INNOVATION AND PRODUCTIVITY GROWTH T.W. Schultz Lecture DALE W. J ORGENSON “The puzzle confronting economists has been that the rate of growth of output that was being observed has been much larger than the rate of increase in the principal resources that were being measured. It is now clear that this puzzle is of our own making because we have been using measures of capital and labor which had been refined and narrowed in ways that excluded many of the improvements that have been made in the quality of these resources.” Theodore W. Schultz (1962). The computer equipment manufacturing of the economy depends critically on the industry comprised only 0.3 percent of U.S. performance of a relatively small number of value added from 1960–2007, but generated sectors, such as agriculture and computers, 2.7 percent of economic growth and 25 per- where innovation takes place. cent of productivity growth. By comparison This paper begins with a brief history agriculture accounted for 1.8 percent of U.S. of productivity measurement. The traditional value added, but only 1.0 percent of economic approach of Simon Kuznets (1971) and growth during this period. This reflects the fact Robert Solow (1970), has been superseded that agriculture has grown more slowly than by the new framework presented in Paul the U.S. economy, while the computer industry Schreyer’s OECD (2001) manual, Measuring has grown thirteen times as fast. However, Productivity. The focus of productivity mea- agriculture accounted for fifteen percent of surement has shifted from the economy as a U.S. productivity growth, indicating a very whole to individual industries like agriculture significant role for agricultural innovation. and computers. The OECD productivity man- The purpose of this paper is to assess the ual has established international standards for role of innovation in U.S. economic growth dur- economy-wide and industry-level productivity ing the period 1960–2007. Productivity growth measurement. is the key economic indicator of innovation. The OECD standards for productivity mea- Despite the importance of innovation in indus- surement are based on the production accounts tries like computers and agriculture, this inno- constructed by Jorgenson, Frank Gollop, and vation accounts for less than twelve percent Barbara Fraumeni (1987). These accounts of U.S. growth. The great bulk of economic have been updated and revised to incorporate growth in the U.S. is due to the replication of investments in information technology hard- existing technologies through investment and ware and software by Jorgenson, Mun Ho, expansion of the labor force. Although inno- and Kevin Stiroh (2005). The EU (European vation contributes only a modest portion of Union) KLEMS (capital, labor, energy, mate- economic growth, this contribution is vital to rials, and services) study, described by Mary gains in the U.S. standard of living. O’Mahony and Marcel Timmer (2009), was The predominant role of replication of completed on June 30, 2008. This landmark existing technologies in U.S. economic growth study presents productivity measurements for is crucial to the formulation of economic 25 of the 27 EU members, as well as Australia, policy. As the U.S. economy recovers from Canada,Japan,and Korea,and data for the U.S. the Great Recession of 2007–2009, economic based on those of Jorgenson, Ho, and Stiroh policy must focus on maintaining the growth (2005). Current data for the participating coun- of employment and reviving investment. tries are available at the EU KLEMS website: Policies that concentrate on enhancing the http://www.euklems.net/. rate of innovation will have a relatively modest The hallmark of the new framework for impact over the intermediate term of 5–10 productivity measurement is the concept years. However, the long-run growth rate of capital services, including the services Amer. J. Agr. Econ. 93(2): 276–296; doi: 10.1093/ajae/aaq191 © The Author (2011). Published by Oxford University Press on behalf of the Agricultural and Applied Economics Association. All rights reserved. For permissions, please e-mail: [email protected] Jorgenson Innovation and Productivity Growth 277 provided by IT equipment and software. growth between substitution and technical Modern information technology is based on change. Constant quality indexes of labor input semiconductor technology used in comput- are discussed detail by Jorgenson, Gollop, and ers and telecommunications equipment. The Fraumeni (1987, Chapters 3 and 8, pp. 69–108 economics of information technology begins and 261–300), and Jorgenson, Ho, and Stiroh with the staggering rates of decline in the (2005, Chapter 6, pp. 201–290). prices of IT equipment used for storage of Griliches and I introduced a constant quality information and computing. The “killer appli- index of capital input by distinguishing among cation” of the new framework for productivity different types of capital inputs. To combine measurement is the impact of IT investment these capital inputs into a constant quality summarized below. The final section sums up index, we identified prices of the inputs with the paper. rental prices, rather than the asset prices used in measuring capital stock used by Solow and Kuznets. This also broadened the concept of The New Framework for Productivity substitution and further altered the allocation Measurement of economic growth between substitution and technical change. A detailed survey of recent research on sources Griliches and I employed a model of capital of economic growth is given in my 2005 paper, as a factor of production I had introduced in my “Accounting for Growth in the Information 1963 article, “Capital Theory and Investment Age”. A survey of earlier research is given in Behavior”.This made it possible to incorporate my 1990 article, “Productivity and Economic differences among depreciation rates on differ- Growth”, presented at the Jubilee of the Con- ent assets,as well as variations in returns due to ference on Research in Income and Wealth, the tax treatment of different types of capital commemorating the fiftieth anniversary of the income, into the rental prices. Constant quality founding of the Conference by Kuznets. Addi- indexes of capital input are presented by Jor- tional surveys are provided by Zvi Griliches’ genson, Fraumeni, and Gollop (1987, Chapters (2000) posthumous book, R& D, Education, 4 and 8, pp. 109–140 and 267–300), and by Jor- and Productivity, and Charles Hulten’s (2001) genson, Ho, and Stiroh (2005, Chapter 5, pp. article, “Total Factor Productivity: A Short 147–200). Biography”. Finally, Griliches and I replaced the aggre- The most serious challenge to the tradi- gate production function employed by Kuznets tional approach to productivity measurement and Solow with the production possibility of Kuznets (1971) and Solow (1970) was frontier introduced in my 1966 paper, “The mounted by my 1967 paper with Griliches, Embodiment Hypothesis”. This allowed for “The Explanation of Productivity Change”. joint production of consumption and invest- Griliches and I departed radically from the ment goods from capital and labor services. measurement conventions of the traditional I used this approach to generalize Solow’s approach. We replaced NNP with GNP as a (1960) concept of embodied technical change, measure of output and introduced constant showing that productivity growth could be quality indexes for both capital and labor interpreted, equivalently, as “embodied” in inputs. Jorgenson (2002) refers to this as the investment or “disembodied”. My 1967 paper production theory approach to measurement. with Griliches removed this indeterminacy by The key idea underlying our constant qual- introducing constant quality price indexes for ity index of labor input was to distinguish investment goods. As a natural extension of among different types of labor inputs. We Solow’s (1956) one-sector neo-classical model combined hours worked for each type into a of economic growth,his 1960 model of embodi- constant quality index of labor input, using ment had only a single output and did not allow labor compensation per hour as weights in the for the introduction of a separate price index index number methodology Griliches (1960) for investment goods. had developed for U.S. agriculture.This consid- Nicholas Oulton (2007) demonstrates that erably broadened the concept of substitution Solow’s model of embodied technical change employed by Solow (1957). is a special case of the model I had proposed While Solow modeled substitution between in 1966. He also compares the empirical results capital and labor inputs, Griliches and I of Solow’s one-sector model and a two-sector extended the concept of substitution to include model with outputs of consumption and invest- different types of labor inputs as well. This ment goods. Jeremy Greenwood and Per Krus- altered, irrevocably,the allocation of economic sell (2007) employ Solow’s one-sector model, 278 April 2011 Amer. J. Agr. Econ. replacing constant quality price indexes for However, BLS retained hours worked as investment goods with “investment-specific” a measure of labor input until July 11, 1994, or embodied technical change. The deflator when it released a new multifactor productivity for the single output, consumption, is used to measure including a constant quality index of deflate investment, conflicting with a require- labor input as well. Meanwhile, BEA (1986) ment of the systems of national accounts dis- had incorporated a constant quality price index cussed below, namely, separate deflators for for computers into the national accounts – consumption and investment. over the strenuous objections of Edward F. Griliches and I showed that changes in the Denison (1989). This index was incorporated quality of capital and labor inputs and the into the BLS measure of output, completing quality of investment goods explained most of the displacement of the traditional framework the Solow residual. We estimated that capital of economic measurement by the conventions and labor inputs accounted for eighty-five per- employed in my paper with Griliches.