Uncertainty and the Evaluation of Public Investment Decisions: Comment
评论Arrow和Lind关于公共投资风险成本为零的观点，指出其依赖的希克斯-卡尔多改进标准存在根本缺陷：无法证明收益确实超过成本，因为缺乏个人自愿同意的信息。
Using Pareto optimality (in the HicksKaldor sense) as their criterion throughotut, Kenneth Arrow and Robert Lind argue in the June 1970, issue of this Review that 1) for public investments the cost of risk-bearing should be regarded as zero because this cost is spread over a large number of persons; 2) consequently, public investment should displace private investment if the expected rate of return exceeds the expected return to private investment minus an adjustment for the cost of risk-bearing; 3) furthermore, project costs borne publicly or benefits accruing to government should be discounted at relatively low rates (because the cost of risk-bearing, is low if spread among large numbers of persons), but project costs borne privately or benefits accruing to private individuals should be discounted at relatively high rates. Arrow and Lind are abstracting from other factors, e.g., externalities, public good characteristics, or ideological preferences for either state or private activity, that might also affect the choice between public and private investments. We wish to emphasize anew the fundamental defect in anv proof that a policy yields a Hicks-Kaldor improvement. We are not referring to the objection to the Hicks-Kaldor criterion-the fact that without actual compensation there will be a redistribution to which one may attach negative value. (Arrow and others have stressed that for this reason one cannot say that a Hicks-Kaldor change is a gain in welfare.)' We are referring rather to the fact that, without actual purchase of everyone's consent, one lacks information about whether the gains exceed the cost, i.e., about whether it would in fact be possible to make some better off without makingr anyone worse off.2 One may judge that a policy, such as public investment, would be a Hicks-Kaldor improvement because he believes the relevant tradeoffs in individuals' preference surfaces are such as to make the gains exceed the costs (as seen by each individual for himself). But he cannot show others that this is so: a Hicks-Kaldor improvement is by definition a change such that one can never demonstrate that it is a Hicks-Kaldor improvement! The Arrow-Lind argument is in difficulty this score because of the alternatives it considers, a public investment financed by taxes versus a private investment. In the latter case, individuals invest voluntarily, taking into account their marginal time preferences as well as their risk preferences. In a public investment financed by taxes, people are forced to invest. One has no observable data on whether all of these individuals would be willing to invest rather than consume, or data on how much thev would have to be paid to invest voluntarily.3 Some of them might much prefer to consume, given the circumstances assumed by Arrow and Lind: At the margin, different