The Gerschenkron Effect: A Re-Examination

Abstract

I NDEX number has richness and complexity of a jungle and includes a number of unexplored spots. In spite of an extensive and high-quality literature, of accuracy of economic measurements is far from being fully explored. It is interesting that economists dealing with market-type systems usually turn their attention to theory of cost-of-living index numbers, whereas students of centrally planned systems are captivated by problem of measuring economic growth. quantity index numbers and nature of their biases have received special attention from a number of Sovietologists, including Nutter [9, 10, 11], Levine [7], Bergson [2], Morsteen [8], and [4, 5, 6]. These people were in agreement that series aggregated by means of preindustrialization weights grow considerably faster than series aggregated by later-day weights. Nutter [9], Levine [7], Morsteen [8], and Bergson [2] used production possibility curves to demonstrate theoretically fact that base-yearweighted quantity indexes tend to drift upwards.' Gerschenkron, using actual data (selected items of United States machinery output in different years), offered empirical evidence to support this position. In order to explain phenomenon, he formulated a hypothesis which is often called Effect. present paper will re-examine this proposition.2 provides an explanation of why a quantity index, in binary comparison of preand post-industrialization situations, weighted with base-year prices, can be expected to be biased upwards. A brief summary of this explanation is that, during process of industrialization, scarcity relations change in such a way that production of highly fabricated goods increases proportionately more than that of slightly fabricated goods; and, at same time, relative prices of highly fabricated goods decrease. These inversely correlated price and quantity movements create a drift upwards in baseyear-weighted quantity index, and extent of divergence from current-year-weighted index depends on these opposite movements. One formulation of this proposition in Gerschenkron's words [6, p. 48] -is the process of industrialization in field of machinery production may be regarded as consisting (1) in an increase within total machinery output of share of highly fabricated goods in terms of low fabricated goods and (2) in a concomitant decline in prices of highly fabricated goods in terms of low fabricated goods. . . . To use . . . [an example] it is tractors whose output at beginning of industrialization process is small in relation to that of sickles. In course of industrialization, output of tractors increases much faster than that of sickles, while price of tractors in terms of sickles declines. I Thus characteristic feature of Gerschenkron-type industrialization process is a negative coefficient of correlation between price and quantity relatives, which is responsible for 'This fact was also shown by use of isoquants by Yoram Barzel, Some Observations on Index Number Problem, Econometrica, XXX, July 1963; and same method was used for interpretation of aggregate output indexes in R. Morsteen and R. Powell, Capital Stock (Homewood, Ill.: Irwin, 1966), p. 5. 2 Gerschenkron Hypothesis and Gerschenkron Effect should be kept apart. former refers to development characteristics in initial period of industrialization in some European countries; it proposes that these can be explained better if reference is made to relative backwardness prior to great spurt. See: A. Gerschenkron, Backwardness in Historical Perspective, in Progress of Underdeveloped Areas, ed. by Bert F. Hoselitz (Chicago: University of Chicago Press, 1962), pp. 3-29; The Early Phases of Industrialization in Russia: Afterthoughts and Counterthoughts, in Economics of TakeOff into Sustained Growth, ed. by W. W. Rostow (New York: St. Martin's Press, 1963), pp. 151-169; The Discipline and I, Journal of Economic History, XXVII (Dec. 1967), pp. 443-459. present paper is concerned only with Effect, or with reasons for upward bias of base-year-weighted quantity index. 'Gerschenkron gives same explanation for index number bias in Chapter IX, Soviet Heavy Industry: A Dollar Index of Output, 1927-1937, Economic Backwardness in Historical Perspective (Cambridge, Mass.: Harvard University Press, 1962), pp. 235-250.