Productivity Growth in the 2000s

J. Bradford DeLong1

University of California at Berkeley and NBER

March 2002

First Draft

Abstract

The

1 I would like to thank the University of California at Berkeley for financial support, and Chad Jones, Dan Sichel, and Larry Summers for helpful discussions.

2

Figure 1: Types of Real Non-Residential Fixed Investment Divided by Real GDP

0%

2%

4%

6%

8%

10%

12%

1987 1992 1997 2002Date

Equipment

Non-ResidentialStructures

3

Figure 2: Log Labor Productivity in Nonfarm Business

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0

0.2

1947 1957 1967 1977 1987 1997

Date

Log NonfarmBusinessOutput perHour1973-1995TrendExtended

4

Figure 3: Log Real GDP per Member of the Labor Force

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

1960 1970 1980 1990 2000Date

Period Productivity Growth (Real Output per Hour)

1947:1-1973:2

0.02875224

1973:2-1995:1

0.0130173

1995:1-2001:2

0.0280261

1995:1-2001:4

0.02690108

(1) Even before the mid-1990s, ICT had a much bigger impact on growth than steam and at least a similar impact to that of electricity in a similar early phase.

5

(2) The Solow productivity paradox stems largely from unrealistic expectations. In the early phases of general purpose technologies their impact on growth is modest because the new varieties of capital have only a small weight relative to the economy as a whole. (3) If there has been an ICT productivity paradox, it comprises an apparent absence of TFP spillovers; in this respect, the contribution made by electricity in the 1920s through its impact on the reorganization of factory work has probably not yet been matched by ICT.

examining the impact of the two previous technological breakthroughs with similar claims to be regarded as general purpose technologies (Bresnahan and Trajtenberg, 1995), namely, steam and electricity, in a growth accounting framework. The results of this exercise are striking: they suggest that the growth contribution of ICT in the past 25 years has exceeeded that of steam and at least matched that of electricity over comparable periods and that the true paradox is why more should have been expected of ICT.

A straightforward generalization of this is used by Oliner and Sichel (2000) which features different varieties of capital (including computer hardware, software and communication equipment as types of ICT capital) whose growth contributions are weighted by their shares in income, and in which TFP growth is decomposed into TFP growth in making ICT capital and in other activities weighted by output shares. The contribution of innovations in ICT is captured through two components: extra TFP growth and through the three additional capital inputs. This is similar to the endogenous innovation based growth accounting for the expanding varieties growth models of Romer (1990) and Grossman and Helpman (1991), as set out in Barro (1999).

A different strand of endogenous growth economics is adopted by Schreyer (2000) in which ICT capital goods are 'special' in that they provide knowledge spillovers or other positive externalities to the economy similar to the formulation in Romer (1986).

Basu, Fernald, and Shapiro (2001) estimate that because of adjustment costs productivity growth in the second half of the 1990s undershot the long-run technology trend by half a percentage point per year because of adjustment costs.

Of the actual 2.86 percent annual growth of output per hour, 0.40 is attributedto a cyclical effect and the remaining 2.46 percent to trend growth, and the latter is 1.04 points faster than the 1972-95 trend. How can this acceleration be explained? A small part on lines 6 and 7 is attributed to changes in price measurement methods and to a slight acceleration in the growth of labor quality. All of the remaining 0.89 points can be directly attributed to computers. The capital-deepening effect of faster growth in computer capital relative to labor in the aggregate economy accounts of 0.60 percentage points of the acceleration (line 9a) and a 0.30-point acceleration of MFP growth in computer and computer-related semiconductor manufacturing account (line 10) sum to an

6

explanation of 0.90 points, compared to the 0.89 acceleration in trend that needs to be explained.

There is no sign of a fundamental transformation of the U. S. economy. Therehas been no acceleration of MFP growth outside of computer production and the rest ofdurable manufacturing. Responding to the accelerated rate of price decline of computers thatoccurred between 1995 and 1998, business firms throughout the economy boosted purchasesof computers, creating an investment boom and "capital deepening" in the form of fastergrowth of capital relative to labor. But computer capital did not have any kind of magical orextraordinary effect ó it earned the same rate of return as any other type of capital.

If, as Gordon says, computer capital “earned the same rate of return” then why did growth accelerate? The answer is that the rate of return that remains the same is the rate of return on a marginal addition to the real capital stock

The Bubble-Boom of the 1990s?If one wishes to minimize the possibility and the magnitude of technology-driven secular productivity acceleration in the second half of the 1990s, there are two analytical strategies to pursue. The first is to stress the cyclical component of 1990s economic growth. The second is to argue that the internet bubble led standard statistical methods to overstate the productivity of the American economy in the second half of the 1990s.

7

__________ and Franco Malerba (1999). "Industrial Dynamics and the Evolution of Firms'and Nations' Competitive Capabilities in the World Computer Industry," in Moweryand Nelson, eds. (1999b), pp. 79-132.Brynjolfsson, Erik (1996). "The Contribution of Information Technology to ConsumerWelfare," Information Systems Research, 7:3, September, pp. 281-300.Eller, Jonathan (2000). "U. S. Inflation and Unemployment in the late 1990s: A Re-examination of the Time-Varying NAIRU," Northwestern University senior honorsthesis, May.Goldin, Claudia (1998). "America's Graduation from High School: The Evolution andSpread of Secondary Schooling in the Twentieth Century." Journal of Economic History, vol.58 (June), pp. 345-74.Gordon, Robert J. (1977). "Can the Inflation of the 1970s Be Explained?" BPEA 1:1977,253-77._________ (1982). "Inflation, Flexible Exchange Rates, and the Natural Rate ofUnemployment." In Workers, Jobs, and Inflation, edited by Martin N. Baily. Washington:Brookings, 88-152._________ (1990). The Measurement of Durable Goods Prices. University of Chicago Press forNBER._________ (1997). "The Time-Varying NAIRU and its Implications for Economic Policy."Journal of Economic Perspectives 11(1): 11-32._________ (1998). "Foundations of the Goldilocks Economy: Supply Shocks and theTime-Varying NAIRU," Brookings Papers on Economic Activity, vol. 29 (no. 2), pp. 297-333.__________ (2000a). "Interpreting the `One Big Wave' in U. S. Long-term ProductivityGrowth," in Bart van Ark, Simon Kuipers, and Gerard Kuper, eds., Productivity,Technology, and Economic Growth, Kluwer Publishers, 2000, pp. 19-65.__________ (2000b). "Does the New Economy Measure Up to the Great Inventions of the

8

Technology and Economic Performance, Page 45Past?" Journal of Economic Perspectives, vol. 14 (Fall), pp. 49-74.Ip, Greg (2000). "Market on a High Wire," Wall Street Journal, January 18, p. C1.Jorgenson, Dale W. and Kevin J. Stiroh (2000). "Raising the Speed Limit: U. S. EconomicGrowth in the Information Age," Brookings Papers on Economic Activity, 31:1, pp. 125-211.Katz, Lawrence F., and Alan B. Krueger (1999). "The High-Pressure U. S. Labor Market ofthe 1990s," Brookings Papers on Economic Activity, 30:1, pp. 1-65.Lum, Sherlene K. S., and Brian C. Moyer (2000). "Gross Domestic Product by Industry for1997-99," Survey of Current Business, vol. 80, no. 12, December, pp. 24-35.Mishel, Lawrence, Jared Bernstein, and John Schmitt (1999). The State of Working America 1998-99. Ithaca NY: Cornell University Press.Mowery, David C. (1999). "The Computer Software Industry," in Mowery and Nelson, eds.(1999b), pp. 133-68.__________ and Richard R. Nelson (1999a). "Explaining Industrial Leadership," in Moweryand Nelson, eds. (1999b), pp. 359-82.__________ and Richard R. Nelson, eds. (1999b). Sources of Industrial Leadership: Studies of SevenIndustries. Cambridge UK: Cambridge University Press. Oliner, Stephen D., and Daniel E. Sichel (2000). "The Resurgence of Growth in the Late1990s: Is Information Technology the Story?" Journal of Economic Perspectives, vol. 14, Fall,pp. 3-22.__________ (2001). "The Resurgence of Growth in the Late 1990s: Is InformationTechnology the Story?" Updated presentation presented at the meetings of theAmerican Economic Association, January 7.Parker, Robert, and Bruce Grimm (2000). "Software Prices and Real Output: RecentDevelopments at the Bureau of Economic Analysis," paper presented at NBERproductivity workshop, March 17.Rosenberg, Nathan (1986). "The Impact of Technological Innovation: A Historical View,"in Ralph Landau and Nathan Rosenberg, eds., The Positive Sum Strategy: Harnessing Technology

9

for Economic Growth. Washington DC: National Academy Press, pp. 17-32.Sichel, Daniel E. (1997). The Computer Revolution: An Economic Perspective. Washington: Brookings.Triplett, Jack E. (1999). "The Solow Computer Paradox: What do Computers do toProductivity?" Candian Journal of Economics, 32:2, April, pp. 309-34.Uchitelle, Louis (2000). "Economic View: Productivity Finally Shows the Impact ofComputers," New York Times, March 12, section 3, p. 4.Wright, Gavin (1990). "The Origins of American Industrial Success: 1879-1940," AmericanEconomic Review, vol. 80 (September), pp. 651-68.

10

The average workweek did not rise above its mid-1994 level, ever

11

THE SOLOW PRODUCTIVITY PARADOX IN HISTORICAL PERSPECTIVE By

Nicholas Crafts (London School of Economics)

November 2001. Abstract

A growth accounting methodology is used to compare the contributions to growth in terms of capital-deepening and total factor productivity growth of three general-purpose technologies, namely, steam in Britain during 1780-1860, electricity and information and communications technology in the United States during 1899-1929 and 1974-2000, respectively. The format permits explicit comparison of earlier episodes with the results for ICT obtained by Oliner and Sichel. The results suggest that the contribution of ICT was already relatively large before 1995 and it is suggested that the true productivity paradox is why economists expected more sooner from ICT.

This paper attempts to fill this gap by examining the impact of the two previous technological breakthroughs with similar claims to be regarded as general purpose technologies (Bresnahan and Trajtenberg, 1995), namely, steam and electricity, in a growth accounting framework. The results of this exercise are striking: they suggest that the growth contribution of ICT in the past 25 years has exceeeded that of steam and at least matched that of electricity over comparable periods and that the true paradox is why more should have been expected of ICT.

A straightforward generalization of this is used by Oliner and Sichel (2000) which features different varieties of capital (including computer hardware, software and communication equipment as types of ICT capital) whose growth contributions are weighted by their shares in income, and in which TFP growth is decomposed into TFP growth in making ICT capital and in other activities weighted by output shares. The contribution of innovations in ICT is captured through two components: extra TFP growth and through the three additional capital inputs. This is similar to the endogenous innovation based growth accounting for the expanding varieties growth models of Romer (1990) and Grossman and Helpman (1991), as set out in Barro (1999). A different strand of endogenous growth economics is adopted by Schreyer (2000) in which ICT capital goods are 'special' in that they provide knowledge spillovers or other positive externalities to the economy similar to the formulation in Romer (1986). To capture this idea in the one sector model, equation (1) is rewritten as Y = AK£\

+ £] L1 - £\ (3)

where £] > 0 represents the impact of the knowledge spillover on output. If the contribution of capital is still weighted by sK = £\, then the standard Solow residual becomes

12

£GY/Y - £GK/K - £GL/L = £GA/A + £]£GK/K (4) so it comprises both exogenous technological change and the growth effect from spillovers.

sK

sL

Nevertheless, the sectoral pattern of labor productivity growth, which was heavily skewed towards contributions from ICT-intensive industries (Stiroh, 2001), and the evidence of micro studies that find important lagged productivity gains from reorganizations of work facilitated by ICT (Brynjolffson and Hitt, 2000) suggest that a significant part of the economy wide TFP acceleration in these years may have been due to TFP spillover effects from ICT investment.

It should be noted that parts of the service sector are among the main users of ICT and that the response of output to ICT may be masked by measurement problems (McGuckin and Stiroh, 2000).

The main results of this benchmarking exercise are: (1) Even before the mid-1990s, ICT had a much bigger impact on growth than steam and at least a similar impact to that of electricity in a similar early phase. (2) The Solow productivity paradox stems largely from unrealistic expectations. In the early phases of general purpose technologies their impact on growth is modest because the new varieties of capital have only a small weight relative to the economy as a whole. (3) If there has been an ICT productivity paradox, it comprises an apparent absence of TFP spillovers; in this respect, the contribution made by electricity in the 1920s through its impact on the reorganization of factory work has probably not yet been matched by ICT.

13

14

15

16

17

TECHNOLOGY AND ECONOMIC PERFORMANCE IN THE AMERICAN ECONOMYRobert J. GordonWorking Paper 8771http://www.nber.org/papers/w8771NATIONAL BUREAU OF ECONOMIC RESEARCH1050 Massachusetts AvenueCambridge, MA 02138February 2002

This paper examines the sources of the U. S. macroeconomic miracle of 1995-2000 and attempts todistinguish among permanent sources of American leadership in high-technology industries, as contrasted withthe particular post-1995 episode of technological acceleration, and with other independent sources of the economicmiracle unrelated to technology. The core of the American achievement was the maintenance of low inflation inthe presence of a decline in the unemployment rate to the lowest level reached in three decades. The post-1995technological acceleration, particularly in information technology (IT) and accompanying revival of productivitygrowth, directly contributed both to faster output growth and to holding down the inflation rate, but inflation wasalso held down by a substantial decline in real non-oil import prices, by low energy prices through early 1999, andby a temporary cessation in 1996-98 of inflation in real medical care prices. In turn low inflation allowed the Fedto maintain an easy monetary policy that fueled rapid growth in real demand, profits, and stock prices, which fedback into growth of consumption in excess of growth in income.The technological acceleration was made possible in part by permanent sources of American advantageover Europe and Japan, most notably the mixed system of government- and privately-funded research universities,the large role of U. S. government agencies providing research funding based on peer review, the strong traditionof patent and securities regulation, the leading worldwide position of U.S. business schools and U. S.-ownedinvestment banking, accounting, and management-consulting firms, and the particular importance of the capitalmarket for high-tech financing led by a uniquely dynamic venture capital industry. While these advantages helpto explain why the IT boom happened in the United States, they did not prevent the U. S. from experiencing adismal period of slow productivity growth between 1972 and 1995 nor from falling behind in numerous industriesoutside the IT sector.

18

The 1995-2000 productivity growth revival was fragile, both because a portion rested on unsustainablyrapid output growth in 1999-2000, and because much of the rest was the result of a doubling in the growth rateof computer investment after 1995 that could not continue forever. The web could only be invented once, Y2Kartificially compressed the computer replacement cycle, and some IT purchases were made by dot-coms that byearly 2001 were bankrupt. As an invention, the web provided abundant consumer surplus but no recipe for mostdot-coms to make a profit from providing free services. High-tech also included a boom in biotech and medicaltechnology, which also provided consumer surplus without necessarily creating higher productivity, at least withinthe feasible scope of output measurement.

Part of this change in perception was an illusion based on a change in the measuring rod. The annualgrowth rate of output per hour for 1972-95 was 1.1 percent per year based on data available prior to 1999 butjumped to 1.5 percent per year as a result of data revisions announced in late 1999.

The essence of the miracle was the conjunction of low unemployment and lowinflation. Fed policy avoided any sharp upward spike in short-term interest rates such as hadhappened during the previous expansion in 1988-89 because of the perception thataccelerating inflation was not a problem, despite a much lower unemployment rate than theminimum achieved in the earlier expansion. Policy reactions were less aggressive in the late1990s than in the late 1980s, because the economy appeared to have experienced a sharpchange in behavior along at least two dimensions. Unemployment could be allowed todecline because inflation remained low. The second change of behavior was in the growthof productivity. After resigned acceptance of the so-called "productivity slowdown," morethan two decades following 1973 when output per hour grew at barely one percent per annum(well under half of the rate experienced before 1973), analysts were astonished to observeproductivity growth at a rate of nearly three percent as the average annual rate for 1996-2000and an unbelievable 5.3 percent in the four quarters ending in mid-2000.3

Falling unemployment, low inflation, and accelerating productivity growth brought

19

many other good things in their wake. In February, 2000, the American economy set a recordfor the longest business expansion since records began in 1850. Profits surged and, at least until early in the year 2000, stock market prices grew even faster than profits, showeringhouseholds with unheard-of increases in wealth that in turned fueled a boom in consumptionand an evaporation of household saving (at least as conventionally measured, excluding capitalgains).

For the first time since the early 1970s, gains in real income were enjoyed byhouseholds in the bottom half of the income distribution, and in April, 2000, theunemployment rates for blacks and Hispanics reached the lowest levels ever recorded.

One way of describing the changing relationship betweentechnology and economic performance is through Robert M. Solow's famous 1987 saying that"we can see the computer age everywhere but in the productivity statistics." For a decadeeconomists took "Solow's paradox" as a truism, noting the continuing co-existence ofexplosive growth in computer use and dismal growth in labor productivity, and they differedonly in their explanations.6 But by 1999-2000 a consensus emerged that the technologicalrevolution represented by the New Economy was responsible directly or indirectly not justfor the productivity growth acceleration, but also the other manifestations of the miracle, including the stock market and wealth boom and spreading of benefits to the lower half ofthe income distribution. In short, Solow's paradox is now obsolete and its inventor hasadmitted as much.

The unemployment rate in 1999-2000 fell to four percent,the lowest rate since the 1966-70 period during which inflation accelerated steadily. Yet in1998 and early 1999, prior to the 1999-2000 upsurge in oil prices, inflation not only failed toaccelerate but rather decelerated.

I have attempted to use acommon framework to explain why the performance of the 1970s was so bad and of the1990s was so good, pointing to the role of adverse supply shocks in the earlier episode and

20

beneficial supply shocks more recently. In my interpretation (1998) inflation in 1997-98 washeld down by two "old" supply shocks, falling real prices of imports and energy, and by two"new" supply shocks, the accelerating decline in computer prices (see Figure 9 below) and asharp decline in the prices of medical care services made possible by the managed carerevolution. In retrospect my analysis, while still valid regarding the role of the supply shocks,did not place sufficient emphasis on the productivity growth revival as an independent sourceof low inflation. Between 1995 and late-2000 wage growth accelerated substantially from 2.1to above 6 percent at an annual rate, thus appearing to validate the Phillips curve hypothesisof a negative tradeoff between unemployment and wage growth. However, soaringproductivity growth during the same period prevented faster wage growth from translatinginto growth in unit labor costs (defined as wage growth minus productivity growth). If productivity growth were to decelerate, then it added one more element to the list oftransitory elements that had held down inflation in the late 1990s. Any of the items on thelist ó falling relative import and energy prices, a faster rate of decline in computer prices,moderate medical care inflation, and the productivity growth revival itself ó could turnaround and put upward rather than downward pressure on the inflation rate.

Thus far we have examined several manifestations of the American economic miracleof the late 1990s without focussing explicitly on the single most important factor which made all of this possible, namely the sharp acceleration in productivity growth that started at the end of 1995 and that was presumably caused entirely or in large part by the technological acceleration that we have labelled the "New Economy." Figure 7 divides the postwar into three periods using the standard quarterly data published by the Bureau of Labor Statistics (BLS), the "golden age" of rapid productivity growth between 1950:2 and 1972:2, the dismal slowdown period extending from 1972:2 to 1995:4, and the revival period since 1995:4.

The average annual growth rate of nonfarm private output per hour during the period 1972:2-1995:4 is 1.42 percent in the newly revised data.

Every major industrialized country experienced a sharp slowdown in productivitygrowth after 1973, and the extent of the slowdown in most countries was greater than in the United States. During 1960-73 growth in productivity (ALP) in the 15 countries of the European Union was double and in Japan quadruple that in the U. S. In the 1970s and 1980s productivity growth slowed down everywhere, but later than in the U. S., and by the first half of the 1990s productivity growth in Europe and Japan had converged to

21

that of the U. S. Thus the productivity slowdown was universal in the developed world rather than being unique to the U. S.

The post-1972 slowdown in the U. S., Japan, and Europe can be traced back to thesources of the "golden age" which began around the time of World War I in the United States (Gordon, 2000a). A set of "great inventions" of unprecedented scope and importance, including electricity and the internal combustion engine, had been developed during the Second Industrial Revolution of 1860-1900 and began the process of diffusion through the structure of the economy and society soon after the turn of the century (Gordon, 2000c). The productivity acceleration of the "golden age" occurred as the electric motor revolutionized manufacturing, as the internal combustion engine revolutionized ground transport and allowed the invention of air transport, and as other innovations in chemicals, petroleum, entertainment, communication, and public health transformed the standard of living in the United States between 1900 and the 1950s. In addition to the original advantages of the United States, particularly economies of scale and a wealth of natural resources (Wright, 1990), the dislocation of the two world wars and the turbulent interwar period delayed the diffusion of many of these innovations in Europe and Japan until after 1945, but then the rich plate of unexploited technology led to a period of rapid catch-up, if not convergence, to the U. S. frontier. This interpretation explains the post-1972 productivity slowdown as resulting from the the inevitable depletion of the fruits of the previous great inventions. The faster productivity growth in Europe and Japan during 1950-72, and the greater magnitude of their slowdowns, and the delayed timing of the slowdown into the 1980s and 1990s, is explained by the late start of Europe and Japan in exploiting the late 19th century "great inventions

How important has the New Economy and IT revolution been in creating the U. S.productivity revival which appears directly or indirectly to be responsible for most otherdimensions of the late-1990s U. S. economic miracle? Fortunately we do not need to explore this question from scratch, since recent academic research has produced a relatively clear answer which is summarized and interpreted in this section. The basic answer is that the acceleration in technical change in computers, peripherals, and semiconductors explains most of the acceleration in overall productivity growth since 1995, but virtually all the progress has been concentrated in the durable manufacturing sector, with surprisingly little spillover to the rest of the economy.

To provide a more precise analysis we must begin by distinguishing between thegrowth in output per hour, sometimes called average labor productivity (ALP), from thegrowth of multi-factor productivity (MFP). The former compares output growth with thatof a single input, labor hours, while the latter compares output with a weighted average ofseveral inputs, including labor, capital, and sometimes others, including materials, energy,and/or imports. ALP always grows faster than MFP, and the difference between them is the contribution of "capital deepening," the fact that any growing economy achieves a

22

growth rate of its capital input that is faster than its labor input, thus equipping each unit of labor with an ever-growing quantity of capital.

How have computers and the New Economy influenced the recent productivitygrowth revival? Imagine a spontaneous acceleration in the rate of technological change in the computer sector, which induces a more rapid rate of decline in computer prices and aninvestment boom as firms respond to cheaper computer prices by buying more computers. In response, since computers are part of output, this acceleration of technical change incomputer production raises the growth rate of MFP in the total economy, boosting the growth rate of ALP one-for-one. Second, the ensuing investment boom raises the "capital deepening" effect by increasing the growth rate of capital input relative to labor input and thus increasing ALP growth relative to MFP growth.

In discussing the New Economy, it is important to separate the computer-producingsector from the computer-using sector. No one denies that there has been a markedacceleration of output and productivity growth in the production of computer hardware. The real issue has been the response of productivity to massive computer investment by the 96 percent of the economy engaged in using computers rather than producing them. If the only effect of the technological breakthrough in computer production on the non-computer economy is an investment boom that accelerates the growth rate of capital input, then non-computer ALP growth would rise by the capital-deepening effect, but there would be no increase in non-computer MFP growth. Let us call this the "direct" effect of the New Economy on the non-computer sector. Sometimes advocates of the revolutionary nature of the New Economy imply that computer investment has a higher rate of return than other types of investment and creates "spillover" effects on business practices and productivity in the non-computer economy; evidence of this "spillover" effect would be an acceleration in MFP growth in the non-computer economy occurring at the same time as the technological acceleration in computer production.

In 1999 nominal final sales of computers and peripherals plus fixed investment in softwarerepresented 3.5 percent of nominal GDP in the nonfarm nonhousing private business economy. Thus the "non-computer part of the economy" represents 96.5 percent of nonfarm nonhousing private business output. Finalsales of computer hardware is an unpublished series obtained from Christian Ehemann of the BEA; the otherseries in this calculation appear in the Economic Report of the President, February 2000, Tables B-10 and B-16.

What is the counterpart of the New Economy in the official output data? Theremarkable event which occurred at the end of 1995 was an acceleration of the rate of pricechange in computer hardware (including peripherals) from an average rate of -12 percent during 1987-95 to an average rate of -29 percent during 1996-98. Computers

23

did not become more important as a share of dollar spending in the economy, which stagnated at around 1.3 percent of the nonfarm private business economy. The counterpart of the post-1995 acceleration in the rate of price decline was an acceleration in the rate of technological progress; apparently the time cycle of Moore's Law shortened from 18 months to 12 months at about the same time

We use the recent results of Oliner and Sichel (2000, 2001) to compute the contribution ofcomputers and semiconductors both to capital deepening and to the MFP acceleration in theoverall economy. Second, we summarize my recent study (Gordon, 2000b) that adds twoelements to the work of Oliner and Sichel.

Of the actual 2.86 percent annual growth of output per hour, 0.40 is attributedto a cyclical effect and the remaining 2.46 percent to trend growth, and the latter is 1.04 points faster than the 1972-95 trend. How can this acceleration be explained? A small part on lines 6 and 7 is attributed to changes in price measurement methods and to a slight acceleration in the growth of labor quality. All of the remaining 0.89 points can be directly attributed to computers. The capital-deepening effect of faster growth in computer capital relative to labor in the aggregate economy accounts of 0.60 percentage points of the acceleration (line 9a) and a 0.30-point acceleration of MFP growth in computer and computer-related semiconductor manufacturing account (line 10) sum to an explanation of 0.90 points, compared to the 0.89 acceleration in trend that needs to be explained.

There is no sign of a fundamental transformation of the U. S. economy. Therehas been no acceleration of MFP growth outside of computer production and the rest ofdurable manufacturing. Responding to the accelerated rate of price decline of computers thatoccurred between 1995 and 1998, business firms throughout the economy boosted purchasesof computers, creating an investment boom and "capital deepening" in the form of fastergrowth of capital relative to labor. But computer capital did not have any kind of magical orextraordinary effect ó it earned the same rate of return as any other type of capital.

The dependence of the U. S. productivity revival on the production and use ofcomputers waves a danger flag for the future. Consider the possibility that the accelerated1995-98 29 percent rate of price decline for computers does not continue. Already in the yearending in 2000:Q4 the rate of price decline slowed from 29 to 12 percent, the same asbetween 1987 and 1995. If in response the growth rate of computer investment were to slowdown to a rate similar to that before 1995, then the main source of the productivity revival

24

identified by Oliner and Sichel (2000, 2001) would disappear, and with it much of the U. S.economic miracle.

In fact, the slope of theprice-quantity relationship was appreciably flatter during 1972-87 than during 1987-95 or1995-99. If the demand curve has not shifted, the inverse of these slopes is the price elasticityof demand, namely -1.96, -1.19, and -1.11 in these three intervals, which can be comparedwith Brynjolfsson's (1996, p. 292) estimated price elasticity of -1.33 over the period 1970-89. The apparent decline in the price elasticity is consistent withthe view that the most important uses of computers were developed more than a decade intothe past, not currently.

A longstanding puzzle in the retardation of Britisheconomic growth after the 1870s is the fact that many inventions initially made by Britishinventors were brought to commercial success in the U. S., Japan, and elsewhere. This issueof who captures the fruits of innovation suggests that the British were not alone in losing out.The U. S. invention of videotape was followed by exploitation of the consumer VCR marketthat was almost entirely achieved by Japanese companies. The Finnish company Nokia tookover leadership in mobile phones from Motorola. Within any economy there are winners andlosers as upstart companies (Intel, Microsoft) seize the advantage in developing technologywhile leaving older competitors (IBM, Wang, Digital Equipment, Xerox) behind. While predicting technological developments in advance is exceedingly difficult, thereis an ample literature which points to particular national characteristics that help to explain,at least in retrospect, why particular inventions and industries came to be dominated byparticular countries.

25

26

27

28

29

30

31

Projecting Productivity Growth:Lessons from the U.S. Growth ResurgenceDale W. JorgensonHarvard UniversityMun S. HoResources for the FutureKevin J. Stiroh°ØFederal Reserve Bank of New YorkDecember 31, 2001

This paper analyzes the sources of U.S. labor productivity growth in the post-1995 period andpresents projections for both output and labor productivity growth for the next decade.Despite the recent downward revisions to U.S. GDP and software investment, we show thatinformation technology (IT) played a substantial role in the U.S. productivity revival. Wethen outline a methodology for projecting trend output and productivity growth. Our basecaseprojection puts the rate of trend productivity growth at 2.24 percent per year over thenext decade with a range of 1.33 to 2.98 percent, reflecting fundamental uncertainties aboutthe rate of technological progress in IT-production and investment patterns. Our centralprojection is only slightly below the average growth rate of 2.36 percent during the 1995-2000period.

This uncertainty is only magnified by the observation that recent productivityestimates remain surprisingly strong for an economy in recession. The Bureau of LaborStatistics (BLS, 2001b) estimates that business sector productivity grew 1.5 percent per yearfrom the third quarter of 2000 to the third quarter of 2001, while business sector output grewonly 0.1 percent per year.

32

We then turn to the future of U.S. productivity growth. Our overall conclusion is thatthe projections of Jorgenson and Stiroh (2000), prepared more than eighteen months ago, arelargely on target. Our new base-case projection of trend labor productivity growth for thenext decade is 2.24 percent per year, only slightly below the average of the period 1995-2000of 2.36 percent per year. Our projection of output growth for the next decade, however, isonly 3.34 percent per year, compared with the 1995-2000 average of 4.60 percent, due toslower projected growth in hours worked.

The starting point for projecting U.S. output growth is the projection of future growthof the labor force. The growth of hours worked of 2.24 percent per year from 1995-2000 isnot likely to be sustainable because labor force growth for the next decade will average only1.10 percent. An abrupt slowdown in growth of hours worked would have reduced outputgrowth by 1.14 percent, even if labor productivity growth had continued unabated. Weestimate that labor productivity growth from 1995-2000 also exceeded its sustainable rate,however, leading to an additional decline of 0.12 percent in the trend rate of output growth, sothat our base-case scenario projects output growth of 3.34 percent for the next decade.

Basu, Fernald, Shapiro (2001) present an alternative approach to estimating trendgrowth in total factor productivity by separately accounting for factor utilization and factoraccumulation. They extend the growth accounting framework to incorporate adjustment costs,scale economies, imperfect competition, and changes in utilization. Industry-level data for the1990s suggests that the post-1995 rise in productivity appears to be largely a change in trendrather than a cyclical phenomenon, since there was little change in utilization in the late

33

1990s. While Basu et al. are clear that they do not make predictions about the sustainability of11Both Roberts (2000) and French (2001) employ the Stock and Watson (1998) method of dealing with the zerobias.these changes, their results suggest that any slowdown in investment growth is likely to beassociated with a temporary increase in output growth as resources are reallocated away fromadjustment and toward production.

34

35

The Economic Impact of Information TechnologyKevin J. Stiroh*

Federal Reserve Bank of New YorkJanuary 18, 2001SummaryOne of the defining features of the U.S. economy in the 1990s is the massive expenditure oninformation technology equipment and software. With the U.S. economy surging over the sameperiod, this raises an obvious and important question: what is the economic impact of informationtechnology? Not surprisingly, there is a large and growing literature on the question and this essayreviews the empirical research done by economists and business analysts. The essay begins with adiscussion of how economists think about information technology and measure it using the hedonictheory of prices. The next section reviews the empirical estimates of the impact of informationtechnology, both at the macroeconomic level of the entire U.S. economy and then at the moremicroeconomic level of individual firms or industries. In both cases, the evidence is building thatinformation technology is having a substantial economic impact. The final section of the essaydiscusses some of the conclusions about information technology and the potential impact on the U.S.economy in the future.

36

Information Technology and the U.S. Productivity Revival:A Review of the EvidenceKevin J. StirohJanuary 10, 2002AbstractAggregate, industry, and firm level studies all point to a strong connection between informationtechnology (IT) and the U.S. productivity revival in the late 1990s. At the aggregate level, growthaccounting studies show a large and growing contribution to productivity growth from both theproduction and the use of IT. At the industry level, industries that produce or use IT mostintensively have shown the largest increases in productivity growth after 1995. At the firm level, ITintensivefirms show better performance than their peers, and several specific case studies show howIT improves real business practices. This accumulation of evidence from a variety of studiessuggests a real productivity impact from IT.

The revival in U.S. productivity growth in the late 1990s is a critical development for theU.S. economy because faster productivity growth contributes to rising living standards, helps keepinflationary pressures under control, and supports business profitability. Understanding the sourcesof productivity growth and why it fluctuates, however, is a difficult task and many factors contributeto its rise and fall.This paper focuses on one specific factor – information technology – and reviews theevidence on whether IT has contributed to the recent period of strong productivity growth. Whilesome discussion remains, there is a growing consensus that IT has played a critical role. At its heart,this is a story of profound technological progress in the production of IT assets like computerhardware, software, and telecommunications equipment. This generates large productivity gains inthe industries that produce IT assets, and provides powerful incentives for other firms to invest in thelatest IT assets. The evidence from aggregate, industry, and firm level studies is building that theseeffects are economically large and have contributed substantially to the U.S. productivity revival.

37

38

39

40

Information Technology and the U.S. Productivity Revival:What Do the Industry Data Say?Kevin J. Stiroh°ØFederal Reserve Bank of New YorkThis Draft: December 4, 2001First Draft: January 24, 2001AbstractThis paper examines the link between information technology (IT) and the post-1995 U.S.productivity revival. Industry-level data show a broad productivity growth resurgence that reflectsboth the production and the use of IT. The most IT-intensive industries experienced significantlylarger productivity gains than other industries and there is a strong link between IT capital shares andthe relative acceleration of labor productivity. A novel decomposition shows that all of the directcontribution to the post-1995 productivity acceleration can be traced to the industries that eitherproduce IT or use IT most intensively, with no net contribution from other industries that are relativelyisolated from the IT revolution.

41

42

43

44

45

46

47

48

The New Economy: Post Mortem or Second Wind?ByMartin Neil BailySenior Fellow,

How Well Does Growth Accounting Explain Shifting Productivity Trends?Table 1 shows the growth accounting estimates made by the Bureau of Labor Statistics(BLS) covering the periods 1948-73 and 1973-95. Increases in capital services per hour worked accounted for a modest 28 percent of the growth in output per hour over the period 1948-73 in non farm business. This figure increased to 50 percent over the period 1973-95 but capital services per hour worked provides almost no explanation of the slowdown in labor productivity. The slowdown in labor productivity growth that took place after 1973 is matched by an equal decline in the unexplained residual item of MFP growth.

A large but inconclusive literature developed in the 1980s attempting to understand why productivity growth slowed after 1973. Growth accounting failed to explain the post-73 acceleration; how does it do in explaining the pick-up in growth after 1995? Table 2 shows three estimates of the decomposition of the increase of productivity growth after 1995, focusing on the differences in growth rates pre and post-1995. The first updates Steven Oliner and Daniel Sichel (2000)1, the second updates the estimates reported in Baily and Lawrence (2001),2 and the third is from Dale Jorgenson, Mun Ho and Kevin Stiroh (2001). The results differ from those previously available largely because therewas a major downward revision of the GDP and hence productivity data made last summer. Much of the revision came when the estimates of software investment were revised down. The first two columns differ for two main reasons. Oliner and Sichel base their non farm business productivity growth only the product side of the National Income accounts, while my preferred estimate averages the income and product figures from the accounts, which are both valid estimates of non farm output. This makes the acceleration of labor productivity larger, because estimated income grew more rapidly than estimated production. Second, the two columns estimate the contribution of the IT sector to MFP differently.3 It is important to note that neither column uses an adjustment for business cycle effects. Jorgenson et al. apply their growth accounting to a broader definition of the economy, so the figures are not directly comparable.

According to both Oliner and Sichel and Jorgenson et al, the speed-up in labor

49

productivity growth after 1995 is largely ‘explained’ by the growth accounting framework, and the explanation is mostly an IT story. The rapid accumulation of IT capital provided a large boost to labor productivity (adding 0.59 percentage point in Oliner-Sichel), more than offsetting the slowing of the contribution of other capital. There was a small effect of labor quality (education and experience effects). And then a boost from faster MFP within the IT sector. Of course that is an MFP residual effect, but it is much less mysterious than the traditional MFP residual. It is well-known that the computer and semiconductor industries increased their pace of productivity advance after 1995 as a result of a faster rate of introduction of new generations of chips and intense competitive pressure. There is a small (0.23 percentage point in Oliner-Sichel) step-up in‘other MFP’, but if some fraction of the productivity acceleration were cyclical, that mystery component could easily vanish, in which case the whole acceleration is explained or at least can be easily described.

Following the data revisions of last summer, therefore, the latest Oliner and Sichel andJorgenson et al. results are consistent with the view that the post-95 acceleration in the economy as a whole was driven by the IT sector. Faster MFP in the IT sector added directly to productivity growth and the resulting decline in the price of IT capital induced rapid capital accumulation that added to labor productivity in the rest of the economy. Robert Gordon also reaches very much the same conclusion.

Second, microeconomic analyses of plants and firms find substantial adjustment costs toinvestment and lags between investment and productivity. The growth accounting approachassumes increases in capital intensity have an impact on productivity in the same year they occur. That is an important issue because the coincident timing of the productivity surge and the IT investment surge is a key reason for thinking the latter caused the former.Third, the growth accounting framework formalizes what is essentially a one-observationcorrelation. Labor productivity accelerated at about the same time that IT investment surged. The assumption is made that the IT capital surge caused the productivity surge, but reverse causality is also possible.

To determine the importance of all these issues we have to go down a level to look atindustry data and case studies. If the growth accounting is correct, then the productivityacceleration should be showing up in the industries that invested in all the IT capital. The availability in 2000 of output by industry data

50

prepared by the Bureau of Economic Analysis using new price deflators spawned a series of analyses of the productivity acceleration that changed the perspective on the slowdown. I will discuss the findings of these studies shortly, but before that it is worth looking at the latest industry numbers since updated and revised figures were released in November 2001. Table 3 shows labor productivity growth estimates from 1989- 95 and 1995-2000. Each industry’s output reflects the value added in that industry (gross productoriginating) and labor input is measured by the number of full-time equivalent employees.

These new figures confirm the important conclusion that emerged in the literature cited above, namely that the increase in labor productivity growth took place in service industries as well as in durable goods manufacturing. In particular there was a surge in productivity in wholesale and retail trade and in the finance sector.

What about the link to IT capital? Baily and Lawrence carried out a simple exercise thatis repeated in the last two rows of Table 3. The industries were divided into those that were more IT-intensive and those that were less IT-intensive, measured by IT capital in 1995 relative to value added. The IT-intensive group had much faster productivity growth throughout and alarger productivity acceleration.Stiroh has gone well beyond this simple calculation and estimated the impact of ITintensity. His strongest results come from defining IT intensity based on the share of IT capitalservices in total capital services in 1995. If that is high, he argues, it “identifies industriesexpending tangible investment on IT and reallocating assets toward high-tech assets.” (page 10).

He also argues that it is important to use a measure that is exogenous, not one that is defined overthe same period as the productivity revival. Hence he uses the IT intensity in 1995. He finds thatindustries that are above the median in their IT intensity, by his measure, have much largerincreases in labor productivity after 1995. His findings are pretty robust, remaining strong evenafter excluding outliers and making other tests. One place that gives weaker results is when ITintensity in 1995 is calculated as IT services per FTE. He notes that industries that had invested

51

heavily in IT to reduce labor in the early 90s would have had difficulty making further laborproductivity gains in the late 90s.The McKinsey Global Institute (2001) recently completed an analysis of the accelerationin US productivity growth. They main part of this study consisted of case studies, which I willdiscuss later, but they also offer an alternative regression analysis with mixed results. Startingwith value added data, they find that if each industry is considered as one observation, there isalmost no correlation between the acceleration in labor productivity growth by industry after1995 and the surge in IT capital per employee by industry. On the other hand, if each industry is8weighted by employment, there is a positive and significant correlation.10 The study arguedfurther that the concentration of the acceleration in a few industries is revealing because theseindustries account for much of the total acceleration, but a much smaller fraction of the totalsurge in IT capital accumulation. This again, they argue, undermines the IT/productivity link.No statistical evidence can be definitive, but there is a pretty strong case for saying thereis some connection between IT and the productivity acceleration. But how close that connectionis and exactly how it works is not certain. Case study evidence can provide more information.Case Study Evidence The main contribution of the McKinsey study was a series of eightindustry case studies, looking in detail at what had happened to productivity in the 90s. Sixindustries that had contributed disproportionately to the productivity acceleration were included,wholesale and retail trade, computers and semiconductors, telecom and securities. And retailbanking and hotels were included as two industries that had invested heavily in IT but hadexperienced no surge in productivity.11Based on these case studies, the study concluded that competitive pressure was the maindriver of the productivity acceleration, by forcing improvements in business operations. In the

52

retail trade case, they found that Wal-Mart played a pivotal role, both because of its own largesize and high productivity, but also because it put competitive pressure on other retailers. In thesemiconductor industry, Intel came under competitive pressure from AMD and responded byshortening its product cycle. There was an accelerated decline in the price of microprocessorsthat was translated into an acceleration of measured productivity in this industry. Going in theopposite direction, the study found that hotels and retail banks were facing weaker competitivepressure and were able to earn large profits without pushing as hard to improve efficiency.In general, the case studies make clear that productivity improvements come from avariety of sources. Improvements in retail productivity came about through organizationalimprovements, the advantages of large-scale “big box” stores and by a shift to higher value goodsassociated with the growth in the number of high-income consumers.There are examples where IT investments yielded little payoff. Banks invested veryheavily in powerful computers that were not really needed for the tasks that most of the bankemployees were performing. However, there are also examples where IT did contributesubstantially to productivity. Two of the industries were high-tech producers. The telecomindustry is a direct application of IT. The Internet allowed much higher productivity in the10 The case for weighting the observations is that the variance in productivity measures may be larger forsmall industries.11 Note, however, that the latest BEA data now find increased growth in banking productivity after 1995.9securities industry, and IT is used heavily in the giant wholesale and retail sectors. Wal-Mart, forexample, has relied heavily on IT as an enabler of its efficient supply chain. It operates thelargest commercial database in the world.The relation between IT and productivity revealed in the cases is a complex one, andoften IT systems developed prior to 1995 were vital facilitators of productivity improvements

53

after 1995. This suggests some of the impact of the surge in spending after 1995 may show up asfaster growth in future years.Conclusions from the Industry and Case Study Data. There are pitfalls in any industrylevelanalysis of productivity because the data are not collected in a way that really allows theresearcher to allocate outputs and inputs by industry, particularly inputs. Deciding whichindustries are using IT capital intensely is problematic, for example. One example of the kind ofproblem posed is business services. This is a large industry that has grown a lot and is a largeinvestor in IT. There is no serious measure of real output or productivity for this industry, so it isvery hard to know the extent to which the industry or its IT may be contributing to theacceleration of productivity in other industries.Another difficulty in drawing strong conclusions about the IT--productivity link from theindustry data results from the fact that a main driver of the surge in IT capital was the acceleratedprice decline that encouraged everyone to invest more. Investing in IT became the thing to do.Ex post, some companies or industries used their IT capital well and others not so well. Someindustries were subject to unrelated productivity shocks and some to swings in measuredproductivity resulting from vagaries in the measurement process. Without controls for these‘other factors’ it is hard to separate out the impact of greater or lesser amounts of IT capitalaccumulation.Despite these difficulties the industry data and case studies are valuable. It does look asif the productivity acceleration, even though it is concentrated, spread outside just the ITproducing sector. There are signs that IT using industries have done relatively well inproductivity growth, but clearly other factors are important. It is hard to look at the industry dataand the detailed case studies and conclude the productivity acceleration was just IT-driven.Other Indicators of Structural Change: What do they Show?10

54

There are three aspects of US macroeconomic performance in the 1990s that give an alternativereading on the economy in the late 90s.12 They are related to productivity, but rely on differentdata.Rising Real Wages and Low Unemployment Average hourly earnings, adjusted forconsumer prices, declined by 0.5 percent a year from 1978-95 and then increased at 2.0 percent ayear from 1995-2000, a difference of 2.5 percent a year. The corresponding acceleration of realconsumption wages was 1.9 percent for compensation per hour and 1.3 percent for theemployment cost index.13 It appears that the strong economic performance and increasedproductivity growth of the late 90s translated into much faster real wage growth. This result isimportant because it shows that the economy in the late 90s did not just generate faster computersand communications equipment, it affected the ability of the economy to deliver consumer goodsin return for hours of work. It changed the path of workers’ real consumption wages, includinghourly employees. Some part of this was likely the result of the very strong labor demand in theperiod, but not that much, because average real wages are not strongly pro-cyclical. Moreover,real wages grew even faster from the third quarter of 2000 to the third quarter of 2001 than 1995-2000, using any of the three measures of wages. So the pattern of solid real consumption wageperformance has continued despite demand weakness in the economy.Based on the experience of the 70s and 80s, estimates of the non-accelerating rate ofunemployment (NAIRU) for the US were 6 to 6_ percent a year. Certainly the experience of the80s expansion seemed to confirm this, where wage and price inflation started to worsen as theunemployment rate fell below 6 percent. By contrast, unemployment fell below 6 percent inSeptember of 1994 and moved to around the 4 percent mark. There was no sign of acceleratinginflation until 2000, in fact core inflation fell over much of the 90s. Several writers have linkedthe improvement in the inflation/unemployment tradeoff to the change in the path of real wages.14

55

Nominal wage changes do not increase immediately when productivity growth improves, and sounit costs fall relative to trend and inflation is held down.In summary: the experience of the labor market provides confirmation of a structuralchange in the US economy in the 90s that is somewhat independent of specific productivitynumbers. The hedonic price indexes for computers have changed the measured rate of12 For a more extended discussion of these issues see Baily (2001).13 In all three cases consumer prices are measured by the chain price index for personal consumptionexpenditure, excluding food and energy. Food and energy do of course affect workers living standards, butthe fluctuations in the prices of these items are largely unrelated to the performance of the nonfarmbusiness sector. Using the core CPI shows smaller, although still strong, increases in real wage growth.See Baily (2001) Table 5. The employment cost index starts only in 1980.14 Blinder (2001), Ball and Moffitt (2001) and Council of Economic Advisers (2001).11productivity growth, but they have relatively little impact on consumer prices and cannot be themain reason why unemployment stayed so low for so long.Market Value of Corporations According to Robert Hall (2001), the ratio of the marketvalue of corporations to the replacement cost of their capital (Tobin’s Q) nearly tripled in the1990s. In part, this increase in market value came about because stocks were valued by differentcriteria at the beginning and end of the decade. In particular, the risk premium may havedeclined. Moreover, stock prices reached unsustainable levels in 1999 (the end point of Hall’scalculation), and have fallen since then. Technology stocks certainly went through a bubble, andthe overall stock market may also have experienced an increase that was not driven by thefundamentals.The surprising thing about the increase in market value, however, is not that it hasweakened from its extreme highs, but how robust it has been. Despite the economic downturn

56

and the uncertainty induced by the September 11 attacks, stock price indexes remain very high.Even if Tobin’s Q were to fall sharply from its level at the end of 1999, it would still be wellabove unity, its value at the beginning of the 90s. This means that markets have a persistentbelief that corporations now hold valuable intangible capital assets.Wall Street pays some attention to productivity data, but not a lot. Market valuations arebased on expectations about profits. Back in the mid-90s before clear signs of improvedproductivity had emerged, Wall Street analysts started to increased their expected profit growthfigures and market valuations, which had been rising strongly since the mid-80s, started to riseeven faster. History may look back at the current period as one where the market turned out to bedramatically over-valued. But for the present, Wall Street seems convinced that there is stillstrong profit growth potential in the corporate sector. It retains the belief that the economy hasexperienced a structural change.Capital Inflow and the Dollar Figure 1 shows the rapid increase in the net inflow offoreign capital into the United States, heavily concentrated in direct foreign investment andsecurities.15As a consequence of the foreign appetite for these US investment opportunities, theportfolio of assets held by foreigners has shifted into equities and direct ownership of UScompanies. These two classes of investment constituted roughly 50 percent of all assets held byforeigners in 1999, up from around 30 percent in 1990.The desire of foreign companies to buy into the United States does not mean they are justbuying high-tech or IT companies. The new economy in the United States is apparently making12traditional industries attractive too. The breakdown of direct foreign investment by industryshows increased investment in high-tech sectors such as electronic components, computerservices and telecom. But there was also a large increase in investment spread throughout a range

57

of different industries, including traditional manufacturing industries and heavy IT users such asinsurance and financial services.16 Equity investment may have been more heavily skewed to thetech sector, but I do not have a breakdown of that capital inflow.When the US economy started to slow dramatically and then move into recession,triggered in part by the terrorist attacks, one might have expected the value of the US dollar to fallas the appetite for US assets declined. Instead, the US dollar has remained very high against boththe Euro and the Yen. The weakness in Japan is a separate issue, but the continued strength ofthe dollar against the Euro and other currencies remains something of a surprise. It appears therest of the world believed in the increased profitability of capital in the US and continues tobelieve that the prospects for asset returns are stronger in the US than in the other main industrialcountries. The inflow of capital and the strength of the dollar can be seen as another piece ofevidence supporting the view of a structural change in the US economy.17Causes and Future ProspectsBased on the data given above and the studies described, the weight of the evidence points clearlyto the conclusion that there was a structural break in the performance of the US economy in the1990s. Despite revisions, the aggregate productivity data still show a break in the productivitytrend. The case study evidence supports the view that the pace of innovation picked up in a rangeof different industries. The changes in the labor market, the rise in the market valuation ofcorporate capital, and the inflow of capital to the US, all support the conclusion that structuralchange took place. This combination of economic outcomes suggests productivity growth andthe return on capital both increased in the 90s.A caveat is in order. Should the improvement in economic performance be described ashaving created a new economy? There are reasons to be skeptical about the term ‘new economy.’After all, the economy is like a large oil tanker, it does not change direction quickly and much

58

economic activity today is similar to what it was in the early 1990s. Most employment and15 The “other private assets” category includes bank loans.16 See the table at http://www.bea.doc.gov/bea/di/fdi-ind.htm17 Jaume Ventura (2001) disagrees with this interpretation. He argues that US residents experienced a verylarge increase in wealth in the 90s and responded by increasing their consumption and borrowing from therest of the world. However the fact that US residents could attract funds as easily as they did suggests aperception by foreigners that returns had increased in the US.13output remain in traditional industries and we learned in 2001 that the business cycle is still alive.In fact the current recession, which involves large swings in inventory and equipment investment,looks rather like the old-style recessions of the period after World War II. While I use the termnew economy, because it captures the idea of a shift in the economy in the 90s, I recognize that itprobably overstates the extent of change.With that terminology disclaimer out of the way, the question is whether there is somecompromise or consensus explanation of the productivity improvement that emerges. Certainlyadvances in IT are changing the way business is done. And as we have seen, the aggregategrowth accounting evidence and some analyses of the industry data make a case that a major partof the productivity acceleration was the result of these advances. But for the following reasons, Ijudge that IT was not the sole reason for the productivity shift after 1995.First, I prefer the growth accounting framework used in Baily-Lawrence. Thisframework indicates there was a significant increase in MFP growth after 1995 outside the IThardware sector. Second, the growth accounting framework itself may overstate the causal linkfrom IT capital to growth if there is reverse causality. Third, the large increase in real wagegrowth, together with the extraordinary growth in the market valuation of corporations does notseem consistent with a model where there is faster MFP growth only in the IT hardware sector

59

that affects other sectors only through IT capital accumulation. Fourth, the case studies providedevidence that innovative business practices, that in some cases were unrelated to IT, contributedheavily to faster productivity. Fifth, the regressions from the industry data and the case studiesindicate that IT capital in place prior to 1995 facilitated productivity growth after 1995. This putsinto question the immediate and direct linkage from rapid IT capital accumulation to rapidproductivity growth that is built into the growth accounting exercise. And finally, the failure ofthe growth accounting framework to explain the slowdown in the 70s leaves some residualskepticism about its ability to fully capture the speed up.Whether the improved economic performance of the 90s was wholly IT driven or onlypartly IT driven, there remains the question of why did the pace of innovation speed up in the 90s.What was it about the economic conditions in the 90s that fostered stronger economicperformance? One simple way to describe the shifts in the productivity trend is that there was apool of innovations and investments to be exploited after World War II, and these generated thefast trend of productivity over the period. There was then a lull as the easy ways to raiseperformance had already been found. During the 1970s and 1980s there was a lot of changetaking place and old capital was being destroyed. There was an ongoing push of economic changeand innovation, and the IT revolution was in the making, but the benefits were not yet showing14up, at least in measured productivity. Some time in the 1990s a new flow of productivityenhancinginnovations came on stream, the economic environment was favorable, and there hasbeen a return to something closer to the postwar trend of faster growth.While there is no proof here, one can see some of the reasons why the environment in theUS economy in the 1990s was favorable both for rapid innovation and rapid diffusion ofinnovation, both IT related and otherwise. There has been heightened competition in an

60

increasingly deregulated economy facing strong international competition. IT innovation is drivenby the demand for improved technologies in the using industries. The United States hascompetitive service industries, often on a global scale, and this encourages them to seek out newtechnologies to improve their own productivity. If the new economy were entirely the result of arandom surge in the flow of innovation, then all countries should have had similar changestogether. The new technologies are available globally. In practice the United States has been wellahead of most of the industrial countries and a reason for this is that the United States hascompetitive markets in the industries that are using IT.When innovations occur in one area, they can bring benefits. But when complementaryinnovations occur together the effects can be greatly increased. The 90s were a particularlyfavorable time because a combination of rapid advances in computing power, software andcommunications capabilities formed such a set of complementary innovations. Large amounts ofdata can be processed and presented in a way non-technical personnel can use and thentransmitted to remote locations within the same firm or to other firms.The policy environment contributed to the creation of the right environment for growthand innovation. Policies to deregulate industries and to maintain domestic competition andincrease international competition have been stressed. Funds have been provided to support basicresearch and education. And the mix of monetary and fiscal policy has lowered interest rates andencouraged investment.Future Prospects If you believe the productivity revival was structural, but it came fromthe surge in productivity growth within the computer sector, plus some cyclical component, thereis a case for a rather pessimistic view of future productivity growth. First, the huge rate of declinein computer prices of the late 90s may not continue indefinitely and, second, there may bediminishing returns to investment in IT capital. The collapse of investment in the past year seems

61

consistent with this view, and has caused a large adjustment in some commercial forecasters’predictions of productivity growth. Macroeconomic Advisers, for example, predicts laborproductivity growth to be fairly strong in the second half of 2002, with a cyclical bounce, but toaverage only 1.6 percent for the four quarters of 2003, because of the collapse of IT investment.15Using ‘conservative’ assumptions, Oliner and Sichel (2002) argue that the productivity growthtrend could fall slightly below 2 percent.The basis for pessimism in the IT-only scenario can be questioned, however. First, ITprices could well continue to fall rapidly since the technological drivers of improved performanceand the intense competition are still at work.18 And even if the price declines do moderate, thereis an offset, a “share effect.”19 The share of IT capital in total capital has been rising rapidly, withthe effect of pushing up the rate of real capital accumulation. The growth rate of the total capitalstock is a weighted average of the growth rates of the components. The faster-growingcomponents (IT equipment) are a growing share of the total. Jorgenson et al. (2001) are actuallyrather optimistic about the future prospects for labor productivity, even though they view therevival as being IT driven. Jorgenson et al. anticipate a growth rate of 2.24 percent a year, notmuch slower than in the late 90s, based on the assumption of continued rapid decline of processorand computer prices. There is a wide error band around their central figure, ranging from apessimistic estimate of 1.33 percent and an optimistic estimate of 2.98. Macroeconomic Advisersis also more optimistic beyond 2003, when they predict a resumption of investment spendingpicks and a productivity trend of well over 2 percent. And Oliner and Sichel (2002) project asmuch 2_ percent as the future productivity trend under more optimistic assumptions.The McKinsey Global Institute, which argues that IT is only a secondary factor in theproductivity surge made a forecast of future labor productivity growth based upon the case

62

studies. They looked at each industry under study and asked to what extent would the drivers ofproductivity in the 90s continue into the future and were there innovations in process that wouldincrease productivity. They estimated growth at 2.0 percent a year going forward with a range of1.6 to 2.5 percent.As I have noted, my view of the determinants of recent productivity puts less weight onIT than Jorgenson et al. or Oliner and Sichel, but somewhat more weight than that ascribed by thecases studies. On this basis, how much of the post-95 growth surge will continue? The basicdrivers of the acceleration, rapid improvements in IT, strong competition and the impact ofglobalization are in place for a continuation of growth. One cannot necessarily tell how newtechnologies will be used, but going forward the Internet is likely to add to productivity, whereasit was not a major factor in the 90s growth. Robert Litan and Alice Rivlin (2001) estimate thatthe Internet will add about a quarter of a percentage point to future growth. And although they do18 For example, John Markoff reports in the June 30th New York Times (2001) on an advance made at Intel“demonstrating that the semiconductor industry will be able to continue shrinking its basic building blocksat a torrid pace at least until the end of this decade.”19 Daniel Sichel has pointed this out.16not allow one to put a specific number on productivity, the “other” indicators of a new economy--the robust stock market and the strong dollar--are still signaling that markets expect continuedhigh returns to capital, which will contribute to growth.Despite a weak economy, labor productivity over the six quarters since mid 2000 grew atabout 1.5 percent at an annual rate. Non farm output declined in the second, third and fourthquarters of 2001 and yet there was positive labor productivity growth in all three quarters. Thishas never occurred before over the postwar period. With the caveat that recent data are oftenrevised, this suggests that the trend of labor productivity is well above 1.5 percent.

63

I take as a starting point the figure that labor productivity grew at about 2.7 percent a year1995-2000, based on an averaging of income and product estimates. At one time, it was thoughtthat the productivity trend was continuously accelerating, but once the output data were revised,that view became hard to sustain. To the contrary, it seems much more likely that some part ofthe growth over this period was either temporary or cyclical. There was a boom in financialmarkets that drove revenues through the roof in the financial services sector and pushed measuredproductivity growth unsustainably high. There was a spending boom that pushed productivity inthe trade sector way up. Some part of that productivity increase does not seem sustainable.Therefore, 2.5 percent is a reasonable upper bound on the trend. The fact that technology andcompetition remain strong, plus the evidence of recent productivity data, suggest 2.0 percentgrowth is a reasonable lower bound to the current labor productivity trend.20A Second Wind but at a Lesser Strength Despite different perceptions of the drivers ofproductivity, it seems that the literature I have reviewed above all concludes that the productivitytrend is likely to be in the range of 2 to 2_ percent a year. That is a strong performance that willincrease wages and living standards and allow potential GDP growth in the range 2.8 to 3.3percent a year. It is certainly enough to say the new economy will get a second wind. That doesnot mean we will go back to the 90s. This rate of GDP growth is far slower than the rate of over4 percent achieved 1997-2000. Moreover, other elements of the new economy may look weaker.20 One negative for productivity growth over the next year or so is the impact of September 11. As othershave noted, this could reduce underlying growth for a period, but should not affect the longer run trend.The attacks of September 11 will have some lasting effects on the level of productivity in the US and hencea short-term effect on productivity growth. One quick way to estimate the impact is as follows: If privatebusiness were to hire 200,000 additional security personnel, about a 50 percent increase over the current

64

amount, and the all-in cost per person were $100,000, then this would add to $20 billion or about 0.2percent of GDP. If there were also extra travel time, higher inventory etc, the result could be a reduction ofproductivity growth by about 0.2 to 0.3 percentage point for a couple of years. The forecasting firm,Macroeconomic Advisers, has suggested the productivity effect of the attacks would be smaller. Theyexpect around 100,000 extra employees at a cost of $54,000 per person, giving a total cost of only $5.4billion. Running this through their macro model, they estimate the productivity effect as 0.01 to 0.02percentage point per year, although the effect was more persistent in their view.17Even though the new economy is likely to continue with faster productivity, this does notimply that stocks or the dollar will remain as strong. Future corporate earnings could justify thecurrent level of the stock market or stimulate further growth, but they may very well not do so.Even a substantial fall in the market would still leave Tobin’s Q well above unity, consistent witha sizeable valuation of intangible capital, and that is the main expectation based on the neweconomy. For the dollar: At some time in the future the dollar is very likely to decline even if thereturns to capital in the US remain high. The dynamics of servicing a rising foreign debt suggestthis, plus new economy opportunities may develop more strongly in other countries, changing therelative attractiveness of investing in the US.The continuation of solid productivity growth, occurring in industries that produceconsumer goods and services as well as those making investment goods, implies real wagesshould continue to grow more rapidly going forward than they did during the post-1973 period.However, there is no guarantee that the NAIRU will stay as low as it seemed to be in the 90s. Inmany specifications of the Phillips curve, an increase in the trend rate of growth of real wagesinduces only a temporary shift in the inflation/unemployment tradeoff.21 I would expect somepersistent benefit in lower unemployment coming from faster productivity growth, largely

65

because the NAIRU was much lower in the 50s and 60s when real wage growth was rapid than inthe 70s and 80s when real wages were weak. But given the wide disagreements in the professionof the nature or even existence of a Phillips curve, there is no definitive evidence to prove thisview.In summary: Even though I conclude that the ‘new economy’ will get a second wind,this does not mean a return to the economic euphoria of the late 1990s.Policy ImplicationsThe high level of competitive intensity in the US economy has been important in stimulatinginnovation in both the high-tech sector and in the old economy, or the traditional industries.Policy must act to sustain this by maintaining open international markets and avoidingprotectionist measures. Maintaining competitive intensity also means pursuing an anti-trustpolicy that allows industry change to take place, including consolidations, but that resists anticompetitiveactions by dominant players that could discourage innovation and encourageinefficiency. Anti-trust policy is not just about improving static efficiency, it is also aboutmaintaining the competitive environment for innovation and productivity growth.21 See Baily (2001) for a more extended discussion of this issue and additional references.18Regulatory policy is just as important for productivity growth. The deregulationmovement of the 70s and 80s surely contributed to the flowering of innovation in the 90s. Goodregulatory policy is not just a matter of getting rid of regulations, however, it should set the rulesof the game for competitors in a way that enhances efficiency and innovation. One example isregulatory policy towards the use of wireless spectrum, an area of potential future technologygrowth. The current allocation of wireless spectrum is very inefficient, and a more economicallyrational allocation of wireless spectrum would enhance the development of this important area ofthe new economy.22

66

There is not the space here to deal with the issue of skills, but I want to acknowledgetheir importance. The returns to education and skill have risen dramatically in the past twentyyears, contributing to a widening of the wage and income distribution. The increased deploymentof IT has also changed the skill requirements of jobs. So it seems evident that education andtraining have become more important. In practice it is hard to create effective programs toenhance skills. Much of the skill acquisition necessary comes in training on the job. But workerswho are better prepared before they enter or re-enter the workforce are more likely to be trainedon the job and are more able to take advantage of training. Since innovation and productivitygrowth often involve the loss of jobs in some firms and industries as others expand, it is essentialthat workers have the opportunity to acquire new skills and become re-employed.Good macroeconomic policy contributed to the favorable economic environment of the90s. Monetary policy encouraged growth in the mid 90s, tightened up at the end of the decade asgrowth started running too fast. And it acted strongly to mitigate the impact of the downturn andrecession of 2001. It may face new challenges going forward, either if the recovery fails tomaterialize or if the turnaround is too strong. As I noted earlier, the combination of lowunemployment and low inflation achieved in the 90s may not be as easily reached over the nextfive years. In the 90s there was an exaggerated view of what monetary policy can accomplishthat may engender disappointment in the future.Fiscal policy in the 90s contributed to growth by keeping interest rates low andstimulating investment. Going forward, it will be necessary to rely more heavily on domesticsaving as a source of investment funds than was the case in the 90s. The US economy needs tobe rebalanced with a higher saving rate in the US, lower growth of imports, and a smaller currentaccount deficit. Maintaining budget surpluses can contribute positively to this rebalancing and

67

sustain the environment for growth. I noted earlier that budget planners consistentlyunderestimated the improvement in the fiscal situation in the 90s, but that lesson now seems to22 See Martin Baily, Robert Willig, Peter Orszag and Jonathan Orszag, (2001).19have been learned too well. The danger today, even if the new economy gets a second wind, isthat budget planners will overestimate tax revenues and underestimate spending, thereby erodingor eliminating the budget surpluses.ConclusionWhen running or playing a racquet sport it is easy to start too strong and then become winded andexhausted. After setting a better pace, the body fortunately seems to get a second wind. Theexhaustion lessens and breathing becomes easier, although one does not have quite the sameenergy as was available in the beginning. That seems an appropriate analogy for the economy inrecent years. The enthusiasm for IT and for the shares of high-tech companies went too far.Growth became unsustainably rapid and a slowdown inevitable. But the excessive optimism ofthe 90s should not give way to excessive pessimism now. Many of the drivers of growth in the90s, those linked to IT and those unrelated to it, remain in place.

68

69

70

71

72

73

74

75

References

Robert Barro (1999), "Notes on Growth Accounting", Journal of Economic Growth, 4, 119- 137.

Martin Baily (2002), “The New Economy: Post-Mortem or Second Wind?” Journal of Economic Perspectives (forthcoming).

Susanto Basu, John Fernald, and Matthew Shapiro (2001), “Productivity Growth in the 1990s: Technology, Utilization, or Adjustment?” (Cambridge: NBER Working Paper 8359).

Mark Blaug (1961), "The Productivity of Capital in the Lancashire Cotton Industry during the Nineteenth Century", Economic History Review, 13, 358-381.

Timothy Bresnahan and Manuel Trajtenberg (1995), "General Purpose Technologies: 'Engines of Growth'?", Journal of Econometrics, 65, 83-108.

Bresnahan, Timothy F., and Robert J. Gordon (1997). "Introduction," The Economics of NewGoods. University of Chicago Press for NBER, pp. 1-26.

Erik Brynjolffson and Loren Hitt (2000), "Beyond Computation: Information Technology, Organizational Transformation and Business Performance", Journal of Economic Perspectives, 14(4), 23-48.

Brookes, M. and Wahhaj, Z. (2000), "Is the Internet Better than Electricity ?", Goldman Sachs Global Economics Paper No. 49.

Bureau of Economic Analysis (1974), Fixed Nonresidential Business Capital in the United States, 1925-1973. Washington, DC: Government Printing Office.

Paul David (1991), "Computer and Dynamo: The Modern Productivity Paradox in a Not-Too-Distant Mirror", in OECD, Technology and Productivity: The Challenge for Economic Policy. Paris: OECD, 315-348.

Paul David and Gavin Wright (1999), "Early Twentieth Century Productivity Growth Dynamics: An Inquiry into the Economic History of 'Our Ignorance'", University of Oxford Discussion Papers in Economic History No. 33.

Phyllis Deane and W.A. Cole (1962), British Economic Growth, 1688-1959. Cambridge: Cambridge University Press.

Walter Devine (1983), "From Shafts to Wires: Historical Perspective on Electrification", Journal of Economic History, 43, 347-372.

76

Charles Feinstein (1988), "National Statistics, 1760-1920", in Charles Feinstein and Sidney Pollard (eds.), Studies in Capital Formation in the United Kingdom, 1750-1920. Oxford: Clarendon Press, 257-471.

Robert Fogel (1964), Railroads and American Economic Growth: Essays in Econometric History. Baltimore: Johns Hopkins University Press.

Robert Gordon (2000), "Does the 'New Economy' Measure Up to the Great Inventions of the Past ?", Journal of Economic Perspectives, 14(4), 49-74.

Gene Grossman and Elhanen Helpman (1991), Innovation and Growth in the Global Economy. Cambridge, Mass.: MIT Press.

C. Knick Harley (1999), "Reassessing the Industrial Revolution: A Macro View", in Joel Mokyr (ed.), The British Industrial Revolution: An Economic Perspective. Oxford: Westview Press, 160-205.

Dale Jorgenson and Kevin Stiroh (2000), "Raising the Speed Limit: US Economic Growth in the Information Age", Brookings Papers on Economic Activity, 1, 125-211.

Dale Jorgenson, Mun Ho, and Kevin Stiroh (2001), “Projecting Productivity Growth: Lessons from the U.S. Growth Resurgence” (Cambridge: Harvard University xerox).

John Kendrick (1961), Productivity Trends in the United States. Princeton: Princeton University Press.

Steven Oliner and Daniel Sichel (2000), "The Resurgence of Growth in the Late 1990s: Is Information Technology the Story?", Journal of Economic Perspectives, 14(4), 3-22.

Nicholas Oulton (2001), "ICT and Productivity Growth in the United Kingdom", Bank of England Working Paper, No. 140.

Paul Romer (1986), "Increasing Returns and Long Run Growth ", Journal of Political Economy, 95, 1002-1037.

Paul Romer (1990), "Endogenous Technological Change", Journal of Political Economy, 98, S71-S102.

Paul Schreyer (2000), "The Contribution of Information and Communication Technology to Output Growth: A Study of the G7 Countries", OECD STI Working Paper No 2000/2.

Daniel Sichel (2001), "The Resurgence of US Growth in the Late 1990s: An Update and What Lies Ahead", paper presented to HM Treasury seminar, London

Kevin Stiroh (1998), "Computers, Productivity, and Input Substitution", Economic Inquiry, 36, 175-191.

77

Kevin Stiroh (2001), "Information Technology and the US Productivity Revival: What Do the Industry Data Say ?", Federal Reserve Board of New York Working Paper No. 115.

Jack Triplett (1999), "The Solow Productivity Paradox: What Do Computers Do to Productivity?", Canadian Journal of Economics, 32, 309-334.

Nicholas von Tunzelmann (1978), Steam Power and British Industrialization to 1860. Oxford: Clarendon Press.

K. Whelan (2000), "Computers, Obsolescence, and Productivity", mimeo, Federal Reserve.

78