Journal of the History of Economic Thought, Volume 20, Number 2, 1998 203 PARETO AND THE WICKSELL-COBB-DOUGLAS FUNCTIONAL FORM BY CHRISTIAN E. WEBER I. INTRODUCTION It is understood that Charles Cobb and Paul Douglas (1928) were not the first to use the production function named after them. Joseph Schumpeter (1954, p. 1042), Carl-Axel Olsson (1971), and Henry Spiegel (1991, p. 816) all note that the production function Y = AK*\J ~x had been used by Knut Wicksell (1901, 1906) more than twenty years before Cobb and Douglas published their study.1 While it is quite possible that Wicksell was the first to use the Cobb-Douglas functional form to study production, he was not the first to apply it to economic analysis in general. Vilfredo Pareto had worked out several implications of a specific version of the Cobb-Douglas utility function as early as 1892. Later, he repeated and extended this analysis in the mathematical appendix to the French translation of his Manual of Political Economy (1909). This note discusses Pareto's use of the Cobb-Douglas functional form. Its purpose is to explain his rationale for adopting it and the results he derived from it, and to point out some interesting omissions and minor errors in his later analysis. II. FUNCTIONAL FORM IN PARETO'S "CONSIDERAZIONI" Pareto (1892a) criticized Alfred Marshall's (1890) assumption that the marginal utility of income2 remains constant when the price of a good Seattle University, USA. I would like to thank Dean Peterson for helpful comments on an earlier draft of this paper and John Chipman for providing me with an English translation of Pareto, 1892a, b, 1893 and 1900a, b. All errors are my own responsibility. 1 Douglas (1976) provides an informative survey on the development of the Cobb-Douglas production function between 1928 and 1976. He claims, without providing a reference, that Wicksteed had also discussed the functional form later named after Cobb and Douglas. 2 Pareto (1892a, b) referred to the marginal utility of income as the "final degree of utility of capital goods." He referred to it as "the elementary index of the good whose price is one, that is, of money" (Pareto, 1909, p. 421). 1042-7716/98/020203-08 © 1998 The History of Economics Society Downloaded from https:/www.cambridge.org/core. IP address: 88.99.165.207, on 13 Jul 2017 at 00:41:52, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S1053837200001863 204 JOURNAL OF THE HISTORY OF ECONOMIC THOUGHT changes.3 He sought to prove that in general, a change in the price of any good must change the marginal utility of income.4 He reasoned as follows:5 Suppose that the household consumes goods b, c, etc. in the quantities, rb, rc, etc. and that the marginal utility of good i, denoted </>,, depends only on the level of consumption of good i: (j)b = <j>b{rb), <j)c = <t>c{rc), etc.6 Goods b, c, etc. are purchased at prices, pb, pc, etc. Finally, assume that the marginal utility of income, m, does not vary with pb, pc, etc. In this case, the first order conditions for maximizing utility, m = (j)b{rb)lpb = <t>c{rc)lpc = ... (1) 7 can be solved for r^ rc, etc. This yields rb = i//b(mpb), rc = ^c(mpc), ... (2) where i/f,(m/?,) = (f>r '(>;)• If qa is the constant income available to the individual to spend buying market goods, then substituting the solutions for n, rc, etc. in equation (2) into the budget constraint yields qa = Pb^bimpb) + Pc^cimpc) + ... (3) Since m is constant by assumption and the pi's are all independent of each other, qa can only be constant if for the given value of qa each term of the form Piif/iimpi) is constant. Pareto denoted the constant value of p,i/',(mp,) by Aj. Substituting this into equation (3) yields qa = Ah + Ac+... (4) Equation (4) implies that total spending on each good must be a constant fraction of income if the marginal utility of income is to be constant. Since Cobb-Douglas utility functions have become popular pedagogical tools, modern economists immediately recognize that Pareto has shown that utility must be a monotone increasing function of the Cobb-Douglas functional form, 3 A number of authors have argued that Marshall only assumed that the marginal utility of income is nearly constant following a change in income. For example, George Stigler (1950) claims that it "seems beyond doubt that Marshall treated the marginal utility of income as approximately, and not rigorously, constant, and fairly clear that it is constant with respect to variations in the price of a commodity whose total cost is not too large a part of the budget." In his discussion of the constancy of the marginal utility of income, Paul Samuelson (1942, p. 79) discussed two different meanings which have been attached to the phrase. One refers to the constancy of the marginal utility of income with respect to a change in any price. The other refers to constancy with respect to income. By showing that the marginal utility of income must be homogeneous of degree minus one in income and all prices, Samuelson, showed that the marginal utility of income cannot be simultaneously constant with respect to income and all prices. 4 Peter Dooley (1983b) presents an interesting account of the debate over the possible constancy of the marginal utility of income following the publication of the first edition of Marshall's Principles in 1890. 5 The notation used here follows Pareto (1892a) closely. 6 That is, Pareto assumed here, as he occasionally did elsewhere, that the total utility function is additively separable. 7 According to P. Dooley (1983b), Marshall (1890) was the first to show how to solve for demand functions from marginal utility functions when the total utility function is additive. Downloaded from https:/www.cambridge.org/core. IP address: 88.99.165.207, on 13 Jul 2017 at 00:41:52, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S1053837200001863 PARETO AND THE COBB-DOUGLAS FUNCTION 205 U = rj,ryc... , if the marginal utility of income is to remain constant when prices change. Although he did not actually write down the particular function of U = r\ryc... which must apply, Pareto did point the reader in the right direction. He noted that since equation (4) implies /?, = A/r,-, the first order conditions in equation (1) can be rewritten (j>b{rb) - mpb = mAb/rb, (j)c(rc) = mpc = mAJrc, ... (5) so that simple integration (which Pareto did carry out) yields the separate utility functions for rb, rc, etc., which are denoted &b, <PC, etc. &b(rb) = mAb\nrb, <Pc(rc) = mAc\nrc, ... (6) The results in equation (6) can then be combined to yield the additive log-linear version of the Cobb-Douglas total utility function U(rb,rc, ...) = 0b(rb) + <Pc(rc) + ...= mAb\nrb + mAc\nrc + ... (7) Thus, by considering marginal utility only (no expression similar to equation (7) appears in Pareto's analysis), Pareto arrived at the fact that Cobb-Douglas utility implies constant budget shares for all goods. Based on this analysis, Pareto rejected the possibility that the marginal utility of income might remain constant when the price of a good changes on the grounds that marginal utilities such as those in equation (5) are too special to hold in the real world:8 "Hence, it would be necessary for the final degree of utility of all the commodities to have that extremely singular expression for the final degree of utility of a capital good to remain constant. Which amounts to saying that it will never be so" (emphasis in the original). Pareto also derived part of another property of Cobb-Douglas utility functions. He noticed that the price elasticity of demand must be constant if marginal utilities take the form in equation (5). However, he did not go so far as to say explicitly that the price elasticity must be unity, nor did he discuss income elasticities at all. III. FUNCTIONAL FORM IN THE MANUAL Pareto returned to the question of the constancy of the marginal utility of income in the Manual (1909, sections 56-59, pp. 426-29). He used essentially the same method of analysis and produced the same results as he had earlier. However, there are several small changes in the appendix to the Manual compared to the earlier analysis, as well as several small flaws in the later analysis which bear mention. In the Manual, as in much of his other analyses, Pareto considered "general equilibrium" demand curves, where the household is endowed not with money income, but with non-negative quantities of some of the goods it consumes. It sells a portion of its endowment to obtain income which it uses to purchase those It is interesting to note that after he had gone to some lengths to show that the marginal utility of income had to change in response to a change in price, several pages later, Pareto discussed cases in which the marginal utility of income would be nearly constant following a change in price. Downloaded from https:/www.cambridge.org/core. IP address: 88.99.165.207, on 13 Jul 2017 at 00:41:52, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S1053837200001863 206 JOURNAL OF THE HISTORY OF ECONOMIC THOUGHT goods which it wants to consume in excess of its endowment. This leads to budget constraints of the form found in Pareto's equation (51) (Pareto, 1909, p. 414): + ... (8) 0 = x- x0 + py(y-yo)+pz(z-zo) where xo, yo, Zo, etc. are the household's initial endowments of the goods, and py, pz, etc. are the relative prices of goods y and z in terms of the numeraire, good x. Using this budget constraint, Pareto derived the following comparative static results for additive utility functions using methods which he had pioneered in Pareto (1892b, 1893):9 T = l/*« + pfeyy + p\l$a +..., = (y-yo + $y/<Pyy)/T, $a, (9a) (9b) (9c) where <P,, is the second derivative of the utility function with respect to good i, and the second-order cross partial derivatives, <P,y with / ±j are all zero, so that the total utility function is once again assumed to be additively separable. Pareto's discussion of the implications of equations (9a)-(9c) was incomplete. Several additional points related to constancy of the marginal utility of income had either already been made by Arthur Berry (1891) and F. Y. Edgeworth (1891) or were added later by Paul A. Samuelson (1942). Two of these points made by other authors merit brief mention. First, Pareto understood by 1900 at the latest that economic theory requires only ordinal utility indices (see Pareto, 1900a, b; Schumpeter, 1954, p. 1062; and Weber, 1997b). Economic theory requires only that consumers be able to rank different consumption bundles according to whether they are indifferent to each other or one is strictly preferred to the other. Pareto reiterated this point in the mathematical appendix to the Manual (sections 4—18, pp. 392-403). What he did not realize was that this methodological improvement was the last nail in the coffin of constant marginal utility of income. Any restriction on the marginal utility of income, including the possibility that it might be constant, clearly requires that utility be cardinally measurable; otherwise such a restriction is meaningless. Once economics had discarded cardinal utility functions, cardinal restrictions on the marginal utility of income had no place in the analysis. If Pareto understood this, he did not discuss it in the Manual. It appears that Samuelson (1942) was the first to make this point. Second, despite his careful discussion of T in equations (9b) and (9c) Pareto did not recognize a second case in which T is infinite. That is the case of the quasi-linear utility function: U(x, y, z, ...) =Ax + u(y, z, ...). where A is a constant, so that utility is linear in exactly one good (Varian, 1992). Since ^.a = 0, T= oo. Here, one of the first-order conditions is: m = Alpx, so that m depends only on px. dm/dp* = 0 for i i1 x. This point had been made by Arthur 9 The reader will notice a slight difference in notation between Pareto's "Considerazioni ..." and the appendix to the Manual. Downloaded from https:/www.cambridge.org/core. IP address: 88.99.165.207, on 13 Jul 2017 at 00:41:52, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S1053837200001863 PARETO AND THE COBB-DOUGLAS FUNCTION 207 Berry (1891) and Edgeworth (1891) and was later reiterated by Samuelson (1942). No doubt, Pareto neglected this case since he assumed diminishing marginal utility for all goods throughout his analysis, both in the appendix to the Manual and elsewhere. In addition to these lacunae, Pareto's analysis is subject to the following criticisms: First, although he showed that if the marginal utility of income remains constant when prices change, then uncompensated cross price effects must be zero, his proof did not use marginal utility functions of the form discussed above, and he apparently failed to recognize that Cobb-Douglas preferences imply zero uncompensated cross price effects. At the very least, he did not state this result explicitly. Instead, he argued using equations (9a)-(9c) above. He recognized from equation (9c) that if the marginal utility of income, m, did not change in response to a change in py, then what we would today call the uncompensated cross price effect, dz/dpy, would necessarily be zero. However, he discussed two conditions under which the marginal utility of income would be constant in the face of a change in py. One is the case in which T in equations (9b) and (9c) is "very large." Pareto stated that constant marginal utility of income implies zero cross price effects following his discussion of the case where T is very large (section 58). He recognized that this provides an argument against the constancy of the marginal utility of income in addition to that quoted above from the "Considerazioni sui principii ..." of 1892. Referring to the hypothesis that the marginal utility of income is constant, he observed: "Hence, the hypothesis is tantamount to assuming that when py varies, only the quantity of y varies, while z, u,... remain constant. This assumption may be admissible in certain cases, but in general it is inadmissible" (Pareto, 1909, p. 427). The other case in which the marginal utility of income is constant is that where the numerator of equation (9b) is zero. Following his discussion of the case where T is very large, Pareto considered this second case. The numerator of equation (9b) will be zero so that the marginal utility of income does not vary with the price of good y if: dm/dpy = 0 if: y - y0 + $>/<£yy = 0, or: 0y = Bliy - y0) where B is an arbitrary constant. Integrating equation (10) yields a logarithmic total utility function in net trades of y, y — y0. Again, this shows that marginal utility must take the reciprocal form of equations (5), so that utility had to be logarithmic. However, Pareto did not mention that in this case the cross price effects would be zero. At least in print, Pareto did not make the connection between Cobb-Douglas preferences and zero cross price effects. Finally, in the Manual, as in the rest of his work, Pareto neglected income effects in his comparative statics analysis of demand. One crucial result of this neglect was that although Pareto pioneered the method of analysis which E. E. Downloaded from https:/www.cambridge.org/core. IP address: 88.99.165.207, on 13 Jul 2017 at 00:41:52, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S1053837200001863 208 JOURNAL OF THE HISTORY OF ECONOMIC THOUGHT Slutsky (1915) later used to derive separate income and substitution effects of a change in price, Pareto did not discover these effects himself.10 For present purposes, however, a more interesting result of this neglect occurs in the Manual (Pareto, 1909, p. 428). From equation (10), Pareto reasoned that constancy of the marginal utility of income required that the marginal utility of good y depend on both the level of consumption of _y, and on yo, the initial endowment of good y, rather than on the actual level of consumption only. Since he found this implication unacceptable, he believed that he had found yet a third argument against a constant marginal utility of money. What he apparently failed to recognize was that the household's consumption of all goods, including good y, depends in general on the market prices of the goods and on the market value of the household's initial endowment of the various goods. In other words, the general form of the demand function for y is: y = y(Py>P* •••;x0,y0,zo, ...). (11) Moreover, defining the market value of the household's initial endowment (measured in terms of the numeraire) as: s = xo + pyyo + pzzo + . . . , (12) the Cobb-Douglas functional form for the utility function implies: y = Bslpy = B(x0 + pyyo + pzzo + ••• )/py, (13) where B is a constant. Equation (13) shows that a change in yo alters y. Thus, at constant prices, the first-order conditions in equation (1) imply that a change in y0 does affect the marginal utility of income. Pareto's attempt to argue that the marginal utility of income should not depend on initial endowments is fundamentally flawed and only two of his three arguments against a constant marginal utility of money stand: A constant marginal utility of money does imply that total utility must be log linear, which is highly restrictive. Also, a constant marginal utility of money implies that uncompensated cross price effects must be zero, which can be rejected empirically. However, one cannot argue that a constant marginal utility of income should be rejected since it implies that marginal utilities depend on endowment levels. IV. CONCLUSION On purely aesthetic grounds, there is no need to rename the Cobb-Douglas utility function the Pareto-Wicksell-Cobb-Douglas utility function." However, it is worth noting that in addition to his other achievements, Pareto did present an early, 10 On the relationship between the comparative statics analyses of Pareto and Slutsky, see Dooley, 1983a. On Pareto's neglect of income effects, see Weber, 1997a. '' It appears that neither Wicksell nor Cobb and Douglas ever used this functional form in connection with consumer preferences. Two of the earliest authors to use Cobb-Douglas utility functions were R. G. D. Allen (1938) and Paul Samuelson (1942). Downloaded from https:/www.cambridge.org/core. IP address: 88.99.165.207, on 13 Jul 2017 at 00:41:52, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S1053837200001863 PARETO AND THE COBB-DOUGLAS FUNCTION 209 if incomplete, discussion of the Cobb-Douglas utility function.12 His conclusions were generally sound, and anticipated closely results which Samuelson (1942) would use half a century later to put to rest forever economists' concerns over the possible constancy of the marginal utility of money. The logic and clarity with which Pareto stated his results and their implications make it all the more disappointing that, as Peter Dooley (1983b, p. 36) notes, Marshall and his defenders largely dominated thinking on the marginal utility of income for four decades until more careful analysis was applied to the question in the 1930s and '40s. REFERENCES Allen, ,R. G. D. 1938. Mathematical Analysis for Economists, St. Martin's Press, New York. Berry,-Arthur. 1891. "Alcune brevi parole sulla teoria del trattato di A. Marshall," Giornale degli Economisti, 2, 549-53. Cobb, Charles and Paul Douglas. 1928. "A Theory of Production," American Economic Review, 18, March, 139-65. Dooley, Peter. 1983a. "Slutsky's Equation is Pareto's Solution," History of Political Economy, 15, Winter, 513-17. Dooley, Peter. 1983b. "Consumer's Surplus: Marshall and His Critics," Canadian Journal of Economics, 16, February, 26-38. Douglas, Paul H. 1976. "The Cobb-Douglas Production Function Once Again: Its History, Its Testing, and Some Empirical Values," Journal of Political Economy, 84, October, 903-15. Edgeworth, Francis Y. 1891. "Osservazioni sulla teoria matematica dell'economia politica con riguardo speciale al principi di economia di Alfredo Marshall," Giornale degli Economisti, 2, _ 233-45. Geary, R. C. 1949-1950. "A Note on a Constant Utility Index of the Cost of Living," Review of Economic Studies, 18, 65-66. Klein, L. and H. Rubin. 1947-1948. "A Constant Utility Index of the Cost of Living," Review of Economic Studies, 15, 84-87. Marshall, Alfred. 1890. Principles of Economics, Macmillan, London. Olsson, Carl-Axel. 1971. "The Cobb-Douglas or the Wicksell Function?," Economy and History, 14, 64-69. Pareto, Vilfredo. 1892a. "Considerazioni sui Principii Fondamentali dell'Economia Politica Pura," Part II, Giornale degli Economisti, 4, June, 485-512. Pareto, Vilfredo. 1892b. "Considerazioni sui Principii Fondamentali dell'Economia Politica Pura," Part III, Giornale degli Economisti, 5, August, 119-57. Pareto, Vilfredo. 1893. "Considerazioni sui Principii Fondamentali dell'Economia Politica Pura," Part V, Giornale degli Economisti, 7, October, 279-321. Pareto, Vilfredo. 1900a. "Sunto di alcuni capitoli di un nuovo trattato di economia pura," Part I, Giornale degli Economisti, 20, March, 216-35. Pareto, Vilfredo. 1900b. "Sunto di alcuni capitoli di un nuovo trattato di economia pura," Part II, Giornale degli Economisti, 20, June, 511-49. Pareto, Vilfredo. 1909. Manual of Political Economy, translated by A. S. Schwier, Augustus M. Kelley, New York, 1971. 12 Pareto (1900b) also wrote down the functional form for the linear expenditure system (LES) utility function, U(x, y) = (.v - af(y - bf, nearly fifty years before L. Klein and H. Rubin (1947-1948), Paul Samuelson (1947-1948), R. C. Geary (1949-1950), and Richard Stone (1954) analyzed it. However, he merely drew the indifference curves for this utility function and noted that they slope downward and are convex to the origin. He did not discuss the empirical implications of this functional form as he had (in part) for the Cobb-Douglas utility function. Downloaded from https:/www.cambridge.org/core. IP address: 88.99.165.207, on 13 Jul 2017 at 00:41:52, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S1053837200001863 210 JOURNAL OF THE HISTORY OF ECONOMIC THOUGHT Samuelson, Paul A. 1942. "Constancy of the Marginal Utility of Income," in O. Lange, ed., Studies in Mathematical Economics and Econometrics in Memory of Henry Schultz. University of Chicago Press, Chicago, 1942. Samuelson, Paul A. 1947-1948. "Some Implications of 'Linearity,'" Review of Economic Studies, 15, 88-90. Schumpeter, Joseph A. 1954. History of Economic Analysis, Oxford University Press, New York. Slutsky, E. E. 1915. "On the Theory of the Budget of the Consumer," Giornale degli Economist, 51, 1-26; reprinted in G. J. Stigler and K. E. Boulding, eds., A.E.A. Readings in Price Theory, Richard D. Irwin, Chicago, 1952. Spiegel, Henry W. 1991. The Growth of Economic Thought, 3d ed., Duke University Press, Durham. Stigler, George J. 1950. "The Development of Utility Theory," Journal of Political Economy, 58, August, 307-27 and October, 373-96; reprinted in G. J. Stigler, ed., Essays in the History of Economics, University of Chicago Press, Chicago, 1965. Stone, Richard. 1954. "Linear Expenditure Systems and Demand Analysis: An Application to the Pattern of British Demand," Economic Journal, 64, 511-27. Varian, Hal R. 1992. Microeconomic Analysis, W. W. Norton, New York. Weber, Christian E. 1997a. "More on Slutsky's Equation as Pareto's Solution," History of Political Economy, forthcoming. Weber, Christian E. 1997b. "Pareto and the 43 Percent Ordinal Theory of Utility," unpublished working paper, Seattle University. Wicksell, Knut. 1901, 1906. Forenla'sningar i nationalekonmi, translated by E. Classen, as Lectures on Political Economy, edited by L. Robbins, Routledge and Kegan Paul, London, 1934-1935. Downloaded from https:/www.cambridge.org/core. IP address: 88.99.165.207, on 13 Jul 2017 at 00:41:52, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S1053837200001863
© Copyright 2024 Paperzz