Tibor Scitovsky
Welfare and Competition
Irwin, Homewood, Illinois, 1971
Chapter 4
The Consumer’s Market and the Notion of Economic Efficiency
(pp. 55-75)
The discussion in the last chapter of the individual consumer's market behavior has enabled us neither to help him plan his expenditures nor to sit in judgment over his behavior; but this, it must be remembered, was not the aim of our analysis. In the economist's eyes the consumer is king; and the aim of our analysis was to see not whether consumers conform to an ideal behavior pattern but whether the economic system conforms or can be made to conform to the consumer's wishes. For this purpose, it has been necessary, first of all, to ascertain the way in which the consumer's wishes - whatever they may be - make themselves felt in the market as demand; and this is why, in the last chapter, we developed a method whereby we might infer the consumer's preferences from his behavior. Much of what follows will be concerned with appraising economic institutions and policies by their conformity with consumers' preferences. In the present chapter we shall apply this criterion to the one institution that, in the absence of a discussion of production, can be so appraised already at this stage: the consumers' market through which consumers' goods and services are distributed.
1. THE EFFICIENCY OF DISTRIBUTION
In order to set up the criteria of an ideal distributive system, we shall assume the available quantity of foods and services to be fixed and start out with the simple case of only two people and two commodities. Let us call the two commodities A and B, and the two people Frank and George or, for short, F and G. Both Frank and George consume A and B and buy these commodities for consumption only. In a diagram in which Frank's consumption of A is measured along the horizontal axis and his consumption of B along the vertical axis, any combination of A and B consumed by Frank can be expressed by a point. In a similar diagram drawn for George, points express George's consumption of commodities A and B. Turning George's diagram around by 180 degrees, so that his consumption of A is measured from right to left and his consumption of B is measured vertically downwards, and superimposing this upside down
55 Top
diagram on Frank's diagram, we get the box diagram shown in Figure 4-1. The width of this box measures the fixed quantity of A, at, which is available to the two people; the height of the box measures the fixed quantity of B, bt, available to them; and any point within the box shows a given distribution of A and B between the two people. For example, the midpoint of the box shows an equal distribution of both A and B the midpoint of the box shows an equal distribution of both A and B between F and G. Point P shows that F gets al of A and bl of B; whereas G gets the remainder, alat of A and blbt of B. A movement in the box diagram from one point to another represents an exchange of the com modities between the two people. For example, a movement from P to
Q, is the geometrical representation of a transaction whereby Frank gives George a1a2 of commodity A in exchange for b2b1 of commodity B.
Whereas the quantities of the goods consumed by the two people are shown by the coordinates of a point referred to the two pairs of axes, the satisfaction the two people derive from consuming these quantities can be represented by indifference curves. F's system of indifference curves has the appearance we are accustomed to from the previous chapter; G's indifference curves are similar, except that they are turned around and upside down. Through every point in the box diagram, two indifference curves can be drawn, showing the two people's levels of satisfaction at that point. For example, at point P, F's satisfaction is shown by the indiffer
56
ence curve f3, G's by the indifference curve g4 F would be in a better position anywhere above the curve f3, whereas G's position would be better anywhere below the curve g4; and it is apparent from the diagram that, starting from point P, both F and G could simultaneously better their positions by moving to any point within the area bounded by the indifference curves f3 and g4. For example, moving from P to Q, which represents an exchange of goods between F and G, would raise the satisfaction of both of them. There is nothing surprising, of course, about an exchange of goods making both parties to the exchange better off: this, after all, is the aim of every market transaction. People would not trade if they did not expect thereby to better their position; and the mere fact that trading occurs proves that it is to mutual advantage.
An inspection of Figure 4-1 will show that from every point in the diagram where two indifference curves cross, it is possible to proceed to a whole range of other points which lie on the concave side of the two intersecting indifference curves and which represent higher levels of satisfaction from both people's point of view. An exchange of goods that involves such a movement, therefore, would benefit both F and G. By contrast, there are other points in the diagram from which no movement is possible that would benefit both parties simultaneously. These are the points where indifference curves only touch each other. Point T, for example, where the indifference curves f4 and g4 are tangential to each other, is such a point. Proceeding from T, either F or G could be brought onto a higher level of satisfaction, but only at the cost of reducing the other person's satisfaction. There is a whole range of such points in the figure, where two indifference curves are tangential to each other, and wherefrom no movement beneficial to both parties is possible. In Figure 4-1, these points have been connected by a curve, which is called the contract curve. Moving along the contract curve from left to right would bring F to successively higher and G to successively lower levels of satisfaction; but at no point on the contract curve is it possible to improve one man's satisfaction without diminishing the other's.
In addition to the points of tangency between indifference curves, there are also other points in the diagram wherefrom no movement is possible that would benefit both parties simultaneously, even though the indifference curves through these points are not tangential. For example, the meeting point of indifference curves f7, and g1 on the right hand vertical axis is such a point. The reason that, starting from this point, we cannot move in such a way as to benefit both people simultaneously is that we cannot go outside the box diagram. If we could prolong the two indifference curves to the right, we would probably find a point that is on the concave side of both curves; but we cannot do this, because going outside the box to the right would mean that G consumes a negative quantity of A, which is impossible and absurd.
57 Top
Since we must imagine every indifference map densely covered with indifference curves (of which only a few have been drawn in our diagrams), it is apparent that there are many points like the meeting point of the f7, and g1 curves; and all these points must lie on the axes, in the sections between the origin and the point where the contract curve reaches the axis. It will be convenient to draw the contract curve so as to include also these points. Accordingly, the contract curve in Figure 4-1 extends from origin to origin and coincides with the axes along these sections.
We are now ready to set up standards for the distribution of the two goods between the two people and to examine how distribution in our economy measures up to these standards. It is obvious, to begin with, that distribution between two people always involves a clash of interests. What F gets, G cannot get; and this immediately raises a moral problem: Is it equitable for F to get this much when G is getting that much? This is an important problem, but it is not the only one. In addition to being equitable, distribution must also be efficient. We shall therefore have to discuss both the ethical and the efficiency aspects of distribution, and it will be convenient to start with the problem of efficiency.
We saw above how a movement from P to Q, in Figure 4-1 represented a redistribution of A and B between F and G that brought both of them onto a higher indifference curve. Such a redistribution of goods involves no redistribution of satisfactions in the sense of benefiting one man at the expense of another. A change that benefits somebody without hurting anybody can objectively be said to be a change for the better; and it is a change of this kind that the movement from P to Q involves. This idea is the basis of the notion of economic efficiency.
We shall say that any change of economic policy or institutions capable of making some people better off without making anyone worse off is a change that improves economic efficiency. A situation in which it would be impossible to make anyone better off without making someone else worse off, therefore, will be called an economically efficient situation.
We have derived this definition of economic efficiency in connection with our discussion of the distribution of goods and services among consumers; but its applicability is much more general. We shall use the same criterion of economic efficiency to appraise all economic institutions and not only the system of distribution among consumers. In particular, we shall use it later to, appraise the efficiency of the labor market, of the firm, and of the productive system in general.
In addition to economic efficiency, we shall also be concerned, at a later stage, with another form of efficiency, called technological efficiency. Technological efficiency, and the difference between it and economic efficiency, will be discussed in detail in Chapter 8. Suffice it here to say that an economically efficient distribution of consumers' goods is one that distributes a given quantity of goods in best conformity with consumers' prefer
58
ences; whereas a technologically efficient distributive system is one that performs the physical task of distribution at a minimum cost in terms of manpower, equipment, and other resources.
Returning now to the particular problem at hand, we can say that the distribution of goods between F and G is economically efficient when it is impossible by a mere redistribution of the fixed quantities of A and B to bring either F or G onto a higher indifference curve without pushing the other person onto a lower one. Therefore, to prove that a distributive system is efficient, we have to show that it enables any pair of consumers to get onto their contract curve.
In Chapter 3, we showed that when the consumer is a price taker, he always equates his marginal rate of substitution between any two goods
he buys to their relative market prices; or, in geometrical terms, he always proceeds to a point in his indifference map where an indifference curve is tangential to his budget line. When all consumers are price takers and they all face the same prices in the market, the budget lines of all consumers have the same slope; and each consumer proceeds to a point in his indifference map where one of his indifference curves has the same slope as the budget lines and indifference curves of all the other consumers. In other words, each consumer adopts a consumption pattern which makes his marginal rate of substitution between any two goods the same as that of any other consumer who also consumes the same two goods.
It is easy to see that such behavior brings any pair of consumers onto their contract curve. Figures 4-2 and 4-3 show the consumption patterns of two consumers who have different incomes and different tastes and
59 Top
who consume different quantities of the two goods, but both of whom face the same market prices and choose a consumption pattern (p and P, respectively) for which their marginal rates of substitution between the two goods are the same. To represent the possibilities of barter between these two people, the two figures must be superimposed upon each other in such a way that the length and width of the resulting box diagram represent the total quantities of the two goods in their joint possession. This has been done in Figure 4-4. It is apparent that this will always cause the two budget lines and points p and P to coincide, from which it immediately follows that the two indifference curves on which p and P are respectively located are tangential to one another at the point at which p and P coincide. This shows that barter to mutual advantage is
impossible between these two people and that they are on their contract curve.
So far, we have been concerned with two people who are both consumers of the two goods considered and are therefore able to equate their marginal rates of substitution to the ratio of market prices. We have yet to show that barter to mutual advantage is also impossible between people who are not both consumers of the two goods. To do this, let us pair off the person whose indifference map and market behavior are shown in Figure 4-2 with a third person who consumes B but cannot afford to consume A, because his marginal valuation of A is lower than its price. Figure 4-5 shows the budget line, indifference curve, and consumption pattern of such a person. In Figure 4-6, this diagram is superimposed on that of Figure 4 2; and it is apparent that, although these two people's marginal rates of substitution between the two goods are not equal, they are neverthe-
60
less unable to barter to mutual advantage. This completes our proof of the proposition that when all consumers are price takers and they all face the same market prices, any two consumers will always be on their contract curve.
Although the above argument has been concerned with only two commodities and only two consumers, the conclusion we have derived from it is perfectly general and applies to any number of commodities and any number of consumers. This is so because we proved our proposition not for a specially selected pair of commodities and a particular pair of consumers but for any two commodities and any two consumers chosen at
61 Top
random. In other words, we started out with a large number of commodities and a large number of consumers; and then we showed that, however a pair of commodities and a pair of consumers are selected from among these, we always get the same result. It is clear that a result so reached must apply to all the consumers and all the commodities from among which the selection was made.
Having proved that perfect competition among consumers results in an efficient allocation of consumers' goods and services, we may consider for a moment what happens when competition among consumers is not perfect. Imperfect competition among consumers may assume a variety of forms. First of all, the consumer may be able to bargain and so exert a conscious influence over price. This seldom happens in our society, where the consumer usually faces set prices. The mere fact, however, that the consumer regards prices as given does not in itself render him a price taker. For the price taker, while regarding prices as given, must also be free to decide what and how much to buy. Hence, the second form of market imperfection occurs when rationing or a shortage of goods deprives the consumer of his freedom to determine his rate of purchases. Rationing keeps some consumers from buying as much as they want of rationed goods and hence from equating their marginal valuation of the rationed goods to the prices of these goods. Therefore different people's marginal valuations of a rationed commodity are likely to be different; and the same is true also of commodities that are in short supply. Even in the absence of rationing and shortages; however, competition among consumers may be imperfect if all consumers, though price takers, do not face the same prices. For if different consumers or groups of consumers pay different prices for the same commodity, their marginal valuations of that commodity will also be different. Price discrimination, therefore, is the third form of market imperfection.
In the following, only the results of this last form of market imperfection will be analyzed. This case is illustrated in Figure 4-7, where on top of F's indifference map, as shown in Figure 4-2, we have superimposed the indifference map of consumer H, who pays for commodity A a higher price than F and whose budget line therefore has a steeper slope than F's. Again, the two diagrams are superimposed in such a way that the distribution of the two commodities between F and H is shown by point p. But now, the two budget lines, having different slopes, intersect at this point; and so do the two people's indifference curves that go through this point and are tangential to their respective budget lines. Accordingly, the distribution of the two commodities between F and H is not efficient; and the same could be proved, with the aid of similar diagrams, also in the case of rationing, shortages, or any other form of imperfect competition among consumers.
We have proved, therefore, not only that perfect competition among
62
consumers results in efficient distribution but also that imperfect competition among them results in an inefficient distribution of consumers' goods and services. This is an important result, because competition among consumers is usually perfect in our economy, and because the resulting efficient distribution of consumers' goods and services is one of the elements of an efficient economic system. But lest we exaggerate the importance of having an efficient distribution of consumers' goods and services, let us examine its limitations and see what exactly it amounts to. One look at Figure 4-1 shows that an efficient distribution of goods is not in itself enough to insure an ideal distribution of goods. For distribution is efficient at any point on the contract curve; and since the contract curve runs diagonally across the whole diagram, efficient distribution appears to be compatible with extreme inequalities in different people's income and welfare. In fact, goods and services can be efficiently distributed for any and every income distribution. Perfect competition among consumers and the consequent efficient distribution of goods and services can be likened to a system of balloting that enables everybody to register his preferences and distributes goods and services according to people's preferences; but such a system is not necessarily democratic because, instead of giving everybody equal votes, it weights people's preferences according to their purchasing power. It follows from this that, for the distribution of consumer's goods and services to be ideal, the distribution of wealth and income the two sources of purchasing power would also have to be ideal.
63 Top
Efficiency, therefore, is only one of two criteria by which economic organization must be appraised. Equity is the other criterion; and it is an equally important one. Unfortunately, the economist cannot set up standards of equity as he sets up standards of efficiency; nor have objective or universally accepted standards of equity been set up by anyone else. This raises the difficult problem as to how the economist can provide the basis for appraising economic organization when he only has one of the two yardsticks by which economic organization must be judged.
Only in one special case is there a simple solution to this problem. If he has to choose between two situations that are exactly equal as far as the distribution of income and wealth is concerned, the economist can pass judgment by the criterion of efficiency alone. This is so because, for any given distribution of wealth and income, it is better to have a more efficient economic organization than a less efficient one. But very rarely is the economist's task as simple as this. As a rule, he has to weigh alternatives that differ both in efficiency and in the distribution of wealth and income; and we shall have to discuss the nature of the problem that is raised by such a choice.
Before doing so, however, we must first explain why we regard the distribution of income and wealth as a matter of equity alone.1 One might define an efficient economic organization as one that maximizes the sum total of human satisfactions. Accordingly, one might argue that taking $100 from a millionaire and giving it to a beggar would not only render income distribution more equitable but would also raise the efficiency of the economic system, since the millionaire's loss of satisfaction would be negligible, the beggar's gain considerable, and the sum of their satisfactions would therefore be increased on balance. This argument appeals to common sense but cannot be proved, because we cannot compare or add one person's satisfaction to another person's. In other words, we have a strong subjective feeling but no objective proof that such a redistribution of income would increase the sum total of satisfactions as, indeed, we can attach no definite and rigorously defined meaning to the idea of a sum of satisfactions. It seems advisable, therefore, to keep objective and provable statements meticulously apart from arguments based on subjective feeling alone. This is the basis of our distinction between efficiency and equity. We regard all arguments based on subjective judgment as matters of equity and use the term "efficiency" only in connection with statements that can be proved.
2. EQUITY AND EFFICIENCY
When the economist has to appraise the relative merits of two alternative systems of distribution among consumers or more generally, of two alterna
1 We are deliberately ignoring here the effect of income distribution on incentive. That topic will be discussed in section 6 of Chapter 5, and again in section 2 of Chapter 15.
65 Top
tive economic policies or forms of economic organization he usually finds that they differ with respect to both efficiency and equity.2 Accordingly, he must be prepared to make his appraisal on both efficiency and equity grounds. On occasion, this may be very difficult; for it may happen that one alternative is preferable on efficiency grounds, whereas the other is better on equity grounds.
It will be convenient to analyze the nature of this problem diagrammatically, with the aid of the box diagram in Figure 4-1. We have discussed the difference between a point on the contract curve and one off it; and this difference has served as the basis for our definition of economic efficiency. We have shown that, to any given point off the contract curve, such as P in Figure 4-1, there corresponds a whole range of points on the contract curve, such as the range between S and T, which is preferable to it from the point of view of both persons. It is important to realize, however, that from this it does not follow that both would prefer any point on the contract curve to any point off the contract curve. Compare, for example, points P and R in Figure 4-1; and think of them as representing the results of two alternative economic systems that would bring these two people to points P and R, respectively. Unlike P and Q, points P and R differ from each other not only in efficiency but also in the two people's relative levels of satisfaction. Distribution is undoubtedly more efficient at R, and F is better off there; but G is less well off at R than at P; and if we were very much concerned with G's welfare, we might well prefer P to R, despite the former's lower efficiency.
It could be argued, of course, that the equity of distribution is a matter of ethics or politics and, as such, is none of the economist's concern. One might say, therefore, that the economist should always favor R as the more efficient measure of the two, leaving equity to be taken care of by whoever is responsible for taking care of it.
This might conceivably be the correct attitude in a socialist economy, or generally in an economy in which the State assumes full control over the regulation of economic affairs and takes full responsibility for maintaining an equitable distribution of income. In such an economy the economist could make policy recommendations on the basis of efficiency considerations alone, because he could, rest assured that if his recommendations were followed and resulted in a redistribution of income which was considered undesirable, this would be corrected as a matter of course by the authority responsible for maintaining an equitable distribution of income. In terms of our diagram, if the economist's recommendations should lead to R and the State should consider R an inequitable position, the latter would tax away some of F's gain, from this compensate G for his loss, and so bring
1In this section, we shall be concerned only with the distribution of goods and services among consumers. But the argument is of general validity and applies pari passu to the connection and possible conflict between equity and efficiency in all fields of economic organization.
65 Top
the two people to, say, point Q in the diagram, enabling both of them to benefit by the superior efficiency of measure R.
In the free enterprise economy, however, we cannot take it for granted that changes in economic policy will be accompanied by a state imposed redistribution of income offsetting any loss of equity which may be caused by such changes. This is so not because the State is unwilling to take action to mitigate inequities in distribution (progressive taxation and social insurance testify to the contrary) but because there is a presumption in such an economy against the State's interfering, except in a general way, with the income distribution brought about by the market mechanism. A change in a country's foreign trade policy, the raising or lowering of tariffs, a change in farm policy, or any other economic change usually benefits some people and harms others; but it seldom happens in our economy that those harmed are compensated for their loss. In other words, the effects of economic policy on efficiency on the one hand and on income distribution on the other hand cannot as a rule be separated, because compensation payments are seldom feasible politically in the free enterprise economy. From this, it follows that in such an economy all policy decisions must be based on considerations both of efficiency and of equity.
Nevertheless, it is sometimes argued that, even in the free enterprise economy, economists should concern themselves with efficiency considerations alone. According to this argument, economists should offer the policy maker their expert advice on matters of inefficiency and warn him at the same time that he must also get expert advice on matters of equity and base his decision on both efficiency and equity considerations.
This argument would be valid if there were experts on equity, whose opinions could be pitted against the economist's expert opinion on efficiency. In our society however there are no such experts. The economist is as good a judge of equity as anyone else; as a matter of fact, he, as a social scientist, is considered by many a better judge. Also, the economist is in the best position to appraise the relative importance of efficiency and equity considerations. In any case, whether the economist is better or merely no worse than others in judging equity, he cannot neglect it. For if he did, and based his recommendations on efficiency considerations alone, he might unwittingly cause these considerations to be given more weight than is their due. If he ignored considerations of equity although the public regarded him, rightly or wrongly, as the best judge of equity, he would give the impression that he considered efficiency a more important criterion than equity. But even if the public did not regard him as the best judge of equity, he could still not ignore equity considerations; for if he did, there would be danger that his expert advice on efficiency would be given more weight than the public's vague feelings concerning equity. In our society therefore the economist must, whether he likes it or not, weigh both efficiency and equity considerations when he tenders his advice on
66
policy decisions. He must make it clear, of course, that his recommendations are based on both criteria; and he must also stress the fact that, on matters of equity, he does not consider himself an expert.
The impossibility in most cases of separating considerations of efficiency from those of equity and the consequent need for considering both together explain one essential difference between economics and, say, engineering. The standards of efficiency are as objective and scientific in economics as they are in engineering; but whereas they are the only standards of the engineer, the economist must weigh, in addition to the objective and scientific standard of efficiency, also the subjective and ethical standard of equity. This is why it is not enough for the economist to be merely a competent technician. Since most of his recommendations are bound to affect the distribution of welfare between individuals and between social classes, he must also have a sense of fairness and economic justice.
Let us now return to our diagram. We are concerned with the problem of how to choose between P and R on the basis both of efficiency and of equity. To emphasize the fact that the choice has to be made on the basis of two entirely separate criteria, we shall make the comparison between P and R in two separate steps, by comparing the two points not to each other but to an intermediate point, Q,. There may be a slight difference in equity between P and Q; but this can be neglected, since Q, is preferable to P from both people's point of view. This part of the comparison therefore can be made on the basis of efficiency considerations alone. Points Q and R are equally efficient, since both lie on the contract curve; but they differ as far as the two people's relative satisfactions are concerned, since F is better off at R, and G is better off at Q. This part of the comparison therefore can be made on the basis of equity considerations alone. In choosing between Q and R, one person's gain must be weighed against the other's loss; and according to one's appraisal of their relative needs and deserts, one may regard either Q or R as representing the more equitable distribution of welfare between the two people; or one may regard them as equally equitable.3 These three possibilities give rise to three different cases, each of which merits closer examination..
Before discussing these three cases in detail, let us summarize them, together with the two simpler, ones, in which the comparison between the alternative situations can be made on the basis of only one criterion. We shall list the two simpler cases first, and the three more complex ones afterwards
1. One of the simpler cases is that in which two alternative situations
3 For an example of two different but equally equitable income distributions, consider a community consisting of two people, Frank and George; and assume that their needs, tastes, and deserts are similar. Then, giving an income of $6,000 to Frank and $4,000 to George is as equitable as giving $6,000 to George and $4,000 to Frank.
67 Top
are equally efficient, differing only in the distribution of income. An example of this would be the choice between two situations that lead our two consumers to positions R and Q, respectively. In choosing between these two situations, we need not worry about efficiency at all and can make our minds on the basis of equity considerations alone.
2. Equally simple is the choice between two situations that differ only in efficiency but not in the distribution of income. This case (mention on page 64) corresponds to a choice between P and Q. In choosing between these alternatives, we can forget about equity, concentrate on efficiency, and on efficiency grounds give Q preference over P.
3. We now come to the first of the three complex cases, where t choice between two alternatives must be made on the basis of both equity and efficiency considerations, and which therefore correspond to a choice between P and R. The simplest of these is the case in which R is considered to represent a more equitable distribution than Q, because, say, F's need and deserts are greater than G's. In this case, R is better than Q (and hence also than P) on equity grounds, P is inferior to Q (and hence also to R) on efficiency grounds, and R is therefore preferable to P of both counts.
4. Almost as simple is the next case, in which R and Q (and hence also P) represent welfare distributions of equal degrees of equity. Since P is inferior to Q (and hence also to R) on efficiency grounds, we conclude in this case that R is to be preferred to P, because it results in greater economic efficiency and in a different distribution of welfare which, i , no better, is at least no worse than that obtaining at P. This case is similar; to case (2) , above, because in both of them a choice has to be made between two equally equitable situations. The difference between them is that, whereas in case (2) the alternatives are equally equitable because they involve the same welfare distribution, in case (4) they involve different but equally equitable welfare distributions.
5. The problem of choice becomes difficult only in the last case, in which R is inferior to Q (and hence also to P) on equity grounds. Since P falls short of Q (and hence also of R) as regards economic efficiency, there is a conflict here between considerations of equity and efficiency. R is inferior on equity, P on efficiency grounds; and in choosing between the two positions, the relative importance of the two criteria must be weighed against each other. Either of them may be considered the more important of the two; accordingly, either P or R may be preferred on balance.
Having listed the five possible cases, we may consider examples of at least some of them. An example of the first case is the payment of relief to the needy out of an income tax levied on taxpayers. Such a redistribution of income redistributes welfare without affecting the efficiency with which consumers' goods are distributed. In other words, when competition among
68
consumers is perfect, we can make any pair of consumers move along their contract curve by taking away part of one man's income and giving it to the other man to spend; and this is exactly what happens when sums of money raised by income tax from some people are paid out as relief to some other people. Hence, relief payments financed by income taxation correspond to a movement from R to Q, in Figure 4-1; and their desirability can be appraised without reference to efficiency considerations. This result is subject to one minor qualification. The administration of relief involves a cost; and in deciding for or against relief, this fact must also be taken into account.
Entirely different is the situation when relief is not paid out in money but made available in some other form. An example of this is the Food Stamp Program, introduced by the U.S. Government in 1939 and again in the mid 1960s, under which needy families are enabled to buy certain foods at half the market price,4 the other half being paid by the Government. This program lowers the efficiency of distribution by making the price of foodstuffs different for, different people; and the resulting loss of economic efficiency can be expressed in terms of money by the difference between the cost of relief to the taxpayer and its value to the relief recipient.
It will be helpful to illustrate this argument graphically. In Figure 4-8, where money is measured on the vertical axis and a foodstuff (say, butter) on the horizontal axis, a person's market opportunities before and after
1 In actual fact, the extent of the price reduction varies and averages somewhat less than half; but for purposes of illustration we assume it to be one half.
69 Top
the introduction of the food subsidy are shown by the budget lines Ie and If, respectively. The slope of the first budget line shows the mark price of butter, that of the second shows the price paid by a person entitle to the subsidy. When the Food Stamp Program enables him to buy butter at half price, he will buy Ob of butter, spending on it Ic. Since the full market value of this much butter is Id, the cost to the government (of taxpayer) of subsidizing this person's butter consumption is dc, the vertical distance between the two budget lines at point b. The money value of the subsidy to the subsidized person is expressed by the equivalent variation IJ, which shows the amount of additional income that in the absence of the subsidy would bring this person to the same indifference curve which the subsidy enabled him to reach. It is apparent from the diagram that Ij, the value of the subsidy to the subsidized person, is smaller than dc, the cost of the subsidy to the government. This is so whatever the shape of a particular indifference curve, as long as it has a smooth curvature.5 The common sense interpretation of this result is that one can make a man happier by giving him cash and letting him spend it as he thinks best than by forcing him to take all his relief in the form of one commodity. Hence, relief payments in cash are preferable to a food subsidy, because they are economically more efficient, giving the relief recipients either a greater gain at the same cost to the government or the same gain at a lower costs.6 A choice between the two forms of relief, therefore, corresponds to case (2) and is represented in Figure 4-1 by a choice between points Q and P.
Such a choice between an efficient and an inefficient form of relief is easy to make, because it raises no problems of equity and can be made on the basis of efficiency considerations alone. It is conceivable, however, that relief in the form of cash payments should be ruled out by political considerations, in which case the choice would lie between an inefficient form of relief and no relief at all. Geometrically, this is a choice between P and R in Figure 4-1; and since relief is presumably more equitable than no relief, this would correspond to case (5) and would have to be decided by weighing the gain in equity against the loss in economic efficiency.
As an example of case (3), consider the choice of raising public revenue (for relief or for any other purpose) either by an excise tax (e.g., the cigarette tax) or by income taxation. An excise tax on a particular commodity raises the price of that commodity to the consumer and thereby
5 Only if the indifference curve had a kink at its point of contact with the budget line would the cost of the subsidy equal its value to the person subsidized.
6 Needless to say, there are also other considerations than those discussed here. For example, the 1939 Food Stamp Program had the very important additional purpose of disposing of agricultural surpluses. For yet other considerations, see pp. 72 73 and Chapter 15.
70
lowers his satisfaction. This is shown in Figure 4-9. The imposition of the excise tax shifts the consumer's budget line from ie to if. If he buys Ob units of the taxed commodity, he will have to pay for it ic, or cd more than he would have had to pay in the absence of the tax. Hence, cd is the amount of tax collected from him by the government. The consumer's loss of satisfaction due to the tax is expressed in terms of money by the equivalent variation ji. That this will always be greater than cd, the amount of tax collected from him by the government, is apparent from the diagram. An excise tax appears therefore as a less efficient way of raising public revenue than the income tax, since ji obtained from this person by income tax would make him no worse off than the payment
of cd, a smaller amount, in the form of an excise tax. The above argument also implies that the amount, cd, levied by an income tax would cause the taxpayer a smaller loss of satisfaction than if it were obtained by an excise tax.7
The difference between an excise and an income tax, however, lies not only in their relative efficiency but also in their incidence. A flat rate income tax taxes everybody in proportion to his income; a progressive income tax taxes away a higher proportion of people's income in the higher income brackets and a lower proportion in the lower brackets. An excise tax is proportional to people's expenditures on the commodity taxed; and its incidence therefore depends on different people's expenditures on this com
7 The above argument, while correct as far as it goes, is subject to serious qualifications, since it tacitly assumes unchanged cost conditions, which it is often not realistic to assume. Cf. the I. M. D. Little article referred to at the end of the Bibliographical Note to this chapter for further discussion.
71 Top
modity. Since the poor always spend a larger proportion of their income than the rich and therefore spend a larger proportion on an average commodity, an excise tax on an average commodity is regressive in the sense that it takes away a larger part of people's income in the lower than in the higher income brackets. Even more regressive is an excise tax on a necessity, the consumption of which is little affected by the consumer's income, so that a tax on it costs the poor as much or almost as much money as it costs the rich. This is why an excise tax on food or cigarettes is regressive. Assuming, therefore, that a progressive or proportional tax is more equitable than a regressive one, income taxation is preferable to most excise taxes on grounds both of equity and of economic efficiency.
The above examples should suffice to illustrate the nature of the connection between equity and efficiency and of the problems raised by it. Throughout all this, however, it is very important to realize that we defined "economic efficiency" as conformity to consumers' preferences and that economic efficiency is therefore a meaningful concept only as long as we accept the consumer's preferences as a datum and regard him as the best judge of what is good for him.8 It would be meaningless to apply the criterion of economic efficiency to a measure aimed deliberately at changing the consumer's preferences or influencing his market behavior.
For example, it appears from Figure 4-8 that a food subsidy will always cause the relief recipient to consume more food than an equivalent payment of cash would. A cash payment of IJ added to his income would prompt him to consume Oa of butter; the equivalent food subsidy, dc, makes him consume Ob; and the difference between the two, ab, will be recognized as the substitution effect of the food subsidy. A food subsidy therefore encourages people's food consumption; and it can be used as a public health measure when the government is more concerned with providing an adequate diet for needy families than with giving them the greatest satisfaction at minimum cost.9
In the same way, Figure 4-9 shows that an excise tax, because of its substitution effect, discourages the consumption of the taxed commodity more than an equivalent income tax would. Hence, an excise tax may be adopted for the sake of the special discouragement it offers, just as
8Throughout this book, we shall often be concerned with the conformity of economic organization to consumers', or the community's preferences. In all these cases, we shall be concerned with the community not in Hegel's sense of a collective entity which is something mole than the sum of its members but with the individual members of the community and with conformity to their individual preferences. It is not intuitively obvious that economic organizations can conform to the different preferences of different people all at the same time; nor is this always possible. But an example is distribution between two persons represented by a point on the contract curve in Figure 4-1, which does conform simultaneously to the preferences of both people. For further discussion of this subject, see pp. 76-77, 169, and 182.
9 For a further discussion of the argument of this paragraph and the next, see Chapters 11 and 15.
72
a food subsidy may be preferred to relief payments in cash for the premium its puts on food consumption. Thus, although an excise tax on alcoholic beverages is inequitable and inefficient as a means of raising revenue, it is useful as a public health measure, because it lowers liquor consumption and discourages drunkenness. In other words, the very measure that we condemn as inefficient for its failure to conform to consumers' preferences may be considered useful if we cease to regard the consumer as the supreme authority on what is good for him. There are many instances of an excise tax being used to sway consumers' preferences and to discourage the consumption of a particular commodity. For example, during World War II the United States government imposed an excise tax on travel, telegrams, and long distance telephone calls, in an effort to discourage the civilian use of these services and so make them available for the use of the armed forces. These taxes were very well suited for that purpose; but their retention after the war for the purpose of raising revenue was objectionable because, for this purpose, they were inefficient and inequitable.
In the chapters that follow, we shall concentrate on the study of efficiency and make few references to the problem, of equity. It must be emphasized, however, that we shall do so not because we believe that efficiency is more important than equity but solely because we can make objective statements about efficiency, whereas everyone must make his own judgment about equity according to his own conscience and ethical norms. It must be understood, therefore, that whenever we say that one situation is more efficient than another, we shall not mean that this situation is necessarily better than the other. The more efficient situation may be preferable, but only if it also happens to be more equitable, or no less equitable, or so little less equitable that the loss in equity is more than offset by the gain in efficiency. In other words, the reader must bear in mind constantly that few judgments in this book are welfare judgments. Usually, we shall say not that one situation is better than another but only that it is more efficient and would be better if it were found acceptable also on equity grounds. This is a serious limitation; but it is a limitation not only of this book but of all economic theory, which can never be more than a partial guide to economic policy.
BIBLIOGRAPHICAL NOTE
I know of no good elementary discussion of the basic principles discussed in this chapter. This is undoubtedly due to the fact that there is no general agreement among economists on the important problems of the economist's function in society, the basis on which he should make his recommendations, the extent to which he,can make recommendations, and so forth. While there is no agreement, there has been considerable controversy; and the reader is referred to the following articles on the subject: R. F. Harrod, "Scope and Method of Economics," Economic Jour
73 Top
nal, Vol. XLVIII (September, 1938), pp. 383 412; N. Kaldor, "Welfare Propositions in Economics," Economic Journal, Vol. XLIX (1939), pp. 549 52; J. R. Hicks, "The Foundations of Welfare Economics," Economic Journal, Vol. XLIX (1939), pp. 699 712; I. M. D. Little, A Critique of Welfare Economics (Oxford: Clarendon Press, 1950); my "The State of Welfare Economics," American Economic Review, Vol. XLI (1951) pp. 302 15; and E. J. Mishan, "A Survey of Welfare Economics, 1939-1959," Economic Journal, Vol. LXX (1960), pp. 197 256.
As to detail, the box diagram, as defined in this chapter, comes from F. Y. Edgeworth, Mathematical Psychics (London: C. Kegan Paul & Co., 1881) and /A. L. Bowley, Mathematical Groundwork of Economics (Oxford: Clarendon Press, 1924 ) . The argument concerning excise taxes comes from M. F. W. Joseph, "Excess Burden of Indirect Taxation," Review of Economic Studies, Vol. VI (1939), pp. 226 31; although it was already stated by Dupuit in 1844, in the paper referred to in the Bibliographical Note to Chapter 13.
So much for sources. The most important next step for the reader to take is to acquaint himself with the general subject of how a group of people, each with something to offer and something desired, can reach a mutually satisfactory agreement, of what the nature of this is and how an increase in the number of people in the group affects it. On all this, see J. Quirk and R. Saposnik, Introduction to General Equilibrium Theory and Welfare Economics (New York: McGraw Hill Book Co., 1968), esp. chapter 4; P. K. Newman, The Theory of Exchange (Englewood Cliffs, N.J.: Prentice Hall Inc., 1965), chapter 3; and V. C. Walsh, Introduction to Contemporary Microeconomics (New York: McGraw Hill Book Co., 1970), chapter 16. On the subject of excise taxes, an important qualification of the simple argument as here presented is contained in L. M. D. Little, "Direct versus Indirect Taxes," Economic Journal, Vol. LXI (1951), pp. 577 84, reprinted in K. J. Arrow & T. Scitovsky (eds.), Richard D. Irwin, Inc., Readings in Welfare Economics ( Homewood, III.: 1969 ).
75
