Smith v. Pareto

Atlantic
Economic
Journal

VOLUME 27
NUMBER 3

September 1999

Pages 254-259

This paper argues that we do not and cannot actually use Paretian criteria, therefore, I recommend that we stop pretending we do. (JEL A1, D0, H0)

To avoid fooling people, I should begin by confessing that this article is one of most modest intentions. I simply want to change the kind of thing people say, not what people do. I want people to stop using Pareto or Pareto optimality in their articles on policy. I am not suggesting that they refrain from Paretian analysis because they never use Paretian analysis anyway. I shall demonstrate this point below.

In a way, I will be arguing that we have been too ambitious. We have reasonably good reason for preferring one policy rather than another. We cannot, however, be apodictically certain. This is nothing remarkable. Doctors are rarely certain that their diagnosis is correct and that the prescription will help. This is also true with other professions in all of life. It is rare for us to have perfect reasons for our decisions. This does not stop us from making these decisions, but it does mean that we are occasionally wrong. We go ahead and usually the decisions work out, but sometimes they are wrong. We can hope for similar results when we advise on economic policy. The desire to be perfect is an expression of hubris. As in a Greek drama, hubris normally is punished. In our case, we do not offend the gods, but we do waste much space and the readers' time. Philosophers have pointed out the errors in our work in this field (see Hausman and McPherson [1996]). We have generally ignored their writings just as the scientists ignored Hume's proof that we cannot induce. Scientists went on inducing and were frequently wrong. Today, we say the proof of the pudding is in the eating. We also say that hypotheses produced by induction must be tested. Economists should be similarly modest.

Nevertheless, we frequently cannot test our deductions. Shortly, I will be talking about free trade. There is no way we can, strictly speaking, test our advice. We do have lots of evidence that countries with low tariffs do better than those with high. Saying that the repeal of a particular tariff will cause more benefit than harm, however, cannot be tested empirically for some reasons which will be discussed below. We should not worry about this any more than the doctor worries because he knows there is some chance his diagnosis is faulty. In the long run, he will save many more patients than he refused to diagnose until he had absolute certainty.

In discussing possible policy changes, Adam Smith simply argued that they were beneficial. He was aware that some people might be injured. He thought, however, that the injury to some people would be more than compensated by the benefit to others. In other words, he aimed at general social improvement at the cost to some people. In essence, he assumed that he could judge whether the people injured by a given change were injured more than the benefit received by those who gained.

Pareto raised the question of whether we can be sure how much people are injured or gain by a change. In other words, can we tell what their subjective feelings are? Smith did not seriously discuss this problem. He apparently thought that we could measure material differences and that would be adequate. Pareto disputed this. He reasoned that people might give radically different weighting to the satisfaction or suffering they received from some policy change.

Before turning to Pareto, however, I would like to give a brief, and what the specialist will think is an inadequate history, of the development in this field. As it happens, for a considerable period of time, the most important economic reform pressed by the radicals like Ricardo was repealing the corn laws. This raised the problem of compensating those who were injured in a particularly severe form. Most of the big landlords in England sat in the House of Lords. Their consent would be necessary to repeal the corn laws. Compensating them then was a real issue.

This led to serious discussion on compensation. John Stewart Mill, for example, argued:

"...if there were nothing in the whole process but a transfer; if whatever is lost by the consumer and by the capitalist were gains by the landlord; there might be robbery, but there would not be waste, there might be a worse distribution of the national wealth, but there would be no positive diminution of the national wealth. But protectionist duties can be derived from them by the receivers of rent; and for every pound which finds its way into the pockets of the landlords...the community is robbed of several" [Irwin, 1996].

If landlords demand this public transfer, Mill stated that a direct tax is less costly than an indirect one and that financial compensation be paid to the landlords:

"It would be better, however, to have a repeal of the Corn Laws, even clogged by a compensation, than not to have it at all; and if this were our only alternative, no one could complain of a change, by which, though an enormous amount of evil would be prevented, no one would lose" [Irwin, 1996].

Mills' position was more or less typical. In essence, compensation was thought to be both morally proper and expedient. In fact, no compensation was paid to the landlords. Their lordships apparently were willing to make some sacrifices for the public good. In any event, discussion of compensation became a decidedly minor subject for a considerable period of time.

In fact, as far as I know, compensation directly has never been seriously proposed. Congress, however, frequently engages in indirect compensation. If the city of Tucson is injured, let us say, by extension of the national forest, impeding further development, then Congress will normally also do something nice for Tucson. For example, they might put an additional entry on the interstate where it will help Tucson.

Although Congress may engage in this kind of informal compensation, they do not engage in any formal computation of the size of the compensation. Indeed, although the city of Tucson will be compensated, it is rather unlikely that individuals injured by the extension of the national forest will be the same individuals as those who gain from the new entry to the interstate. This is not because Congress consciously wants to compensate Tucson rather than select individuals, but because it would be difficult to select the right people and calculate the appropriate sum.

It seems likely that these compensations are actually efforts to get as many favorable votes from the appropriation as they lose from the expansion of the forest. In other words, they are compensating the congressman for the votes he lost by giving him some other votes. This is not the type of compensation we normally refer to when we talk about compensating someone who has been injured. It is true that the congressman has been injured and he is compensated, but the citizens are not.

The difficulty of compensating the people injured is partly because we do not know how much they are injured. Traditional England contained many villages dominated by a squire. There were many people who thought this was a desirable situation. Even if the squire were compensated for the loss in value of his land, he might move from the village and thus change the social order. How much compensation should we give a professor at Cambridge who favors the old order and dislikes the new?

This kind of emotional injury is quite common. At one point in my class, I was bemoaning certain regulations that, if applied, would hinder Swiss capital from buying factories in the U.S., thus meaning that we had less capital. One of my students immediately objected that he did not want to work in a foreign-owned factory. I am sure he was sincere although he probably could have been easily compensated.

Some time ago, the farmers of Iowa were up in arms about foreigners buying Iowa farmland. They pressed the legislature to make such foreign purchases illegal. Note that in doing so, they were reducing the sale value of their own land. Clearly, this was a case where financial values were thought to be less important than some sort of spiritual value.

We will return to this type of problem later, but let us temporarily deal with directly economic gains and losses. Grain prices have been low recently. As a result, farmers physically blockade roads from Canada to the U.S. to prevent the important exportation of wheat. Clearly, these farmers could have been compensated and, in fact, Congress did provide funds to compensate farmers.

Once again however, it was not clear that the compensation actually hit the right people. The farmers who blockaded the roads in North Dakota did not get most of the money. Further, it was not restricted to grain farmers. Once again, Congress was watching the voters and not the actual economic damage.

Actually computing the damage would be a major problem even if we confined our attention to material damage. Suppose a reduction in tariffs causes certain plants to close. We could probably figure out the damage to the owners of the plants. The real cost to the employees would be harder. The damage to real estate owners and shopkeepers in the vicinity of the plant would be even harder.

Note that calculating the gain would be still harder. The increased imports would be matched, presumably in the long run, by increased exports. Which factories expanded and which laborers got new jobs would be an even more difficult calculation. Further, there is, of course, the changes in real estate value and an increase in sales around these factories. Altogether, detailed calculations would be impossible.

All we actually do is make a general estimate of both the gain and the loss. In essence, we would direct attention to the dead weight loss and point out that this could be eliminated. We would not inquire as to whether the net social gain was properly distributed. Certainly, we would pay no attention to the secondary gains or losses. Further, we would not inquire whether there were some people who just do not like foreigners and would not like to trade with them. As we know, there are such people, but we tend to think that they are wicked and should not be compensated. Perhaps they should be punished.

Even at this stage, we are unable to make perfect calculations. Deferring Pareto to a few paragraphs down, we already are unable to be sure. Nonetheless, like a doctor, we are willing to make a decision and take action. This is not an improper proceeding. In essence, when we make such a decision, which we think is fairly close to accurate, we are giving the beneficiaries or the injured a ticket in a lottery. It is an unusual lottery, however, because the net payoff will be greater than the net loss.

In most lotteries, the purchasers of the tickets pay in more than the total of the prizes. We think that in this lottery, the prizes will exceed the total paid in. However, we do not mean that every decision will have a larger positive payoff to every person. Any particular policy decision could injure some people and benefit others. We would hope that the total benefit would be greater than the total loss. If this were not true for the given policy decision, we would hope that the losses and benefits would be brought into a favorable balance by other policy decisions. Thus, the prizes in this lottery would exceed the costs, but for any individual, it would be possible that the losses would exceed the gains.

At this point, Pareto comes in. At a time when he was contemplating a switch to sociology, he pointed out that people's internal preferences are beyond our reach or measure. It seems to me that his objective may have been simply to make economics appear less scientific. I do not know his politics but people at least vaguely on the left or traditional conservatives tend to object to "bloodless calculation by economists." On the left, there is a desire to engage in more aid to the poor. The traditional conservatives want simply to stick to tradition. Which, in this case, means keeping the tariff.

Pareto perhaps offered a way out. He suggested that a change that benefitted at least one person and injured no one would be desirable. The problem with this is that we have exactly the same problem of calculation that we mentioned before. Indeed, we have both halves. Actually allocating the financial cost to be sure that no one is injured and only one gains would be as difficult a problem as our previous calculation. Further, the calculation of internal evaluation remains the same.

Altogether, it seems that Pareto was simply arguing that economic methods did not work. Since he became a sociologist, perhaps this was merely an opening gun in a general campaign against economics. Sociologists generally do not like economics and, I must admit, the feeling is more than reciprocated. Thus, Pareto would be in one mainstream if he had this motive.

Pareto was a distinguished economist even though his training was in another field. Thus, any pronouncement he made would be given attention. As far as I can see, his criticism did not stop any economist from continuing to use conventional methods in their calculations. It did, however, lead them to worry.

If they had simply worried and done nothing more over time, then the whole issue would have been gradually forgotten. Unfortunately, they did not do that. Instead, they invented an apparent way out. As a first stage, they suggested compensation. This immediately raises the calculation problems mentioned above. Indeed, Pareto here has made calculations even more difficult.

The principal effect was that more thought was given to such calculations. This did not mean that the problems were solved. What actually happened was that new problems were discovered. For example, if losers are compensated by tax on the winners, then it might be that the change in the wealth of the two groups would lead to enough change in their preferences so that the former losers could purchase a partial reversal of the decision. Clearly, this was merely an unnecessary refinement. The problem was already insolvable.

At this point, most economists more or less simply went back to their original methods. Thus, there was little or no real effect. Nonetheless, a slogan was invented that purported to solve the problem. It was pointed out that if everyone agreed, then the move would injure no one. This, of course, assumes that the subjects of the change would make correct calculations. From our previous discussion, it can be seen that this was dubious.

Nonetheless, many economists began at least bowing in the direction called unanimous consent. I regret to say that I, particularly in joint work with Buchanan, did this myself. I apologize. In practice, this involved mere lip service. Reforms and changes were suggested, and references to obtaining unanimous consent were merely ceremonial. For example, no one suggested that we should await unanimous consent to reduce tariffs.

I have to admit that my work with Buchanan made this error. We suggested constitutional changes and said that they needed unanimous consent. I do not think, however, that any of our readers were fooled. We actually advocated some radical changes. The reference to unanimous consent was not very vigorous and probably merely ceremonial.

In this, we followed the mainstream. If you read any economic journal, you will find statements that various things are Pareto optimal. In most cases, this is merely a slogan. No effort is made to solve the computation problems given above. In general, we are back to Adam Smith or, perhaps better, Marshall. This is no criticism. Whether he intended to sabotage economics or genuinely thought he was making a step forward, Pareto would have paralyzed the science if he had been taken seriously.

There is one case in which occasionally more than lip service is given to Pareto. Some economists, mainly associated with the University of Chicago, have said that the status quo is Pareto optimal. Strictly speaking, this is correct. If there are gains from a change in which no one will be injured and at least one will gain with the cost of making the change included in the calculation and there are still gains, then the change would be made. Since the status quo persists, we can correctly say that it is Pareto optimal. Sometimes economists, mainly from Chicago, do actually say this. Mostly, however, this possibility is ignored.

Thus, my argument is concluded. I suggest we drop the words Pareto optimality and thus shorten our articles by perhaps five-tenths of a percent. Otherwise, we should continue as we are. This is a very minor reform as I said at the beginning. Nonetheless, even minor reforms are worth making.

Hausman, Daniel M.; McPherson, Michael S. Economic Analysis and Moral Philosophy, Cambridge, United Kingdom: Cambridge University Press, 1996, pp. 67-117.

Irwin, Douglas A. Against the Tide: An Intellectual History of Free Trade, Princeton, NJ: Princeton University Press, 1996, p. 93.

1. George Mason University--U.S.A. Presidential Address presented at the Forty-Seventh International Atlantic Economic Conference, Vienna, Austria, March 16-23, 1999.
(Return to Text)