Jump to ContentJump to Main Navigation
An Inquiry into Well-Being and Destitution$

Partha Dasgupta

Print publication date: 1995

Print ISBN-13: 9780198288350

Published to Oxford Scholarship Online: November 2003

DOI: 10.1093/0198288352.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2019. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 25 June 2019

Population and Savings: Normative Considerations

Population and Savings: Normative Considerations

Chapter:
(p.377) 13 Population and Savings: Normative Considerations
Source:
An Inquiry into Well-Being and Destitution
Author(s):

Partha Dasgupta (Contributor Webpage)

Publisher:
Oxford University Press
DOI:10.1093/0198288352.003.0019

Abstract and Keywords

The main part of this chapter discusses normative considerations on population and savings. It has five sections. The first discusses parental concerns on the well‐being of their children in relation to savings. The second discusses the Genesis Problem (which in its purest form asks how many lives there should be, enjoying what standards), and the Repugnant Conclusion (which, in Parfit's formulation states that ‘For any population of at least ten billion people, all with a very high quality of life, there must be some larger imaginable population whose existence, if other things are equal, would be better, even though its members have lives that are barely worth living). Section (3) questions whether the Repugnant Conclusion is repugnant when applied to comparisons of well‐being in the Genesis Problem, and section 4 argues that the Genesis Problem is irrelevant in real life, which addresses actual problems. Section (5) looks at population ethics. An extra and separate section (designated Chapter *13) gives theoretical presentations on classical utilitarianism in a limited world.

Keywords:   Genesis Problem, normative theory, population, population ethics, quality of life, Repugnant Conclusion, savings, utilitarianism, well‐being

13.1 Parental Concerns

In developing his contractual theory of justice among generations, Rawls (1972: 284–94) writes:

The parties do not know to which generation they belong or, what comes to the same thing, the stage of civilization of their society . . . Thus the persons in the original position are to ask themselves how much they would be willing to save . . . at any given phase of civilization with the understanding that the rates they propose are to regulate the whole span of accumulation . . . Since no one knows to which generation he belongs, the question is viewed from the standpoint of each and a fair accommodation is expressed by the principle adopted. All generations are virtually represented in the original position, since the same principle would always be chosen . . . Moreover, it is immediately obvious that every generation, except possibly the first, gains when a reasonable rate of saving is maintained . . . The process of accumulation, once it is begun, and carried through, is to the good of all subsequent generations. Each passes on to the next a fair equivalent in real capital as defined by a just savings principle . . . Only those in the first generation do not benefit . . . for while they begin the whole process, they do not share in the fruits of their provision. Nevertheless, since it is assumed that a generation cares for its immediate descendants, as fathers say care for their sons, a just savings principle . . . would be acknowledged. (Italics mine)

In this passage Rawls is concerned with intergenerational savings, not population policies. There is no suggestion in his book that he regards unalloyed contractualism to be capable of providing a basis for the latter. One may doubt that it is.1 But public policies bearing on fertility and savings decisions can't be kept independent of each other: desirable investment policies are a function of demographic profiles, and defendable population policies depend upon investment rates. The two need to be discussed simultaneously.2

(p.378) A number of authors have expressed the thought that, external effects aside, population and savings decisions don't involve social ethics. They have argued that considerate parents take into account the well‐being of their children when choosing their family size and deciding how much to save. If they are in addition thoughtful parents, they would know that the welfare of their children will depend upon the well‐being of their grandchildren, that the welfare of their grandchildren will in turn depend upon the well‐being of their children, and so on, down the generations. In short, there is a natural recursion of well‐being interests along a family line. Thoughtful parents could be expected to take account of their distant descendants, even when they are directly interested only in their own children. A variety of such recursive formulae have been put to work in economic models for the purposes of studying their implied investment rates (see Phelps and Pollak, 1968; Arrow, 1973b; Dasgupta, 1974a, b; Calvo, 1978; Rodriguez, 1981). Recursive formulae have also been used for a combined study of their implied fertility and investment rates (see Dasgupta, 1969, 1974c; Barro and Becker, 1989). It will prove useful to see what the argument amounts to.

Earlier, we took parental utility to depend upon parental consumption (c), the number of children (n), and the average quality of their children (z). We now reinterpret z to be the average welfare of the children, and for simplicity we identify welfare with well‐being. Consider generation t (≥0) of a family line. We assume there are N t members. Each person is to be thought of as a completely autonomous agent. I denote the representative member's welfare by W t. Writings by c t generation t's average ‘consumption’ stream, and by n t the number of children each member of this generation has, we can write

W t = W t ( c t , n t , W t + 1 ) .
(13.1)
It makes obvious sense to simplify and assume that W t doesn't depend explicitly on t. We can then simplify further by supposing that parents keep the worth of their consumption distinct from the other determinants of their welfare. An obvious form of this is
W t = U ( c t ) + δ n t L ( n t ) W t + 1 , where  1 δ > 0 .
(13.2)
In this expression δ, a constant, is a time discount factor, and L(n t) is a (concave) function of n t, given to assuming positive values only.

Repeated use of expression (13.2) implies that

W 0 = Σ δ t N t Q t U ( c t )
(13.3)
where Q t= ΠL(n τ), and N t is the size of generation t of this dynasty.3 (p.379) There is a problem with this though. Even when parents admit to a concern with the well‐being of their descendants through such a recursive formula, it doesn't follow that they award the right weights to the well‐being of their descendants. We began with thoughtful parents. Such folk ask themselves what are the correct arguments to use when choosing family size and the amount to save or dissave. That they are thoughtful means only that they will ask the question, it doesn't imply they will have an answer ‘wired’ into them. This is one weakness with Professor Rawls's theory of intergenerational justice, a weakness embedded in the concluding sentences of the passage with which we began. It is a theory concerning how generations might be expected to save, not about how they ought to save. Theories of optimum population and saving address this latter question. But as we will see, we are nowhere near to having a persuasive answer.4 Of course, in situations of severe economic stress there may not be an ethically right answer.

13.2 The Genesis Problem and the Repugnant Conclusion5

‘Utilitarian’ theories of optimum population and savings have broadly speaking been of two kinds, based on aggregate utility functions reflecting average and total utility.

The average view (attributable to Mill, Cannan, Wicksell, Robbins, and (p.380) Wolfe; see Gottlieb, 1945) is at once problematic, in that it does not specify if we are to maximize the intertemporal sum of each generation's average level of utility, or if we are to maximize the ratio of the intertemporal sum of each generation's total utility to the total number of all who are ever born.6 The former has been explored by Pitchford (1974); so we know theoretically what it implies in the way of policy. The problem is that the principle lacks philosophical foundations: it is ad hoc, and it does not reduce to a defendable theory of just savings in those situations where population is not subject to choice.7

The latter interpretation, of maximizing the ratio of the intertemporal sum of each generation's total utility to the total number of all who are ever born, can be given a rationale. (Which island would you choose among islands of varying population sizes and levels of individual utility, if you were not to know which person's shoes you would occupy in any island, and were to attribute ‘equi‐probability’ to each such position?)8 However, programmes that maximize such an objective are intergenerationally incoherent (see fn. 4 above). This means that, if any generation were to set such a programme into motion, it would be revoked by the next generation. As the earlier generation would know this in advance, it would hardly wish to set the programme into motion.

But this is only one difficulty; there is a prior problem with the formulation. It is questionable whether the thought‐experiment of choosing among islands has much to do with the problem in hand, which is to determine a defendable future population size (see Dasgupta, 1988c). Average utilitarianism, whichever way we define ‘average’, would seem to have fundamental problems with it.

Unlike the ‘average’ view, the ‘total’ view isn't so readily vulnerable to scrutiny. It has in any case an impeccable pedigree, namely classical utilitarianism:

For if we take Utilitarianism to prescribe, as the ultimate end of action, happiness as a whole, and not any individual's happiness, unless considered as an element of the whole, it would follow that, if the additional population enjoy on the whole positive happiness, we ought to weigh the amount of happiness gained by the extra number against the amount lost by the remainder. So that, strictly conceived, the (p.381) point up to which, on Utilitarian principles, population ought to be encouraged to increase, is not that at which average happiness is the greatest possible . . . but that at which the product formed by multiplying the number of persons living into the amount of average happiness reaches its maximum. (Sidgwick, 1907: 415–16)

This formulation was revived in the important work of Meade (1955). It was subsequently developed for an intertemporal economy in Dasgupta (1969), Lane (1977), and Gigliotti (1983), among others.9 As an exploration into a deep and difficult set of issues, this literature has something to commend it, but not much. The theory's weakness is its insistence on casting the problem of optimum population and savings as a Genesis Problem, not as an actual problem. This has been the source of a number of seeming paradoxes, much discussed in the recent philosophical literature.10 It will pay to look at the more striking ones.

In the Genesis Problem there are no actual people. All persons are potential. In its purest form, the Genesis Problem asks how many lives there should be, enjoying what living standards. Now the application of classical utilitarianism to the Genesis Problem in a world with finite resources can imply a ‘large’ population size. By this I mean that optimum average welfare, even though positive, can be ‘low’. So long as average welfare falls slowly enough when the number of individuals increases, population size under classical utilitarianism is encouraged to grow indefinitely no matter how low the average has fallen (see Dasgupta, 1969: 307; Rawls, 1972: 162–3). Parfit (1982, 1984) finds this repugnant. So he has a term for it: the Repugnant Conclusion.

Personal identities ought not to matter in the Genesis Problem. One may argue that they cannot matter, since in the Genesis Problem all persons are potential. In a comparison of possible worlds there is no privileged position; no particular agent's point of view, no family's point of view, no generation's point of view. Consider a possible world of M persons which, if created, would be one where each person enjoys a welfare (or well‐being; I will use these terms synonymously in this chapter) level equal to W *. Using the notation developed in Chapter 3, we may express aggregate well‐being in this possible world as W(W *, M), where the second argument has been introduced to indicate that there is to be a comparison of possible worlds of different population sizes. Now imagine another possible world, of M + 1 persons, in which if created each person would enjoy the same welfare level, W *. Aggregate well‐being in this world is then W(W *, M + 1). I shall now confine myself to that class of (p.382) ethical theories in which there is a unique value of W *, such that, for all M ≥ 0, W(W *, M) = W(W *, M + 1). I calibrate this W * as zero. This defines the zero level of welfare.11 Levels of well‐being in excess of this reflect good states of affairs, and such theories as I am discussing here state that it is good that people enjoy a good quality of life. Contrariwise, levels falling short of this (i.e. negative levels of well‐being) reflect bad states of affairs, and such theories as I am discussing here will also hold that it is an undesirable world wherein the quality of lives is bad.

Consider two possible worlds, (W 1, . . . , W M) and (W 1, . . . , W M, W M+1). Call them X and Y respectively. They differ solely in the feature that Y would have an additional person (labelled M + 1), with well‐being W M+1. The identities of the first M labels (e.g. their genetic makeup) in the two worlds may not correspond, but within the class of theories we are restricting ourselves to here it is of no consequence in the Genesis Problem (X and Y are only possible worlds.) The question is: how should X and Y be ranked?

One can argue that X is the better world if W M+1 is negative, a guiding principle in such theories as we are discussing being that, ceteris paribus, it would be wrong to bring into existence a person whose life is to be bad. But if W M+1 is positive, what then?

In a thoughtful essay, Sikora (1978: 42) has reasserted the classical thesis that ‘it is prima facie wrong to prevent the existence of anyone with reasonable prospects of happiness’, the implication being that, in the event W M+1 is positive, Y is a better possible world than X. Sikora calls theories based on this thesis Obligation Theories. His wording is curious; I mean the idea of preventing the existence of someone. It suggests an image of potential immigrants to a place of reasonable plenty condemned instead to suspension in an eternal limbo. It would be an error to regard potential persons as a special sort of people. The recognition that W M+1 is positive involves no more than a comparison of the level of well‐being of person M + 1 in Y with the worst state such that it is not a positively bad thing that a person should live in such a state. It would be an odd thing to say that, were Y instead of X to come about, person M + 1 would be benefited. Certainly, it would not convey the sense we usually impute to the term ‘benefiting’. Above all, we must avoid the error of regarding zero well‐being as the point at which a person is indifferent between dying and continuing to live. Subsequently we will need to come back to this point.

Ethically, the only relevant difference between X and Y in the Genesis Problem is that Y would have an additional person enjoying a positive level (p.383) of well‐being. Call the conception that says that therefore Y is the better world the Pareto‐plus Principle. The principle is so appealing that many philosophers have felt no need to justify it (see e.g. Sikora, 1978). But there would seem to be a problem with it: under fairly weak conditions the Pareto‐plus Principle implies the Repugnant Conclusion. Parfit (1984) calls this implication the Mere Addition Paradox.12

I will argue in the next section that this is no paradox: there is nothing repugnant about the Repugnant Conclusion. I will then argue (Section 13.4) that the Genesis Problem is in any case a wrong problem for us. Even if the Mere Addition Paradox were a paradox, nothing of consequence to ethics would have followed from it.

13.3 Is the Repugnant Conclusion Repugnant?

Recall our definition of the zero level of well‐being. This isn't a standard arrived at through a comparison with ‘non‐existence’. Such comparisons can't be made. The ‘unborn’ aren't a class of people. It makes no sense to attribute a degree of well‐being, low or high or nil, to the ‘state of not being born’. Non‐existence is like nothing for us, not even a very long night, because there is no us to imagine upon. One can't be asked what it would be like to experience one's own non‐existence, for there is no subject of experience in non‐existence.13 The impossibility of imagining our own non‐existence gives spurious credence to the view that non‐existence must be a long dismal night from which we must try to rescue people. We can, of course, feel grateful to the persons who created us for doing just that, (p.384) not because they rescued us from anything, but because they are responsible for all this experience. To say that a person has a wretched life, a dismally low standard of living, is not at all to say that the person would have been better off unborn. It is to say only that it is bad that her standard of living is what it is. No doubt it is enormously difficult to make such an assessment (e.g. where are we to draw the line separating positive and negative levels of well‐being?). This does not mean we can avoid making it, nor that we ought to even if we could. Possible people aren't actual (or future) people, any more than clay by the river bank is a mud hut. It is actual persons who have feelings, aspirations, needs, claims, projects, and a sense of justice. In short, it is actual persons who are moral agents. When we revere the memory of deceased persons it is to their memory that we show reverence, not to ‘them’. When we debate at what stage in the development of a foetus we ought to regard the foetus a person, we recognize that there is something akin to a discontinuity in the process of each person's creation. The debate no doubt shows the notion to be fuzzy, even more than, and intrinsically a good deal more important than, the notion of a heap of stones (how many stones are needed to form a heap?), but this doesn't mean that the notion is spurious, nor that it depends upon mere convention. Social convention, possibly backed by formal legislation, dictates how in fact we resolve the issue of when a foetus becomes a person. This does not mean the resolution is right, it only means we think there is something to resolve. In this we are right.

It is for these reasons permissible to say that a person has a wretched existence (or, to put it sharply, a low, negative level of well‐being), and that it is bad that she should be in such a state, and yet to insist that the person has moral worth, that her life has value, that her existence has value, because, if nothing else, it is her life. ‘Better if you hadn't existed’ is a different judgement from ‘better if an additional life isn't created’. This delinking of the notion of zero well‐being from the worth of an actual person's life implies that the quality of life at which well‐being is zero isn't what is conventionally urged upon us by philosophers working on normative population theory. Consider, for example, Parfit's framing of the Repugnant Conclusion: ‘For any possible population of at least ten billion people, all with a very high quality of life, there must be some larger imaginable population whose existence, if other things are equal, would be better, even though its members have lives that are barely worth living’ (Parfit, 1984: 388).

One would not deny that this is repugnant, but then one should not contest that it is rigged. We are first tempted with a population size about twice the world's current population, a figure almost certainly to be reached by the middle of the next century, and a figure that, given current and expected future technology and resources, many think can in principle (p.385) be sustained at reasonable material comfort.14 This is at once followed by a picture of a vastly overcrowded earth, where people scramble for resources so as to eke out an existence, leading lives ‘barely worth living’. But the underlying logic in zero well‐being is a far cry from this. A person whose life is barely worth living has a very low, negative living standard. She is one of the wretched of the earth, and there are hundreds of millions of such people alive today, disfranchised, malnourished, prone to illness—but surviving, and tenaciously displaying that their lives are worth living by the persistence with which they continue to wish to live. When the Conclusion is stated as Parfit states it, it is repugnant. But it is not a conclusion to which the Pareto‐plus Principle leads us if the principle is applied to comparisons of well‐being in the Genesis Problem. There is nothing repugnant about a very large imaginable population, all enjoying positive well‐being. As well‐being would be positive, their lives would be good; they would be more than just worth living. There is nothing morally repugnant in judging that in the Genesis Problem sufficient numbers can compensate for average well‐being, so long as average well‐being is positive; that is, so long as lives are good.

13.4 Actual Problems and an Underlying Asymmetry

We have seen that the Pareto‐plus Principle, when applied to the Genesis Problem, does not imply a large population suffering lives not worth living. The Mere Addition Paradox is therefore no paradox within the confines of the Genesis Problem. It is not even a problem. But this provides no excuse for studying the Genesis Problem. The Genesis Problem is the wrong problem to investigate. We should instead be studying actual problems.

In an actual problem there are actual people, real persons whom I shall for simplicity call the current generation, who deliberate over future population sizes and future living standards. They are by the nature of things the decision‐makers. Actual parents are members of the current generation, and as thoughtful parents they grapple with actual problems, not with the Genesis Problem. This leads us back full circle to fertility decisions, the subject of the previous chapter. The size of the current generation is given, it is a datum.

Consider the following problem. A couple have a newly born daughter, whose well‐being over her entire life is firmly expected to be nil unless additional resources (for example, additional health care and education in her early years) are diverted to her needs. Option X is to make available such resources as will raise her well‐being level to W **. Option Y is for (p.386) the couple to create an additional child, with the understanding that resources will be diverted to this new child sufficient to enable it to enjoy a lifetime standard of living equal to W **; however, under Y the little girl's well‐being over her entire life will be nil. What should the couple do?

If, as Sidgwick (1907) would have it, pleasure or agreeable consciousness is the sole good, and if the fact that something good would be the result of one's action is the basic reason for doing anything (the ground of binding reasons), then the couple in question should be indifferent between X and Y.15 But classical utilitarianism presupposes a conception of persons quite unsuitable for analysing so personal a problem as this. As a model for obtaining the ends of personal action, the theory won't do.16

There are many considerations the parents can legitimately bring to bear in choosing between X and Y. How many children do they already have? What is the source of the additional resources under the two options? What are the implications of their decisions on the family? What is their motivation in having children? And so on. It is thus tempting to insert a ceteris paribus clause in the example, so as to let it pose the problem of choice in a sharp way. In the Genesis Problem it is possible to do this; so the literature on optimum population is littered with the ceterus paribus clause (see e.g. Parfit, 1982, 1984, 1990). In actual problems it isn't possible to do so.

One reason (there are others) why it isn't possible is that the newly born daughter is part of what constitutes the couple's family, whereas the possible further child under option Y is not. A theory of obligation which invites the idea of a family, and more generally of a community, to play a role will provide a reason to the couple for choosing X over Y. This reason does not of course settle the matter. (The little girl may be their only child; this may be the last opportunity for having another child; three may not conform to the couple's conception of a family; and so forth). What it does do is expose the fact that family members have a special claim upon one another.17 Potential persons don't have this claim. ‘They’ are not members of the community.

Each of us, to be sure, belongs simultaneously to many communities involving varying strengths of ties and commitments; and in Chapter 5 I invoked the idea of a wide community when appealing to the differing (p.387) spheres of citizenship and their implied obligations on the part of the State towards its citizens. Here I am thinking of the family as a nuclear community. But it is a community with so very many special properties that it is unlike any other community we belong to.18 Among other things, it is special in that a child is never a party to the decision that leads to her birth. It is also special in that, assuming happy circumstances, her creation is the decision of a loving couple. Parents, by virtue of their act, acquire an obligation towards their offspring that no others have. People of course don't have an obligation to become parents, but they acquire an obligation if they do choose to become parents. By the same token, children have a type of claim on their parents which no one else has.

In the example with which we started, this special claim of the little girl on her parents has a number of implications. From our perspective here, one such implication is that thoughful parents will not, and should not, attach the same weight to the little girl's well‐being as to the potential well‐being of an additional child. The special claim provides a prima facie case for choosing X over Y. The case is, of course, not decisive. But it must play a role in the couple's decision.19 The problem is, this asymmetry leads to a seeming paradox: an intransitivity of ethical relations. I turn to this and its resolution.

Consider now a couple with three children. Option A facing them is not to have any more children and to enjoy a lifetime living standard equal to 10 units for each of the five members of the family (see Table 13.1).20 Options B and C are to have yet another child, followed by two different resource allocations. Under B the existing members of the family would still enjoy lifetime living standards equal to 10 units, and the new child would enjoy 3 units. Under C each of the six members would enjoy 9 units. We take it that the couple cannot pre‐commit, in that both B and C remain viable options even after the third is born. (Of course, once the third child is born option A is no longer available.) What should the couple do? (p.388)

Table 13.1 Fertility choice of a five‐member family

A

B

C

Father

10

10

9

Mother

10

10

9

First child

10

10

9

Second child

10

10

9

Third child

10

10

9

Fourth child

3

9

Notice first that the group comprising the family on whose behalf the couple chooses among these three options is different from the group comprising this same family on whose behalf the couple would choose among B and C were the fourth child to be born. This matters. But first we will see how the desired asymmetry leads to a seeming intransitivity of the underlying moral relation.

We will assume for concreteness that the living standards of actual people count for thrice the living standards of possible people. This isn't of course how thoughtful parents reason; they reason more qualitatively. The way a moral dilemma is framed can matter, and in any case not everything of moral significance can be articulated. There is more than mere vulgarity in attaching explicit weights to the well‐being of different people. But our purpose here isn't to resolve a dilemma, it is to illustrate a point. So we imagine that it is possible to use quantitative weights in making fertility decisions.

We will assume next for simplicity that the couple in our example evaluate alternatives on the basis of the weighted sum of living standards, the weights applied to actual persons being always equal. Notice that this implies they subscribe to the Pareto‐plus Principle. We will see though that in spite of this there is no whiff of the Mere Addition Paradox. Consider then how the couple might reason. Take first the initial evaluation of A, B, and C. As there are five actual members and one possible addition, aggregate living standard under A is 3(10 + 10 + 10 + 10 + 10) = 150; under B it is 3(10 + 10 + 10 + 10 + 10) + 3 = 153; and under C it is 3(9 + 9 + 9 + 9 + 9) + 9 = 144. From the point of vew of the five‐membered family, the ranking is therefore ‘B over A over C’. The couple would now be well‐advised to have a fourth child, but for one thing: it knows in advance that once the new child is born the ranking of B relative to C will reverse itself. (What kind of family are we that allows our littlest to enjoy a living standard of only 3 while each of the rest of us enjoys 10 each?) In reality, this is what happens: Once the fourth child is born, the composition of the family changes, and there are six actual people. The couple now assesses aggregate living standards under B and C to be (p.389) 3(10 + 10 + 10 + 10 + 10) + 3 × 3 = 159 and 3(9 + 9 + 9 + 9 + 9 + 9) = 162, respectively. This reversal of ranking violates the well‐known ‘independence of irrelevant alternatives’ axiom in social choice theory.21 The axiom may have ethical bite when population size is not subject to choice, but it has no bite in the present context (see below). When A, B, and C are all feasible options, the fourth child is only a potential child. But when B and C are the only alternatives, the fourth child is an actual child; option A is not an ‘irrelevant’ alternative.

As the couple can't pre‐commit (and it is important to recognize that they will desire not to pre‐commit), they will know in advance when faced with A, B, and C that after the birth of the fourth child they will reverse their decision and opt for C, the least desirable option from the point of view of the existing five‐membered family.

This looks incoherent (it looks as though the couple suffers from intransitive moral preferences), and in an influential paper it has in effect been used by Parfit (1976) as an argument for rejecting the distinction between actual and possible people.22 But there is no incoherence here. It would have been an incoherent state of affairs were arguments provided to demonstrate that the ranking of options ought to be independent of the family's composition (which is the case in the Genesis Problem). But I have seen none provided by anyone, at least none that is itself coherent. When family size is the object of choice there can be no overall moral ordering of options, and it is a mistake to search for one. Moral perspective has to be from somewhere, it can't be from absolutely nowhere. This is why the Genesis Problem offers such a very misleading substitute framework for thinking about actual problems. It explains why so much of the literature on normative population theory is divorced from life. That the ranking of options based on a family's well‐being changes when its membership increases is no paradox. With the addition of the fourth child the family's perspective changes. There is a new member now, and she must count.

The notion of impartiality in social ethics, the idea that we should seek to peer at matters from no one particular person's viewpoint (as in Harsanyi's notion of impartial preferences (Harsanyi, 1955), and Rawls's reasoning behind the veil of ignorance), has force when future numbers are not subject to choice. In such situations we, the actual people, can deliberate over options affecting ourselves and future people. We can look at the world not only from our perspective, but also from the perspective (p.390) of future people as and when they appear. The veil of ignorance provides us with a reason for doing so.

The problem here is different. Future numbers are a matter of decision. Neither Harsanyi's nor Rawls's construct can get a grip on the matter here. It isn't possible to assume the perspective of possible people. The veil can be worn for a pure savings problem, where future numbers are given. It can do no work for the joint savings and population problem. For the joint problem an overall ordering can only be conceived for each generation of actual people. The moral viewpoint is thereby generation‐relative.23 As generations change with the appearance of newer and newer people, the point of view changes. This means that the ordering itself changes. What appears to be intransitivity isn't intransitivity because the perspective changes.24

How then ought the couple to reason? Rationality dictates the familiar backward‐induction reasoning. The couple have a reason for choosing A, the second‐ranked option among A, B, and C. This is because the couple knows that were it to have a fourth child it would be guided by a different ordering, which would result in the eventual choice of C.

Earlier I argued that the Repugnant Conclusion is not repugnant once we recognize that a life involving zero well‐being ought not to be equated to a life barely worth living. From this we concluded that the Pareto‐Plus Principle when applied to the Genesis Problem does not imply a vast population. The distinction between the Genesis Problem and actual problems reduces even further the possibility that desirable population sizes are ‘large’. In an actual problem Parfit's Mere Addition Paradox cannot be constructed out of the Pareto‐Plus Principle. Redefine A to be the option, open to the current generation, of enjoying a moderately high standard of living and of not adding to its numbers; redefine B to be the option where the current generation maintains its moderately high living standard and adds a number of new persons with a low but positive (p.391) standard of living; and, finally, redefine C to be the option where these new people are created and everyone shares the earth's resources, so that the current generation's living standard, though it remains positive, is much reduced. It is perfectly coherent for the current generation to maintain that, while it recognizes C to be a better world than B were the additional people to be born, and while from its ethical perspective B would be a better world than A, it will nevertheless choose A because, from its perspective again, A is a better world than C.25 There is nothing in ethical reasoning which requires of us to create a world with large numbers of people, all having a very low standard of living.

These ideas extend themselves to the more complex question of savings and fertility decisions across the whole sequence of generations. As I have argued, the question can only be seen from the perspective of actual and future people. With the passage of time some potential lives become actual lives as the world unfolds along a path determined by choices made by Mother Nature and actual people of the past. No doubt the present generation plays God in choosing the next generation's size and its resource and capital base. But there is no unique present generation. Each future generation in turn becomes the present and has to choose. So long as there are future generations, no generation is privileged in this sense. Just as we have to peer into the future, each generation in turn peers into future possibilities having accepted the resource and capital base it has inherited from the past. Given the asymmetry we have identified for fertility decisions, each generation awards a higher weight to its own living standard when proposing the sizes of all future generations and choosing the size of the next generation.

This isn't the place to develop the formal argument. (For this see Dasgupta, 1974c.) The numerical example we have just studied indicates a way we might proceed. For each decision‐maker (I have been calling them ‘generations’ here) fertility and savings decisions need to be made sequentially, not simultaneously. Corresponding to each possible demographic profile, the present generation deliberates over alternative savings programmes. For our purposes here it doesn't matter how choice over these programmes is arrived at. It could be based on contractarian notions, or utilitarian notions, or whatever. What is obtained is a savings rule, a mapping from demographic profiles into savings programmes.26 Given its (p.392) own perspective, the choice of future numbers is then made by the present generation, they having kept in mind that the programme must be acceptable to all who are born in the course of its duration. This is, of course, hopelessly non‐substantive. But it seems to offer the right grammar for constructing a substantive theory of optimal population and savings. For the moment it would seem to be the most we may expect.

13.5 Rational Ends

Population ethics has for long been an underdeveloped branch of moral philosophy. That it has remained backward has much to do with the insistence of philosophers writing on the subject on ignoring the ethical relevance of parental desires, and the related question of what gives meaning to us concerning our own lives. That my neighbour is not as close to me as are my daughters and son is a genetic fact, but that isn't quite the point here. More to the mark is that my children provide me with a means of self‐transcendence, the widest avenue open to me of living through time. Mortality is necessary if we are to imbue life with a sense of urgency. Without it time would be costless, and so life would be shorn of one essential value. But life's achievements are rendered durable by the possibility of procreation. The ability to leave descendants enables us to invest in projects that will not cease to have value once we are gone, projects that justify life rather than merely serve it. These projects include not only the creation of ideas and artefacts; more pervasively, they include the formation of personal values. Thus the questions, ‘what kind of person ought I to try and be; what should I value?’ do not presume the questioner to own a specific set of talents, abilities, or resources (anyone can, and must, ask them); they presume only that they play a role in any reasoned answer.

Procreation is a means of making one's values durable. We imbue our children with values we cherish not merely because we think it is good for them, but also because we desire to see our values survive. It seems to me that our descendants do something supremely important for us here: they add a certain value to our lives which our mortality would otherwise deprive them of. Alexander Herzen's remark, that human development is a kind of chronological unfairness, since those who live later profit from the labour of their predecessors without paying the same price, and Kant's view, that it is disconcerting that earlier generations should carry their burdens only for the sake of the later ones, and that only the last should have the good fortune to dwell in the completed building, or in other words, the thought that we can do something for posterity but it can do nothing for us (see Rawls, 1972: 291), is a reflection of an extreme form of alienation—alienation from one's own life.

(p.393) This viewpoint, of seeing ourselves as part of a delegation of generations, has roots reaching far back, in many cultures; and in recent years it has found its deepest expression in Schell (1982) and Heyd (1992). We act upon this perspective most often with no explicit verbalization to accompany it. We assume parenthood quite naturally; we don't make a big intellectual meal of it. It's the sort of thing we take responsibility for in the normal course of events. Of course, special circumstances may deflect us; we may have more urgent projects and purposes. Here, the fact of a general assumption of parenthood is of importance. An artist, for example, may regard his work as more important than parenting; but he is able to do so only because others are assuring him by their actions that there will be a next generation to bestow durability to the value of his work. The springs that motivate the general run of humankind to assume parenthood are deep and abiding. The genetic basis of the matter merely explains the existence of this motivation, it doesn't justify it. Justification has to be sought elsewhere, and any reasonable answer must come allied to the viewpoint that every generation is a trustee of the wide range of capital stocks (be it cultural or moral, manufactured or natural) it has inherited from the past. Looking backward, it acknowledges an implicit contract with the previous generation, of receiving the capital in return for its transmission, modified suitably in the light of changing circumstances and of increasing knowledge. Looking forward, it offers an implicit contract to the next generation, of bequeathing its stocks of capital in return that they be modified suitably by it and then passed on to the following generation. The idea of intergenerational exchange is embedded in the perspective of eternity. But the intellectual source of such exchange is a far cry from the conception that balked Herzen in his effort at locating mutually beneficial terms of trade.

Recent attempts by social thinkers in Western industrial countries at creating an environmental ethic draw their strength from something like this conception (see e.g. Schell, 1982). But it does not provide enough of an apparatus for them to succeed. Finally, there is no avoiding the question, ‘what should I value?’ if we are to see ourselves living through time, rather than in time. It is, for example, a mistake to try to justify the protection of the giant redwoods—or of a seemingly trivial species such as the hawksbill turtles—or, more widely, the preservation of ecological diversity solely on instrumental grounds; on the grounds that we know they are useful to us, or that they may prove useful to our descendants. Such arguments have a role, but they are not all. Nor can the argument rely on the ‘welfare’ of the members of such species (it doesn't account for the special role that species preservation plays in the argument); or indeed on the ‘rights’ of animals. A full justification must base itself also on how we see ourselves, on what kind of people we ought to try to be, on what our (p.394) rational desires are. In examining our values, and thus our lives, we have to ask if the destruction of an entire species‐habitat for some immediate gratification is something we can live with comfortably. The mistake is to see procreation and ecological preservation as matters of personal and political morality. It is as much a matter of ethics.

Population ethics is rightly regarded a difficult field of inquiry. In this chapter I have tried to argue that the kinds of difficulty that have intrigued philosophers in recent years are insubstantial. Real difficulties lie elsewhere. They lie in deep conceptual problems actual people are faced with when they contemplate the desirable size of their family and the amount of savings that should accompany it. They lie in particular in the problems that poor households in poor countries repeatedly face when deliberating on this.

Notes:

(1) See Dasgupta (1974c, 1988c) and Barry (1977).

(2) Dasgupta (1969), Lane (1977), and Gigliotti (1983) have applied the classical utilitarian calculus to address this joint exercise. See also Meade (1955), who noted the connection but did not analyse it.

(3) That is, N t=N 0 n 0 n 1,. . . , n t−1. As an example, suppose L(n t)=1. Then Q t=1, and we are left with a well‐being function which is a pristine form of classical utilitarianism. See Dasgupta (1969).

(4) See Meade (1966) and Phelps and Pollak (1968) for theories of savings based, respectively, on complete and incomplete impartiality across generations. Dasgupta (1974b) was an attempt at interpreting Rawls literally and exploring the idea that his theory requires parental preferences to be taken as the sole basis for intergenerational justice. The article was an exercise in intergenerational (non‐cooperative) Nash equilibrium savings rules. It was shown that, unless parental preferences extend sufficiently into the future, Nash equilibrium savings rules yield consumption programmes that are Pareto‐inefficient across generations. As a basis for justice among generations, this will not do. Arrow (1973b) and Solow (1974), on the other hand, interpreted Rawls's theory to be the intergenerational extension of his lexicographic maxi‐min principle. They and Dasgupta (1974b) proved that, unless parental preferences extend sufficiently into the future, the principle implies either a stagnant economy, or a programme of savings and dissavings which would be revoked by the generation following any that were to pursue it. The programme is therefore intergenerationally ‘incoherent’. One should contrast this with the corresponding implications of classical utilitarianism, which are (i) that it is coherent, and (ii) that in plausible economies the optimum rate of savings is of the order of 40–5% of national income (see Mirrlees, 1967). A third possible interpretation of Rawls's principle of just savings is intergenerational bargaining behind a ‘veil of ignorance’. This was explored by Dasgupta (1974b). The problem with this route is that a number of well‐known bargaining solutions admit to far too many outcomes in plausible economies. The principle therefore has no cutting power. For a substantive theory of justice, this matters. For example, it was shown that both the intergenerational α‐core and β‐core are to all intents and purposes as large as the set of all intergenerationally Pareto‐efficient consumption programmes. This is hardly a guiding principle: there is usually an infinity of Pareto‐efficient consumption programmes.

(5) The remainder of this chapter is based on Dasgupta (1974c, 1988c, 1989c), which were in turn much influenced by the many discussions I had in the early 1970s with Simon Blackburn.

(6) Letting N t denote the size of generation t and W t the average level of its well‐being (alternatively, welfare), the former takes the form Σδt W t, and the latter takes the form Σδt N t W t/Σδt N t; where δ, a constant (0<δ≤1), reflects a simplified view of the conditional probability rate of extinction.

(7) I am grateful to Professor Kenneth Arrow for this last observation.

(8) See Harsanyi (1955) and Vickrey (1960). I have qualified equi‐probability in the text because it makes no sense when the future has no termination. To give it sense we must suppose that the probability of extinction over the indefinite future is unity. We may then talk of equi‐probability of the conditionals. See Dasgupta and Heal (1979, ch. 9) for elaboration of this.

(9) Blackorby and Donaldson (1985) and Hammond (1988) have offered axiomatic bases for the classical utilitarian view of population and savings.

(10) Parfit (1976, 1982, 1984, 1987, 1990) is the source. See also Bayles (1976), Sikora and Barry (1978), McMahan (1981), Hurka (1983), Blackorby and Donaldson (1985), Sterba (1987), Temkin (1987), Cowen (1989), Ng (1989), Hauser (1990), and Heyd (1992).

(11) Theories in this class will of course differ as to the kind of life at which the level of welfare is zero. I shall return to this important point later. We should note that classical utilitarianism, in which the W function is additive in individual welfares, belongs to this class of theories.

(12) See also Parfit (1982) and Blackorby and Donaldson (1985). The reasoning is as follows. Suppose X 0 is a potential world with M persons, each enjoying a level of well‐being equal to W 0, where W 0 is positive. Assuming that W(W 0, M) is continuous, the Pareto‐plus Principle implies that X 0 is exactly as good a world as X 1, where X 1 is a potential world with M persons, each enjoying W 0, and an additional person whose level of well‐being is nil. But then there is a positive level of well‐being, say W 1, such that X 1 would be exactly as good as a potential world in which each of the M+1 persons would enjoy a level of well‐being equal to W 1. Now, any conception of aggregate well‐being for a fixed number of people which is ‘more egalitarian’ than the lexicographic maxi‐max will have it that W 1<W 0. Let us assume this. Next, construct X 2 from X 1 in the same way as X 1 was constructed from X 0, and define W 2 analogously. Then W 2<W 1, and so W 2<W 1<W 0. Proceeding in this way, we can create more and more populous worlds. In particular, for the kth extension, 0<W k<W k−1<. . .<W 2<W 1<W 0. This means that W k tends to a limit as k tends to infinity. If the limit is zero we have the Repugnant Conclusion; if not, we don't. We finally note that, if the aggregate well‐being function for each given number of people is additive over individual well‐beings, or is a function more equality‐conscious, then W 0≥(M+k)W k/M. This means that W k tends to zero as k tends to infinity, which is the Repugnant Conclusion.

(13) I am talking of non‐existence in the text, not death, which is a different matter altogether. In talking of death we talk of some existing person's death. Thus T. Nagel's (1986a) claim that imagining one's death is no different from imagining oneself unconscious isn't directly relevant to our discussion.

(14) Many experts, on the other hand, do not: see e.g. Ehrlich and Ehrlich (1990).

(15) For this example I am assuming implicitly that well‐being is a measure of ‘agreeable consciousness’.

(16) We need not rehearse the reasons why it is so; the literature identifying classical utilitarianism's weaknesses is now vast. See e.g. Williams (1985). It is a striking feature of procreation that as an activity it is at once intensely personal and social.

(17) I am grateful to Paul Seabright for discussions on this point. In Seabright (1989) he has, by showing how potent is the idea of a community in population policy, provided a non‐rights‐based rationale for the argument I used in Dasgupta (1982b, 1989c) when developing this example. In the earlier work I appealed to the rights of the existing child.

(18) These properties include the fact of close genetic linkage between members of the nuclear community. I am stressing the special nature of the family so as not to give any suggestion that the modern State with its attendant parts is merely a large family. It is so far from it that it can do no useful work for the purposes of developing just rules governing the basic structure of society. There is no inconsistency in our appealing to what is an explicit communitarian consideration here when peering inside the household's reproductive decisions, and in avoiding it when developing the ends of public policy.

(19) Neo‐Utilitarians too have reached this conclusion of asymmetric treatment, but by a different route: ‘We are in favour of making people happy, but neutral about making happy people’ (Narveson, 1973: 73). The pioneering paper on what I call actual problems is Narveson (1967). He labelled the version of utilitarianism that accommodates this asymmetry ‘person‐affecting utilitarianism’. See also Narveson (1978) and Warren (1978). For a farreaching critique of classical utilitarianism when applied to reproductive decisions, see Heyd (1992).

(20) I revert to living standards rather than the more comprehensive well‐being (or welfare), because it has greater immediacy when discussing the problem in hand.

(21) I am grateful to John Broome for drawing my attention to this.

(22) I say ‘in effect’ only because Parfit's objective was to reject Narveson's person‐affecting utilitarianism.

(23) Compare this with the agent‐relativity of ethical reasoning demonstrated by Bernard Williams in his essay in Smart and Williams (1973), and by Nagel (1986b), among others. For convenience I am now identifying generations with actual people. In this account generations change every time a person is born. For convenience of exposition, I am also assuming that future numbers are totally subject to choice, so that there are no exogenously given number of future people to reckon with.

(24) Taken on their own, contractual theories cannot get off the ground when fertility is subject to choice. They need to be embedded within a larger theory embracing reproductive choice. For every possible population profile one can develop a contractual theory of intergenerational justice; but across different profiles it makes no sense to try and do so. (Who are the contractees?) This is one reason why Hare's criticism of Rawls's theory, that it has embarrassing consequences when possible people are included among parties in the Rawlsian ‘original position’, is off the mark (see Hare, 1973: 245–6). For an attempt at constructing a method of making fertility decisions consistent with a contractual view of intergenerational savings, see Dasgupta (1974c, 1988c).

(25) A number of additional paradoxes of population have been presented in the recent literature. See especially Parfit (1984, 1990) and Temkin (1987). Each of them can be resolved in much the same way as the Mere Addition Paradox, provided of course that we address actual problems.

(26) For an account of optimal utilitarian savings rates in aggregative models of economics, see Cass (1965), Koopmans (1965, 1967), Mirrlees (1967), Chakravatry (1969), and Arrow and Kurz (1970). In each of these works, population size is assumed to be given.