Ian Cafferky
Junior Sophister

The purpose of the reply to the essay Welfare Economics : For light or for fruit is to inject a sense of realism into the study of welfare economics as I understand it. This is not intended as a slight on the author, to whom I address this reply, but it is a personally held criticism of the tired standard approach to the subject. Implicit in my reply is the commonly held assumption that welfare economics should be a policy prescriptive subject and thus its aim is for fruit rather than for light.

The approach to welfare economics to which I refer can be found in any standard textbook on welfare economics and is cogently proposed in Mortons essay. It begins with a discussion regarding the measurement of welfare and then enters into abstract theoretical efficiency conditions for a Pareto Optimum. That aside, the question of distribution is discussed and overall the framework would appear to be plausible. Next market failures are introduced as a necessary step to make the theory acceptable for real world situations. Unfortunately, this step into the real world requires a great mental leap as the apparent consequence of theoretical welfare economics is the applied welfare economics of cost- benefit analysis.

My proposal is straightforward. Theoretically, many aspects of welfare economics are not entirely sound, and furthermore, when the theory is faced with real world dilemmas it collapses. I also believe that the applied branch, cost benefit analysis, is not necessarily a product of welfare economics. Thus, it can be concluded that the most notable achievement of welfare economics is its own redundancy.

Criticism of the standard textbook approach

Standard welfare economics begins with the proposal that the measurement of welfare can be closely proxied by the measurement of utility. Any observer would duly ask: When does utility deviate from welfare? Welfare is the well being of an individual whereas utility is the satisfaction an individual derives from the consumption of goods and services. The justification for the synonymity is the claim that if an individual chooses X over Y, then he or she can be assumed to be happier in X than in Y. This is not always true. Consider the irrationality of preferences which can be due to many factors. Ng (1979) notes that habits/customs, advertising and ignorance are clear examples of possible causes of irrationality of preferences. None the less it is generally assumed that the individual is the best judge of his or her own welfare. This, I would agree is an acceptable assumption. There is one final criticism of the claim to equality of welfare and utility. That is, that welfare is an ex-post measure and utility is an ex-ante measure. Thus, utility as a measure of welfare, is in fact an expectation of welfare. This logically makes the point that ex-ante and ex-post measures are not necessarily equal. It is pedantic to labour these points as there is no absolute measure of welfare and nor is there ever likely to be. It is, as Ng (1979), proposes, convenient to accept these divergences and to continue the study. This criticism of welfare economics is in fact minor, in that it will never be solved and must be accepted. The ultimate criticism lies in the inapplicability of the theory that follows.

Accepting utility as a proxy for welfare the theory of welfare advances taking a two plank approach. On the efficiency side there is the apparent positive economics of Pareto, and on the distribution side there enters the explicitly normative side of welfare in the quest for a social welfare function. I shall consider each of these in turn.

The efficiency side of the theory is especially open to criticism due to its positive approach. Using Paretian analysis, Morton introduces the first order conditions necessary for a social optimum. These conditions, however, are completely and absolutely inapplicable to the real world. The form of competition necessary to achieve this optimum is a theoretical abstraction, perfect competition. In practice, the world suffers from market failures such as monopolies, public goods and externalities that cause a deviation from this perfect competition theory.

Then, given this reality constraint, what does the theory tell us to do? Perhaps we should try to achieve the Paretian first order conditions wherever possible? To answer this Lipsey and Lancaster (1956) attempted to derive similar first order conditions for an optimum when market distortions exist. Their results, in the form of the theory of second best, were conclusive. The resulting conclusions were far too complex to utilise, involving second order cross partials for substitutes and complements. Furthermore, the analysis showed that fulfilling the Paretian first order conditions for some sectors and not for other sectors could in fact lead to a situation where we are worse off in efficiency terms than before. Thus we must analyse the entire economy and take every problem into account to guarantee an improvement. The dilemma is stated succinctly by Ng, We must leap right to the summit to be sure of an improvement. But it is clear that this task is epistemologically, administratively and politically impossible.[8] The summit being referred to is the fulfilment of each and every first and second order condition necessary for an optimum. It is apparent that the entirety of theoretical efficiency analysis is inapplicable. The purpose may have been to produce conditions for the entire economy to achieve an optimum, yet the result is the inadequacy of the theory to adapt to real world scenarios. Perhaps the discussion of market failures, under the theory of welfare, should be changed to the discussion of welfare theory failure when applied to the market.[9]

The other plank of the theoretical welfare economics approach is the distribution issue. Welfare economics aims to rank all possibilities and thus it must not only consider beyond efficiency but also whether a distribution is good or bad. The crucial problem in this side of the analysis arises in the aggregation of the necessary ordinal utility functions. Arrows Impossibility Theorem shows that from individual orderings of social states, a social ordering consistent with some reasonable condition can not be found in general. [10]Without digressing, it simply eliminates the possibility of a social welfare function based on ordinal utility functions. Furthermore the theorem has similar implications for the voting process which is often the mechanism by which our individual preferences may be revealed.[11]

Finally, even if there were an acceptable method of aggregation we must introduce the reality constraint once more by reminding ourselves that utility is not directly observable and that at best, we will end up using a proxy for utility which is a proxy for welfare. This is clearly an inexact theory, but we are attempting to measure such qualitative concepts as happiness and so it is an inevitable problem.

I believe it is clear that the standard textbook teachings are of little use to anyone. The efficiency conditions are redundant and to put it mildly, the distribution issue suffers from aggregation and measurability problems. It appears that the bulk of welfare economics analysis is fruitless. This brings us back to the title of the essay. There can be little doubt that the intention was to use the abstract efficiency theory as a base from which assumptions may be relaxed and reality introduced. This approach has failed. The theory has merely illustrated the complexity of the problem. Perhaps welfare economics has been for light rather than for fruit?