Dale Jorgenson

Samuel W. Morris University Professor

PRODUCTIVITY: Volume Two -- Preface

This volume is devoted to international comparisons of economic growth among industrialized countries and is the second of two volumes containing my empirical studies of economic growth. The first volume, Postwar U.S. Economic Growth, shows that investment is the predominant source of U.S. growth since World War II. The objective of this second volume is to compare postwar U.S. experience with the growth of other industrialized countries. The focus is on comparisons among the G-7 countries -- Canada , France , Germany , Italy , Japan , the United Kingdom , and the United States .

Investment is the commitment of current resources in the expectation of future returns. The distinctive feature of investment as a source of economic growth is that the returns can be internalized by the investor. The most straightforward application of this definition is to investments that create property rights, such as investments in tangible assets like plant, equipment, and inventories. However, empirical research has broadened the meaning of capital formation to include investment in human capital through education and training. While these investments produce returns that can be internalized, they do not create property rights that can be transferred.

By contrast, the defining characteristics of productivity as a source of economic growth is that benefits generated by higher productivity are external to the activities that create growth. These benefits "spill over" to income recipients not involved in these activities, severing the connection between the creation of growth and the resulting incomes. Since the benefits of policies to create externalities or "spill overs " cannot be appropriated, these policies typically involve government programs or activities supported through public subsidies.

The wide diversity of experiences among industrialized countries provides a badly needed perspective for debates over policies to stimulate future growth of the U.S. economy. My paper "Productivity and Economic Growth in Japan and the United States ," reprinted as Chapter 8 below, shows that the Asian model of development exemplified by Japan relies primarily on investment as a source of economic growth. For example, during the period 1960-1979 sixty percent of Japanese economic growth can be attributed to investments in tangible assets, while these investments account for forty percent of U.S. growth.

Investment in tangible assets was especially critical for growth of the Japanese economy during the period 1960-1973, when output grew at more than ten percent annually. However, investment in human capital through upgrading of the education and training of the Japanese labor force was also an important contributor to these double digit growth rates. Although Japan and the U.S. are often portrayed as economic adversaries, postwar experiences in both countries support economic policies that give high priority to capital formation.

This volume also presents assessments of competitiveness among nations that are essential for designing international trade policies. Following the Smithsonian Agreements of 1970, there have been very substantial changes in international competitiveness among industrialized countries. Since the U.S. trade balance has moved from surplus to deficit during this period, it has been natural for economic journalists to assume that U.S. competitiveness has deteriorated. In fact, the U.S. has gained considerably relative to other industrialized countries, where competitiveness is defined in terms of comparisons of product prices in a common currency.

The Smithsonian Agreements replaced the Bretton Woods regime of fixed exchange rates with flexible exchange rates. Subsequent fluctuations in competitiveness have been driven largely by exchange rate changes, especially the steady devaluation of the dollar relative to the German mark and the Japanese yen. An important secondary influence has been the growth of wage rates in other industrialized countries relative to the U.S. In periods affected by increases in petroleum prices, like 1973, and 1979, changes in these prices have further strengthened U.S. international competitiveness.

Since the facts about U.S. gains in international competitiveness are not available from official sources, the business press has been left without an anchor in reality for speculations about the sources of changes in the U.S. trade balance. During the 1980's and 1990's a large literature developed, presenting a broad panoply of mainly fanciful ideas about the alleged decline in U.S. competitiveness and its relationship to the trade balance. An important role was assigned to the alleged shortfall in U.S. productivity, relative to that of other industrialized countries.

Readers of the business press are still regularly bombarded with anecdotal evidence of the low level of U.S. productivity relative to Japan and Germany . However, the research presented in this volume shows that relative productivity levels have been fairly sable since around 1980 and that Japan and Germany have emerged as productivity laggards, relative to the U.S. , with productivity somewhere between 85-90 percent of the U.S. level. This relationship has been unaffected by changes in the yen-dollar and mark-dollar exchange rates and fluctuations of U.S.-Japan and U.S.-German trade balances.

While the growth of productivity in the U.S. has been slower than that of other industrialized countries for much of the postwar period, the U.S. began the postwar period with an enormous productivity advantage over its competitors. Although gains in productivity by Germany and Japan in the early postwar period were very dramatic, these gains slowed markedly after 1973. Somewhat paradoxically, relative productivity levels between the U.S. and other industrialized countries have moved in the opposite direction from change in international competitiveness since the collapse of the Bretton Woods regime.

The origin of confusion about international comparisons in the business media is that these comparisons are not a settled matter among economists. For decades economists have employed two competing approaches to the analysis of economic growth. National output provides a measure of production and can be used to allocate the sources of economic growth between investment and productivity. National income provides a measure of welfare and can be divided between current consumption and future consumption through saving. These components of income can be identified with alternative uses of economic growth.

The starting point for my research on international comparisons of economic growth is a unified system of U.S. national accounts for income, product, and wealth that I presented in a series of papers with Laurits Christensen (1969, 1970, 1973a, 1973b). In a paper reprinted in Postwar U.S. Economic Growth Christensen and I (1973a) utilized data on inputs and outputs from a production account for the U.S. economy to allocate the sources of economic growth between investment and productivity. We employed data on income, consumption and saving from an income and expenditure account to divide the uses of growth between present and future consumption.

Christensen and I also succeeded in unifying wealth accounts with income and production accounts by introducing a system of vintage accounts for stocks of assets and their prices. These vintage accounts provide the basis for implementing both welfare and production approaches to the analysis of economic growth. Saving is linked to the asset side of the wealth account by capital accumulation equations for each type of asset. These equations provide a perpetual inventory of assets accumulated at different points of time. Prices for different vintages are linked to rental prices of capital inputs through a parallel set of asset pricing equations.

Mitsuo Ezaki and I (1973) constructed a parallel system of national accounts for Japan in a paper reprinted in Chapter 2 below. This paper employed the same conceptual framework as my work with Christensen and presented aggregate accounts for Japan covering the period 1951-1968. These accounts were based on a system of vintage accounts for assets and their prices. Ezaki (1977) updated our results through 1971 and compared our system of accounts with the official national accounts for Japan . Klaus Conrad and I presented a comparable system of accounts for West Germany in our 1975 book, Measuring the Performance in the Private Economy of the Federal Republic of Germany, 1950-1973.

After production accounts have been constructed for individual countries, the next problem is to link them. My paper with Nieko Nishimizu (1978), reprinted in Chapter 3 below, presents a methodology for bilateral comparisons of output, input, and productivity. These comparisons require purchasing power parities for outputs and inputs. Purchasing power parities give relative prices at which goods and services produced in different countries can be traded, while exchange rates provide relative prices for trading currencies. Competitiveness is defined in terms of relative prices of goods and services in a common currency, so that measures of competitiveness are obtained by dividing purchasing power parities by the corresponding exchange rates.

Nishimizu and I developed estimates of purchasing power parities for both outputs and inputs for Japan and the U.S. , covering the period 1952-1974. These parities are well established for outputs in the official statistics, thanks f\to the work of Irving Kravis , Alan Heston , and Robert Summers (1982). Levels of output can be compared between countries by using purchasing power parities to express these outputs in a common currency. Unfortunately, outputs are often compared between countries by using exchange rates. These comparisons may be highly inaccurate, since purchasing power parities can exchange rates may differ considerably.

Purchasing power parities available from official sources only deal with the output side of the production account. International comparisons of productivity also require purchasing power parities for inputs. Since labor force composition by characteristics such as age, sex, and educational attainment of workers differs substantially between countries like Japan and the U.S. , these differences must be taken into account in measuring purchasing power parities for labor inputs. Each type of labor input must be weighted by its marginal product, just as in comparisons over time for a single country.

A similar issue arises for a capital input. Since capital goods differ substantially in marginal productivity, it is necessary to focus on the flow of capital services rather than the stock of capital. Comparisons between countries require purchasing power parities for different types of capital input like those Nishimizu and I have constructed for Japan and the U.S. Each type of capital input must be weighted by its marginal product. The marginal products for different types of capital must be broken down by legal form of organization and class of asset, just as in comparisons over time for a single country.

In order to account for substitutions among different types of capital inputs in comparisons over time for a given country, each capital good must be weighted by its marginal product. Important progress has been made in measuring capital input in the official statistics for the U.S. compiled by the Bureau of Labor Statistics (1983). Unfortunately, the approach implemented for OECD countries by Steven Englander and Axel Mittelstadt (1988) uses unweighted capital stocks. Productivity comparisons based on these data do not adequately account for substitutions among different types of capital inputs. This results in a highly distorted view of capital as a source of economic growth and a contributor to differences in production levels between countries.

In papers reprinted in Chapters 4 and 5 below Christensen, Dianne Cummings, and I (1980, 1981) extended international comparisons of economic growth and relative levels of economic activity to Canada, France, Germany, Italy, Japan, Korea, the Netherlands, the United Kingdom and the United States. These comparisons covered the period 1947-1973 and included estimates for all nine countries for the period 1960-1973. The methodology for multilateral comparisons was developed by Douglas Caves , Christensen, and Erwin Diewert (1982a, 1982b), generalizing the methodology for bilateral comparisons presented in my paper with Nishimizu .

The key results for international comparisons of economic growth are presented in Chapter 4. During the period 1960-1973, every country except for the United Kingdom grew more rapidly than the U.S. For Japan and Korea annual growth rates were close to ten percent, while growth rates for Canada , France , Germany , Italy , and the Netherlands clustered around five percent. For the U.S. and the U.K. growth rates were around four percent. These variations in growth rates of output are associated with the corresponding variations in growth rates of capital and labor inputs.

Measures of capital and labor inputs based on unweighted capital stocks and hours worked would have resulted in a highly misleading view of the relative importance of growth in input and productivity. For example, growth in hours worked is negative or zero for all five European countries for the period 1960-1973, while measures of labor input are positive. This is due to upgrading of the quality of labor input through substitution of hours with higher marginal productivity for those with lower productivity. Similarly, growth in capital input would be substantially under-estimated by omitting substitution among different types of capital stock.

The comparisons of relative productivity levels presented in Chapter 5 showed that output per capita and productivity converged toward U.S. levels during the postwar period. However, in 1970 the U.S. had the highest levels of output per capita and productivity among the nine countries considered. Only Canada had achieved parity with the U.S. in its capital-labor ratio. Japan 's capital-labor ratio was only one-quarter of the U.S. , while Korea 's ratio was only eleven percent of the U.S. By comparison Japan had attained more than three-quarters of the U.S. productivity level, while Korea had reached more than thirty percent of the U.S. level.

Christensen, Cummings, and I have compared our results for the period 1947-1973 with those of Edward Denison (1967) for the period 1950-1962. Denison employed the concept of income as a basis for productivity comparisons, appealing to the welfare approach to these comparisons. This concept requires data on income and its allocation between consumption and saving and is inappropriate for productivity measurement. These data are provided by the income and expenditure account of the Christensen-Jorgenson accounting system, while productivity comparisons are based on measures of output and input from the production account.

The second stage of my research on international comparisons of economic growth and relative levels of economic activity was to disaggregate the results to the level of individual industries. My 1980 paper on "Accounting for Capital" extended the vintage accounting system I had developed with Christensen to the industry level. For the U.S. this was implemented in my papers with Frank Gollop (1980, 1983) and Barbara Fraumeni (1980, 1986) and our 1987 book, Productivity and U.S. Economic Growth. The results are updated and summarized in my paper, "Productivity and Economic Growth," reprinted as Chapter 1 below.

The critical issue in linking productivity to international competitiveness is the definition of output for individual industries. The older literature on productivity measurement, especially that associated with the work of John Kendrick (1973), uses the concept of value added as a measure of industry output. Value added is defined as the difference between the value of output and the value of intermediate inputs. The value added measure of industry output has the convenient property that national product is the sum of industry-level measures of value added. However, by "simplifying" productivity measurements the value added approach severs the connection between productivity and international competitiveness.

One of the important advances in industry-level productivity measurement described in Chapter 1 has been to utilize gross output rather than value added as a measure of product at the industry level. Industry output is especially advantageous for international comparisons, since measures of competitiveness are based on product prices rather than prices of value added. Another advantage of industry output is that intermediate inputs can be treated symmetrically with inputs of capital and labor services in measuring productivity. These important advantages are acquired at some cost, however, since a fully satisfactory implementation requires the integration of interindustry accounts with national income and product accounts for each of the countries involved in an international comparison.

In Chapter 7 Masahiro Kuroda, Nishimizu , and I employed industry-level gross outputs in international comparisons of productivity between Japan and the U.S. For this purpose we developed annual time series of interindustry accounts in current and constant prices for both Japan and the U.S. We supplemented these data by extensive information on capital and labor inputs for both countries. Completion of this arduous task was essential for relating productivity to international competitiveness. In Chapter 6 Conrad and I presented trilateral comparisons for West Germany, Japan and the U.S., using data for Japan from my paper with Kuroda and Nishimizu and U.S. data from my book with Gollop and Fraumeni . A more extensive discussion is provided by Conrad (1985).

After measures of output have been constructed for individual industries, the next problem is to link the results for different countries. This requires purchasing power parities for outputs at the industry level based on producers' prices rather than the purchasers' prices employed at the aggregate level. Duroda , Nishimizu , and I have transformed purchasers' prices to producers' prices for Japan and the U.S. , using interindustry accounts for both countries to eliminate trade and transportation margins and indirect taxes. A similar approach is employed in the trilateral comparisons involving West Germany in my paper with Conrad.

The second issue that arises in international comparisons of productivity at the industry level is the comparison of labor inputs between countries. Unweighted hours worked are commonly employed for this purpose, but this is highly inappropriate for international comparisons. Hours worked for each type of labor input must be weighted by the corresponding marginal product to capture this substitution. Since the labor force composition of individual industries differes substantially among Germany, Japan, and the U.S. by characteristics such as age, sex, and educational attainment, this is a fundamental issue in comparing labor inputs among the three countries.

The third, and final, issue in comparing productivity at the industry level among countries is the comparisons of capital inputs. Unweighted capital stocks are inappropriate for this purpose, since each type of capital input must be weighted by the corresponding marginal product. Capital stocks differ in composition among Germany, Japan, and U.S. by class of asset and legal form of organization and these characteristics must be taken into account in measures of capital inputs employed in productivity comarisons among the three countries.

The research required for productivity comparisons relevant to international competitiveness poses formidable challenges to economists. These comparisons require a system of national accounts for each country that successfully integrates interindustry accounts with national income and product accounts. Even for Japan and the U.S. two countries with highly developed satistical systems, productivity comparisons have required the development of new data bases for output, and intermediate, capital, and labor inputs. Extending productivity comparisons to Germany poses many additional problems. Finally, data for these countries must be linked by purchasing power parities for inputs and outputs.

An important objective of international comparisons of productivity among Germany, Japan, and the U.S. is to generate data for economemtric modeling of producer behavior at the sectoral level. In 1981 Fraumeni and I developed a general equilibrium model of production for the U.S., including econometric models of production for thirty-five industrial sectors. Kuroda, Kanji Yoshioka, and I developed a similar model for thirty-five industrial sectors of the Japanese economy in 1984. Studies presenting models of this type are summarized in my 1986 paper, "Econometric Methods for Modeling Producer Behavior". These papers are reprinted in an accompanying volume, Econometrics and Producer Behavior.

In 1990 Kuroda and I updated our paper with Nishimizu and presented new comparisons of competitiveness between Japanese and U.S. industries. This paper is reprinted in Chapter 9. Hikaro Sakuramoto , Yoshioka, Kuroda and I employed the results in constructing a bilateral model of production for Japanese and U.S. industries. This model combined data for both countries in estimating parameters that describe the technology for each industry. This paper is reprinted in the volume, Econometrics and Producer Behavior.

An important issue in econometric modeling of producer behavior is the role of economies of scale. For the past decade trade theory has been dominated by models of production that incorporate monopolistic competition and economies of scale. Since monopolistic competition uses a zero profit condition, the accounting framework is very similar to the one used in national income and product measurements and interindustry accounting under perfect competition. An implication of monopolistic competition is that economies of scale can generate possibilities for strategic interactions among traders. The question that remains is: Are economies of scale important?

The principal conclusion of the econometric literature reviewed in my 1986 survey paper is that there are two industries in which economies of scale are important. In long distance telecommunications there are sizable economies of scale associated with the indivisibility of such facilities as underwater cables and satellites. These indivisibilities give rise to opportunities for monopoly or, if there is a free entry, monopolistic competition. International telecommunications is an important component of trade in services.

A second industry in which economies of scale are important is the electric utility sector. Again, there are important indivisibilities associated with power generation and transmission facilities. The electric utility sector also plays a role in international trade, for examle , between Canada and the U.S. or between France and Germany, but this industry is insignificant for Japanese trade. By contrast, firm level data for manufacturing seems to suggest that economies of scale are not very important. For most industries expansion occurs by increasing the number of manufacturing plants, not by expanding the output of a small number of plants. Under these conditions there is little role for economies of scale.

If there is only a modest role for economies of scale among the determiants of international competitiveness, how should we proceed in constructing econometric models for analyzing the consequencesof changes in trade policy? Since economies of scale will be important only for a relatively small range of issues, the best approach is to retain the featuresof the Heckscher -Ohlin model of trade, especially constant returns to scale. Since growth is a very important focus for trade policy, the Heckscher0Ohlin model of trade, especially constant returns to scale. Since growth is a very important focus for trade policy, the Heckscher -Ohlin approach must be implemented within a dynamic setting. Mun Sing Ho and I have developed a model of this type for the U.S. and Kuroda and his colleaguesare developing a similar model for Japan.

One of the critical assumptions of the Heckscher -Ohlin theory is that technologies are identical for all countries. That is a very appealing assumption, since it has been difficult to find a rationale for failures of countries to achieve the same level of technical sophistication. However, data on relative productivity levels for German, Japanese, and U.S. industries presented in this volume reveal that the assumption of identical technologies is untenable. There is no evidence for the emergence of a regime in which the Heckscher -Ohlin assumption of identical technologies would be appropriate.

Second, product mixes for U.S. and Japanese industries are highly dissimilar, so that the assumption of perfect substitutability of commodities produced in different countries is also inappropriate. John Chipman has suggested that this is the consequence of the relatively high degree of aggregation, even at the two-digit or three-digit level that is practical in empirical trade modeling. We conclude that the appropriate point of departure for econometric modeling of international competitiveness is a model with perfect competition, constant returns to scale, technologies that are not identical across countries, and products of identical industries that are not perfect substitutes.

Could a model with the characteristics I have described be used in the analysis of the issues that motivated interest in international competitiveness in the first place? The answer to this question is definitely in the affirmative. However, changes in competitiveness reflect the evolution of exchange rates, so that the initial approach to these issues must be macroeconomic in character and must deal with the interactions between domestic and international savings and investment balances. There is an obvious role here for macro-econometric modeling, especially in analyzing the relationships between exchange rates and trade balances.

The top priority for future research is better measurement of international competitiveness and relative productivity levels. It is crucial to get the numbers right and the results reported in this volume provide evidence of substantial progress. The initial step is to measure international competitiveness, using purchasing power parities based on OECD and Eurostat data. Second, it is essential to generate relative productivity levels within an accounting framework that includes purchasing power parities for capital, labor, and intermediate inputs, broken down in considerable detail.

I have found that international competitiveness has changed very rapidly under the post-Smithsonian regime of floating exchange rates. The speed of these variations in competitiveness obviously has had a wrenching impact on individual countries. By contrast productivity trends are much more gradual and influence international competitiveness only in the long run. In a period as short as a decade or so, productivity growth is of secondary importance. This is an important conclusion since the absence of detailed information on sources of international competitiveness has produced an intellectual vacuum. Until now this vacuum has been filled by purely speculative ideas with little connection with underlying economic realities.

Productivity measurement requires a system of national accounts that successfully integrates interindustry accounts with national incomes and product accounts. This is one of the most important features of the United Nations (1968) System of National Accounts. However, the U.N. system is less successful in integrating capital accounts with those for income flows. For productivity measurement economic researchers must construct data for investment flows, capital stocks, and capital services. These must be linked, using purchasing power parities.

There is a great deal for economists to do in improving economic measurements and using the resulting data in analyzing problems of trade policy. The framework appropriate for this purpose and the results available to date are summarized in my paper, "Productivity and Economic Growth," reprinted as Chapter 1. This paper provides a self-contained introduction to the national accounting concepts required for productivity measurement. These concepts are illustrated in detail by empirical studies of U.S. economic growth over the period 1947-1985. The paper also gives a brief summary of alternative econometric approaches for modeling production.