productivity growth has halved due to both weaker capital deepening and lower TFP growth. At the eve of the financial crisis, the U.S. economy seemed to have returned broadly to a historical growth path. However, given the demographic pressures on labor force participation rates and the secular downward trend in hours worked per employee (both documented in the last section of this chapter), potential output growth in the United States was set to decline slowly.

C. Weaker Financial Intermediation

6. As a result of the crisis, financial intermediation will be weaker in the years ahead, contributing to a protracted recession and adding to demographic pressures on potential growth. Lower financial sector activity would affect potential growth directly through lower growth in output of finance, insurance and real estate (FIRE, which contributed a fourth of aggregate output growth between 2005 and 2007) and indirectly through reduced activity elsewhere in the economy, as credit scarcity raises the cost of capital and limits the types of projects being undertaken. That is even more important during a period of sectoral restructuring, as production of some nontradable goods (say, housing) shrinks and resources need to be transferred to the tradable goods sector. These are important factors behind the weak investment flows projected by the staff.

7. Indeed, past financial crises had severe and long-lasting effects on potential output.3 While this is true for a broad cross-section of financial crises, a review of the U.S. Savings and Loan (S&L) crisis of the late-1980s and early-1990s illustrates the effect of a financial crisis on the direct contributions of FIRE output to GDP growth. The S&L crisis lasted from 1986 to 1995, when 943 thrifts with a combined $519 billion in assets failed, and it was a key factor explaining the early 1990s recession in the United States. In the event, the contribution of FIRE output to GDP growth decreased (Figure 1).

Figure 1.
Figure 1.

Contribution of Finance, Insurance and Real Estate to GDP Growth

Citation: IMF Staff Country Reports 2009, 229; 10.5089/9781451839746.002.A001

Sources: Haver Analytics and Fund staff estimates.Note: The shaded areas correspond to NBER recessions.

8. By comparison, the current U.S. crisis is much more severe than the Savings and Loan episode. The combined assets of Washington Mutual, which was closed by the OTS, and Lehman Brothers, which filed for bankruptcy protection, totaled about $1 trillion. Other systemic financial institutions, with assets totaling $1.3 trillion (Wachovia with $764 billion and Merrill Lynch with $569 billion) have had to merge, while other financial institutions have required capital injections amounting to almost $500 billion since September 2007. These developments have had unprecedented implications for growth; notably, the contribution of FIRE to 2008 growth was negative for the first time in the history of the series.

9. Reflecting the relatively higher magnitude of financial sector distress, this chapter considers two scenarios, one in which the contribution of FIRE to overall growth goes from the historic average (1948 to 2008) of 0.6 percentage point to zero over the next several years (in line with the experience of Sweden over 1991–95, the lack of growth in U.S. savings institutions’ assets for several years even after the very large decline in the middle of the S&L crisis (Figure 2), and the wider impact of the recent crisis) and one in which the contribution declines only by a third of the historic value or 0.2 percentage point. The thought experiment takes these reductions as exogenous and driven by a needed adjustment in the balance sheet of financial institutions and possibly stricter regulations on the sector. To this exogenous shock we add indirect effects on other sectors, obtained using historic correlations between FIRE output and output in every other major sector.4 Overall, GDP growth would be reduced by a percentage point in the more pessimistic scenario and by about 0.35 percentage point in the less pessimistic scenario. Applying these markdowns to our estimates for potential output growth in the 2005–08 period of 2.4 percent (see below and Table 3) would produce potential growth rates between 1.4 percent and about 2.0 percent for the 2009–14 period.

Figure 2.
Figure 2.

Assets of Savings Institutions

Citation: IMF Staff Country Reports 2009, 229; 10.5089/9781451839746.002.A001

Source: Board of Governors of the Federal Reserve System.

10. Sectors that depend more on external financing will suffer the brunt of the adjustment. Indeed, taking the 20 manufacturing industries with readily available data on external finance dependence as an example, sectors that have higher correlations with activity in FIRE tend to have stronger dependence on external finance, which corroborates the exercise above.5 As a result, the relative growth of manufacturing sectors highly dependent on external financing would probably fall in the years ahead, an effect that is already present in the data for the post-2005 period. Table 2 ranks industries according to their dependence on external finance and groups the top-five most dependent industries and the top-five least dependent industries. Industries that are highly dependent on external financing tend to grow faster than industries that are less dependent on external financing during the whole period. However, the ratio of growth in highly dependent industries to that in less dependent industries increased significantly during the high productivity/easy financing boom period (1995–2004) and returned to its historical average after 2005 as financial conditions tightened.

Table 2.

United States: Industries’ Reliance on External Funding and Average Growth

(in percent, unless otherwise indicated)

article image
Sources: Board of Governors of the Federal Reserve System; Haver Analytics; and Fund staff estimates.

External funding need is calculated by Rajan and Zingales (1998) and is the fraction of capital expenditures not financed with cash flow from operations. Where necessary, external funding needs were averaged to reconcile SIC industrial classifications to NAICS definitions.

D. A Simple Framework to Decompose and Project Potential Output Growth

11. Aggregate data and trends show that potential growth has been slowing down since the early 2000s and more markedly after 2004. Potential GDP growth can be decomposed into changes in: (i) capital stock; (ii) equilibrium capital utilization; (iii) trend hours of work per employee; (iv) the equilibrium rate of unemployment (or the non-accelerating inflation rate of unemployment, NAIRU); (v) trend labor force participation rate; (vi) working-age population; and (vii) trend TFP.6 Figure 4 and Table 3 show actual data and trends (when relevant) for these variables, with trend values obtained by applying a Hodrick-Prescott filter to each data series. The data show clearly that, even before the crisis, labor force participation, growth in working-age population, and hours worked per employee had been declining, while the NAIRU had begun to increase. Together, these factors were forcing reductions in trend labor input growth in the last years of the sample. The filtered TFP growth varies little but captures the increase in the period 1995–2004, also observed in Oliner and Sichel (2008) for the nonfarm business sector, and the return to historic growth rates since then.

Figure 3.
Figure 3.

Potential GDP

Citation: IMF Staff Country Reports 2009, 229; 10.5089/9781451839746.002.A001

Source: Fund staff estimates.
Figure 4.
Figure 4.

United States: Decomposing Potential Output Growth

Citation: IMF Staff Country Reports 2009, 229; 10.5089/9781451839746.002.A001

Sources: Aaronson et al. (2006); Economic Outlook, OECD; Haver Analytics; World Economic Outlook, IMF; and staff calculations.
Table 3.

United States: Path for Potential Output Growth Components 1/

article image
Sources: Economic Outlook, OECD; Haver Analytics; World Economic Outlook, IMF; and Fund staff calculations.

Output-labor elasticity assumed to be 0.7 and output-capital elasticity assumed to be 0.3.

Trend capacity utilization in the U.S. industry calculated by the Federal Reserve Board obtained by using an HP filter.

Non-accelerating inflation rate of unemployment. HP filter of civilian unemployment rate, 16 years and over (seasonally adjusted).

Trend labor force participation rate calculated by applying the HP filter of the ratio between civilian labor force and working age population.

Trend changes in annual hours work per employee is calculated by applying the HP filter of annual hours worked per employee in the total economy.

Working-age population refers to U.S. population 16 years of age and over.

12. Looking ahead, staff have allocated further reductions in potential growth across slower capital accumulation, higher NAIRU, lower labor force participation rates, continuing declines in hours worked per employee, and slightly reduced TFP growth in the next couple of years. The exact path for each of these variables should be seen as illustrative and will be subject to further research by the staff in the coming year. Key factors determining future potential growth are:

  • The WEO forecast has gross private fixed investment declining by 21.1 percent in 2009 (by far, the largest decline in the post-WWII period) and by 2.7 percent in 2010, as corporations feel the full sting of the recession. Following these sharp declines, investment is set to return slowly to near pre-crisis ratios to output by the end of the forecast horizon. Using a perpetual inventory method, including by accounting for a historical rate of depreciation of 5 percent a year,7 we obtain the path for the growth in the capital stock shown in Figure 4 and Table 3. The dramatic investment dynamics are key to the overall shape of potential growth going forward and consistent with the overall view that financing conditions will remain tight for a while, and that aggregate demand growth will remain sluggish.

  • Staff calculated a measure of the NAIRU by fitting an HP filter to actual unemployment rates in the United States. This procedure has the advantage of being simple but is devoid of economic structure and can misrepresent changes in structural labor market functioning. Indeed, when compared to results from Gianella et al. (2008)8 our NAIRU estimates have wider swings and may be overestimating the degree of NAIRU dependence on the actual unemployment rate path, a matter for further research. Going forward, staff (i) forecast a path for the NAIRU that mirrors historic relationships between the filtered series and the actual unemployment rate, and (ii) complete OECD estimates originally ending in 2010 with milder NAIRU dynamics for the 2011–14 period (see top left chart in Figure 4). Despite some differences between the two NAIRU series, the shape of potential growth estimates using the two series (and the same methods for all remaining variables) are similar (shown in the middle chart of Figure 5), although the OECD-consistent measure has slightly higher growth rates in 2009–10 and slightly lower rates in 2011–14.

  • After a brief pause in the next two years as older individuals remain in or return to the working force to rebuild lost savings, the labor force participation rate is assumed to contract slowly. This path is more benign than the trend based on labor force characteristics and demographic changes produced by Aaronson et al (2006). The gap between the two paths for labor force participation suggests that in the absence of crisis-related changes in labor supply attitude, demographic trends will put even larger pressure on potential output growth going forward than we currently assume.9

  • Growth in working-age population (individuals at least 16 years old) is set to decline slowly in the next few years, according to projections from the U.S. Census Bureau.

  • Average hours of work is assumed to continue its long-term downward trend in the coming years.

  • Equilibrium capacity utilization in U.S. industry (a proxy for overall U.S. capital utilization) seems to have declined for more than a decade beginning at the end-1960s. After somewhat of a rebound in the 1990s and consistent declines since then, we assume that it will remain constant throughout the forecast period.

  • Trend aggregate TFP growth is relatively smooth, and going forward we assume only a slight contraction of TFP growth in the coming years to account for disturbances from ongoing structural changes in the U.S. economy. Trend TFP growth is assumed to go back to pre-crisis rates by the end of the forecast horizon.10

Figure 5.
Figure 5.

United States: Potential Output Growth, and Output and Unemployment Gaps

Citation: IMF Staff Country Reports 2009, 229; 10.5089/9781451839746.002.A001

Sources: Economic Outlook, OECD; Haver Analytics; World Economic Outlook, IMF; and staff calculations.1/ Unemployment gap is the difference between unemployment rate and the natural rate of unemployment.2/ Output gap is the difference between actual output and potential output as percent of potential output.

13. Estimated potential output growth lies between 1 and 2 percent for the next five years (Figure 5, top chart, and Table 3). After being hit severely by the capital growth dynamics, potential growth converges slowly towards 2 percent in 2014—a level somewhat lower than the one estimated for 2005–08, but consistent with estimates of potential growth in the absence of the crisis. 11 The average potential growth rate for 2009–2014 turns out to be 1½ percent, about ½ percentage point below our estimates for potential growth in the absence of the crisis. The resulting paths for the output gap and the unemployment gap (defined as the difference between actual unemployment rates and the NAIRU) are shown in the bottom chart of Figure 5. The output gap reaches its widest point in 2010 at about similar levels as in the early-1980s recession.

14. Ultimate losses in potential output are in the ball park of previous research. By 2014, potential output is expected to be about 5¾ percent below the counterfactual level produced by assuming potential output grows from 2009 to 2014 at the same average rate observed in 2005–08 (Figure 3). This gap is a bit larger than the ones observed after previous financial crises as documented in OECD (2009) but is closer to the lower end of the interval estimated by Cerra and Saxena (2008) of 4 percent to 16 percent permanent output loss following financial crisis in developed and less developed economies.

E. Conclusion and Policy Implications

15. The financial crisis will likely weigh on potential growth going forward, which will impose a constraint to economic policy. In particular, public debt-to-GDP ratios will trend up faster than otherwise in the following years, although the exact path will depend on the behavior of interest rates in this lower-growth (but high-debt-accumulation) environment. Going beyond the medium-term period analyzed here, there is even larger uncertainty about key determinants of potential output, but demographic forces will likely limit economic growth in outer years, raising the stakes for fiscal consolidation.

References

  • Aaronson, S., B. Fallick, A. Figura, J. Pingle, and W. Wascher, 2006 “The Recent Decline in the Labor Force Participation Rate and Its Implications for Potential Labor Supply”, Brookings Papers on Economic Activity, Vol. 1, pp. 69134.

    • Search Google Scholar
    • Export Citation
  • Cerra, V. and S. C. Saxena, 2008, “Growth Dynamics: The Myth of Economic Recovery”, American Economic Review, Vol. 98 (1), pp. 439457.

    • Search Google Scholar
    • Export Citation
  • Furceri, D. and A. Mourougane, 2009, “The Effect of Financial Crises on Potential Output: New Empirical Evidence from OECD Countries”, OECD Economics Department Working Papers, No. 699, OECD, Paris, France.

    • Search Google Scholar
    • Export Citation
  • Gianella, C., I. Koske, E. Rusticelli, and O. Chatal, 2008, “What Drives the NAIRU? Evidence from a Panel of OECD Countries”, OECD Economics Department Working Papers, No. 649, OECD, Paris, France.

    • Search Google Scholar
    • Export Citation
  • Haugh, D., P. Ollivaud, and D. Turner, 2009, “The Macroeconomic Consequences of Banking Crises in OECD Countries”, OECD Economics Department Working Papers, No. 683, OECD, Paris, France.

    • Search Google Scholar
    • Export Citation
  • Oliner, S. D., D. E. Sichel, and K. J. Stiroh, 2007, “Explaining a Productive Decade”, Brookings Papers on Economic Activity, Vol. 1, pp. 81135.

    • Search Google Scholar
    • Export Citation
  • Oliner, S. D. and D. E. Sichel, 2008, “Explaining a Productive Decade: An Update”, Presentation at the Federal Reserve Bank of San Francisco, San Francisco, CA.

    • Search Google Scholar
    • Export Citation
  • Rajan, R. G. and L. Zingales, 1998, “Financial Dependence and Growth”, American Economic Review, Vol. 88 (3), pp. 559586.

1

Potential growth is defined here as the level of output that can be produced without undue strains on productive resources, i.e. without inflationary impact.

2

Oliner and Sichel (2008) breaks down productivity growth as:

(1)dylαIdkIl+αOdkOl+(1αIαO)dq+dtfpI+dtfpO

where dx refers to the difference of the logarithm of x, yl is labor productivity, kl is the capital-labor ratio, dkl represents capital deepening, and α is the elasticity of output to capital. The superscripts I and O represent information technology and other types of equipment, respectively. Thus, the first two terms in the right-hand side represent capital deepening in these two types of equipment; tfpI and tfpO refer to TFP growth in sectors producing information technology equipment (and embedded semiconductors) and in other sectors, respectively, already scaled for their share in total production value and corrected for adjustment costs and factor utilization; and q refers to labor quality.

3

Previous research corroborates the view that potential output growth would decline for several years after a financial crisis, although its longer-term path depends on other key variables, like institutional reforms and political stability. For examples, see Cerra and Saxena (2008), and Furceri and Mourougane (2009).

4

The resulting rule of thumb is that the indirect effect corresponds to two-thirds of the direct shock.

5

The correlation between (i) the output correlation between manufacturing sectors and finance and insurance, and (ii) the ranking in external finance dependence is 0.35. Data on dependence in external finance for each manufacturing sector are from Rajan and Zingales (1998).

6

Equations (2) and (3) were used to calculate U.S. potential growth:

(2)dtfp=dyαdksαdku(1α)dl

(3)dy*=αdks+αdku*+(1α)dh*+(1α)d(1u*)+(1α)dlfp*+(lα)dwap+dtfp*

where dx refers to difference of the logarithm of x, y is output, ks is the capital stock, ku is capital utilization, l is total labor hours, h is average hours of work, u is the unemployment rate, lfp is the labor force participation rate, wap is working-age population, and tfp is total factor productivity. Variables with a * are trend values. All variables refer to the whole U.S. economy and were constructed using information from Haver Analytics and the Organization for Economic Co-Operation and Development (OECD). TFP growth is obtained by applying formula (2) using the level of capacity utilization in the industry calculated by the Federal Reserve Board as a proxy for economy-wide capital utilization and the average share of labor compensation in value added in the past 30 years (about 0.7 in the United States) as a proxy for (l-α). Once TFP growth is obtained, HP filters were applied to capture trend values. Other trend values were obtained by using an HP filter and formula (3) is used to calculate potential GDP growth (Figure 4). The HP filter used for all series assumes a smoothness parameter, λ, of 100—the traditional value for annual-frequency data.

7

This is consistent with an average duration of capital stock in the United States of about 20 years, as reported in the official statistics.

8

The authors estimate time-varying NAIRUs for a panel of OECD economies using Phillips curve equations and Kalman filter techniques.

9

The authors estimated a detailed model combining developments in labor force attachment for particular age-sex groups and demographic changes. Their results show a faster contraction in labor force participation than forecasts done at the time by the Congressional Budget Office, the Bureau of Labor Statistics, and the Social Security Administration Office, although these agencies also anticipated a noticeable downward trend in participation rates.

10

Projecting trend TFP becomes more uncertain during extreme episodes like the current turmoil. In one hand, the sharp contraction of the housing and financial sector induces costly factor reallocations that, at least in the short term, will hurt total factor productivity. On the other hand, exceptional circumstances triggered by the financial meltdown favor structural reforms that may enhance TFP in the medium to long run. As pointed out by Haugh et al. (2009), financial crises usually have negligible effects on trend TFP for industrial countries over horizons of 5 to 10 years. Nonetheless, the authors stress that trend TFP was adversely affected in Japan during its banking crisis in the 1990s (probably due to the protracted resolution of the difficulties faced by the financial system and the associated inefficiencies resulting from the misallocation of factors).

11

Assuming that in the absence of the crisis labor force participation rates would progress as in Aaronson et al (2006) but other factors would behave as shown in Table 3 for the 2005-2008 period, we obtain an average potential growth rate of 2 percent for the 2009–2014 period.

United States: Selected Issues
Author: International Monetary Fund