World Bank Document
Total Page:16
File Type:pdf, Size:1020Kb
FILE COPY A Survey and Critique of World Bank Supported Research on International Comparisons of Real Product Public Disclosure Authorized SWP365 World Bank Staff Working Paper No. 365 P.C.C. 1 December 1979 Public Disclosure Authorized Public Disclosure Authorized Prepared by: Robin Marris (Consultant) Economic Analysis & Projections Department Copyright ( 1979 The World Bank 1818 H Street, N.W. Public Disclosure Authorized Washington, D.C. 20433, U.S.A. The views and interpretations in this document are those of the author and should not be attributed to the World Bank, to its affiliated organizations, or to any individual acting in their behalf. Che views and interpretations in this document are those of the author and should not be attributed to the World Bank, to its iffiliated organizations or to any individual acting in their behalf. WORLD BANK Staff Working Paper No.365 December 1979 A SURVEY ANID CRITIQUE OF WORLD BANK SUPPORTED RESEARCH'ON INTERNATIONAL COHPARISONS OF REAL PRODUCT This paper describes the nature and content of the statistical data generated by the project on International Comparisons of Real Product (ICP). It analyzes their theoretical implications, investigates more generally the problems of international comparisons of economic welfare, discusses and criticizes the methods used by the ICP to compare internationally expendi- ture in the services sectors, reconsiders the applied theory of the rela- tionship between price-structure, economic development and purchasing power exchange rates. Prepared by: Robin Marris (Consultant) Economic Analysis & Projections Department Copyright® 1979 The World Bank 1818 H Street, N.W. Washington, D.C. 20433 U.S.A. INTRODUCTION During the middle of the decade 1970-80, the World Bank, the United Nations, the Ford Foundation and other institutions provided substantial finan- cial and administrative support for a major research project on international comparisons of real product undertaken by Irving Kravis, Alan Heston and Roy Summers at the University of Pennsylvania0 Irving Kravis was, of course, the creator of data on the same subject relating to France, Italy, the U.K. and the U.S., published by the OECD in the 1950's. From time to time, as pro- posals for extending or continuing the earlier work came up, Foundations receiving grant applications in the field became accustomed to consulting Kravis, and in due course, when it had been decided to fund new work in the UN, he was the natural choice for leader. The Bank became intimately involved when it decided to agree to a request that it become a co-sponsor and major financial supporter. The results of the new project have been in process of publication since the middle of 1975. The first volume appeared in 1975, the second in 1978, and a third is expected in 1980.* In addition to these publications, Kravis, Heston and Summers published an article in 1978* giving the results of an indirect method for estimating the results that could have been expected for GDP per capita for over a hundred countries if all these countries had been intensively studied by the direct method. This Staff Paper is the result of a consultancy undertaken by the author in 1978-79 to summarize, survey and criticize the methods and results of Phases I and Io. A Summary of its contents follows. * For references, see footnote, p. below. SUMMARY Section I, Nature and Content of ICP Data, summarizes the data published in Phases I and II, namely quantities for 153 commodities in, eventually, 34 countries, for benchmark years 1970 and 1973, and in some countries, 1975, with resulting calculations of overall real GDP per capita and various sub-aggregates according to Laspeyres-type, Paasche-type, Ideal and "multilateral" index numbers. Section II, A More Formal Account, provides a mathematical summary of the basic data and of the conceptual character of the various international comparisons that may be derived from them. It sets out potential intransitivi- ties and weighting instabilities contained in the system of binary comparisons, together with a more precise definition of the multilateral system. Section III, Purposes and Uses of the Data. In a First Account of Purposes, this section lists six different types of pragmatic uses of these data, including aid policy, research on economic growth, development planning, studies in income distribution and growth, etc. In the subsection "Product Comparisons or Welfare Comparisons," the section questions the validity of the traditional distinction between measurements of production and measurements of welfare on the grounds that, in international comparisons, one is effectively treating each country as if it contained only one (representative) individual. It also digresses on some questions in the logic of evaluating world income distribution. A subsection called "More General Discussions of Purposes" discusses the meaning of international comparisons at a more abstract level and con- cludes that, whether one likes it or not, the fundamental purpose of these data is to support cardinal international comparisons of economic welfare. It goes on to defend this position against conventional "ordinalist" cri- ticism. Section IV, Theoretical Problems in International Welfare Compari- sons, investigates various kinds of quantity index-numbers in the interna- tional context, with special reference to the problem of taste differences. Some of the theoretical material is original. A new empirical relation between price dispersion and per capita GDP levels is reported. This creates the possibility of non-monotonic bias in the ICP multilateral indices. Section V, Problems of Quality Differences and the Services Sectors, describes the methods used in the ICP to deal with quality differences in the "hardware" sectors, such as automobiles. The section then continues to a critique of the methods used to deal with the analogous problem of measuring output (or consumption) of the services sectors, especially those concerned with public goods or publicly provided goods. It argues that, for about 15% of the typical country's GDP, the methods used by the ICP do not represent measures of output, but rather of input, and therefore must not be used for productivity comparisons. A statistical investigation is conducted of the - ii effects, which are showu to be potentially a:,nkfAcant0 Possible alternative solutions to an admittedly extremely difficult problem are discussed. The general argument concerning services is illustrated by detailed comparisons of the structure of the ICP measures for Health Care and Public Administra- tion, as between India, West Germany, the U.K. and the U.S. Section VI, Price Structure, Econowic Development and Exchange Rates. Sultan Ahmad, Bela Balassa, Christopher Clague, Irving Kravis, Stephen Marris, Vito Tanzi, Paul Samuelson and others have iuvestigated the empirical rela- tionship between national price structures, levels of development and exchange deviations. This section develops an Analysis of Variance of prices and pro- ductivities in the course of economic development, on which some alternative models, or V'scenarios" are based, in the hope of helping to explain the empiri- cal relationship between economic development and price dispersion reported in Section IV. Initially these models abstract from international trade, but are then married with trade models to review the classic contributions of Balassa, et al. It is concluded that the theory behind the association between exchange deviations and GDP levels is quite complex and that, as argued by Clague and Tanzi, though empirically supported, is not inevitable. It is considered possible, though unlikely, that the empirical phenomenon is mainly due to underestimates, in the ICP methodology, of the comparative productivity of services in rich countrieso Acknowledgements I wish to acknowledge the benefit of conversations with Prof. Christopher Clague, University of Maryland; Professors Irving Kravis, Alan Heston, and Roy Summers, of the University of Pennsylvania; Dr. Sultan Ahmand, Dr. Helen Hughes and Dr. R. McPheeters of the World Bank. None of these individuals, however, is in any way responsible for errors. -1- I. NATURE AND CONTENT OF THE ICP DATA As already indicated, the results of the project have been orgauized in three phases. Phase I, published in 1975, related to ten countries: six developed, three developing, and one "centrally planned" (Hungary). Phase II, published in 19789 added six countries, including four developing countries. Phase III is expected to add 18 more countries of various types. In addi- tion, an important article, referred to below as "KHS," providing indirect estimates of total real per capita GDP for over a hundred countries was also published by Kravis, Heston and Summers, in 1978. 1/ Most data presented in Phase I are also available in the Phase II publication. Some of the methodological discussion found in Phase I is not, however, repeated in Phase II. For the various countries represented in the several Phases, the results as a whole provide, inter alia: (a) Binary real GDP comparisons for the years 1970 and 1973 of each country with the United States, using own-country price weights, and U.S. price weights (Phase II, page 8, commodity detail, pp. 173-230). (b) Ideal-index comparisons between all pairs of coun- tries (Phase II, page 12). 1/ The full title of the Phase I publication was A System of International Comparisons of Gross Product and Purchasing Power, by Irving B. Kravis and associates, Johns Hopkins Press, Baltimore, for United Nations and the World Bank, 1975. Phase II is called International Comparisons of Real Product and Purchasing Power, same authors, publishers, etc., 1978. KHS is called "Real GDP Estimates for More than One Hundred Countries," by Irving Kravis, Alan Heston, and Robert Summers, Economic Journal, June 1978. - 2 - (c) Extrapolations, 1/ based mainly on national data, for the same countries to the years 1965-1975, based on national series adjusted for changes in the terms of trade in such a way that real GGP is treated as an "income" rather than as a "production" concept 2/ (Phase II, p.