Comparative Study of Long-Term Records of Economic Growth

These notes deal with the attempt of the Council’s Committee on Economic Growth to secure, for a number of countries, long-term estimates of national income, wealth, and their components—as basic data in a comparative study of the economic growth of nations.1The Committee on Economic Growth, whose members are Simon Kuznets of Johns Hopkins University (chairman), Richard Hartshorne of the University of Wisconsin, Melville J. Herskovits of Northwestern University, Edgar M. Hoover of the Office of Population Research, Bert F. Hoselitz of the University of Chicago, Wilbert E. Moore of Princeton University, and Joseph J. Spengler of Duke University, was appointed in 1949 to explore means of improving research on long-term changes in the magnitude and structure of nations or other major economic regions. Its program of foreign studies was initiated in 1953 and has been developed with support granted to the Council by the Ford Foundation for the committee’s activities. The present paper is adapted from a memorandum prepared for the guidance of the participants in this program. For earlier brief reports on it, see Items, December 1953, pp. 46–47, and March 1954, p. 8. The committee considers such estimates indispensable for an adequate study of economic growth. This view may be challenged by saying that if reliable estimates of such countrywide aggregates and their components can be derived, there must be back of them a rich supply of economic data of various descriptions; and that it is not certain that this particular form of aggregation is the most useful one for the analysis of economic growth.

Without discussing this argument in detail, one may urge that the requirement of comprehensiveness followed in national income and wealth estimation provides assurance that no hasty conclusions will be derived from partial data; that the close relation of these aggregative concepts to the body of economic theory and analysis makes these sets of estimates particularly useful; and that given an effort to provide as many of the structural breakdowns as possible (by industrial origin, classes of income recipients, character of economic units engaged in production, type of use, domestic and foreign sources of origin, regions within the country, etc.), the resulting quantitative framework should be useful in studying not only the rate of economic growth but also the structural changes that accompanied it.

It should be admitted, of course, that the long-term estimates of national income, wealth, and their components sought in this study will be properly interpretable only if they are: (a) supplemented by a variety of other basic statistics relating to population, labor force, output indexes for major industries or products, foreign trade, etc.; (b) read in the light of knowledge of economic history of the countries, i.e., of the record of successive secular decisions made in them and of the historical changes in the international conditions within which the countries had to exist and grow. Two practical conclusions are implicit. First, in working on long-term estimates of national income, wealth, and their components, an effort should be made to provide also the important long-term records on other relevant aspects of economic growth. Second, the critical review, collation, and preparation of long-term estimates of national income, wealth, and their components should be followed, for each country, by analysis of these estimates and other relevant and available long-term records, in an attempt to interpret them in the light of the country’s economic history.

Scope of the program

The present scope of the committee’s program is limited to the countries for which long-term estimates of national income, wealth, and their components appear to be obtainable largely by critical review and collation of series already available; and for which the preparation of completely new estimates is facilitated by the known stock of economic statistics. At the time of writing, the committee has projects under way or contemplated in the United Kingdom, France, Germany, Italy, the Netherlands, Sweden, Norway, Denmark, and Japan. There is hope that Australia can be added to the list; and current work on Russia should eventually permit addition of that country as well.

The work under way relates to somewhat different historical periods in the various countries and is at different stages of progress. In each country present work is under the guidance of resident scholars who have access to whatever data are available within the country.

In most countries included in the program, interest in countrywide estimates of income, wealth, and their components is of long standing; and there have been several past attempts to prepare such estimates—if of varying quality with respect to clarity of concepts, adequacy of statistical data, and completeness of description of the bases of estimation. It would be most useful to have a thorough and critical review of such past work in each country, partly to permit the investigator to take advantage of work already done; partly to help him orient himself by the views taken by his predecessors, early or late, in interpreting the available statistical evidence. The results of such a review should be presented fully in whatever report eventuates, because many of the early publications in this field are not accessible to scholars in other countries, who could profit greatly from having competent accounts of the early work.

No country possesses such rich data, properly digested and cross-checked, as to permit preparation of long-term estimates of national income, wealth, and all the possible sets of components, for the full period in which we are interested, i.e., the two or two and a half centuries back to the late seventeenth. Since time, intellectual resources, and financial means are limited, restrictive choices will have to be made—choices that should be made by the investigator in each country on the basis of his knowledge of the data and of the possibilities which they offer. There follow some common-sense observations on criteria that might be followed in the choices; these criteria reflect the common interests of the project.

Criteria and procedure in preparation of estimates

The period of coverage should be as long as possible. For many aspects of comparative analysis of economic growth even a period of 50 years is far too short. Often the total period covered can be stretched to its maximum by sacrificing detail either with respect to time continuity or to coverage of components. It is obviously more useful to have a series of estimates covering 100 years at decennial intervals than one covering 10 years at annual intervals, provided that the single years given for each decade are not so materially affected by transitory factors as to obscure the path of secular growth over the century. Therefore, if necessary, one should sacrifice continuity over time to total time span. Such sacrifice is a genuine one: without a continuous series covering a long period, it would be quite difficult to distinguish and time properly the successive phases in a country’s economic growth.

Alternatively, one may be able to obtain a longer period of coverage for some of the broader aggregates but not for many of the components. In this case, the longest coverage for the more comprehensive aggregates is obviously to be chosen, even though some of the divisions by components will be available only for a shorter time span (e.g., total national income for 100 years, but distribution between capital formation and flow of goods to consumers for only 50 years). Or it may be possible to secure estimates for the more comprehensive aggregate (e.g., national income) on a continuous basis (i.e., annually or quinquennially), while the industrial distribution can be estimated only every tenth year. In such cases it would be valuable to have a continuous long series for the more comprehensive aggregates, while being content with a decennial series, or one with even longer time intervals, for the components.

In choosing among the various aspects to be covered by the estimates, the following tentative scale of priority can be suggested: (1) As between national income and national wealth, one would urge the former; for national wealth, as it is ordinarily measured, is but a cumulative total of a part of national income in earlier periods. It is most important to have the longest and most continuous series of national income (or some related gross product) aggregates, in both current and constant prices. (2) The distributions by industrial origin and by type of use are particularly to be stressed, largely because of their availability for most countries; the distributions of income among various social groups or income classes are most difficult, except in one or two countries, and emphasis on them might result in devoting too great resources to them at this stage. (3) At the present stage of the committee’s program, the main purpose is to organize into comparable records as much as can easily be skimmed from the available data and estimates. This means that once the indispensable series on aggregate product—national income, or domestic product, or gross national product—have been secured in both current and constant prices, the choice among several component distributions should be determined largely by the availability of data within each country.

With respect to national product in constant prices, it would be best to adjust for price changes in parts; in this case price indexes based on a relatively recent year are to be preferred. If the adjustment for price changes is global, i.e., is carried through by applying a single price index to an undivided countrywide aggregate in current prices, one should use a price index with early year quantity weights, as well as choose a price index as close as possible to the levels of prices of final products (retail rather than wholesale, etc.). In any case, experimentation with different price indexes would be most illuminating; and if these indexes are available, the minor volume of additional calculation thus involved is warranted.

A complete description of the sources of the data used and the methods of deriving all estimates should be prepared by each investigator. It is the committee’s hope that the results will be widely used, not only by scholars closely associated with this project, but by all interested in the economic growth of nations. These scholars need adequate descriptions of sources and methods so that they may have a clear view of how the estimates have been obtained, and what margins of error to attach to them. It may prove useful to have the estimates grouped into classes with specified limits of margin of error. This has been done in some publications in the field in the past; and the feasibility and the possible value of such a procedure should be considered.

Need for preliminary analysis

Since the first phase of work is limited to assembling as much of the long-term record of national income, wealth, and their components as possible on the basis of easily available data and past work, without committing large resources to finding new primary data in the archives and laboring on raw, if basic, economic statistics, its completion within the reasonably near future can be anticipated. Of course the results will not be complete, in the sense that they will not comprise all the long-term estimates within the field that could be secured, if much longer time and much larger resources were devoted to the task. Yet incomplete as the results will be in that sense, they will be sufficiently rich to permit some preliminary analysis; and this analysis should be undertaken as the next phase, rather than attempting to obtain additional estimates. Such analysis should be extremely useful, first, as a check on the sets of estimates already derived and, second, as an indication of the major gaps that would suggest fruitful possibilities of further work on the estimates proper.

The hope is that the long-term estimates resulting from the first phase of work (or otherwise already at hand), in combination with other basic long-term records ordinarily available for these countries, will permit rather simple analysis directed at some questions commonly discussed in the literature on the economic growth of nations. But simple as such analysis may be, it will require thorough familiarity with the basic estimates themselves; good knowledge of the economic history of the country; and some acquaintance with the literature on economic growth. The notes that follow illustrate the questions that may guide the analysis. In the nature of the case they cannot constitute a complete list, and perhaps the kind of review and discussion of estimates suggested should not be called analysis, but it would seem to be a highly useful stage of further work.

Illustrative questions for analysis

Given the long-term records of national product and of population, the first set of questions concerns the possibility of establishing phases in economic growth of the country, distinguished by marked changes in the rate of overall growth of either total product or product per capita. The full determination of such phases or periods in the country’s growth must await examination of structural changes as revealed by shifts in weight of various significant components of the economy. But in a preliminary fashion, such “phasing” can be discerned on the basis of the countrywide aggregates alone. This is particularly important as an introduction to any further analysis since the first task is to place the period covered by the income estimates within the longer perspective of the country’s economic history. The “phasing,” hence, would necessitate combining the evidence provided by the countrywide aggregate with what is known about the country’s economic history. Periods of more rapid or slower growth could then be distinguished; the relation to the timing of industrialization processes indicated; and the interplay between growth of population, total product, and per capita product examined. At this point also the bearing of any “long cycles,” i.e., long alternations in the rate of growth, would have to be examined (here the value of continuous annual data is particularly great).

If a distribution of national product by different industries of origin is available over the long period, simple analysis can be undertaken provided that long-term estimates of the industrial distribution of the labor force or of capital formation are also at hand. One could then ask to what extent growth in real income per capita was due to shifts in the distribution of the labor force in favor of industries with higher than average levels of product per worker; and to what extent it was due to intra-industry growth of product per worker. Further, if capital formation (or reproducible wealth) is available by industrial destination or affiliation, one could study for the major industrial sectors the association between differences in product per worker and in capital per worker—either in industrial comparisons for successive points of time, or within each industrial sector in the movements of product per worker and capital per worker over time. In the analysis so briefly suggested above, quantity production indexes for some basic industrial sectors (such as agriculture or manufacturing) available for many countries could be utilized, recognizing the tricky problem of grossness and netness in using such indexes, which are ordinarily gross, in conjunction with the countrywide aggregates that constitute the basic series, which are more nearly net.

If long-term records on distribution of national income are available for either socioeconomic groups or income classes, an effort should be made to utilize them. Information on long-term trends in the distribution of income among recipients is quite scarce, except that flowing directly from the distribution by industrial origin. Hence in the few countries in which such data are available, even if defective, efforts to subject them to some analysis are warranted. Again, perhaps one should not call it analysis since the primary task is to treat the estimates in such a way that the findings they suggest become clear—findings relating to the national economy as a whole, not to some vaguely and variably defined segment of it. Thus, if long-term series on numbers and incomes of filers of income tax returns are available, the series in themselves have little meaning unless they can be converted into estimates of the shares of some specified ordinal group in the countrywide income distribution. Likewise, calculations of the share of “labor” are meaningless, unless we also know whether the population groups that receive labor shares form increasing or decreasing proportions of the total; the same is true for shares of “farming,” etc. Nothing more precise can be suggested at this stage; but the specific tasks in the countries that do have some long-term data on this subject will differ from country to country, and will require more detailed discussion than can be given here.

In the distributions of product by type of use, long-term records on the proportions devoted to capital formation should show whether any distinct phases emerge (early periods of high proportions, later periods of lower proportions), and in combination with other data would possibly indicate the sources of savings used to finance capital formation (and hence some idea of the trends in the nationwide savings proportions). There would also be interest in composition of capital formation—residential housing, other construction, durable equipment, inventories, etc.—even if there were no way of showing long-term trends in industrial destination. Much of the discussion of these questions would be related to the analyses suggested in the first paragraphs of this section and, in general, there are obvious interrelations among questions and findings suggested throughout the section.

The other and by far the dominant component, viz., flow of goods to consumers, should be subjected to much more scrutiny than has been customary in both estimation and theoretical discussion. The changing structure of ultimate consumption is linked to secular shifts in industrial structure and contributes much to its explanation. Furthermore, without examination of the secular changes in composition of ultimate consumption, the very growth in real income per capita remains too much of a mystery. It is important, therefore, to get some view of changes in ultimate consumption; and if the overall production data do not permit such a long-term view, some effort to use results of sample consumption and expenditure surveys would be warranted. Once the trends have been observed, comparing them with differences in the consumption structure of income classes at a given time would prove illuminating.

Long-term records of foreign trade are available for each country; and there may be similar information, if at longer time intervals, on other economic flows across their boundaries. The contribution of these flows across the boundaries of a country to its economic growth is of deep interest, even though at the suggested stage of analysis one cannot urge more than the obvious comparisons. How large was this flow across the boundaries compared with the magnitudes for the domestic economy? Has the proportionate contribution of imports or exports to a relevant measure of aggregate output shown an upward, downward, or constant trend? What about the relation of capital imports or exports to domestic capital formation or domestic savings? Have trends in trade or capital dependence upon foreign areas differed for the major sectors in the economy? Has the shifting structure of imports and exports meant secular shifts in dependence of the economy on different areas in the world, either as sources of supply or as markets for products? And if so, to what extent has the economic growth of a given country been part of the growth of a wider area?

Incomplete as the questions and suggestions made so far have been, omitting for example those resulting from distribution of product among different types of economic unit—individual firm, corporation, regulated public utility, nonprofit, government—or among regions, they may suffice to indicate what we mean by the next stage of analysis.

Conclusion

As already suggested, the choice of particular aspects to emphasize in preparing estimates depends largely upon the availability of data, but it is hoped that for each country the record of national product will be long and otherwise adequate enough to permit the kind of discussion suggested in the preceding section. But beyond this, there may be such differences among the several countries in the character of the available data that the choice of the investigator will be limited. For most countries the distribution by industrial origin will be available, but not for all; distribution by type of use will be available for some countries but not for all.

Much of the “analysis” that has been suggested is really an attempt to state the findings which the basic estimates and related data suggest. Such an attempt is indispensable if only as a test of what the scholar responsible for the estimates, or one who understands their derivation, thinks of their reliability in indicating this or that secular trend. Clearly, in saying that we find a more rapid rate of growth during the period X to Y than during period S to T, we imply that the difference between the two periods shown by the estimates is reliable enough to warrant such a statement. Statements of findings of the sort suggested often require not only some experimentation with the specific series proper, but also comparisons with others, based upon some reasonably assumed expectations of relations, in turn founded upon some analytical hypotheses.

The proposed type of analysis, while related to some examples in the past, is still in process of development, and certainly in the present connection is in the nature of an experiment. No rigid rules can be proposed now; and the Committee on Economic Growth counts upon exchange of information and ideas as well as the free play of the mind of each investigator. It is the committee’s hope that in such exchange all participants in the program will be learning much from each other.

This essay originally appeared in Items Vol. 9, No. 4 in the winter of 1955. Visit our archives to view the original as it first appeared in the print editions of Items.