Information Technology and the U.S. Economy

Тип работы:
Реферат
Предмет:
Экономические науки


Узнать стоимость

Детальная информация о работе

Выдержка из работы

ВОПРОСЫ ТЕОРИИ
Автор предлагаемой статьи Дейл Йоргенсон является одним из наиболее значимых современных экономистов. Его блестящие работы в различных областях экономики принесли ему заслуженное уважение коллег во всем мире. В своей стране профессор Д. Йорген-сон избирался председателем Американского эконометрического общества в 1987 г. В настоящее время он является президентом Американской экономической ассоциации.
Значителен вклад профессора Д. Йоргенсона в воспитание молодых экономистов в знаменитых университетах: Беркли с 1959 по 1969 гг. и с 1969 г. — в Гарварде.
Профессор Д. Йоргенсон любезно предоставил «Экономическому журналу ВШЭ» право опубликовать свое ежегодное обращение к членам Американской экономической ассоциации от 6 января 2001 г. Одновременно с «Экономическим журналом ВШЭ» это обращение будет напечатано в «American Economic Review» (Vol. 91, № 1, р. 1−32). Во время предстоящего визита в ГУ-ВШЭ, в середине мая 2001 г., профессор Д. Йоргенсон выступит с открытой лекцией, тема которой развивает содержание данной публикации.
Information Technology and the U.S. Economy
Presidential Address to the American Economic Association, New Orleans, Louisiana, January 6, 2001 by
Dale W. Jorgenson1)
The resurgence of the American economy since 1995 has outrun all but the most optimistic expectations. Economic forecasting models have been seriously off track and growth projections have been revised to reflect a more sanguine outlook only recently2). It is not surprising that the unusual combination of more rapid
1)1 Department of Economics, Harvard University, 122 Littauer Center, Cambridge, MA 2 138−3001. I am indebted to the Program on Technology and Economic Policy at Harvard University provided financial support. I am also indebted to Kevin J. Stiroh for our joint research and his helpful comments, Jon Samuels for excellent research assistance, Mun S. Ho for the labor data, as well as useful comments. J. Steven Landefeld, Clinton McCully, and David Wass-hausen of the Bureau of Economic Analysis provided valuable data on information technology. Colleagues far too numerous to mention have contributed useful suggestions and advice. I am grateful to all of them but retain sole responsibility for any remaining deficiencies.
2) See Congressional Budget Office (2000) on official forecasts and Economics and Statistics Administration (2000), p. 60, on private forecasts.
growth and slower inflation in the 1990'-s has touched off a strenuous debate among economists about whether improvements in America'-s economic performance can be sustained.
The starting point for the economic debate is the thesis that the 1990'-s are a mirror image of the 1970'-s, when an unfavorable series of «supply shocks» led to stagflation — slower growth and higher inflation3). In this view, the development of information technology (IT) is one of a series of positive, but temporary, shocks. The competing perspective is that IT has produced a fundamental change in the U.S. economy, leading to a permanent improvement in growth prospects4).
The relentless decline in the prices of information technology equipment has steadily enhanced the role of IT investment as a source of American economic growth. Productivity growth in IT-producing industries has gradually risen in importance and a productivity revival is now underway in the rest of the economy. Despite differences in methodology and data sources, a consensus is building that the remarkable behavior of IT prices provides the key to the surge in economic growth.
In the following section I show that the foundation for the American growth resurgence is the development and deployment of semiconductors. The decline in IT prices is rooted in developments in semiconductor technology that are widely understood by technologists and economists. This technology has found its broadest applications in computing and communications equipment, but has reduced the cost of a wide variety of other products.
A substantial acceleration in the IT price decline occurred in 1995, triggered by a much sharper acceleration in the price decline of semiconductors in 1994. Although the decline in semiconductor prices has been projected to continue for at least another decade, the recent acceleration could be temporary. This can be traced to a shift in the product cycle for semiconductors from three years to two years that took place in 1995 as the consequence of intensifying competition in markets for semiconductor products.
In Section II I outline a framework for analyzing the role of information technology in the American growth resurgence. Constant quality price indexes separate the change in the performance of IT equipment from the change in price for a given level of performance. Accurate and timely computer prices have been part of the U.S. National Income and Product Accounts (NIPA) since 1985. Unfortunately, important information gaps remain, especially on trends in prices for closely related investments, such as software and communications equipment.
The cost of capital is an essential concept for capturing the economic impact of information technology prices. Swiftly falling prices provide powerful economic incentives for the substitution of IT equipment for other forms of capital and for labor services. The rate of the IT price decline is a key component of the cost of capital, required for assessing the impacts of rapidly growing stocks of computers, communications equipment, and software.
In Section III I analyze the impact of the 1995 acceleration in the information technology price decline on U.S. economic growth. I introduce a production possibility frontier that encompasses substitutions between outputs of consumption and investment goods, as well as inputs of capital and labor services. This frontier treats IT
3) Gordon (1998, 2000) — Bosworth and Triplett (2000).
4) Greenspan (2000).
equipment as part of investment goods output and the capital services from this equipment as a component of capital input.
Capital input has been the most important source of U.S. economic growth throughout the postwar period. More rapid substitution toward information technology has given much additional weight to components of capital input with higher marginal products. The vaulting contribution of capital input since 1995 has boosted growth by nearly a full percentage point. The contribution of IT accounts for more than half of this increase. Computers have been the predominant impetus to faster growth, but communications equipment and software have made important contributions as well.
The accelerated information technology price decline signals faster productivity growth in IT-producing industries. In fact, these industries have been the source of most of aggregate productivity growth throughout the 1990'-s. Before 1995 this was due to the decline of productivity growth elsewhere in the economy. The IT-producing industries have accounted for about half the surge in productivity growth since 1995, but faster growth is not limited to these industries.
I conclude that the decline in IT prices will continue for some time. This will provide incentives for the ongoing substitution of IT for other productive inputs. Falling IT prices also serve as an indicator of rapid productivity growth in IT-producing industries. However, it would be premature to extrapolate the recent acceleration in productivity growth in these industries into the indefinite future, since this depends on the persistence of a two-year product cycle for semiconductors.
In Section IV I outline research opportunities created by the development and diffusion of information technology. A voluminous and rapidly expanding business literature is testimony to the massive impact of IT on firms and product markets. Highest priority must be given to a better understanding of the markets for semiconductors. Although several models of the market for semiconductors already exist, none explains the shift from a three-year to a two-year product cycle.
The dramatic effects of information technology on capital and labor markets have already generated a substantial and growing economic literature, but many important issues remain to be resolved. For capital markets the relationship between equity valuations and growth prospects merits much further study. For labor markets more research is needed on investment in information technology and substitution among different types of labor.
I. The Information Age
The development and deployment of information technology is the foundation of the American growth resurgence. A mantra of the «new economy» — faster, better, cheaper — captures the speed of technological change and product improvement in semiconductors and the precipitous and continuing fall in semiconductor prices. The price decline has been transmitted to the prices of products that rely heavily on semiconductor technology, like computers and telecommunications equipment. This technology has also helped to reduce the cost of aircraft, automobiles, scientific instruments, and a host of other products.
Modern information technology begins with the invention of the transistor, a semiconductor device that acts as an electrical switch and encodes information in binary form. A binary digit or bit takes the values zero and one, corresponding to the off
and on positions of a switch. The first transistor, made of the semiconductor germanium, was constructed at Bell Labs in 1947 and won the Nobel Prize in Physics in 1956 for the inventors — John Bardeen, Walter Brattain, and William Shockley5).
The next major milestone in information technology was the co-invention of the integrated circuit by Jack Kilby of Texas Instruments in 1958 and Robert Noyce of Fairchild Semiconductor in 1959. An integrated circuit consists of many, even millions, of transistors that store and manipulate data in binary form. Integrated circuits were originally developed for data storage and retrieval and semiconductor storage devices became known as memory chips6
The first patent for the integrated circuit was granted to Noyce. This resulted in a decade of litigation over the intellectual property rights. The litigation and its outcome demonstrate the critical importance of intellectual property in the development of information technology. Kilby was awarded the Nobel Prize in Physics in 2000 for discovery of the integrated circuit- regrettably, Noyce died in 19 907).
A. Moore'-s Law
In 1965 Gordon E. Moore, then Research Director at Fairchild Semiconductor, made a prescient observation, later known as Moore'-s Law8 Plotting data on memory chips, he observed that each new chip contained roughly twice as many transistors as the previous chip and was released within 18−24 months of its predecessor. This implied exponential growth of chip capacity at 35−45 percent per year! Moore'-s prediction, made in the infancy of the semiconductor industry, has tracked chip capacity for thirty-five years. He recently extrapolated this trend for at least another decade9).
In 1968 Moore and Noyce founded Intel Corporation to speed the commercialization of memory chips10). Integrated circuits gave rise to microprocessors with functions that can be programmed by software, known as logic chips. Intel'-s first general purpose microprocessor was developed for a calculator produced by Busicom, a Japanese firm. Intel retained the intellectual property rights and released the device commercially in 1971.
The rapidly rising trends in the capacity of microprocessors and storage devices illustrate the exponential growth predicted by Moore'-s Law. The first logic chip in 1971 had 2,300 transistors, while the Pentium 4 released on November 20, 2000, had 42 million! Over this twenty-nine year period the number of transistors increased by thirty-four percent per year. The rate of productivity growth for the U.S. economy during this period was slower by two orders of magnitude.
5 On Bardeen, Brattain, and Shockley, see: http: //www. nobel. se/ physics/laureates/1956/.
6) Petzold (2000) provides a general reference on computers and software.
7) On Kilby, see: http: //www. nobel. se/physics/laureates/2000/. On Noyce, see: Wolfe (2000), pp. 17−65.
8) Moore (1965). Ruttan (2001), pp. 316−367, provides a general reference on the economics of semiconductors and computers. On semiconductor technology, see:
http: //euler. berkeley. edu/~esrc/csm.
9) Moore (1997).
10) Moore (1996).
B. Semiconductor Prices
Moore'-s Law captures the fact that successive generations of semiconductors are faster and better. The economics of semiconductors begins with the closely related observation that semiconductors have become cheaper at a truly staggering rate! Chart 1n) gives semiconductor price indexes constructed by Bruce T. Grimm (1998) of the Bureau of Economic Analysis (BEA) and employed in the U.S. National Income and Product Accounts since 1996. These are divided between memory chips and logic chips. The underlying detail includes seven types of memory chips and two types of logic chips.
Between 1974 and 1996 prices of memory chips decreased by a factor of 27,270 times or at 40.9 percent per year, while the implicit deflator for the gross domestic product (GDP) increased by almost 2.7 times or 4.6 percent per year! Prices of logic chips, available for the shorter period 1985 to 1996, decreased by a factor of 1,938 or 54.1 percent per year, while the GDP deflator increased by 1.3 times or 2.6 percent per year! Semiconductor price declines closely parallel Moore'-s Law on the growth of chip capacity, setting semiconductors apart from other products.
Chart 1 also reveals a sharp acceleration in the decline of semiconductor prices in 1994 and 1995. The microprocessor price decline leapt to more than ninety percent per year as the semiconductor industry shifted from a three-year product cycle to a greatly accelerated two-year cycle. This is reflected in the 2000 Update of the International Technology Road Map for Semiconductors12^ prepared by a consortium of industry associations.
C. Constant Quality Price Indexes
The behavior of semiconductor prices is a severe test for the methods used in the official price statistics. The challenge is to separate observed price changes between changes in semiconductor performance and changes in price that hold performance constant. Achieving this objective has required a detailed understanding of the technology, the development of sophisticated measurement techniques, and the introduction of novel methods for assembling the requisite information.
Ellen R. Dulberger (1993) of IBM introduced a «matched model» index for semiconductor prices. A matched model index combines price relatives for products with the same performance at different points of time. Dulberger presented constant quality price indexes based on index number formulas, including the Fisher (1922) ideal index used in the in the U.S. national accounts13). The Fisher index is the geometric average of the familiar Laspeyres and Paasche indexes.
W. Erwin Diewert (1976) defined a superlative index number as an index that exactly replicates a flexible representation of the underlying technology (or preferences). A flexible representation provides a second-order approximation to an arbitrary technology (or preferences). A.A. Konus and S.S. Byushgens (1926) first showed that the Fisher ideal index is superlative in this sense. Laspeyres and Paasche indexes are not superlative and fail to capture substitutions among products in response to price changes accurately.
11) All charts and tables see in Appendix.
12) On International Technology Roadmap for Semiconductors (2000), see: http: //public. itrs. net/.
13) See Landefeld and Parker (1997).
Grimm (1998) combined matched model techniques with hedonic methods, based on an econometric model of semiconductor prices at different points of time. A hedonic model gives the price of a semiconductor product as a function of the characteristics that determine performance, such as speed of processing and storage capacity. A constant quality price index isolates the price change by holding these characteristics of semiconductors fixed.
Beginning in 1997, the Bureau of Labor Statistics (BLS) incorporated a matched model price index for semiconductors into the Producer Price Index (PPI) and since then the national accounts have relied on data from the PPI. Reflecting long-standing BLS policy, historical data were not revised backward. Semiconductor prices reported in the PPI prior to 1997 do not hold quality constant, failing to capture the rapid semiconductor price decline and the acceleration in 1994.
D. Computers
The introduction of the Personal Computer (PC) by IBM in 1981 was a watershed event in the deployment of information technology. The sale of Intel'-s 80 868 088 microprocessor to IBM in 1978 for incorporation into the PC was a major business breakthrough for Intel14). In 1981 IBM licensed the MS-DOS operating system from the Microsoft Corporation, founded by Bill Gates and Paul Allen in 1975. The PC established an Intel/Microsoft relationship that has continued up to the present. In 1985 Microsoft released the first version of Windows, its signature operating system for the PC, giving rise to the Wintel (Windows-Intel) nomenclature for this ongoing collaboration.
Mainframe computers, as well as PC'-s, have come to rely heavily on logic chips for central processing and memory chips for main memory. However, semiconductors account for less than half of computer costs and computer prices have fallen much less rapidly than semiconductor prices. Precise measures of computer prices that hold product quality constant were introduced into the NIPA in 1985 and the PPI during the 1990'-s. The national accounts now rely on PPI data, but historical data on computers from the PPI, like the PPI data on semiconductors, do not hold quality constant.
Gregory C. Chow (1967) pioneered the use of hedonic techniques for constructing a constant quality index of computer prices in research conducted at IBM. Chow documented price declines at more than twenty percent per year during 19 601 965, providing an initial glimpse of the remarkable behavior of computer prices15). In 1985 the Bureau of Economic Analysis incorporated constant quality price indexes for computers and peripheral equipment constructed by Rosanne Cole, Y.C. Chen, Joan A. Barquin-Stolleman, Ellen R. Dulberger, Nurthan Helvacian, and James H. Hodge (1986) of IBM into the NIPA. Jack E. Triplett (1986) discussed the economic interpretation of these indexes, bringing the rapid decline of computer prices to the attention of a very broad audience.
The BEA-IBM constant quality price index for computers provoked a heated exchange between BEA and Edward F. Denison (1989), one of the founders of national accounting methodology in the 1950'-s and head of the national accounts at BEA from 1979 to 1982. Denison sharply attacked the BEA-IBM methodology and argued
14) See Moore (1996).
15) Further details are given by Berndt (1991), pp. 102−149.
vigorously against the introduction of constant quality price indexes into the national accounts16). Allan Young (1989), then Director of BEA, reiterated BEA'-s rationale for introducing constant quality price indexes.
Dulberger (1989) presented a more detailed report on her research on the prices of computer processors for the BEA-IBM project. Speed of processing and main memory played central roles in her model. Triplett (1989) provided an exhaustive survey of research on hedonic price indexes for computers. Robert J. Gordon (1989, 1990) gave an alternative model of computer prices and identified computers and communications equipment, along with commercial aircraft, as assets with the highest rates of price decline.
Chart 2 gives BEA'-s constant quality index of prices of computers and peripheral equipment and its components, including mainframes, PC'-s, storage devices, other peripheral equipment, and terminals. The decline in computer prices follows the behavior of semiconductor prices presented in Chart 1, but in much attenuated form. The 1995 acceleration in the computer price decline parallels the acceleration in the semiconductor price decline that resulted from the changeover from a three-year product cycle to a two-year cycle in 1995.
E. Communications equipment and software
Communications technology is crucial for the rapid development and diffusion of the Internet, perhaps the most striking manifestation of information technology in the American economy17). Kenneth Flamm (1989) was the first to compare the behavior of computer prices and the prices of communications equipment. He concluded that the communications equipment prices fell only a little more slowly than computer prices. Gordon (1990) compared Flamm'-s results with the official price indexes, revealing substantial bias in the official indexes.
Communications equipment is an important market for semiconductors, but constant quality price indexes cover only a portion of this equipment. Switching and terminal equipment rely heavily on semiconductor technology, so that product development reflects improvements in semiconductors. Grimm'-s (1997) constant quality price index for digital telephone switching equipment, given in Chart 3, was incorporated into the national accounts in 1996. The output of communications services in the NIPA also incorporates a constant quality price index for cellular phones.
Much communications investment takes the form of the transmission gear, connecting data, voice, and video terminals to switching equipment. Technologies such as fiber optics, microwave broadcasting, and communications satellites have progressed at rates that outrun even the dramatic pace of semiconductor development. An example is dense wavelength division multiplexing (DWDM), a technology that sends multiple signals over an optical fiber simultaneously. Installation of DWDM equipment, beginning in 1997, has doubled the transmission capacity of fiber optic cables every 6−12 months18).
16) Denison cited his 1957 paper, «Theoretical Aspects of Quality Change, Capital Consumption, and Net Capital Formation», as the definitive statement of the traditional BEA position.
17) A general reference on the Internet is Choi and Whinston (2000). On Internet indicators see: http: //www. internetindicators. com/.
18) Rashad (2000) characterizes this as the «demise» of Moore'-s Law. Hecht (1999) describes DWDM technology and provides a general reference on fiber optics.
Both software and hardware are essential for information technology and this is reflected in the large volume of software expenditures. The eleventh comprehensive revision of the national accounts, released by BEA on October 27, 1999, re-classified computer software as investment19). Before this important advance, business expenditures on software were treated as current outlays, while personal and government expenditures were treated as purchases of nondurable goods. Software investment is growing rapidly and is now much more important than investment in computer hardware.
Robert P. Parker and Grimm (2000) describe the new estimates of investment in software. BEA distinguishes among three types of software — prepackaged, custom, and own-account software. Prepackaged software is sold or licensed in standardized form and is delivered in packages or electronic files downloaded from the Internet. Custom software is tailored to the specific application of the user and is delivered along with analysis, design, and programming services required for customization. Own-account software consists of software created for a specific application. However, only price indexes for prepackaged software hold performance constant.
Parker and Grimm (2000) present a constant quality price index for prepackaged software, given in Chart 3. This combines a hedonic model of prices for business applications software and a matched model index for spreadsheet and word processing programs developed by Steven D. Oliner and Daniel D. Sichel (1994). Prepackaged software prices decline at more than ten percent per year over the period 1962−1998. Since 1998 the BEA has relied on a matched model price index for all prepackaged software from the PPI- prior to 1998 the PPI data do not hold quality constant.
BEA'-s prices for own-account software are based on programmer wage rates. This implicitly assumes no change in the productivity of computer programmers, even with growing investment in hardware and software to support the creation of new software. Custom software prices are a weighted average of prepackaged and own-account software prices with arbitrary weights of 75 percent for own-account and 25 percent for prepackaged software. These price indexes do not hold the software performance constant and present a distorted picture of software prices, as well as software output and investment.
F. Research Opportunities
The official price indexes for computers and semiconductors provide the paradigm for economic measurement. These indexes capture the steady decline in IT prices and the recent acceleration in this decline. The official price indexes for central office switching equipment and prepackaged software also hold quality constant. BEA and BLS, the leading statistical agencies in price research, have carried out much of the best work in this area. However, a critical role has been played by price research at IBM, long the dominant firm in information technology20).
It is important to emphasize that information technology is not limited to applications of semiconductors. Switching and terminal equipment for voice, data, and video communications have come to rely on semiconductor technology and the empirical evidence on prices of this equipment reflects this fact. Transmission gear em-
19) Moulton (2000) describes the 11th comprehensive revision of NIPA and the 1999 update.
20) See Chandler (2000), Table 1. 1, p. 26.
ploys technologies with rates of progress that far outstrip those of semiconductors. This important gap in our official price statistics can only be filled by constant quality price indexes for all types of communications equipment.
Investment in software is more important than investment in hardware. This was essentially invisible until BEA introduced new measures of prepackaged, custom, and own-account software investment into the national accounts in 1999. This is a crucial step in understanding the role of information technology in the American economy. Unfortunately, software prices are another statistical blind spot with only prices of prepackaged software adequately represented in the official system of price statistics. The daunting challenge that lies ahead is to construct constant quality price indexes for custom and own-account software.
II. The Role of Information Technology
At the aggregate level IT is identified with the outputs of computers, communications equipment, and software. These products appear in the GDP as investments by businesses, households, and governments along with net exports to the rest of the world. The GDP also includes the services of IT products consumed by households and governments. A methodology for analyzing economic growth must capture the substitution of IT outputs for other outputs of goods and services.
While semiconductor technology is the driving force behind the spread of IT, the impact of the relentless decline in semiconductor prices is transmitted through falling IT prices. Only net exports of semiconductors, defined as the difference between U.S. exports to the rest of the world and U.S. imports appear in the GDP. Sales of semiconductors to domestic manufacturers of IT products are precisely offset by purchases of semiconductors and are excluded from the GDP.
Constant quality price indexes, like those reviewed in the previous section, are a key component of the methodology for analyzing the American growth resurgence. Computer prices were incorporated into the NIPA in 1985 and are now part of the PPI as well. Much more recently, semiconductor prices have been included in the NIPA and the PPI. Unfortunately, evidence on the prices of communications equipment and software is seriously incomplete, so that the official price indexes are seriously misleading.
A. Output
The output data in Table 1 are based on the most recent benchmark revision of the national accounts, updated through 199 921). The output concept is similar, but not identical, to the concept of gross domestic product used by the BEA. Both measures include final outputs purchased by businesses, governments, households, and the rest of the world. Unlike the BEA concept, the output measure in Table 1 also includes imputations for the service flows from durable goods, including IT products, employed in the household and government sectors.
The imputations for services of IT equipment are based on the cost of capital for IT described in more detail below. The cost of capital is multiplied by the nominal value of IT capital stock to obtain the imputed service flow from IT products.
21) See Jorgenson and Stiroh (2000b), Appendix A, for details on the estimates of output.
In the business sector this accrues as capital income to the firms that employ these products as inputs. In the household and government sectors the flow of capital income must be imputed. This same type of imputation is used for housing in the NIPA. The rental value of renter-occupied housing accrues to real estate firms as capital income, while the rental value of owner-occupied housing is imputed to households.
Current dollar GDP in Table 1 is $ 9.8 trillions in 1999, including imputations, and real output growth averaged 3. 46 percent for the period 1948−99. These magnitudes can be compared to the current dollar value of $ 9.3 trillions in 1999 and the average real growth rate of 3. 40 percent for period 1948−99 for the official GDP. Table 1 presents the current dollar value and price indexes of the GDP and IT output. This includes outputs of investment goods in the form of computers, software, communications equipment, and non-IT investment goods. It also includes outputs of non-IT consumption goods and services as well as imputed IT capital service flows from households and governments.
The most striking feature of the data in Table 1 is the rapid price decline for computer investment, 17.1 percent per year from 1959 to 1995. Since 1995 this decline has almost doubled to 32.1 percent per year. By contrast the relative price of software has been flat for much of the period and began to fall only in the late 1980'-s. The price of communications equipment behaves similarly to the software price, while the consumption of capital services from computers and software by households and governments shows price declines similar to computer investment.
The top panel of Table 2 summarizes the growth rates of prices and quantities for major output categories for 1990−5 and 1995−9. Business investments in computers, software, and communications equipment are the largest categories of IT spending. Households and governments have also spent sizable amounts on computers, software, communications equipment and the services of information technology. Chart 4 shows that the output of software is the largest IT category as a share of GDP, followed by the outputs of computers and communications equipment.
B. Capital Services
This section presents capital estimates for the U.S. economy for the period 1948 to 199 922). These begin with BEA investment data- the perpetual inventory method generates estimates of capital stocks and these are aggregated, using service prices as weights. This approach, originated by Jorgenson and Zvi Griliches (1996), is based on the identification of service prices with marginal products of different types of capital. The service price estimates incorporate the cost of capital23).
The cost of capital is an annualization factor that transforms the price of an asset into the price of the corresponding capital input24). This includes the nominal rate of return, the rate of depreciation, and the rate of capital loss due to declining prices. The cost of capital is an essential concept for the economics of information technology25), due to the astonishing decline of IT prices given in Table 1.
22) See Jorgenson and Stiroh (2000b), Appendix B, for details on the estimates of capital input.
23) Jorgenson (2000) presents a model of capital as a factor of production. BLS (1983) describes the version of this model employed in the official productivity statistics. For a recent update, see: http: //www. bls. gov/news. release/prd3. nr0. htm. Hulten (2001) surveys the literature.
24) Jorgenson and Yun (1991), p. 7.
25) Jorgenson and Stiroh (1995), pp. 300−303.
The cost of capital is important in many areas of economics, especially in modeling producer behavior, productivity measurement, and the economics of taxation26). Many of the important issues in measuring the cost of capital have been debated for decades. The first of these is incorporation of the rate of decline of asset prices into the cost of capital. The assumption of perfect foresight or rational expectations quickly emerged as the most appropriate formulation and has been used in almost all applications of the cost of capital27).
The second empirical issue is the measurement of economic depreciation. The stability of patterns of depreciation in the face of changes in tax policy and price shocks has been carefully documented. The depreciation rates presented by Jorgen-son and Kevin J. Stiroh (2000b) summarize a large body of empirical research on the behavior of asset prices28). A third empirical issue is the description of the tax structure for capital income. This depends on the tax laws prevailing at each point of time. The resolution of these issues has cleared the way for detailed measurements of the cost of capital for all assets that appear in the national accounts, including information technology29).
The definition of capital includes all tangible assets in the U.S. economy, equipment and structures, as well as consumers'- and government durables, land, and inventories. The capital service flows from durable goods employed by households and governments enter measures of both output and input. A steadily rising proportion of these service flows are associated with investments in IT. Investments in IT by business, household, and government sectors must be included in the GDP, along with household and government IT capital services, in order to capture the full impact of IT on the U.S. economy.
Table 3 gives capital stocks from 1948 to 1999, as well as price indexes for total domestic tangible assets and IT assets — computers, software, and communications equipment. The estimate of domestic tangible capital stock in Table 3 is $ 35.4 trillions in 1999, considerably greater than the $ 27.9 trillions in fixed capital estimated by Shelby W. Herman (2000) of BEA. The most important differences reflect the inclusion of inventories and land in Table 3.
Business IT investments, as well as purchases of computers, software, and communications equipment by households and governments, have grown spectacularly in recent years, but remain relatively small. The stocks of all IT assets combined account for only 4. 35 percent of domestic tangible capital stock in 1999. Table 4 presents estimates of the flow of capital services and corresponding price indexes for 1948−99.
The difference between growth in capital services and capital stock is the improvement in capital quality. This represents the substitution towards assets with higher marginal products. The shift toward IT increases the quality of capital, since computers, software, and communications equipment have relatively high marginal products. Capital stock estimates fail to account for this increase in quality and substantially underestimate the impact of IT investment on growth.
26) Lau (2000) surveys applications of the cost of capital.
27) See, for example, Jorgenson, Gollop, and Fraumeni (1987), pp. 40−9, and Jorgenson and Griliches (1996).
28) Jorgenson and Stiroh (2000b), Table B4, pp. 196−7 give the depreciation rates employed in this study. Fraumeni (1997) describes depreciation rates used in the NIPA. Jorgenson (2000) surveys empirical studies of depreciation.
29) See Jorgenson and Yun (2001). Diewert and Lawrence (2000) survey measures of the price and quantity of capital input.
The growth of capital quality is slightly less than twenty percent of capital input growth for the period 1948−1995. However, improvements in capital quality have increased steadily in relative importance. These improvements jumped to 44.9 percent of total growth in capital input during the period 1995−99, reflecting very rapid restructuring of capital to take advantage of the sharp acceleration in the IT price decline. Capital stock has become progressively less accurate as a measure of capital input and is now seriously deficient.
Chart 5 gives the IT capital service flows as a share of gross domestic income. The second panel of Table 2 summarizes the growth rates of prices and quantities of capital inputs for 1990−5 and 1995−9. Growth of IT capital services jumps from 11. 51 percent per year in 1990−5 to 19. 41 percent in 1995−9, while growth of non-IT capital services increases from 1. 72 percent to 2. 94 percent. This reverses the trend toward slower capital growth through 1995.
C. Labor Services
This section presents estimates of labor input for the U.S. economy from 1948 to 1999. These incorporate individual data from the Censuses of Population for 1970, 1980, and 1990, as well as the annual Current Population Surveys. Constant quality indexes for the price and quantity of labor input account for the heterogeneity of the workforce across sex, employment class, age, and education levels. This follows the approach of Jorgenson, Frank M. Gollop, and Barbara M. Fraumeni (1987). The estimates have been revised and updated by Mun S. Ho and Jorgenson (2000)30).
The distinction between labor input and labor hours is analogous to the distinction between capital services and capital stock. The growth in labor quality is the difference between the growth in labor input and hours worked. Labor quality reflects the substitution of workers with high marginal products for those with low marginal products. Table 5 presents estimates of labor input, hours worked, and labor quality.
The value of labor expenditures in Table 5 is $ 5.8 trillions in 1999, 59.3 percent of the value of output. This share accurately reflects the concept of gross domestic income, including imputations for the value of capital services in household and government sectors. As shown in Table 7, the growth rate of labor input accelerated to 2. 18 percent for 1995−9 from 1. 70 percent for 1990−5. This is primarily due to the growth of hours worked, which rose from 1. 17 percent for 1990−5 to 1. 98 percent for 1995−9, as labor force participation increased and unemployment rates plummeted.
The growth of labor quality has declined considerably in the late 1990'-s, dropping from 0. 53 percent for 1990−5 to 0. 20 percent for 1995−9. This slowdown captures well-known demographic trends in the composition of the work force, as well as exhaustion of the pool of available workers. Growth in hours worked does not capture these changes in labor quality growth and is a seriously misleading measure of labor input.
III. The American Growth Resurgence
The American economy has undergone a remarkable resurgence since the mid-1990'-s with accelerating growth in output, labor productivity, and total factor productivity. The purpose of this section is to quantify the sources of growth for 1948−99 and various sub-periods. An important objective is to account for the sharp
30) See Jorgenson and Stiroh (2000b), Appendix C, for details on the estimates of labor input. Gollop (2000) discusses the measurement of labor quality.
acceleration in the level of economic activity since 1995 and, in particular, to document the role of information technology.
The appropriate framework for analyzing the impact of information technology is the production possibility frontier, giving outputs of IT investment goods as well as inputs of IT capital services. An important advantage of this framework is that prices of IT outputs and inputs are linked through the price of IT capital services. This framework successfully captures the substitutions among outputs and inputs in response to the rapid deployment of IT. It also encompasses costs of adjustment, while allowing financial markets to be modeled independently.
As a consequence of the swift advance of information technology, a number of the most familiar concepts in growth economics have been superseded. The aggregate production function heads this list. Capital stock as a measure of capital input is now longer adequate to capture the rising importance of IT. This completely obscures the restructuring of capital input that is such an important wellspring of the growth resurgence. Finally, hours worked must be replaced as a measure of labor input.
A. Production Possibility Frontier
The production possibility frontier describes efficient combinations of outputs and inputs for the economy as a whole31). Aggregate output Y consists of outputs of investment goods and consumption goods. These outputs are produced from aggregate input X, consisting of capital services and labor services. Productivity is a «Hicks-neutral» augmentation of aggregate input.
The production possibility frontier takes the form:
(1) Y (I, I, I, I, C, C) = A ¦ X (K, K, K, K, L),
n c s t n c n c s t
where the outputs include non-IT investment goods In and investments in computers Ic, software Is, and communications equipment It, as well as non-IT consumption goods and services Cn and IT capital services to households and governments Cc. Inputs include non-IT capital services Kn and the services of computers Kc, software Ks, and telecommunications equipment Kt, as well as labor input L32). Total factor productivity (TFP) is denoted by A.
The most important advantage of the production possibility frontier is the explicit role that it provides for constant quality prices of IT products. These are used as deflators for nominal expenditures on IT investments to obtain the quantities of IT outputs. Investments in IT are cumulated into stocks of IT capital. The flow of IT capital services is an aggregate of these stocks with service prices as weights. Similarly, constant quality prices of IT capital services are used in deflating the nominal values of consumption of these services.
Another important advantage of the production possibility frontier is the incorporation of costs of adjustment. For example, an increase in the output of IT investment goods requires foregoing part of the output of consumption goods and non-IT investment goods, so that adjusting the rate of investment in IT is costly. However, costs
31) The production possibility frontier was introduced into productivity measurement by Jorgenson (1996), pp. 27−28.
32) Services of durable goods to governments and households are included in both inputs and outputs.
of adjustment are external to the producing unit and are fully reflected in IT prices. These prices incorporate forward-looking expectations of the future prices of IT capital services.
B. Aggregate Production Function
The aggregate production function employed by Robert M. Solow (1957, 1960) and, more recently, by Jeremy Greenwood, Zvi Hercowitz, and Per Krusell (1997, 2000), Hercowitz (1998), and Arnold C. Harberger (1998) is a competing methodology. The production function gives a single output as a function of capital and labor inputs. There is no role for separate prices of investment and consumption goods and, hence, no place for constant quality IT price indexes for outputs of IT investment goods.
Greenwood, Hercowitz, and Krusell employ a price index for consumption to deflate the output of all investment goods, including information technology.
Confronted by the fact that constant quality prices of investment goods differ from consumption goods prices, they borrow the concept of embodiment from Solow (1960) in order to convert investment goods output into an appropriate form for measuring capital stock33). Investment has two prices, one used in the measuring output and the other used in measuring capital stock. This inconsistency can be removed by simply distinguishing between outputs of consumption and investment goods, as in the national accounts and Equation (1). The concept of embodiment can then be dropped.
Perhaps inadvertently, Greenwood, Hercowitz, and Krussell have revisited the controversy accompanying the introduction of a constant quality price index for computers into the national accounts. They have revived Denison'-s (1993) proposal to use a consumption price index to deflate investment in the NIPA. Denison found this appealing as a means of avoiding the introduction of constant quality price indexes for computers. Denison'-s approach leads to a serious underestimate of GDP growth and an overestimate of inflation.
Another limitation of the aggregate production function is that it fails to incorporate costs of adjustment. Robert E. Lucas, Jr., (1967) presented a production model with internal costs of adjustment. Fumio Hayashi (2000) shows how to identify these adjustment costs from James Tobin'-s (1969) Q-ratio, the ratio of the stock market value of the producing unit to the market value of the unit'-s assets. Implementation of this approach requires simultaneous modeling of production and asset valuation. If costs of adjustment are external, as in the production possibility frontier (1), asset valuation can be modeled separately from production34).
C. Sources of Growth
Under the assumption that product and factor markets are competitive producer equilibrium implies that the share-weighted growth of outputs is the sum of the share-weighted growth of inputs and growth in total factor productivity:
(2) w A ln I + w AI + w AI + w AI + w C + w A ln C =
I, n n I, c c I, s s I, t t C, n n C, c c
v A ln K + v A ln K + v A ln K + v A ln K + v A ln L + A InA,
K, n n K, c c K, s s K, t t L
33) Whelan (1999) also employs Solow'-s concept of embodiment.
34) See, for example, Campbell and Shiller (1998).
where w and v denote average value shares. The shares of outputs and inputs add to one under the additional assumption of constant returns,
w + w + w + w + w + w = v + v + v + v + v = 1.
I, n I, c I, s I, t C, n C, c K, n K, c K, s K, t L
Equation (2) makes it possible to identify the contributions of outputs as well as inputs to U.S. economic growth. The growth rate of output is a weighted average of growth rates of investment and consumption goods outputs. The contribution of each output is its weighted growth rate. Similarly, the growth rate of input is a weighted average of growth rates of capital and labor services and the contribution of each input is its weighted growth rate. The contribution of TFP, the growth rate of the augmentation factor A in Equation (2), is the difference between growth rates of output and input.
Table 6 presents results of a growth accounting decomposition, based on Equation (2), for the period 1948−99 and various sub-periods, following Jorgenson and Stroh (1999, 2000b). Economic growth is broken down by output and input categories, quantifying the contribution of information technology to investment and consumption outputs, as well as capital inputs. These estimates identify computers, software, and communications equipment as distinct types of information technology.
Rearranging Equation (2), the results can be presented in terms of average labor productivity (ALP), defined as y = Y/H, the ratio of output Y to hours worked
H, and k = K/H is the ratio of capital services K to hours worked: (3) Dlny = vKDlnk + vL (AlnL -DlnH)+DlnA.
Equation (3) allocates ALP growth among three sources. The first is capital deepening, the growth in capital input per hour worked, and reflects the capital-labor substitution. The second is improvement in labor quality and captures the rising proportion of hours by workers with higher marginal products. The third is TFP growth, which contributes point-for-point to ALP growth.
D. Contributions of IT Investment
Chart 5 depicts the rapid increase in the importance of IT services, reflecting the accelerating pace of IT price declines. In 1995−9 the capital service price for computers fell 24. 81 percent per year, compared to an increase of 36. 36 percent in capital input from computers. As a consequence, the value of computer services grew substantially. However, the current dollar value of computers was only 1.6 percent of gross domestic income in 1999.
The rapid accumulation of software appears to have different sources. The price of software services has declined only 2. 04 percent per year for 1995−9. Nonetheless, firms have been accumulating software very rapidly, with real capital services growing 16. 30 percent per year. A possible explanation is that firms respond to computer price declines by investing in complementary inputs like software. However, a more plausible explanation is that the price indexes used to deflate software investment fail to hold quality constant. This leads to an overstatement of inflation and an understatement of growth.
Although the price decline for communications equipment during the period 1995−9 is comparable to that of software, investment in this equipment is more in line
with prices. However, prices of communications equipment also fail to hold quality constant. The technology of switching equipment, for example, is similar to that of computers- investment in this category is deflated by a constant-quality price index developed by BEA. Conventional price deflators are employed for transmission gear, such as fiber-optic cables. This leads to an underestimate of the growth rates of investment, capital stock, capital services, and the GDP, as well as an overestimate of the rate of inflation.
Charts 6 and 7 highlight the rising contributions IT outputs to U.S. economic growth. Chart 6 shows the breakdown between IT and non-IT outputs for sub-periods from 1948 to 1999, while Chart 7 decomposes the contribution of IT into its components. Although the importance of IT has steadily increased, Chart 6 shows that the recent investment and consumption surge nearly doubled the output contribution of IT. Chart 7 shows that computer investment is the largest single IT contributor in the late 1990'-s, but that investments in software and communications equipment are becoming increasingly important.
Charts 8 and 9 present a similar decomposition of IT inputs into production. The contribution of these inputs is rising even more dramatically. Chart 8 shows that the contribution of IT now accounts for more than 48.1 percent of the total contribution of capital input. Chart 9 shows that computer hardware is the largest IT contributor on the input side, reflecting the growing share and accelerating growth rate of computer investment in the late 1990'-s.
Private business investment predominates in the output of IT, as shown by Jor-genson and Stiroh (1999, 2000b)35). Household purchases of IT equipment and services are next in importance. Government purchases of IT equipment and services, as well as net exports of IT products, must be included in order to provide a complete picture. Firms, consumers, governments, and purchasers of U.S. exports are responding to relative price changes, increasing the contributions of computers, software, and communications equipment.
Table 2 shows that the price of computer investment fell by more than 32 percent per year, the price of software 2.4 percent, and the price of communications equipment 2.9 percent, and the price of IT services 11.8 percent during the period 1995−9, while non-IT prices rose 2.2 percent. In response to these price changes, firms, households, and governments have accumulated computers, software, and communications equipment much more rapidly than other forms of capital.
E. Total Factor Productivity
The price or «dual» approach to productivity measurement makes it possible to identify the role of IT production as a source of productivity growth at the industry level36). The rate of productivity growth is measured as the decline in the price of output, plus a weighted average of the growth rates of input prices with value shares of the inputs as weights. For the computer industry this expression is dominated by two terms: the decline in the price of computers and the contribution of the price of semiconductors. For the semiconductor industry the expression is dominated by the decline in the price of semiconductors37).
35) Bosworth and Triplett (2000) compare the results of Jorgenson and Stiroh (2000b) with those of Oliner and Sichel (2000).
36) The dual approach is presented by Jorgenson, Gollop, and Fraumeni (1987), pp. 53−63.
37) Dulberger (1993), Triplett (1996), and Oliner and Sichel (2000) present models of the relationships between computer and semiconductor industries. These are special cases of the Domar (1961) aggregation scheme.
Jorgenson, Gollop, and Fraumeni (1987) have employed Evsey Domar'-s (1961) model to trace aggregate productivity growth to its sources at the level of individual industries38). More recently, Harberger (1998), William Gullickson and Michael J. Harper (1999) and Jorgenson and Stiroh (2000a, 2000b) have used the model for similar purposes. Productivity growth for each industry is weighted by the ratio of the gross output of the industry to GDP to estimate the industry contribution to aggregate TFP growth.
If semiconductor output were only used to produce computers, then its contribution to computer industry productivity growth, weighted by computer industry output, would precisely cancel its independent contribution to aggregate TFP growth. This is the ratio of the value of semiconductor output to GDP, multiplied by the rate of semiconductor price decline. In fact, semiconductors are used to produce telecommunications equipment and many other products. However, the value of semiconductor output is dominated by inputs into IT production.
The Domar aggregation formula can be approximated by expressing the declines in prices of computers, communications equipment, and software relative to the price of gross domestic income, an aggregate of the prices of capital and labor services. The rates of relative IT price decline are weighted by ratios of the outputs of IT products to the GDP. Table 8 reports details of this TFP decomposition for 1990−5 and 1995−9- the IT and non-IT contributions are presented in Chart 10. The IT products contribute 0. 50 percentage points to TFP growth for 1995−9, compared to 0. 25 percentage points for 1990−5. This reflects the accelerating decline in relative price changes resulting from shortening the product cycle for semiconductors.
F. Output Growth
This section presents the sources of GDP growth for the entire period 1948 to 1999. Capital services contribute 1. 70 percentage points, labor services 1. 14 percentage points, and TFP growth only 0. 61 percentage points. Input growth is the source of nearly 82.3 percent of U.S. growth over the past half century, while TFP has accounted for 17.7 percent. Chart 11 shows the relatively modest contributions of TFP in all sub-periods.
More than three-quarters of the contribution of capital reflects the accumulation of capital stock, while improvement in the quality of capital accounts for about one-quarter. Similarly, increased labor hours account for 80 percent of labor'-s contribution- the remainder is due to improvements in labor quality. Substitutions among capital and labor inputs in response to price changes are essential components of the sources of economic growth.
A look at the U.S. economy before and after 1973 reveals familiar features of the historical record. After strong output and TFP growth in the 1950'-s, 1960'-s and early 1970'-s, the U.S. economy slowed markedly through 1990, with output growth falling from 3. 99 percent to 2. 86 percent and TFP growth declining from 0. 92 percent to 0. 25 percent. Growth in capital inputs also slowed from 4. 64 percent for 1948−73 to 3. 57 percent for 1973−90. This contributed to sluggish ALP growth — 2. 82 percent for 194 873 and 1. 26 percent for 1973−90.
Relative to the early 1990'-s, output growth increased by 1. 72 percent in 1995−9. The contribution of IT production almost doubled, relative to 1990−5, but still accoun-
38) See Jorgenson, Gollop, and Fraumeni (1987), pp. 63−66, 301−322.
ted for only 28.9 percent of the increased growth of output. Although the contribution of IT has increased steadily throughout the period 1948−99, there has been a sharp response to the acceleration in the IT price decline in 1995. Nonetheless, more than 70 percent of the increased output growth can be attributed to non-IT products.
Between 1990−5 and 1995−9 the contribution of capital input jumped by 0. 95 percentage points, the contribution of labor input rose by only 0. 24 percent, and TFP accelerated by 0. 51 percent. Growth in ALP rose 0. 92 as more rapid capital deepening and growth in TFP offset slower improvement in labor quality. Growth in hours worked accelerated as unemployment fell to a 30-year low. Labor markets have tightened considerably, even as labor force participation rates increased39).
The contribution of capital input reflects the investment boom of the late 1990'-s as businesses, households, and governments poured resources into plant and equipment, especially computers, software, and communications equipment. The contribution of capital, predominantly IT, is considerably more important than the contribution of labor. The contribution of IT capital services has grown steadily throughout the period 1948−99, but Chart 9 reflects the impact of the accelerating decline in IT prices.
After maintaining an average rate of 0. 25 percent for the period 1973−90, TFP growth fell to 0. 24 percent for 1990−5 and then vaulted to 0. 75 percent per year for 1995−9. This is a major source of growth in output and ALP for the U.S. economy (Charts 11 and 12). While TFP growth for 1995−9 is lower than the rate of 1948−73, the U.S. economy is recuperating from the anemic productivity growth of the past two decades. Although only half of the acceleration in TFP from 1990−5 to 1995−9 can be attributed to IT production, this is far greater than the 4. 26 percent share of IT in the GDP.
G. Average Labor Productivity
Output growth is the sum of growth in hours and average labor productivity. Table 7 shows the breakdown between growth in hours and ALP for the same periods as in Table 6. For the period 1948−99, ALP growth predominated in output growth, increasing just over 2 percent per year for 1948−99, while hours increased about 1.4 percent per year. As shown in Equation (3), ALP growth depends on capital deepening, a labor quality effect, and TFP growth.
Chart 12 reveals the well-known productivity slowdown of the 1970'-s and 1980'-s, emphasizing the acceleration in labor productivity growth in the late 1990'-s. The slowdown through 1990 reflects reduced capital deepening, declining labor quality growth, and decelerating growth in TFP. The growth of ALP slipped further during the early 1990'-s with a slump in capital deepening only partly offset by a revival in labor quality growth and an up-tick in TFP growth. A slowdown in hours combined with slowing ALP growth during 1990−5 to produce a further slide in the growth of output. In previous cyclical recoveries during the postwar period, output growth accelerated during the recovery, powered by more rapid growth of hours and ALP.
Accelerating output growth during 1995−9 reflects growth in labor hours and ALP almost equally40^ Comparing 1990−5 to 1995−9, the rate of output growth jumped by 1. 72 percent — due to an increase in hours worked of 0. 81 percent and another
39) Katz and Krueger (1999) analyze the recent performance of the U.S. labor market.
40) Stiroh (2000) shows that ALP growth is concentrated in IT-producing and IT-using industries.
increase in ALP growth of 0. 92 percent. Chart 12 shows the acceleration in ALP growth is due to capital deepening as well as faster TFP growth. Capital deepening contributed 0. 60 percentage points, offsetting a negative contribution of labor quality of 0. 20 percent. The acceleration in TFP added 0. 51 percentage points.
H. Research Opportunities
The use of computers, software, and communications equipment must be carefully distinguished from the production of IT41). Massive increases in computing power, like those experienced by the U.S. economy, have two effects on growth. First, as IT producers become more efficient, more IT equipment and software is produced from the same inputs. This raises productivity in IT-producing industries and contributes to TFP growth for the economy as a whole. Labor productivity also grows at both industry and aggregate levels.
Second, investment in information technology leads to growth of productive capacity in IT-using industries. Since labor is working with more and better equipment, this increases ALP through capital deepening. If the contributions to aggregate output are captured by capital deepening, aggregate TFP growth is unaffected42). Increasing deployment of IT affects TFP growth only if there are spillovers from IT-producing industries to IT-using industries.
Top priority must be given to identifying the impact of investment in IT at the industry level. Stiroh (1998) has shown that this is concentrated in a small number of IT-using industries, while Stiroh (2000) shows that aggregate ALP growth can be attributed to productivity growth in IT-producing and IT-using industries. The next priority is to trace the increase in aggregate TFP growth to its sources in individual industries. Jorgenson and Stiroh (2000a, 2000b) present the appropriate methodology and preliminary results.
IV. Economics on Internet Time
The steadily rising importance of information technology has created new research opportunities in all areas of economics. Economic historians, led by Alfred D. Chandler (2000) and Paul A. David (2 0 00)43), have placed the information age in historical context. The Solow (1987) Paradox, that we see computers everywhere but in the productivity statistics44), has provided a point of departure. Since computers have now left an indelible imprint on the productivity statistics, the remaining issue is whether the breathtaking speed of technological change in semiconductors differentiates this resurgence from previous periods of rapid growth?
Capital and labor markets have been severely impacted by information technology. Enormous uncertainty surrounds the relationship between equity valuations and future growth prospects of the American economy45^ One theory attributes ris-
41) Economics and Statistics Administration (2000), Table 3. 1, p. 23, lists IT-producing industries.
42) Baily and Gordon (1988).
43) See also: David (1990) and Gordon (2000).
44) Griliches (1994), Brynjolfsson and Yang (1996), and Triplett (1999) discuss the Solow Paradox.
45) Campbell and Shiller (1998) and Shiller (2000) discuss equity valuations and growth prospects. Kiley (1999), Brynjolfsson and Hitt (2000), and Hall (2000), present models of investment with internal costs of adjustment.
ing valuations of equities since the growth acceleration began in 1995 to the accumulation of intangible assets, such as intellectual property and organizational capital. An alternative theory treats the high valuations of technology stocks as a bubble that burst during the year 2000.
The behavior of labor markets also poses important puzzles. Widening wage differentials between workers with more and less education has been attributed to computerization of the workplace. A possible explanation could be that high-skilled workers are complementary to IT, while low-skilled workers are substitutable. An alternative explanation is that technical change associated with IT is skill-biased and increases the wages of high-skilled workers relative to low-skilled workers46).
Finally, information technology is altering product markets and business organizations, as attested by the large and growing business literature47), but a fully satisfactory model of the semiconductor industry remains to be developed48). Such a model would derive the demand for semiconductors from investment in information technology in response to rapidly falling IT prices. An important objective is to determine the product cycle for successive generations of new semiconductors endogenously.
The semiconductor industry and the information technology industries are global in their scope with an elaborate international division of labor49). This poses important questions about the American growth resurgence. Where is the evidence of a new economy in other leading industrialized countries? An important explanation is the absence of constant quality price indexes for semiconductors and information technology in national accounting systems outside the U.S. 50) Another conundrum is that several important participants — Korea, Malaysia, Singapore, and Taiwan — are «newly industrializing» economies. What does this portend for developing countries like China and India?
As policy-makers attempt to fill the widening gaps between the information required for sound policy and the available data, the traditional division of labor between statistical agencies and policy-making bodies is breaking down.
In the mean time monetary policy-makers must set policies without accurate measures of price change. Similarly, fiscal policy-makers confront on-going revisions of growth projections that drastically affect the outlook for future tax revenues and government spending.
The stagflation of the 1970'-s greatly undermined the Keynesian Revolution, leading to a New Classical Counter-revolution led by Lucas (1981) that has transformed macroeconomics. The unanticipated American growth revival of the 1990'-s has similar potential for altering economic perspectives. In fact, this is already foreshadowed in a steady stream of excellent books on the economics of information technology51). We are the fortunate beneficiaries of a new agenda for economic research that could refresh our thinking and revitalize our discipline.
46) Acemoglu (2000) and Katz (2000) survey the literature on labor markets and technological change.
47) See, for example, Grove (1996) on the market for computers and semiconductors and Christensen (1997) on the market for storage devices.
48) Irwin and Klenow (1994), Flamm (1996), pp. 305−424, and Helpman and Trajtenberg (1998), pp. 111−119, present models of the semiconductor industry.
49) The role of information technology in U.S. economic growth is discussed by the Economics and Statistics Administration (2000) — comparisons among OECD countries are given by the Organisation for Economic Co-operation and Development (2000).
50) The measurement gap between the U.S. and other OECD countries was first identified by Wykoff (1995). Schreyer (2000) has taken the initial steps to fill this gap.
51) See, for example, Shapiro and Varian (1999), Brynjolfsson and Kahin (2000), and Choi and Whinston (2000).
* * *
REFERENCES
1. Acemoglu, Daron, & quot-Technical Change, Inequality, and the Labor Market,& quot- Cambridge, MA: Department of Economics, Massachusetts Institute of Technology, July 2000.
2. Baily, Martin N., and Gordon, Robert J., & quot-The Productivity Slowdown, Measurement Issues, and the Explosion of Computer Power,& quot- Brookings Papers on Economic Activity, 1988, 2, pp. 347−420.
3. Berndt, Ernst R., The Practice of Econometrics: Classic and Contemporary, Reading, MA: Addison-Wesley, 1991.
4. Bosworth, Barry P., and Triplett, Jack E., & quot-What'-s New About the New Economy? IT, Growth and Productivity,& quot- Washington, DC: The Brookings Institution, October 20, 2000.
5. Brynjolfsson, Erik, and Hitt, Lorin M., & quot-Beyond Computation: Information Technology, Organizational Transformation and Business Performance,& quot- Journal of Economic Perspectives, 14(4), Fall 2000, pp. 23−48.
6. Brynjolfsson, Erik, and Kahin, Brian, Understanding the Digital Economy, Cambridge, MA: The MIT Press, 2000.
7. Brynjolfsson, Erik, and Yang, Shinkyu, & quot-Information Technology and Productivity: A Review of the Literature,& quot- Advances in Computers, 43(1), February 1996, pp. 179−214.
8. Bureau of Labor Statistics, Trends in Multifactor Productivity, 1948−1981, Washington, DC: U.S. Government Printing Office, 1983.
9. Campbell, John Y., and Shiller, Robert J., & quot-Valuation Ratios and the Long-run Stock Market Outlook,& quot- Journal of Portfolio Management, 24(2), Winter 1998, pp. 11−26.
10. Chandler, Alfred D., Jr., & quot-The Information Age in Historical Perspective,& quot- in Alfred D. Chandler and James W. Cortada, A Nation Transformed by Information: How Information Has Shaped the United States from Colonial Times to the Present, New York, Oxford University Press, 2000, pp. 3−38.
11. Choi, Soon-Yong, and Whinston, Andrew B. The Internet Economy: Technology and Practice, Austin, TX: SmartEcon Publishing, 2000.
12. Chow, Gregory C, & quot-Technological Change and the Demand for Computers,& quot- American Economic Review, 57(5), December 1967, pp. 1117−30.
13. Christensen, Clayton M., The Innovator'-s Dilemma, Boston, Harvard Business School Press, 1997.
14. Cole, Rosanne, Chen, Y.C., Barquin-Stolleman, Joan A., Dulberger, Ellen R. Hel-vacian, Nurthan, and Hodge, James H., & quot-Quality-Adjusted Price Indexes for Computer Processors and Selected Peripheral Equipment,& quot- Survey of Current Business, 66(1), January 1986, pp. 41−50.
15. Congressional Budget Office, The Budget and Economic Outlook: An Update, Washington, DC: U.S. Government Printing Office, July 2000.
16. Council of Economic Advisers, Annual Report, Washington, DC: U.S. Government Printing Office, February 2000.
17. David, Paul A., & quot-The Dynamo and the Computer: An Historical Perspective on the Productivity Paradox,& quot- American Economic Review, 80(2), May 1990, pp. 355−61.
18. _, & quot-Understanding Digital Technology'-s Evolution and the Path of Measured
Productivity Growth: Present and Future in the Mirror of the Past,& quot- in Brynjolfsson and Ka-hin (2000), pp. 49−98.
19. Denison, Edward F., & quot-Theoretical Aspects of Quality Change, Capital Consumption, and Net Capital Formation,& quot- in Conference on Research in Income and Wealth, Problems of Capital Formation, Princeton, NJ: Princeton University Press, 1957, pp. 215−61.
20. _, Estimates of Productivity Change by Industry, Washington, DC: Brookings
Institution Press, 1989.
21. _, & quot-Robert J. Gordon'-s Concept of Capital,& quot- Review of Income and Wealth,
39(1), March 1993, pp. 89−102.
22. Diewert, W. Erwin, & quot-Exact and Superlative Index Numbers,& quot- Journal of Econometrics, 4(2), May 1976, pp. 115−46.
23. Diewert, W. Erwin, and Lawrence, Denis A., & quot-Progress in Measuring the Price and Quantity of Capital,& quot- in Lau (2000), pp. 273−326.
24. Domar, Evsey, & quot-On the Measurement of Technological Change,& quot- Economic Journal, 71(284), December 1961, pp. 709−29.
25. Dulberger, Ellen R., & quot-The Application of a Hedonic Model to a Quality-Adjusted Price Index for Computer Processors,& quot- in Jorgenson and Landau (1989), pp. 37−76.
26. _, & quot-Sources of Decline in Computer Processors: Selected Electronic Components,& quot- in Murray F. Foss, Marilyn E. Manser, and Allan H. Young, eds., Price Measurements and Their Uses, Chicago: University of Chicago Press, 1993, pp. 103−24.
27. Economics and Statistics Administration, Digital Economy 2000, Washington, DC: U.S. Department of Commerce, June 2000.
28. Fisher, Irving, The Making of Index Numbers, Boston: Houghton-Mifflin, 1922.
29. Flamm, Kenneth, & quot-Technological Advance and Costs: Computers versus Communications,& quot- in Robert C. Crandall and Kenneth Flamm, eds., Changing the Rules: Technological Change, International Competition, and Regulation in Communications, Washington, DC: Brookings Institution Press, 1989, pp. 13−61.
30. _, Mismanaged Trade? Strategic Policy and the Semiconductor Industry,
Washington, DC: Brookings Institution Press, 1996.
31. Fraumeni, Barbara M., & quot-The Measurement of Depreciation in the U.S. National Income and Product Accounts,& quot- Survey of Current Business, 77(7), July 1997, pp. 7−23.
32. Gollop, Frank M, & quot-The Cost of Capital and the Measurement of Productivity,& quot- in Lau (2000), pp. 85−110.
33. Gordon, Robert J., & quot-The Postwar Evolution of Computer Prices,& quot- in Jorgenson and Landau (1989), pp. 77−126.
34. _, The Measurement of Durable Goods Prices, Chicago, University of Chicago
Press, 1990.
35. _, & quot-Foundations of the Goldilocks Economy: Supply Shocks and the Time-
Varying NAIRU,& quot- Brookings Papers on Economic Activity, 1998, 2, pp. 297−333.
36. _, & quot-Does the '-New Economy'- Measure Up to the Great Inventions of the Past,& quot-
Journal of Economic Perspectives, 14(4), Fall 2000, pp. 49−74.
37. Greenspan, Alan, & quot-Challenges for Monetary Policy-Makers,& quot- Washington, DC: Board of Governors of the Federal Reserve System, October 19, 2000.
38. Greenwood, Jeremy, Hercowitz, Zvi, and Krusell, Per, & quot-Long-run Implications of Investment-specific Technological Change,& quot- American Economic Review, 87(3), June 1997, pp. 342−62.
39. _,_, and_, & quot-The Role of Investment-specific Technological Change in
the Business Cycle,& quot- European Economic Review, 44(1), January 2000, pp. 91−115.
40. Griliches, Zvi, & quot-Productivity, R& amp-D, and the Data Constraint,& quot- American Economic Review, 94(2), March 1994, pp. 1−23.
41. Grimm, Bruce T, & quot-Quality Adjusted Price Indexes for Digital Telephone Switches,& quot- Washington, DC: Bureau of Economic Analysis, May 20, 1997.
42. _, & quot-Price Indexes for Selected Semiconductors: 1974−96,& quot- Survey of Current
Business, February 1998, 78(2), pp. 8−24.
43. Grove, Andrew S., Only the Paranoid Survive: How to Exploit the Crisis Points that Challenge Every Company, New York, Doubleday, 1996.
44. Gullickson, William, and Harper, Michael J., & quot-Possible Measurement Bias in Aggregate Productivity Growth,& quot- Monthly Labor Review, 122(2), February 1999, pp. 47−67.
45. Hall, Robert E., & quot-e-Capital: The Link between the Stock Market and the Labor Market in the 1990'-s,& quot- Brookings Papers on Economic Activity, 2000, 2, forthcoming.
46. Harberger, Arnold C., & quot-A Vision of the Growth Process,& quot- American Economic Review, 88(1), March 1998, pp. 1−32.
47. Hayashi, Fumio, & quot-The Cost of Capital, Q, and the Theory of Investment Demand,& quot- in Lau (2000), pp. 55−84.
48. Hecht, Jeff, City of Light, New York: Oxford University Press, 1999a.
49. Helpman, Elhanan, and Trajtenberg, Manuel, & quot-Diffusion of General Purpose Technologies,& quot- in Elhanan Helpman, ed., General Purpose Technologies and Economic Growth, Cambridge, MA: The MIT Press, pp. 85−120.
50. Hercowitz, Zvi, & quot-The '-Embodiment'- Controversy: A Review Essay,& quot- Journal of Monetary Economics, 41(1), February 1998, pp. 217−24.
51. Herman, Shelby W., & quot-Fixed Assets and Consumer Durable Goods for 1925−99,& quot- Survey of Current Business, 80(9), September 2000, pp. 19−30.
52. Ho, Mun S, and Jorgenson, Dale W., & quot-The Quality of the U.S. Workforce, 1948−99,& quot- Cambridge, MA: Department of Economics, Harvard University, 2000.
53. Hulten, Charles R., & quot-Total Factor Productivity: A Short Biography,& quot- in Charles R. Hulten, Edwin R. Dean, and Michael J. Harper, eds., New Developments in Productivity Analysis, Chicago, University of Chicago Press, 2001, forthcoming.
54. International Technology Roadmap for Semiconductors, 2000 Update, Austin, TX: Sematech Corporation, December 2000.
55. Irwin, Douglas A., and Klenow, Peter J., & quot-Learning-by-Doing Spillovers in the Semiconductor Industry,& quot- Journal of Political Economy, 102(6), December 1994, pp. 1200−27.
56. Jorgenson, Dale W., Postwar U.S. Economic Growth, Cambridge, MA: The MIT Press, 1996.
57. _, Capital Theory and Investment Behavior, Cambridge, MA: The MIT Press,
1997.
58. _, Econometrics and Producer Behavior, Cambridge, MA: The MIT Press, 2000.
59. Jorgenson, Dale W., Gollop, Frank M., and Fraumeni, Barbara M., Productivity and U.S. Economic Growth, Cambridge, MA: Harvard University Press, 1987.
60. Jorgenson, Dale W., and Griliches, Zvi, & quot-The Explanation of Productivity Change,& quot- in Jorgenson (1996), pp. 51−98.
61. Jorgenson, Dale W., and Landau, Ralph, eds., Technology and Capital Formation, Cambridge, MA: The MIT Press, 1989.
62. Jorgenson, Dale W., and Stiroh, Kevin J., & quot-Computers and Growth,& quot- Economics of Innovation and New Technology, 3(3−4), 1995, pp. 295−316.
63. _ and _, & quot-Information Technology and Growth,& quot- American Economic Review, 89(2), May 1999, pp. 109−15.
64. _and_, & quot-U.S. Economic Growth and the Industry Level,& quot- American Economic Review, 90(2), May 2000a, pp. 161−7.
65. _ and _, & quot-Raising the Speed Limit: U.S. Economic Growth in the Information Age,& quot- Brookings Papers on Economic Activity, 2000b, 1, pp. 125−211.
66. Jorgenson, Dale W., and Yun, Kun-Young, Tax Reform and the Cost of Capital, New York: Oxford University Press, 1991.
67. _ and _, Lifting the Burden: Tax Reform, the Cost of Capital, and U.S.
Economic Growth, Cambridge, MA: The MIT Press, 2001.
68. Katz, Lawrence F., & quot-Technological Change, Computerization, and the Wage Structure,& quot- in Brynjolfsson and Kahin (2000), pp. 217−44.
69. Katz, Lawrence F., and Krueger, Alan, & quot-The High-Pressure U.S. Labor Market of the 1990'-s & quot- Brookings Papers on Economic Activity, 1999, 1, pp. 1−87.
70. Kiley, Michael T., & quot-Computers and Growth with Costs of Adjustment: Will the Future Look Like the Past?& quot- Washington, DC: Board of Governors of the Federal Reserve System, July 1999.
71. Konus, A. A., and Byushgens, S. S, & quot-On the Problem of the Purchasing Power of Money,& quot- Economic Bulletin of the Conjuncture Institute, Supplement, 1926, pp. 151−72.
72. Landefeld, J. Steven, and Parker, Robert P., & quot-BEA'-s Chain Indexes, Time Series, and Measures of Long-Term Growth,& quot- Survey of Current Business, 77(5), May 1997, pp. 58−68.
73. Lau, Lawrence J., ed., Econometrics and the Cost of Capital, Cambridge, MA: The MIT Press, 2000.
74. Lucas, Robert E., Jr., & quot-Adjustment Costs and the Theory of Supply,& quot- Journal of Political Economy, 75(4), part 1, August 1967, pp. 321−34.
75. _, Studies in Business-Cycle Theory, Cambridge, MA: The MIT Press, 1981.
76. Moore, Gordon E., & quot-Cramming More Components onto Integrated Circuits,& quot- Electon-ics, 38(8), April 19, 1965, pp. 114−7.
77. _, & quot-Intel — Memories and the Microprocessor,& quot- Daedalus, 125(2), Spring 1996,
pp. 55−80.
78. _, & quot-An Update on Moore'-s Law,& quot- Santa Clara, CA: Intel Corporation, September
30, 1997.
79. Moulton, Brent R., & quot-Improved Estimates of the National Income and Product Accounts for 1929−99: Results of the Comprehensive Revision& quot- Survey of Current Business, 80(4), April 2000, pp. 11−7, 36−145.
80. Organisation for Economic Co-operation and Development, A New Economy? Paris: Organisation for Economic Co-operation and Development, 2000.
81. Oliner, Stephen D., and Sichel, Daniel E., & quot-Computers and Output Growth Revisited: How Big is the Puzzle?& quot- Brookings Papers on Economic Activity, 1994, 2, pp. 273−317.
82. _ and _, & quot-The Resurgence of Growth in the Late 1990'-s: Is Information
Technology the Story?& quot- Journal of Economic Perspectives, 14(4), Fall 2000, pp. 3−22.
83. Parker, Robert P., and Grimm, Bruce T., & quot-Recognition of Business and Government Expenditures on Software as Investment: Methodology and Quantitative Impacts, 1959−98,& quot- Washington, DC: Bureau of Economic Analysis, November 14, 2000.
84. Petzold, Charles, Code: The Hidden Language of Computer Hardware and Software, Redmond, WA: Microsoft Press, 1999.
85. Rashad, Rick, & quot-The Future — It Isn'-t What It Used to Be,& quot- Seattle, WA: Microsoft Research, May 3, 2000.
86. Ruttan, Vernon W., & quot-The Computer and Semiconductor Industries,& quot- Technology, Growth, and Development, New York: Oxford University Press, 2001, pp. 316−67.
87. Schreyer, Paul, & quot-The Contribution of Information and Communication Technology to Output Growth: A Study of the G7 Countries,& quot- Paris: Organisation for Economic Co-operation and Development, May 23, 2000.
88. Shapiro, Carl, and Varian, Hal R., Information Rules, Boston: Harvard Business School Press, 1999.
89. Shiller, Robert, Irrational Exuberance, Princeton, NJ: Princeton University Press,
2000.
90. Solow, Robert M., & quot-Technical Change and the Aggregate Production Function,& quot- Review of Economics and Statistics, 39(3), August 1957, pp. 312−20.
91. _, & quot-Investment and Technical Progress,& quot- in Kenneth J. Arrow, Samuel Karlin,
and Patrick Suppes, Mathematical Methods in the Social Sciences, 1959, Stanford, CA: Stanford University Press, 1960, pp. 89−104.
92. _, & quot-We'-d Better Watch Out,& quot- New York Review of Books, July 12, 1987.
93. Stiroh, Kevin J. & quot-Computers, Productivity, and Input Substitution,& quot- Economic Inquiry, 36(2), April 1998, pp. 175−91.
94. _, & quot-Information Technology and the U.S. Productivity Revival: What Does the
Industry Data Say?& quot- New York, NY: Federal Reserve Bank of New York, December 2000.
95. Tobin, James, & quot-A General Equilibrium Approach to Monetary Theory,& quot- Journal of Money, Credit and Banking, 1(1), February 1969, pp. 15−29.
96. Triplett, Jack E., & quot-The Economic Interpretation of Hedonic Methods,& quot- Survey of Current Business, 66(1), January 1986, pp. 36−40.
97. _, & quot-Price and Technological Change in a Capital Good: Survey of Research on
Computers,& quot- in Jorgenson and Landau (1989), pp. 127−213.
98. _, & quot-High-tech Industry Productivity and Hedonic Price Indices,& quot- in Organisation for Economic Co-operation and Development, Industry Productivity, Paris: Organisation for Economic Co-operation and Development, 1996, pp. 119−42.
99. _, & quot-The Solow Productivity Paradox: What Do Computers Do to Productivity?& quot-
Canadian Journal of Economics, 32(2), April 1999, pp. 309−34.
100. Whelan, Karl, & quot-Computers, Obsolescence, and Productivity,& quot- Washington, DC: Board of Governors of the Federal Reserve System, September 1999.
101. Wolfe, Tom, & quot-Two Men Who Went West,& quot- Hooking Up, New York: Farrar, Straus, and Giroux, 2000, pp. 17−65.
102. Wykoff, Andrew W., & quot-The Impact of Computer Prices on International Comparisons of Productivity,& quot- Economics of Innovation and New Technology, 3(3−4), 1995, pp. 277−93.
103. Young, Allan, & quot-BEA'-s Measurement of Computer Output,& quot- Survey of Current Business, 69(7), July 1989, pp. 108−15.
Appendix
Table 1: Information Technology Output and Gross Domestic Product
Year Computer Software Communications IT Services Total IT Gross Domestic Product
Value Price Value 1 Price Value 1 Price Value 1 Price Value 1 Price Value 1 Price
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
0,0 662,98
0,2 662,98 0,1 0,58
0,3 497,23 0,2 0,59
0,3 350,99 0,2 0,59
0,8 262,69 0,5 0,59
1,0 218,30 0,6 0,57
1,3 179,45 0,9 0,58
1,9 126,16 1,2 0,54
2,1 102,41 1,5 0,58
2,1 87,48 1,6 0,58
2,7 79,16 2,3 0,63
3,0 71,13 3,1 0,70
3,1 54,17 3,2 0,69
3,9 43,67 3,7 0,70
3,9 41,39 4,3 0,72
4,3 33,80 5,3 0,77
4,0 31,27 6,6 0,83
4,9 26,12 7,1 0,85
6,3 22,72 7,5 0,87
8,5 15,44 9,2 0,90
11,4 12,81 11,9 0,94
14,0 9,97 14,5 1,00
19,2 8,75 17,8 1,08
22,0 7,80 21,1 1,12
28,8 6,44 24,9 1,13
37,4 5,24 30,4 1,15
39,6 4,48 35,2 1,15
45,9 4,45 38,5 1,13
48,6 3,93 43,7 1,14
54,1 3,72 51,2 1,15
56,9 3,52 61,4 1,13
52,4 3,09 69,3 1,12
52,6 2,85 78,2 1,13
54,9 2,44 83,9 1,06
54,8 2,02 95,5 1,06
57,6 1,80 104,6 1,04
70,5 1,41 115,7 1,03
78,3 1,00 131,0 1,00
86,0 0,73 158,1 0,97
86,9 0,53 193,3 0,94
92,4 0,39 241,2 0,94
1,8 0,81 0,4 3,26 2,3 2,47 307,7 0,19
1,7 0,81 0,4 2,19 2,0 2,29 297,0 0,18
1,9 0,83 0,6 2,38 2,5 2,38 339,0 0,19
2,2 0,86 0,8 2,30 3,0 2,43 370,6 0,19
2,7 0,84 1,1 2,50 3,9 2,43 387,4 0,19
3,0 0,80 1,5 2,56 4,5 2,38 418,2 0,20
2,7 0,81 1,3 1,86 3,9 2,15 418,3 0,20
3,0 0,81 1,8 2,25 4,7 2,30 461,3 0,20
3,7 0,82 2,0 2,27 5,7 2,33 484,7 0,21
4,3 0,85 1,9 1,79 6,2 2,22 503,6 0,21
3,8 0,86 2,1 1,84 5,9 2,25 507,2 0,22
4,7 0,86 2,7 2,14 7,4 2,37 551,9 0,22
5,1 0,84 2,8 1,99 8,2 2,28 564,9 0,22
5,6 0,82 2,8 1,88 9,0 2,19 581,8 0,22
6,2 0,82 3,3 1,99 10,0 2,20 623,3 0,22
6,2 0,81 3,3 1,81 10,8 2,08 666,9 0,23
6,9 0,79 3,6 1,76 12,1 2,01 726,5 0,24
8,1 0,78 4,7 1,99 15,0 2,03 795,1 0,25
9,7 0,76 5,2 1,85 18,0 1,88 871,3 0,25
10,7 0,76 5,0 1,50 19,3 1,75 918,2 0,26
11,6 0,78 5,4 1,40 20,7 1,71 973,0 0,26
13,0 0,79 5,8 1,31 23,8 1,70 1045,8 0,27
14,4 0,81 6,7 1,34 27,1 1,73 1105,2 0,29
14,7 0,83 8,1 1,47 29,0 1,73 1178,8 0,30
15,6 0,85 9,0 1,48 32,2 1,72 1336,2 0,32
18,2 0,86 12,1 1,78 38,4 1,82 1502,5 0,34
19,9 0,90 10,9 1,45 40,4 1,73 1605,9 0,37
21,3 0,96 12,0 1,46 43,9 1,79 1785,8 0,41
23,8 0,98 14,2 1,58 50,0 1,83 2017,5 0,44
28,1 0,97 22,5 2,28 64,4 2,02 2235,7 0,46
32,7 0,99 20,3 1,86 70,6 1,85 2517,7 0,49
38,4 1,02 26,5 2,18 88,2 1,92 2834,9 0,54
43,9 1,07 23,5 1,73 95,9 1,80 2964,5 0,57
48,6 1,13 22,4 1,46 108,0 1,76 3285,2 0,62
50,9 1,17 25,6 1,49 119,5 1,77 3445,4 0,66
55,0 1,17 29,5 1,50 138,1 1,71 3798,8 0,70
62,9 1,18 33,3 1,44 163,9 1,63 4288,1 0,74
69,9 1,17 38,5 1,44 183,1 1,57 4542,6 0,75
72,7 1,17 42,7 1,36 199,7 1,54 4657,4 0,74
74,9 1,15 50,3 1,37 217,5 1,50 5078,1 0,78
82,1 1,14 59,3 1,40 246,7 1,48 5652,0 0,83
85,1 1,13 63,0 1,31 266,3 1,43 5988,8 0,85
86,5 1,12 68,5 1,28 276,6 1,38 6284,9 0,88
83,9 1,12 67,5 1,13 282,2 1,32 6403,3 0,90
88,1 1,11 77,3 1,15 304,1 1,27 6709,9 0,92
92,6 1,09 84,7 1,11 327,6 1,21 6988,8 0,93
102,6 1,07 96,6 1,12 361,4 1,17 7503,9 0,96
112,4 1,03 108,7 1,10 407,2 1,11 7815,3 0,97
120,1 1,00 115,1 1,00 444,5 1,00 8339,0 1,00
131,5 0,98 123,0 0,90 498,7 0,91 9009,4 1,04
140,4 0,95 131,9 0,79 552,5 0,82 9331,1 1,03
158,1 0,92 140,9 0,69 632,6 0,75 9817,4 1,04
Notes: Values are technology output
in billions of current dollars. is gross domestic product by
Price are normalized type of product.
to one in 1996. Information
Table 2: Growth Rates of Outputs and Inputs
1990−95 1995−99
Prices Quantities Prices Quantities
Outputs
Gross Domestic Product 1,99 2,36 1,62 4,08
Information Technology -4,42 12,15 -9,74 20,75
Computers -15,77 21,71 -32,09 38,87
Software -1,62 11,86 -2,43 20,80
Communications Equipment -1,77 7,01 -2,90 11,42
Information Technology Services -2,95 12,19 -11,76 18,24
Non-Information Technology Investment 2,15 1,22 2,20 4,21
Non-Information Technology Consumption 2,35 2,06 2,31 2,79
Inputs
Gross Domestic Income 2,23 2,13 2,36 3,33
Information Technology Capital Services -2,70 11,51 -10,46 19,41
Computer Capital Services -11,71 20,27 -24,81 36,36
Software Capital Services -1,83 12,67 -2,04 16,30
Communications Equipment Capital Services 2,18 5,45 -5,90 8,07
Non-Information Technology Capital Services 1,53 1,72 2,48 2,94
Labor Services 3,02 1,70 3,39 2,18
Notes: Average annual percentage rates of growth.
Table 3: Information Technology Capital Stock and Domestic Tangible Assets
Year Computer Software Communications Total IT Total Domestic Tangible Assets
Value Price Value Price Value Price Value Price Value Price
1948 4,7 0,81 4,7 1,37 711,7 0,13
1949 5,9 0,82 5,9 1,37 750,5 0,13
1950 7,3 0,84 7,3 1,41 824,5 0,13
1951 9,0 0,87 9,0 1,46 948,1 0,14
1952 10,6 0,84 10,6 1,41 1017,5 0,14
1953 12,2 0,81 12,2 1,36 1094,9 0,15
1954 13,7 0,81 13,7 1,37 1146,9 0,15
1955 15,2 0,81 15,2 1,36 1238,4 0,15
1956 17,5 0,82 17,5 1,38 1373,2 0,16
1957 20,7 0,86 20,7 1,44 1494,1 0,17
1958 22,5 0,86 22,5 1,45 1562,3 0,17
1959 0,2 752,87 0,1 0,54 24,7 0,86 25,0 1,45 1655,7 0,18
1960 0,2 752,87 0,1 0,54 26,5 0,84 26,8 1,42 1755,3 0,18
1961 0,5 564,66 0,3 0,55 28,8 0,83 29,5 1,39 1854,8 0,18
1962 0,6 398,58 0,4 0,55 31,7 0,83 32,7 1,38 1982,7 0,19
1963 1,1 298,31 0,8 0,56 33,8 0,81 35,7 1,34 2088,5 0,19
1964 1,6 247,90 1,1 0,55 36,4 0,79 39,1 1,31 2177,3 0,19
1965 2,2 203,79 1,6 0,55 40,0 0,78 43,8 1,28 2315,4 0,20
1966 2,9 143,27 2,3 0,52 44,5 0,76 49,7 1,22 2512,1 0,20
1967 3,7 116,30 3,2 0,56 50,8 0,77 57,6 1,22 2 693,3 0,21
1968 4,3 99,34 3,8 0,56 57,7 0,79 65,7 1,23 2 986,0 0,22
1969 5,3 89,90 5,1 0,61 65,4 0,80 75,7 1,25 3 319,1 0,24
1970 6,2 80,77 7,0 0,68 74,4 0,83 87,5 1,29 3595,0 0,25
1971 6,3 61,52 7,9 0,67 82,1 0,84 96,3 1,28 3922,6 0,26
1972 7,3 49,59 9,1 0,67 90,6 0,86 107,0 1,29 4396,8 0,28
1973 8,6 47,00 10,7 0,69 99,9 0,88 119,2 1,31 4960,3 0,31
1974 9,1 38,38 13,2 0,75 112,8 0,91 135,0 1,35 5391,6 0,32
1975 9,7 35,51 16,3 0,80 128,7 0,98 154,6 1,43 6200,5 0,36
1976 10,4 29,66 18,3 0,82 142,1 1,01 170,7 1,45 6750,0 0,38
1977 12,4 25,81 20,4 0,84 152,3 0,99 185,1 1,42 7574,4 0,41
1978 14,1 17,46 23,5 0,87 171,8 1,02 209,4 1,42 8644,9 0,46
1979 19,3 14,47 28,7 0,91 195,0 1,04 243,0 1,43 9996,7 0,51
1980 24,2 11,27 35,3 0,97 225,7 1,09 285,2 1,47 11 371,0 0,56
1981 33,6 9,90 43,6 1,04 260,9 1,15 338,1 1,53 13 002,5 0,63
1982 42,4 8,84 52,0 1,08 290,0 1,19 384,3 1,55 13 964,7 0,66
1983 52,6 7,32 60,6 1,09 314,3 1,20 427,5 1,53 14 526,0 0,68
1984 66,2 5,95 72,3 1,11 344,8 1,20 483,3 1,50 15 831,0 0,71
1985 77,7 5,08 84,2 1,11 375,0 1,20 537,0 1,46 17 548,6 0,77
1986 86,0 4,34 94,9 1,10 404,3 1,18 585,1 1,41 18 844,3 0,80
1987 94,1 3,71 108,5 1,11 434,8 1,17 637,4 1,37 20 216,2 0,84
1988 107,2 3,45 125,2 1,12 467,7 1,16 700,0 1,35 21 880,1 0,89
1989 121,0 3,23 144,4 1,11 499,7 1,15 765,1 1,33 23 618,7 0,93
1990 122,3 2,89 165,2 1,10 527,1 1,14 814,5 1,29 24 335,1 0,94
1991 124,6 2,58 189,9 1,10 548,3 1,13 862,8 1,27 24 825,7 0,95
1992 128,2 2,17 203,8 1,04 569,7 1,11 901,7 1,21 25 146,8 0,95
1993 135,6 1,82 231,8 1,05 589,5 1,10 956,9 1,17 25 660,4 0,95
1994 150,4 1,61 255,8 1,02 612,8 1,07 1019,0 1,13 26 301,0 0,95
1995 170,3 1,33 286,7 1,03 634,1 1,03 1091,1 1,07 27 858,4 0,98
1996 181,6 1,00 318,1 1,00 659,3 1,00 1158,9 1,00 29 007,9 1,00
1997 198,7 0,76 365,2 0,97 695,8 0,98 1259,7 0,94 30 895,3 1,04
1998 210,0 0,55 431,2 0,95 730,9 0,94 1372,1 0,87 32 888,5 1,07
1999 232,4 0,41 530,6 0,95 778,5 0,90 1541,5 0,81 35 406,9 1,11
Notes: Values are in billions of current dollars. Prices are normalized to one in 1996. Domestic tangible assets include fixed assets and consumer durable goods, land, and inventories.
Table 4: Information Technology Capital Services and Gross Domestic Income
Year Computer Software Communications Total IT Gross Domestic Income
Value Price Value Price Value Price Value Price Value Price
1948 1,7 1,20 1,7 4,31 307,7 0,14
1949 1,3 0,79 1,3 2,83 297,0 0,14
1950 1,8 0,91 1,8 3,27 339,0 0,15
1951 2,1 0,90 2,1 3,21 370,6 0,15
1952 2,6 0,94 2,6 3,36 387,4 0,15
1953 3,2 0,96 3,2 3,46 418,2 0,15
1954 2,7 0,70 2,7 2,49 418,3 0,15
1955 3,6 0,85 3,6 3,05 461,3 0,16
1956 4,2 0,87 4,2 3,12 484,7 0,17
1957 3,7 0,68 3,7 2,44 503,6 0,17
1958 4,1 0,68 4,1 2,45 507,2 0,17
1959 0,2 444,36 0,1 0,63 5,2 0,80 5,5 2,87 551,9 0,18
1960 0,2 433,59 0,1 0,62 5,4 0,75 5,6 2,68 564,9 0,18
1961 0,3 637,21 0,1 0,58 5,6 0,71 6,0 2,59 581,8 0,18
1962 0,4 508,68 0,2 0,62 6,6 0,76 7,2 2,71 623,3 0,19
1963 0,6 311,81 0,3 0,58 6,5 0,67 7,3 2,34 666,9 0,20
1964 0,8 211,28 0,4 0,60 7,1 0,67 8,3 2,26 726,5 0,21
1965 1,3 182,17 0,6 0,59 9,1 0,78 11,0 2,52 795,1 0,22
1966 2,2 173,57 1,0 0,64 9,6 0,73 12,8 2,40 871,3 0,23
1967 2,3 110,97 1,1 0,50 9,8 0,66 13,2 2,01 918,2 0,23
1968 2,6 87,05 1,6 0,60 10,2 0,61 14,5 1,86 973,0 0,24
1969 2,8 68,23 1,7 0,52 11,3 0,61 15,8 1,76 1045,8 0,25
1970 3,6 65,38 2,3 0,56 13,3 0,65 19,1 1,83 1105,2 0,26
1971 5,2 72,48 3,7 0,77 14,9 0,67 23,9 1,99 1178,8 0,27
1972 4,9 48,57 4,0 0,71 16,6 0,69 25,4 1,85 1336,2 0,30
1973 4,4 33,06 4,5 0,71 22,8 0,88 31,7 2,04 1502,5 0,32
1974 6,6 38,82 5,1 0,70 20,3 0,72 32,0 1,84 1605,9 0,34
1975 5,9 28,43 6,7 0,80 23,2 0,77 35,7 1,85 1785,8 0,37
1976 6,6 26,07 7,7 0,81 25,0 0,78 39,2 1,84 2017,5 0,41
1977 7,0 20,69 8,4 0,82 41,8 1,20 57,2 2,40 2235,7 0,44
1978 11,8 22,49 9,7 0,86 35,5 0,93 57,0 2,07 2517,7 0,47
1979 11,6 13,33 11,6 0,90 47,9 1,14 71,1 2,15 2834,9 0,51
1980 16,6 11,81 13,6 0,91 42,0 0,90 72,2 1,82 2964,5 0,53
1981 17,7 7,89 15,5 0,90 40,5 0,79 73,6 1,53 3285,2 0,58
1982 19,6 5,93 17,6 0,89 43,1 0,77 80,3 1,41 3445,4 0,60
1983 26,4 5,46 20,6 0,91 49,4 0,82 96,4 1,43 3798,8 0,66
1984 36,1 4,87 25,4 0,96 54,3 0,83 115,7 1,41 4288,1 0,71
1985 39,6 3,70 30,6 0,99 63,1 0,89 133,3 1,35 4542,6 0,73
1986 43,1 3,04 35,3 0,99 69,3 0,89 147,6 1,27 4657,4 0,73
1987 53,4 2,93 42,1 1,04 86,5 1,02 181,9 1,36 5078,1 0,77
1988 52,7 2,31 50,5 1,10 104,1 1,14 207,3 1,36 5652,0 0,81
1989 57,6 2,08 60,4 1,13 105,8 1,07 223,8 1,29 5988,8 0,84
1990 64,7 2,01 67,2 1,08 109,8 1,04 241,7 1,25 6284,9 0,86
1991 64,2 1,76 70,8 1,00 104,2 0,93 239,2 1,12 6403,3 0,88
1992 71,7 1,66 89,9 1,11 112,2 0,96 273,7 1,16 6709,9 0,91
1993 77,8 1,45 90,4 0,98 126,9 1,03 295,1 1,11 6988,8 0,92
1994 80,1 1,19 109,5 1,05 142,4 1,10 331,9 1,10 7503,9 0,96
1995 99,3 1,12 115,5 0,99 160,7 1,16 375,6 1,09 7815,3 0,96
1996 123,6 1,00 131,9 1,00 149,0 1,00 404,5 1,00 8339,0 1,00
1997 134,7 0,76 156,2 1,02 157,1 0,98 448,1 0,92 9009,4 1,04
1998 152,5 0,59 178,2 0,97 162,0 0,93 492,6 0,82 9331,1 1,04
1999 157,7 0,42 204,4 0,91 175,3 0,91 537,4 0,72 9817,4 1,06
Notes: Values are in billions of current dollars. Prices are normalized to one in 1996.
Table 5: Labor Services
Year Labor Services Employment Weekly Hourly Hours
Price | Quantity | Value | Quality Hours Compensation Worked
1948 0,08 1924,6 156,1 0,75 61 536 39,1 1,2 125 127
1949 0,09 1860,0 171,5 0,75 60 437 38,5 1,4 121 088
1950 0,09 1961,0 179,2 0,76 62 424 38,5 1,4 125 144
1951 0,10 2133,0 214,4 0,78 66 169 38,7 1,6 133 145
1952 0,10 2197,2 227,2 0,79 67 407 38,5 1,7 135 067
1953 0,11 2254,3 241,8 0,80 68 471 38,3 1,8 136 331
1954 0,11 2190,3 243,9 0,81 66 843 37,8 1,9 131 477
1955 0,11 2254,9 256,7 0,81 68 367 37,8 1,9 134 523
1956 0,12 2305,0 275,0 0,82 69 968 37,5 2,0 136 502
1957 0,13 2305,1 295,5 0,83 70 262 37,0 2,2 135 189
1958 0,14 2245,3 309,1 0,83 68 578 36,7 2,4 130 886
1959 0,14 2322,1 320,1 0,84 70 149 36,8 2,4 134 396
1960 0,15 2352,2 344,1 0,84 71 128 36,5 2,5 135 171
1961 0,15 2378,5 355,0 0,86 71 183 36,3 2,6 134 451
1962 0,15 2474,1 376,7 0,87 72 673 36,4 2,7 137 612
1963 0,15 2511,4 386,2 0,88 73 413 36,4 2,8 139 050
1964 0,16 2578,1 417,6 0,88 74 990 36,3 3,0 141 447
1965 0,17 2670,6 451,9 0,89 77 239 36,3 3,1 145 865
1966 0,18 2788,5 500,3 0,89 80 802 36,0 3,3 151 448
1967 0,19 2842,4 525,5 0,90 82 645 35,7 3,4 153 345
1968 0,20 2917,0 588,3 0,91 84 733 35,5 3,8 156 329
1969 0,22 2992,1 646,6 0,91 87 071 35,4 4,0 160 174
1970 0,23 2938,6 687,3 0,91 86 867 34,9 4,4 157 488
1971 0,26 2924,9 744,5 0,90 86 715 34,8 4,7 156 924
1972 0,27 3011,7 817,6 0,91 88 838 34,8 5,1 160 873
1973 0,29 3135,0 909,4 0,91 92 542 34,8 5,4 167 271
1974 0,31 3148,2 988,5 0,91 94 121 34,2 5,9 167 425
1975 0,35 3082,9 1063,9 0,92 92 575 33,8 6,5 162 879
1976 0,38 3174,4 1194,0 0,92 94 922 33,9 7,1 167 169
1977 0,41 3277,4 1334,5 0,92 98 202 33,8 7,7 172 780
1978 0,44 3430,3 1504,2 0,92 102 931 33,8 8,3 180 842
1979 0,47 3554,7 1673,2 0,92 106 463 33,7 9,0 186 791
1980 0,52 3535,7 1827,9 0,92 107 061 33,3 9,9 185 591
1981 0,55 3563,8 1968,8 0,93 108 050 33,2 10,6 186 257
1982 0,60 3519,7 2096,3 0,93 106 749 32,9 11,5 182 772
1983 0,63 3586,7 2269,8 0,94 107 810 33,1 12,2 185 457
1984 0,66 3786,7 2499,1 0,94 112 604 33,2 12,9 194 555
1985 0,69 3882,9 2679,0 0,95 115 205 33,1 13,5 198 445
1986 0,75 3926,3 2931,1 0,95 117 171 32,9 14,6 200 242
1987 0,74 4075,1 3019,7 0,96 120 474 32,9 14,6 206 312
1988 0,75 4207,7 3172,2 0,96 123 927 32,9 15,0 211 918
1989 0,80 4348,4 3457,8 0,97 126 755 33,0 15,9 217 651
1990 0,84 4381,5 3680,8 0,97 128 341 32,9 16,8 219 306
1991 0,88 4322,0 3800,2 0,98 127 080 32,5 17,7 214 994
1992 0,94 4353,9 4086,9 0,98 127 238 32,6 19,0 215 477
1993 0,96 4497,4 4297,7 0,99 129 770 32,8 19,5 221 003
1994 0,96 4628,3 4453,1 0,99 132 799 32,9 19,6 226 975
1995 0,98 4770,7 4660,5 1,00 135 672 33,0 20,0 232 545
1996 1,00 4861,7 4861,7 1,00 138 018 32,8 20,6 235 798
1997 1,03 4987,9 5122,0 1,00 141 184 33,0 21,1 242 160
1998 1,08 5108,8 5491,5 1,00 144 305 33,0 22,2 247 783
1999 1,12 5204,8 5823,4 1,00 147 036 32,9 23,1 251 683
Notes: Value is in billions of current dollars. Quantity is in billions of 1996 dollars. Price and quality are normalized to one in 1996. Employment is in thousands of workers. Weekly hours is hours per worker, divided by 52. Hourly compensation is in current dollars. Hours worked are in millions of hours.
Table 6: Sources of Gross Domestic Product Growth
1948−99 1948−73 1973−90 1990−95 1995−99
Outputs
Gross Domestic Product 3,46 3,99 2,86 2,36 4,08
Contribution of Information Technology 0,40 0,20 0,46 0,57 1,18
Computers 0,12 0,04 0,16 0,18 0,36
Software 0,08 0,02 0,09 0,15 0,39
Communications Equipment 0,10 0,08 0,10 0,10 0,17
Information Technology Services 0,10 0,06 0,10 0,15 0,25
Contribution of Non-Information Technology 3,06 3,79 2,40 1,79 2,91
Contribution of Non-Information Technology Investment 0,72 1,06 0,34 0,23 0,83
Contribution of Non-Information Technology Consumption 2,34 2,73 2,06 Inputs 1,56 2,08
Gross Domestic Income 2,84 3,07 2,61 2,13 3,33
Contribution of Information Technology Capital Services 0,34 0,16 0,40 0,48 0,99
Computers 0,15 0,04 0,20 0,22 0,55
Software 0,07 0,02 0,08 0,16 0,29
Communications Equipment 0,11 0,10 0,12 0,10 0,14
Contribution of Non-Information Technology Capital Services 1,36 1,77 1,05 0,61 1,07
Contribution of Labor Services 1,14 1,13 1,16 1,03 1,27
Total Factor Productivity 0,61 0,92 0,25 0,24 0,75
Notes: Average annual percentage rates of growth. The contribution of an output or input is the rate of growth, multiplied by the value share.
Table 7: Sources of Average Labor Productivity Growth
1948−99 1948−73 1973−90 1990−95 1995−99
Gross Domestic Product 3,46 3,99 2,86 2,36 4,08
Hours Worked 1,37 1,16 1,59 1,17 1,98
Average Labor Productivity 2,09 2,82 1,26 1,19 2,11
Contribution of Capital Deepening 1,13 1,45 0,79 0,64 1,24
Information Technology 0,30 0,15 0,35 0,43 0,89
Non-Information Technology 0,83 1,30 0,44 0,21 0,35
Contribution of Labor Quality 0,34 0,46 0,22 0,32 0,12
Total Factor Productivity 0,61 0,92 0,25 0,24 0,75
Information Technology 0,16 0,06 0,19 0,25 0,50
Non-Information Technology 0,45 0,86 0,06 -0,01 0,25
Addendum
Labor Input 1,95 1,95 1,97 1,70 2,18
Labor Quality 0,58 0,79 0,38 0,53 0,20
Capital Input 4,12 4,64 3,57 2,75 4,96
Capital Stock 3,37 4,21 2,74 1,82 2,73
Capital Quality 0,75 0,43 0,83 0,93 2,23
Notes: Average annual percentage rates of growth. Contributions are defined in Equation (3) of the text.
Table 8: Sources of Total Factor Productivity Growth
1948−99 1948−73 1973−90 1990−95 1995−9
Total Factor Productivity Growth 0,61 0,92 0,25 0,24 0,75
Contributions to TFP Growth:
Information Technology 0,16 0,06 0,19 0,25 0,50
Computers 0,09 0,02 0,12 0,15 0,32
Software 0,02 0,00 0,02 0,05 0,09
Communications Equipment 0,05 0,03 0,06 0,05 0,08
Non-Information Technology 0,45 0,86 0,06 -0,01 0,25
Relative Price Changes:
Information Technology -6,16 -4,3 -7,4 -7,2 -11,5
Computers -23,01 -23,5 -21,1 -18,0 -34,5
Software -3,29 -3,0 -3,2 -3,9 -4,8
Communications Equipment -3,71 -3,1 -4,2 -4,0 -5,3
Non-Information Technology -0,41 -0,9 0,0 0,1 -0,1
Average Nominal Shares:
Information Technology 2,07 1,09 2,60 3,46 4,26
Computers 0,40 0,10 0,61 0,81 0,94
Software 0,51 0,08 0,60 1,30 1,84
Communications Equipment 1,16 0,91 1,39 1,34 1,48
Non-Information Technology 97,20 98,46 96,55 95,35 94,35
Notes: Average annual rates of growth. Prices are relative to the price of gross domestic income. Contributions are relative price changes, weighted by average nominal output shares.
Chart 1: Relative Prices of Computers and Semiconductors, 1959−1999
100 000,0 T
10 000,0 —
1 000,0
100,0
10,0 —
1,0
0,1
0,0
-•- Computers -^- M emory -•- Logic Note: All price indexes are divided by the output price index.
10 000
1 000
100
& lt-o
ЕЯ О
10
Chart 2: Relative Prices of Computers, Communications, Software, and Services, 1948−99
Computers
Communications
— Software
Services
1
0
Note: All price indexes are divided by the output price index.
Chart 3: Relative Prices of Computers, Communications, and Software, 1959−1999
10 000 -,
1 000
о & lt-n
100
10

A Computers M Central Office Switching Equipment Note: All price indexes are divided by the output price index.
Prepackaged Software
1
0
Chart 4: Output Shares of Information Technology by Type, 1948−99
Computers -•-Communications -A- Software -¦- Services -¦- Total Note: Share of current dollar gross domestic product.

ПоказатьСвернуть
Заполнить форму текущей работой