Approaches to the measurement of national product

This book aims to provide an alternate foundation for the measurement of the production of nations. The framework developed here is applied to the U.S. economy for the postwar period. The patterns that result are significantly different from those derived within conventional systems of national accounts.

National accounts give systematic empirical form to the structure, patterns, and performance of an economy (Young and Tice 1985). In the modern world, they provide the objective basis for judging the level and progress of the wealth of nations and for identifying the causes of success and failure.

Conventional systems of national accounts include the United Nations System of National Accounts, the United States National Income and Product Accounts, and various forms of input–output accounts. It is our contention that these types of accounts seriously distort the levels and trends of the national product, the surplus product, productivity, and other major aggregate economic variables. Because measurement and analysis are inextricably intertwined, our understanding of intertemporal and international economic development is correspondingly affected.

Criticisms of official national accounts are not new. Debates about their purpose and structure have gone on from the very start (Eisner 1988, p. 1611). In recent times, there has been a renewed flurry of questions about their adequacy. Such criticisms come from a variety of quarters, ranging from official agencies such as the United Nations to a variety of prestigious economists.

The Marxian measure of productivity q* is the primary one. It is derived by deflating our measure of total product TP* by the GNP price deflator py (in 1982 dollars), and then dividing by hours of productive labor Hp to obtain hourly productivity. Because orthodox measures of productivity are based on value added rather than total product, we also calculate a quasi-Marxian measure y*, which is real Marxian value added per productive worker hour. This differs greatly in level from q*, but has essentially the same trend, because the proportion CVTP* of circulating constant capital to total product is extremely stable (see Section 5.2, Figure 5.6, and Appendix E). These Marxian productivity measures are listed in Table J.1.

Orthodox measures of productivity vary considerably. The most common one is real GDP per employee hour, which we call y. Also available is the BLS measure y1 of real GDP originating in the non farm business sector by their estimate of hours of persons engaged in production (employees plus self-employed persons) in this same sector. Since the BLS measure is only available in index-number form, we calculate an equivalent measure of productivity y2 in the non farm business sector, as the ratio of real GDP to total hours of all persons engaged. The total-hours measure H2 is in turn calculated by multiplying the average hours per full-time equivalent employee in domestic industries (from NIPA) by estimated non farm total employment (based on our employment data in Table F.1).

The purpose of this chapter is to develop a mapping between Marxian categories and those of conventional national accounts. The essential points of the argument will be presented here, with all further detail reserved for Appendix A.

National income and product accounts (NIPA) are the traditional base for national accounting. But input–output (IO) tables provide a more general framework in that they encompass both interindustrial flows and national income–product flows. We will therefore use IO tables as our basic theoretical and empirical foils.

Although IO accounts provide a superior description of the economy for our purposes, they suffer from the drawback of being available only for benchmark years (1947, 1958, 1963, 1967, 1972, 1977) for the United States. We will therefore use these tables only to provide comprehensive benchmark estimates, which will then be expanded into annual series using NIPA data.

Our project requires distinguishing between three different sets of measures, for which we adopt different notation: Marxian labor value measures such as constant capital, variable capital, and surplus value (C, V, S); their money forms (C*, V*, S*); and corresponding IO–NIPA aggregates such as intermediate inputs, wages, and profit-type (unearned) income (M, W, P). Since all accounts in question will be double-entry accounts, each of the revenue-side flows will have use-side counterparts (such, as U for the Marxian labor value of the inputs used in production, U* for the corresponding Marxian money value, and M for the orthodox measure of intermediate inputs on both revenue and use sides).

The previously derived mappings describe the relation between input–output accounts and the money value form of Marxian categories, on both the revenue and use sides of the accounts. We now turn to the corresponding calculation of the labor value form of these same categories. In what follows, we will outline a procedure first developed in Shaikh (1975) and extended and applied by Khanjian (1989). Only the basic elements will be presented here, since a fuller development is beyond the scope of this book.

Calculating labor value magnitudes

Let us begin by recalling our money value mapping previously summarized in Figure 3.11 and Table 3.12. Figure 4.1 is a simplified version of Figure 3.11. In it, we have explicitly labeled elements of the production and trade rows so as to facilitate later discussion. Thus the (purchaser) price of the total intermediate input of the productive sectors Mp = (Mp)p + (Mp)t, where (Mp)p represents the total producer price of the commodities used as intermediate input in the productive sectors and (Mp)t represents the trading margin on these same goods. The same breakdown holds for all input and final-demand elements. In addition, final demand has been considered into two main categories: the consumption of productive workers CONWp; and surplus demand SD, which is the remainder of final demand – equal to the sum of the consumption of unproductive (trade and royalties) workers and capitalists, investment, net exports, and government expenditures.

National economic accounts provide the empirical foundation for economic theory and policy. And at the core of all national accounts lie the production accounts, for it is the production of new wealth which has been, at least so far, the real foundation of modern economic success.

The classical economists were deeply concerned with the factors that accounted for the economic success or failure of nations. Their analysis of the structure of production led them to the recognition that not all activities resulted in a product. On the contrary, they classified certain activities – such as wholesale/retail trade, military and police, and administration – as forms of social consumption rather than of production. It followed from this that one had to distinguish between production and nonproduction labor. As we have emphasized throughout, such a distinction does not imply that one of the forms of labor is more necessary, more meritorious, or more politically correct. Nor, in spite of Adam Smith, does it imply that services cannot be production activities. The basic distinction arises from the difference between outcome and output; not all outcomes are outputs. We noted in Chapters 1 and 2 that one way to formalize the difference would be in terms of a vector of characteristics associated with each commodity, some of which would affect its status as an object of social use, and others its status as an object of distribution, et cetera.

The primary database from which the condensed tables in Section 5.1 and Appendix C are derived is composed of input–output tables that correspond to the 85-industry IO tables, in that the final versions have 82 rows and 88 columns. All the input–output tables used are originally derived from the benchmark tables produced by the BE A (Bureau of Economic Analysis), which is part of the U.S. Department of Commerce. In an effort to improve the compatibility of the data for different years and from different sources, various adjustments and modifications were carried out on the data series used. Most significantly, Juillard (1988) carried out several adjustments on the input-output tables so as to have as consistent a methodology as possible for all the benchmark years. Several additional adjustments and aggregations were conducted in order to render data sets from different sources more consistent. Among these, the major adjustments were: (1) reversing the methodology of force account construction used with the BEA IO tables; (2) adjusting the BLS (Bureau of Labor Statistics, Department of Labor) capital stock and depreciation matrices for oil and gas exploration costs; and (3) modifying employment and employee compensation data so as to maximize consistency within the adjusted IO tables for all benchmark years.

The Bureau of Economic Analysis of the Department of Commerce has successively published benchmark input–output tables for 1947, 1958, 1963, 1967, 1972, and 1977 (see respectively, BEA 1970, 1965, 1969, 1974, 1979, 1984). Table A.I lists industries and commodities for the 1972 table.

The classical and Marxian traditions share the distinction between production and nonproduction labor. But Marx was particularly concerned with that portion of production labor which is productive of capital, since only this labor creates surplus value. The rest of labor is unproductive of capital, even though it may be wage labor (in distribution and state activities) or production labor (productive of value or of use value). Marx himself does not imply that productive labor is in any way superior to, or more necessary than, unproductive labor. But as we pointed out in Chapter 2, not all Marxists proceed in the same way. Most notably, Baran (1957, p. 32) redefines productive labor as labor that would be necessary under a “rationally ordered” (socialist) society. Marx's definition of productive labor is thereby replaced with a definition based on necessity, and the concept of surplus value is replaced with the concept of “surplus” – defined as the excess of the total product over essential personal and public consumption.

This chapter will analyze the various attempts to measure Marxian categories. In order to make the account manageable, we restrict ourselves to studies published in English, and to estimates of the rate of surplus value. Sharpe (1982a) covers some of the literature available in French, but a comprehensive worldwide survey remains to be done. The Japanese are pioneers in this regard.

This book has been a long time in the making. The interest in providing an empirical framework that would correspond to Marxian categories dates back to 1972–73, when Anwar Shaikh first discovered Shane Mage's pathbreaking work and developed an alternate schema and an alternate set of estimates based on Mage's own data.

In 1974 Shaikh came across Edward Wolff's working paper on input output-based estimates of the rate of surplus value in Puerto Rico. This added a new dimension to the problem. Mage's work emphasized the significance of the distinction between productive and unproductive labor, but it was restricted to only the value-added side of national income accounts. On the other hand, whereas Wolff's work was located within the more comprehensive double-entry framework of input-output accounts, it did not distinguish between productive and unproductive labor. This led Shaikh to attempt to develop a comprehensive framework for Marxian categories which made both distinctions simultaneously.

The procedure that emerged in 1975 was essentially the same one used in this book: a mapping between Marxian and input-output categories illustrated by means of a continuing numerical example in which both total price (the sum of purchasers' prices) and the magnitudes of the aggregate value flows (total value and its basic components) were held constant, while the associated money forms became ever more complex as more concrete factors were considered.

In order to measure the net impact of state expenditures and taxes on the rate of surplus standard of living of wage and salary earners, we need to examine the net transfer between these wage and salary earners and the state. This net transfer is defined as benefits received by workers minus taxes paid. In this appendix we briefly outline the methods of allocating various government expenditures and taxes to labor, and then report estimates of the net transfer for the period 1952–85. Because the net transfer is negative over most of this period, it actually represents a net tax on workers. What follows is a brief summary of the methodology in Shaikh and Tonak (1987).

In determining the portion of state expenditures directed toward workers, we begin by classifying them into three major groups. The first group consists of items such as labor training and services, housing and community services, income support, social security, and welfare (except for the small items called military disability and military retirement, which we treat as a cost of war). These services are assumed to be received entirely by workers either in money or in commodity form. The second group includes such conventional categories as education, health and hospitals, recreational and cultural activities, energy, natural resources, transportation, and postal services; these are treated as social consumption in general, and the workers' share in them is estimated by multiplying the group total by the share of total labor income in personal income. The last group comprises two kinds of expenditures.

The distinction between production and nonproduction activities

Marxist national accounts depend crucially on the distinction between labor which is productive of capital and that which is not. Because this distinction is so often presented in a confused and contradictory manner, we will not begin from it. Instead, this section will focus on the prior and more general distinction between production and nonproduction activities. The next section will then develop this into more concrete distinctions between labors which are and are not productive of capital. As we shall see, our derivation will enable us to arrive at a definition that corresponds exactly to the one Marx uses. Deriving the definition from first principles, rather than simply beginning with it, allows us to endow it with considerably greater depth.

Mistaken conceptions of the distinction

It is useful to begin by emphasizing what the distinction between production and nonproduction labor does not refer to. In the first place, it is not a distinction between necessary and unnecessary activities. We intend no connotation that production activities are either more (or less) necessary than nonproduction activities. The dividing line does not rest on either technical or social standards of efficiency, though of course such standards may well be applied to either set of activities. Baran and Sweezy's argument that some labor under capitalism would be unnecessary in a “rationally ordered” (i.e. socialist) economy (Baran 1957, p. 32) is one such efficiency approach.

Our empirical analysis of the U.S. economy will be set out in several sections. Sections 5.1 and 5.2 will utilize suitably modified input–output tables to develop benchmark year estimates of Marxian measures of the total, intermediate, and final product, and then use NIPA data to interpolate between benchmark estimates to create an annual series for each of these measures. Sections 5.3 and 5.4 develop the estimates of annual employment, wages, variable capital V*, surplus value S*, surplus product SP*, and the rate of surplus value S*/V*, and compare them to the more conventional measures such as profit-type income and the profit/wage ratio. Section 5.5 measures the Marxian rate of profit, and compares it with the average observed rate (net of those parts of surplus value which are absorbed into nonproduction expenses) and the observed corporate rate. Section 5.6 measures the rate of exploitation of unproductive workers, and compares it to that of productive labor; Section 5.7 compares Marxian and conventional measures of productivity. Sections 5.8 and 5.9 examine the impact of the state on accumulation, through its absorption of surplus value and through the effects of taxes and social expenditures on the rate of surplus value. Section 5.10 examines the effects of price–value deviations on aggregate Marxian measures, and Section 5.11 develops a technique that allows us to approximate the rate of surplus value in a relatively simple manner.