Author(s): Paul A. Samuelson Reviewed work(s): Source: The American Economic Review, Vol. 75, No. 2, Papers and Proceedings of the Ninety- Seventh Annual Meeting of the American Economic Association (May, 1985), pp. 166-172 Published by: American Economic Association Stable URL: http://www.jstor.org/stable/1805590 . Accessed: 13/02/2013 17:10 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp . JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org. . American Economic Association is collaborating with JSTOR to digitize, preserve and extend access to The American Economic Review. http://www.jstor.org This content downloaded on Wed, 13 Feb 2013 17:10:36 PM All use subject to JSTOR Terms and Conditions FRONTIERS IN DEMOGRAPHIC ECONOMICSt Modes of Thought in Economics and Biology By PAUL A. SAMUELSON* Sociobiology, particularly in connection with its obsessive interest in Hamiltonian altruism, has veered closer towards econom- ics. Reactions within economics against highfalutin borrowings of the methodology of mathematical physics led Alfred Marshall and a host of later writers to hanker for a "biological" approach to political economy. A dispassionate audit of what resulted from this urge would have to report a disappoint- ing paucity up to now of fruitful finds or insights. But independently of this anti-physicism strain in economics there developed within economics an "evolutionary approach." It is not social Darwinism that I am referring to (although an insidious temptation in that direction is sometimes involved). Rather, several writers explored the notion that- what survives under competition, whether in the jungle or the marketplace, may resemble what is achieved in a maximum problem's solution-even though no participants in the struggle may have any perception and aware- ness that they as individuals are maximizing anything or awareness that the group ends up doing so. Some names associated with this notion are Armen Alchian (1950), Mil- ton Friedman (1953), and Sidney Winter and Richard Nelson (1982). My Foundations (1947) discussions of the stability of surviv- ing forms, borrowed explicitly from the physiologist philosopher L. J. Henderson (1917), is somewhat in this same vein. There is really no trace of anti-physicism in such an approach. Indeed, the case where soap bubbles unconsciously form the shapes that maximize elaborate integrals in the calculus of variations is very much like the case where leaves and branches of a tree, competing for access to the sunlight, achieve maximal shapes. When Edward Wilson's manifesto for soci- ology first came out (1975), I predicted that it would meet a resonant response in Chicago-school economics. Everything has proceeded on schedule to validate my fore- cast. A hard-boiled economist who shares his income with his child can now cease to feel guilty about this superficial violation of the doctrine of "no free lunch." Hamiltonian inclusive fitness shows that this is just a sel- fish gene's way of making more of itself- no less respectable a process than that in which I cannily outsell or outproduce my neighbor. I. Population as Part of Economics Population analysis is at the intersection of social science and biology. Once upon a time, throughout the heyday of classical eco- nomics, demography belonged to political economy. The supply of labor was one of the important endogenous variables in the sys- tems of Smith, Malthus, Mill, and Marx. I have expressed in many places how super- ficial and empty the concept of a subsistence wage has always been. Although economies in the pre-industrial revolution era differed widely in their per capita levels of real in- come, writers blithely embraced the para- digm of a horizontal SS supply curve for labor, as if the cost of production and repro- duction of labor was an identifiable exoge- nous parameter. There must be something very tempting about this vacuous notion, since so many different scholars did succumb tDiscussant: Gordon Tullock, Public Choice Center, Virginia Polytechnic Institute and State University. *Department of Economics, Massachusetts Institute of Technology, Cambridge, MA 02139. I owe thanks to the Sloan Foundation for nartial sunnort- 166 This content downloaded on Wed, 13 Feb 2013 17:10:36 PM All use subject to JSTOR Terms and Conditions VOL. 75 NO. 2 FRONTIERS IN DEMOGRAPHIC ECONOMICS 167 to it. And, by the age of Mill and Marx-in- deed, even in Ricardo's time and Malthus's later editions-the level of the subsistence wage rate was freed from physiological di- mensions and became a dummy variable shot through with conventional standards of life hankered after by (some) workers. At this stage a meaningful false theory had been replaced by a meaningless nontheory that could never be vindicated or rejected by any pattern of historical facts. II. Paradigm Lost and Regained One feature of neoclassical economics that distinguishes it from the classical version is the removal of population as a variable subject to economists' equilibrium analysis. Knut Wicksell, an autodidact in a backwater of northern Europe, was the last neoclassicist to preoccupy himself with demography. Marshall, Edwin Cannan, Frank Knight, and A. C. Pigou all were content to have the trend of population growth imported from the fields of sociology and technical demog- raphy. The pioneers of modern demography -Alfred J. Lotka, Robert R. Kuczynski, David Glass, Kingsley Davis, Ansley J. Coale, Nathan Keyfitz, and others-were not primarily economists and made little pre- tence to be. The classical subsistence wage theory is dead forever- dead as a theory of constancy of wage even as applied to the poor regions of Africa and Asia where gross reproduction rates are still enormously high and where population density perceptibly reduces per capita productivity. So the real wage rate is definitely not an exogenous vari- able for today's economics. Nevertheless, in recent years economists have begun to in- filtrate the field of demography. This is part of the imperialist movement in which we economists try to apply our methodologies to everything- to the law, to the sociology of the family (courtship, marriage, divorce, cohabitation). I have in mind writings by such economists as Richard Easterlin, Paul Schultz, Gary Becker and a whole Chicago school, Ronald Lee, Harvey Leibenstein, and many others. As might be expected, noneconomists have sometimes resented invasion of their turf. Demographers are not as titillated as we are by letting the decision to have or not have a child be likened to the decision to buy or not buy a new car. Easterlin aside, the econo- mists' models did not predict in advance the magnitude of the post-1939 baby boom; nor the post-1957 resumption of the trend to- ward small family sizes. Slutsky-like analysis of the income-and-family size relationship labors like a lion with concepts of quality and alternative earning power of spouses, only to produce the mouse that sometimes these days it is fashionable to have 1 or 2 children and sometimes to have 2 or 3. Weak and hard-to-reproduce evidences for ex- trasensory perception, people find boring. Similarly, a weak tendency for people to have more or to have less of their children during cyclical recessions does not excite noneconomists. Economists are smart and hard-working people, now being produced in copious supply relative to the traditional problems feasible to make progress on. Like lemmings oppressed by the workings out of the law of diminishing returns, economists will con- tinue to swarm into the area of demography. Hard work and intelligence will get them- selves heard, even when somewhat hamstrung by unpromising initial hypotheses and meth- ods. So, in the end, economists' demography will carve out an ecological niche for itself. III. Anecdotes and Anecdotes A Kuhnian expert in the dynamics of sci- entific schools will observe how sociobiology extrudes into the traditional grounds of the social sciences. Francis Galton, Karl Pear- son, and Ronald Fisher sought genetic ex- planations for differences in performance and class position. Although R. A. Fisher was a genius in genetics and mathematical statis- tics, his 1930 classic, The Genetic Theory of Natural Selection, seems naive in the degree to which he hypothecates a genetic basis for observed differences in behavior. No cogent evidence is given for his views and indeed it is not clear what evidence would be cogent. We are beginning to identify today, by means of biochemistry and molecular bi- ology, particular diseases with particular sin- This content downloaded on Wed, 13 Feb 2013 17:10:36 PM All use subject to JSTOR Terms and Conditions 168 A EA PAPERS AND PROCEEDINGS MA Y 1985 gularities in DNA data. But just how the courage observed in offspring is related to the genes for courage inherited from their ancestors is as yet a completely nonoper- ational question. The same has to be said of " altruism," "intelligence," "sexuality," and most of the attributes discussed in the final chapters of Wilson. These final chapters, I noticed at first reading, depend largely upon hypothetical anecdotes and case studies. Ini- tially, I thought this a weakness of the material dealing with man and with higher primates, an understandable let down from the higher level of scientific rigor characteriz- ing the earlier chapters dealing with the solid facts of animal and plant biology. But on further examination and reflection, I dis- covered that much of what goes under the name of natural history involves a similar marshaling of selective anecdotes. Chemistry and meteorology are not like that. And neither is economics. In economics we have data aplenty. Our subject is inexact but we do apply quantitative analysis (statistical and otherwise) to formulate and test our hy- potheses. When you read about Volterra's struggle for existence between predators and prey, or Wilson's programs for a genetic- based ethics, you are not in the constrained world of fact and interpretative hypothesis but rather in the imaginative realm of specu- lative possibility. To be sure, more and better observations will be made available by future scholars working in this vineyard. But the point to be made is how far away we now are from a satisfactory state of knowledge in these be- tween-disciplines domains. Rather than belabor these points about realism and relevance, I wish for the rest of the present endeavor to deal with purely deductive matters, to illustrate by some ex- amples involving sex ratios how unalike (and alike) are the deductive paradigms common to economics (and perhaps sociology) and those of sociobiology and genetic demog- raphy. IV. Different Moods There are important differences between a typical genetic process and an economic or sociological process. Genetic evolutionary selection, as when there is survivability ad- vantage in a malarial environment for the genotype that is heterozygous for the sickle cell trait, involves no consciousness or overt purpose: any teleological aspect is an "as if" phenomenon. Ofttimes genetic change is gla- cially slow, involving hundreds of genera- tions. So to speak, there is no "action at a distance" in demographic genetics: funeral by funeral, mating by mating, each allele must make its way. By contrast, in econom- ics what one firm innovates to do now can in principle sweep the whole industry by quick imitation. As a typical process in biology, consider a society that relies solely on the rhythm method for its birth control (a specification already shot through with sociology and con- vention). Given enough time the strategy will partially or wholly self-destruct-as those genetically disposed to have irregular men- strual cycles leave more of their offspring to be represented in subsequent populations. But the biological time scales involved run into hundreds of years, by which time there will have occurred scores of alternations of sociological ideologies. The sex ratio is a useful topic for our comparison. V. Balanced Sex Ratio Human babies are almost as likely to be girls as boys: typically males are in small excess, 106 to 100; but females have lower mortality so that by adulthood the sex ratio is in virtual balance. Animal populations, whether monogamous or not, are observed often to have nearly balanced sex ratios at birth, with whatever unbalance that implies in maturity. R. A. Fisher gives an ingenious argument why individual natural selection might be expected to evolve toward a balanced sex ratio at birth or conception. Here is my attempt to paraphrase his logic: Suppose all genetic strains but yours tend to produce an excess of male over female births. If your strain tends to produce a more balanced sex ratio of This content downloaded on Wed, 13 Feb 2013 17:10:36 PM All use subject to JSTOR Terms and Conditions VOL. 75 NO. 2 FRONTIERS IN DEMOGRAPHIC ECONOMICS 169 births, other things equal, each of you will be better represented with grand- children than each of the other crowd -this for the reason that, for pro- ducing the grandchildren generation, the totality of all sons have exactly equal representation with the totality of all daughters (even when the respec- tive totals are unequal!), with the impli- cation that each daughter brings you more in relative grandchildren repre- sentation than each son does. So, in each generation, the unbalance of the sex ratio at birth tends to be reduced. Only a sex ratio at birth in balance is immune to evolutionary drift. To see how mathematical geneticists put this heuristic argument on a rigorous analyti- cal basis, refer to I. Eshel (1975), Joel Yellin and myself (1977), and the Gibbs Lecture of Samuel Karlin (1984). To illustrate the logic it suffices for me here to sketch the rock-bot- tom simplest case. VI. Darwin's Mindless Invisible Hand Adult male and females, [M, F] mate to produce this period's births, B. The fractions [g, 1- g] determine the breakdown of new- borns between males and females. The mortality fractions [ Pm, Pf I specify how many of the newborn males and females survive to be adults in the next period. With mortality specified, our model is complete once we specify the mating-fertility function deter- mining the output of births out of the inputs of adult males and females. I write this as (1) B(t)=M(t)F(t)b[M(t),F(t)J, where MFb[M, F] is much like the econo- mists' production function, being monotone increasing (and perhaps concave). Examples would be cMF/(M + F), cMF/(-M + 2F), cMF/M1/3F1/2. Combine (1) with our mortality and sex- ratio assumptions (2) M(t) = pr [ gB(t -1)], F(t)-=pf [(I1-g)B(t -1)] to derive an autonomous first-order dif- ference equation determining births and all demographic variables: (3) B(t)=g(1-g)pmpfB(t - 1)2 x b [ pmgB(t-1), pf (l-g)B(t-1)] -=4[B(t-1)]J, t 21. Thus, if MFb[M, F] is first-degree homo- geneous, we find Malthus-Lotka exponential growth: (4) B(t)=B(O)Rt, B (O) = M(O) F(O)b [ M(O), F(O)J 1 R = g(1 - g)pmpfb[gpm(1 - g)pf] If both sexes played a symmetric role in the MFb function, maximal group fitness would require that any superiority of females in mortality ought to be compensated by hav- ing an excess of males at birth g > 2. The above Fisherine argument denies that the Invisible Hand of individual selection will lead to this group fitness state, leading in- stead to balanced sex ratio at birth (g= =2 whatever that implies for unbalance in the adult sex ratio! Now introduce genetic selection. The old genetic strain that (M, F, B, g) above be- longed to, I now write as (MA, FA, BA, gA). A mutation creates also a new genetic strain, with (Ma, F, Ba, ga). (To keep the discus- sion simplest, I don't deal with diploid in- heritance, which would require us to intro- duce [MAA, Maa, MAa, FAA, ...I , etc.) The axioms of the Fisherine syllogism are very strict: 1. The genotypes differ only in sex ratio: gA * ga- 2. The same survival fractions [Pm' PfI apply to both genotypes. 3. Where fertility-mating is concerned, an individual of genotype A has exactly the same properties as any like-sexed, like-aged individual of genotype a. This implies ran- dom or nonassortative mating and that total This content downloaded on Wed, 13 Feb 2013 17:10:36 PM All use subject to JSTOR Terms and Conditions 170 AEA PAPERS AND PROCEEDINGS MA Y 1985 births, B = BA + Ba, are determined as be- fore by total (M, F) numbers: (5) BA+ Ba= (MA+ Ma)(FA+ Fa) X b[MA+ Ma, FA+ Fa]. For short, the last factor on the right will be written as b. 4. The last axiom specifies that offspring of parents both of the same genotype are of that genotype; half the offspring of parents of different genotype average out to be of the father's genotype, half of the mother's, and sampling irregularities are assumed ignorable in a large enough population. The axioms translate in a straightforward way to give the following first-order dif- ference equation system determining how the vector [BA(t), Ba(t)] = [BA, Ba] transforms itself into [BA(t + 1), Ba(t + 1)] = [BA, Ba]: (6A) B= { ga(1-gaI)Ba [gA(1-ga) + ga(l-gA )]BABa { pmpfb}- (6a) B = { gA(lgA)BA 2 [gA(l ga) ? ga(-gA)] BABa }{pm pfb; Suppose we are interested only in ratios of genotypes, B,/BA = x(t), and in the ratio of the sexes at birth, as measured by g(t) the properly weighted average of [gA, ga]. We can divide (6a) by (6A) and cancel out com- pletely both any differences in the sexes' mortality fractions and any special features of the b[M, F] function that happen to ob- tain! We then get the autonomous first-order difference equation in [x(t), x(t + 1)]: (7) x(t+1)=x(t)X 9al(1-g.)X(t)+ 12 [9A(1 9a)+9,(1 gA)] gA(1-gA)+X(t)2 [gA(1-ga)+ga(19gA)] It can then be shown graphically or analyti- cally that, for 0 <x(O) < o, as t -x oo the overall sex ratio g(t) -2 ' if (gA, ga) straddle 4; if they are both on one side of 4, g(t) the (ggA, g) nearest to 4. Warning: Balance of the sexes at birth would obtain in this model even if males lived longer than females, Pm > Pf, and even if females were more important for nurturing and if a small fraction of the available males would suffice for purposes of procreation. Fitness of the group, when it calls for bal- anced sex ratio of adults or such adult unbal- ance as implies a strong unbalance at birth, will be sacrificed to that maximizing individ- ual fitness. Economists understand well that what's good for each may be bad for all sellers-as in Prisoner's Dilemma, or in per- fect competition.1 VII. Economic and Sociological Processes New techniques are making it possible to know in advance what will be the sex of a baby. With knowledge may come power to control. Thus, if one learns by amniocentesis that an embryo is female, and if one prefers a male, an abortion might be decided on. If such customs become common, some far- reaching sociological changes could occur. And these would have economic ramifica- tions. Nor is all this new. Animal breeders and folklore have held that the probabilities of the different sexes can be affected by timing patterns within the menstrual cycle. Indeed, it is reported in Too Many Women? by M. Guttenberg and P. F. Secord (1983) that the 1See Fisher (pp. 142-43), W. D. Hamilton (1967), Wilson (pp. 316-17), R. L. Trivers and D. E. Willard (1973), for alternative models that alter conclusions about balanced sex ratios by bringing in considerations of " parental investment" and other interactions between genotype and fertility or mortality functions and param- eters. Where an economist would say: " Parents will invest in one sex or the other up to the point of equality of marginal advantage," biologists more often speak of equality of investments in the two sexes. Equality of derivatives and equality of ordinates are not always the same thing! Actually, selection would lead to two-thirds of newborns being male if the (M, F) symbols in (1)-(7), instead of meaning (males, females), had to be inter- preted to be (pairs of males, single females)-as when each female uses up twice the "energy" or "mass" of one male. This content downloaded on Wed, 13 Feb 2013 17:10:36 PM All use subject to JSTOR Terms and Conditions VOL. 75 NO. 2 FRONTIERS IN DEMOGRAPHIC ECONOMICS 171 Talmudic practices followed by nineteenth- century orthodox Jews of Eastern Europe resulted in a sex ratio as distorted as 146/100. Also, the mystery of the alleged excess of male births occurring in the World War I epoch may even be resolvable by prosaic factors having to do with scheduling of army leaves and discharges during that period. Whatever the mixture of science fiction and fact characterizing present knowledge, it is not unlikely that better control over gender is just around the corner for reasons that have nothing to do with genetic change. How do economists model such a process? I shall deal only with an archetypal example. Scarcity tends toward value. The more women there are, other things equal, the lower might be expected to be their relative wages and lifetime earnings. Feminists might therefore look with some approval upon a future trend toward more sons rather than daughters. Working against this in the political sphere is that scarcity leads to weakness of voting position and respect. "Expand thy numbers," is the injunction each identity group re- sponds to. Here I shall stay with the economist's case where per capita real in- comes are hurt by abundance of numbers. The simplest case is where the real na- tional product, Q, is produced by the male and female adult labor forces, (M, F), with the imputed real wage to each being de- termined by respective marginal products: (8) Q= Q[M, F] = Mq[F/M], q' > O > q" Wm = dQ[M, F]/dM= q'[M/F]; Wf = dQ[M, F]/dF= 4P(Wm), '<O. Warning: a polar feminist model that sup- posed male and female inputs to be identical factors of production would make Q be c(M + F) and necessitate qualifying much of the following analysis. If some people indulge a strong enough preference for sons, that will skew the adult sex ratio upward, tending to raise female wage rates and lower male rates. If women are no less productive than men, in the sense that Q[M, F] is a symmetric function equalling Q[F, M], then the female wage will exceed the male. Such an elevation in relative remuneration may serve partially to reverse the sex ratio imbalance. One past reason for preferring sons may have been their superior earning power. Al- though daughters may perhaps be counted on to be more nurturing to you in your old age, sons may have the greater wherewithal to support you. Or the vanity of having the family name carried on may, under our patriarchal culture, be better served by sons. But now that the contrived scarcity of females raises their earnings, you have a new eco- nomic motive to indulge your preference for males less. The new equilibrium, under the postulated symmetry of Q(M, F) and specified prefer- ence bias toward males, will be toward an excess of males but an excess limited by their induced impoverishment. The approach to the new equilibrium could involve successive over- and undershoots (as in the economist's "cobweb" model, where each person doesn't realize how many others are also doing the same thing). Or the approach may be grad- ual in the fashion of adaptive expectations. To keep up with the latest fad, we could even fabricate a rational expectations model: in it, people make a best guess about what the future development of the sex ratio will be; they take account of what is implied by this for relative earnings of their sons, daughters, grandsons and granddaughters, and in terms of this make their decisions; and, miracle of miracles, when all do this they together con- trive what it is that they each expect. Again we discover the possibility of a con- flict between group (economic) selection and individual (economic) selection. If two un- friendly societies compete, the more pros- perous one (Sparta) may eliminate the other. However, the M/(F + M) ratio that maxi- mizes total Q (of Athens) may not be the sex ratio that best pleases its representative citi- zen. So each following self-interest may achieve the destruction of all. Observation: genetic evolution can work to offset economic unbalancing of the sex ratio. To see this best, suppose pf = Pm and This content downloaded on Wed, 13 Feb 2013 17:10:36 PM All use subject to JSTOR Terms and Conditions 172 A EA PAPERS AND PROCEEDINGS MA Y 1985 that only a subset of the population exercise gender control by abortion or otherwise. The groups who do and do not control gender need not be genetically defined. None the less, the dynamic Equation (7) could still apply, provided only we can suppose that when both parents belong to one group, so will the offspring. And that when the parents belong to different groups, half the offspring will be like one parent and half like the other. What will follow? The Fisher logic will then apply to this non-particulate-inheri- tance setup. With what result? Our defined process of sociological selection will lead to a rebalancing of the sexes! Indeed, the process can lead to a serendipitous balancing if there are two groups, one favoring sons and one daughters: each can then get their heart's desire, with their numerical weightings evolv- ing to keep the overall birth ratio in balance though no one is aware of the equilibrating process. Mere deduction cannot prove anything definite about real world sex ratios. If we alter the razor's-edge symmetries of the Fish- erine premises- admit nonrandom matings, etc., etc.-we of course can no longer deduce balanced sex ratios. Realistic economic mod- els are not too likely to honor the precise "other things equal" premises implicit in the Fisherine deduction. VIII. Finale There is much territory between econom- ics and biology that is still virgin ground. It will be tilled increasingly in the future. We should not be surprised if the first explora- tions are both crude and pretentious. Wis- dom and maturity are the last settlers to arrive in pioneering communities. REFERENCES Alchian, Armen A., " Uncertainty, Evolution and Economic Theory," Journal of Politi- cal Economy, June 1950, 58, 211-21. Eshel, I., " Selection on Sex-Ratio and the Evolution of Sex-Determination," Hered- ity, 1975, 34, 351-61. Fisher, R. A., The Genetical Theory of Natural Selection, Oxford: Clarendon Press, 1930. Friedman, Milton, Essays in Positive Econom- ics, Chicago: Chicago University Press, 1953. Guttenberg, M. and Secord, P. F., Too Many Women?- The Sex Ratio Question, Bev- erly Hills: Sage Publications, 1983. Hamilton, W. D., "Extraordinary Sex Ratios," Science, 1967, 156, 477-88. Henderson, L. J., The Order of Nature: An Essay, Cambridge: Harvard University Press, 1917. Karlin, S., "Mathematical Models, Problems, and Controversies of Evolutionary Theo- ry," Bulletin (New Series) of The Amer- ican Mathematical Society, April 1984, 10, 221-73. Samuelson, Paul A., Foundations of Economic Analysis, Cambridge: Harvard University Press, 1947. Trivers, R. L. and Willard, D. E., "Natural Selection of Parental Ability to Vary the Sex Ratio of Offspring," Science, 1973, 179, 90-92. Wilson, Edward, O., Sociobiology, Cambridge: Belknap Press, 1975. Winter, Sidney G. and Nelson, Richard R., An Evolutionary Theory of Economic Change, Cambridge: Belknap Press, 1982. Yellin, J. and Samuelson, P. A., "Genetic Fitness and the Sex Ratio: Natural Selection in Nonlinear Models of Population Growth," unpublished NIH paper 1977. This content downloaded on Wed, 13 Feb 2013 17:10:36 PM All use subject to JSTOR Terms and Conditions