Improving citizens’ well-being requires increasing productivity over time – the efficiency of converting resources such as labor, land, and physical plant and equipment into useful goods and services.
U.S. productivity has slowed dramatically during the last decade, largely due to slower innovation and reduced growth of capital per worker.
The productivity slowdown will make funding government programs more challenging. Public policies that encourage additional capital accumulation and reward innovation could reverse at least some of the recent productivity declines.
Changes in American Economic Productivity
Until the laws of thermodynamics are repealed, I shall continue to relate outputs to inputs - i.e. to believe in production functions.
- Paul Samuelson, Collected Scientific Papers, 1972, p.174
Economic growth results from the growth of productive inputs — mainly labor and capital — and from improvements in production technology. These factors depend on decisions made by individuals to invest in education, to save money, and to work, combined with the decisions of firms to invest and innovate. Policies that impact any of these choices, therefore, impact economic growth.
Economists generally measure the three main factors' contributions to economic growth — capital, labor and technology — using an “aggregate production function.”1 This function expresses the relationship between inputs and outputs for the economy as a whole, thereby allowing us to see the contribution of each factor. During the past decade, real growth has decreased substantially, from 3.6 percent per year to just 1.2 percent per year. Most of the declines have been in a reduction in innovation (reducing growth by 2.08 percent) and capital (reducing growth by 0.41 percent). While recent growth trends might be temporary, considerable headwinds remain in returning growth rates back to their historical levels. Going forward, the retirement of baby boomers will likely reduce growth even more.
Standard measures of a nation’s output typically focus on Gross Domestic Product — the market value of all domestically produced goods and services. However, that measure misses the output of the government and nonprofit sectors. These sectors are, however, included in a newer measure called “total economy output,” which provides a more complete accounting of the economy’s productive capacity.2
Figure 1 plots the growth rate of the total economy output since 1988, the earliest year of a published data series that is consistent and compatible with the total economy output measure.3 A short period of negative growth marks the relatively mild economic recession during the early 1990s; a dip in the growth rate marks the even milder economic recession during the early 2000s; and a large period of negative growth corresponds to the deeper Great Recession of 2008-2009.
Figure 1: Total Economy Output Growth
Source: Bureau of Labor Statistics.
Figure 2 shows the contribution of each of the main factors that have contributed to economic growth over time: Hours worked, labor efficiency, capital, and multifactor productivity (MFP). Hours worked measures the total hours provided in the economy, regardless of who provides the work. Labor efficiency adjusts for the fact that some people are more productive than others. Capital is the amount of buildings, machines, computers and other tools used by labor. Multifactor productivity generally corresponds to the technology that is embedded in capital as well as process improvements for labor — both effects allow more output to be produced within a given amount of capital and labor. Each of these four factors play a key role in generating economic growth.
Figure 2: Contributions to Growth in Real Output (Percentage Points)
Source: PWBM calculations based on Bureau of Labor Statistics.
Contributions from Capital
From 1987 to 2013, the total economy output grew at an average annual rate of 2.7 percent. Growth of capital services was the single largest contributor to growth during this period, accounting for more than 40 percent of the rise in output. As shown in Figure 3, in the late 1990s, output was buoyed by a surge in investment related to the spread of information technology. Rapid technological progress in the production of information capital goods such as computers and communication equipment had produced steady declines in the prices of these goods since at least the 1960s. In 1995, this process accelerated sharply, triggering a burst of investment in information capital.4 Following the dot-com crash in 2000, investment in information capital declined, and capital’s contribution to output growth returned to levels observed before 1995. During the past decade, however, capital intensity has decreased even more, causing annual growth to fall by 0.41 percent.
Figure 3: Growth of Information Capital Services (Percent)
Information capital includes computers, software, communications equipment, and certain other capital goods. Only information capital owned by the private business sector is included.
Source: Bureau of Labor Statistics.
Why has capital intensity fallen? Part of the recent decline is likely due to the recession. Firms make less investment in capital when sales prospects are reduced. Capital intensity has also fallen in other OECD countries.5 In the United States, growth in capital services fell rapidly, initially as a result of falling residential investment following the collapse of the housing market in 2006 but was soon followed by declines in business investment as the recession intensified in 2008. Government investment, which is normally mostly unaffected by the business cycle, also saw sharp declines due to concerns over budget deficits at the federal, state, and local levels. Although business investment has experienced a partial recovery since 2010, growth in housing and government capital remains very weak.
Contributions from Labor
Contributions from labor supply come from the total hours worked and the effectiveness of labor. Between 1987 and 2013, hours worked increased steadily due to population growth, increasing labor force participation among women, and the decline in unemployment during the 1990s. At the same time, gains in educational attainment led to a more productive workforce as the share of workers with a college degree continued its decades-long climb. Altogether, growth in (efficiency adjusted) labor services contributed about 1.4 percentage points on average to annual output growth during these years.
As Figure 4 shows, however, growth in hours worked effectively ceased after 2000, largely due to demographic reasons. Growth in the prime working age population (persons between the ages of 25 and 54) slowed substantially during the early 2000s, as the baby boomers began to age out of their prime working years. In addition, women’s labor force participation rate peaked in 1999 at 60 percent (compared with 75 percent among men) and then began to fall. Although labor efficiency continued to rise, the weak recovery in employment following the recession of 2001 meant that labor contributed little to growth in output during the early 2000s. Compounding the effects of these demographic pressures, hours worked declined precipitously from 2007 to 2009, as unemployment rose during the Great Recession. While hours have recovered somewhat in the years since, they remain more than 2 percent below the pre-recession peak.
Figure 4: Labor Supply (1987 = 100)
Source: Bureau of Labor Statistics.
Going forward, labor is likely to contribute substantially less to economic growth relative to the past. Figure 5 shows projections of total population growth and the share of the population that is 65 or older. The rate of population growth is expected to continue slowing over the next several decades. At the same time, the elderly share is projected to increase until at least 2030, rising from less than 15 percent to more than 20 percent. Compounding this effect, PWBM projections suggest that the efficiency of labor will also decline.
The aging of the population also means rising beneficiary roles for Social Security and Medicare. Thus, growth in the federal government’s fiscal commitment is accelerating at the same time the resources available to meet those commitments are growing more slowly. Immigration and longer careers by workers to better prepare for retirement may somewhat ease the population-aging related fiscal challenge, but these developments appear unlikely to fully eliminate it.
Figure 5: Projected Population Growth and Percent Aged 65 or Older (Percent)
Source: U.S. Census Bureau.
During 1987-2013, MFP accounted for about a quarter of the observed growth in total economy output. But there have been a lot of changes along the way. Rapid productivity gains in information-capital-producing industries boosted growth in MFP, which rose from 0.3 percent per year during 1987-1995 to 0.8 percent per year during 1995-2000.6 From 1987 to 2000, however, growth in output was driven primarily by the expansion of the labor force and growth in capital services, rather than productivity.
However, from 2000 to 2005, MFP grew at an average annual rate of 2.3 percent, accounting for nearly two thirds of all output growth during this period. The reasons for the jump in MFP growth are not well understood. Unlike the rise in MFP growth in the late 1990s, it occurred mainly in capital-producing industries unrelated to information technology (IT). Indeed, productivity growth in the IT capital sector declined slightly. It may reflect the role of the earlier spread of information capital as an enabler of innovation across the economy, what is referred to as a “general purpose technology.” Alternatively, it may be the result of unmeasured investment in intangible capital such as research and development or brand equity. There is also evidence that productivity gains were greatest in industries experiencing increased competitive pressures as a result of globalization.7
After 2005, productivity growth collapsed and was soon followed by the deep recession and slow recovery that began after 2007. Between 2005 and 2013, output growth averaged only at 1.2 percent per year. Growth in MFP was negative from 2007 to 2009. This decline in measured productivity growth was primarily a consequence of the recession, as outputs fell much faster than inputs.8 As output has recovered, productivity growth has returned to more normal rates in recent years but remains well below the highs observed in the early 2000s.
Forecasting future changes in long-term productivity is challenging because it depends on the rate of discovery and technological growth. Past discoveries, like the invention of the transistor that lead to the computer revolution, were generally unpredictable. However, productivity growth also depends on the number of researchers involved in research and development.9 As population growth slows, fewer workers will be available to work in knowledge-producing sectors, potentially hindering the economy’s long-run growth potential.10
Economic growth is driven by increases in capital, labor and multifactor productivity. All three factors have shrunk during the past decade, producing lower growth. While the Great Recession is clearly playing a significant role in recent years, these factors will likely not rebound to their historical highs. All of these changes are occurring at the same time that entitlement programs, including Social Security and Medicare, face major shortfalls. These programs reduced the amount of capital available to firms by making pay-as-you-go benefits substitutable for household savings. Reforming these programs, therefore, will likely have a positive impact on both sides of the government’s ledger. Increasing capital will spur more innovation and growth, leading to more revenues. At the same time, reforming these programs will reduce the strains that these programs place on future taxpayers.
See, for example, the classic article: Robert M. Solow, "Technical Change and the Aggregate Production Function," The Review of Economics and Statistics 39, no. 3 (August 1957): 312-320, available at: http://www.jstor.org/stable/1926047. ↩
Output produced by the government and by nonprofit institutions serving households is not sold in markets and therefore cannot be directly measured. As a result, output in these sectors is only partially reflected in the standard definition of GDP. To fill in these gaps, the Bureau of Economic Analysis and Bureau of Labor Statistics have prepared a prototype production account for the total U.S. economy which includes estimates of the unmeasured output of the government and nonprofit sectors. Specifically, total economy output is equal to GDP plus the imputed returns to capital owned by government and by nonprofit institutions serving households. See Michael J. Harper, Brent R. Moulton, Steven Rosenthal, and David B. Wasshausen, "Integrated GDP-Productivity Accounts," American Economic Review 99, no. 2 (May 2009): 74-79, available at: https://www.aeaweb.org/articles.php?doi=10.1257/aer.99.2.74. For a general discussion of the approach to production account measurement used to construct these accounts, see Bureau of Labor Statistics, “Technical Information About the BLS Multifactor Productivity Measures,” (September 2007). Available at: http://www.bls.gov/mfp/mprtech.pdf. ↩
See Dale W. Jorgenson, "Information Technology and the U.S. Economy," American Economic Review 91, no. 1 (March 2001): 1-32, available at: https://www.aeaweb.org/articles.php?doi=10.1257/aer.91.1.1. ↩
Jason Furman, “Productivity Growth in the Advanced Economies: The Past, the Present, and Lessons for the Future.” Council of Economic Advisers, 2015. ↩
See Dale W. Jorgenson, Mun Ho, and Kevin Stiroh, “A Retrospective Look at the U.S. Productivity Growth Resurgence,” Journal of Economic Perspectives 22, no. 1 (2008): 3-24, http://www.jstor.org/stable/27648221. ↩
See Stephen D. Oliner, Daniel E. Sichel, and Kevin J. Stiroh “Explaining a Productive Decade,” Brookings Papers on Economic Activity 2007:1 (Spring 2007): 81-137, available at: http://www.brookings.edu/~/media/projects/bpea/spring-2007/2007a_bpea_oliner.pdf ↩
To some extent, the decline in productivity reflects mismeasurement of the capital services actually used in production: while capital goods are rarely destroyed in a recession, they may be used less intensely, and this margin is not accounted for in this framework. For an example of a similar framework that does account for variable factor utilization, see Susanto Basu, John Fernald, and Miles Kimball, “Are Technology Improvements Contractionary?” American Economic Review 96, no. 5 (December 2006): 1418–1448. ↩
See Charles I. Jones, “R&D-Based Models of Economic Growth,” Journal of Political Economy 103, no. 4 (August 1995): 759–784, available at: https://ideas.repec.org/a/ucp/jpolec/v103y1995i4p759-84.html. ↩
The effect of slow population growth on the number of researchers can be offset by raising the share of the workforce involved in knowledge production. However, this channel cannot operate indefinitely since this share is bounded from above at one. Alternatively, if knowledge can be shared across borders then reduced innovation in the US might be offset by increased research activity in other countries. But this is true only of countries that are sufficiently close to the world’s technological frontier that they can contribute to the discovery of new ideas, and most of these countries are also experiencing reductions in population growth. ↩
Year,Real output growth 1988,4.7 1989,6.0 1990,0.6 1991,-2.0 1992,2.4 1993,2.8 1994,4.3 1995,3.7 1996,3.3 1997,4.0 1998,4.3 1999,4.2 2000,3.6 2001,1.8 2002,2.1 2003,4.1 2004,4.4 2005,5.6 2006,2.7 2007,1.1 2008,-1.2 2009,-4.1 2010,2.3 2011,6.4 2012,0.2 2013,2.7
,MFP,Capital services,Labor efficiency,Hours worked 1987-2013,0.72,1.13,0.29,0.51 1987-1995,0.28,1.12,0.4,0.98 1995-2000,0.79,1.63,0.21,1.23 2000-2005,2.26,1.16,0.22,-0.05 2005-2013,0.18,0.75,0.27,0.02
Year,Growth of information capital services 1988,10.15490534 1989,9.268092105 1990,7.465944156 1991,5.273478535 1992,5.741085684 1993,8.021390374 1994,9.772859639 1995,12.36205433 1996,15.42638587 1997,17.69277971 1998,19.24226625 1999,20.55325599 2000,19.30991126 2001,14.5714691 2002,9.196398564 2003,7.145402708 2004,6.965211738 2005,6.996466431 2006,7.77675033 2007,8.447424222 2008,7.575808949 2009,5.061881448 2010,3.953 2011,3.685319327 2012,2.950345135 2013,2.839659709
,Hours worked,"Prime working-age population, 25-54" 1987,100,100 1988,102.9864551,102.1817119 1989,105.8363479,104.3644315 1990,106.0169783,106.5935063 1991,104.5322253,108.3429067 1992,104.6198771,110.1799786 1993,107.1399862,111.9102324 1994,110.4707529,113.4873128 1995,113.1874731,114.9061813 1996,114.6073347,116.3976057 1997,117.994276,117.7338412 1998,120.5695911,118.3566116 1999,122.944421,119.1103856 2000,124.5991506,121.5873592 2001,123.0480535,122.5426769 2002,121.4959879,123.0193281 2003,120.9090601,124.2406836 2004,122.2654831,124.3626177 2005,124.0814725,125.1335228 2006,126.3299096,125.8479956 2007,127.2485581,126.6672713 2008,125.9303919,126.6219239 2009,119.0218839,126.5342524 2010,119.000092,126.2571296 2011,120.8417474,125.6666062 2012,122.9323144,125.2735957 2013,124.4785689,125.3663059
,Population growth rate (left),Share of population 65 and over (right) 2001,0.994818579,12.38380315 2002,0.932031203,12.35010865 2003,0.863276836,12.36229266 2004,0.929653784,12.36420143 2005,0.926213692,12.40199379 2006,0.968810593,12.4552584 2007,0.955492996,12.55714053 2008,0.950433388,12.75197801 2009,0.880648747,12.91610708 2010,0.840363527,13.071925 2011,0.766769032,13.27047818 2012,0.766708798,13.74191371 2013,0.759601671,14.13057902 2014,0.745344362,14.50273947 2015,0.787813973,14.8832028 2016,0.815071356,15.28569934 2017,0.81328018,15.6806048 2018,0.807854136,16.07330771 2019,0.798909982,16.46908205 2020,0.786559984,16.87309232 2021,0.776665082,17.29509124 2022,0.769293549,17.73127379 2023,0.758774072,18.16677984 2024,0.745201965,18.58724998 2025,0.728667296,18.97879569 2026,0.714139376,19.34993518 2027,0.702002144,19.71059749 2028,0.68759076,20.04960273 2029,0.670966057,20.35609817 2030,0.652184438,20.61952911 2031,0.633403264,20.83628181 2032,0.616571962,21.01476549 2033,0.600529025,21.16249929 2034,0.585251026,21.28679636 2035,0.570715961,21.3947799 2036,0.554902819,21.48161876 2037,0.539009034,21.54329342 2038,0.525593669,21.58786094 2039,0.514600505,21.6231933 2040,0.505977105,21.65699242 2041,0.495514427,21.68276285 2042,0.483733393,21.69564377 2043,0.475018453,21.70551744 2044,0.469309539,21.72207583 2045,0.466549696,21.7548293 2046,0.462211539,21.80278184 2047,0.456021514,21.85938709 2048,0.452294307,21.92567926 2049,0.450985356,22.00266904 2050,0.45205138,22.09134181 2051,0.452495431,22.19154467 2052,0.451360838,22.30237154 2053,0.451122064,22.42365661 2054,0.451763091,22.55523179 2055,0.45326809,22.69692596 2056,0.455330722,22.85083907 2057,0.456773615,23.01680365 2058,0.457314789,23.19129558 2059,0.456970402,23.37085899 2060,0.45575646,23.55210595