Since the early 1950s there has been a marked increase in the use of benefitcost analysis in designing public projects and in choosing between alternative public expenditures. One of the important problems emerging from this use of benefit-cost analysis concerns the determination of the true social cost of labor and other resources. Theory indicates that during periods of full employment market prices are the appropriate indications of social opportunity costs. When there is substantial unemployment of factors, however, the market price may overstate the true opportunity cost of using a factor in a public works project. Haveman and Krutilla attempted to overcome this problem in a 1967 article later expanded to book length (Haveman, Krutilla and Steinberg 1968). The present paper is an empirical analysis of the hypothesis introduced by Haveman and Krutilla: Does the nominal labor wage in an area of high unemployment and low income overstate the real cost to society of employing that labor? Specifically, the study is directed toward an analysis of the on-site labor force associated with constructing a large multipurpose dam in southcentral Pennsylvania. Several authors have addressed the problem of adjusting benefit-cost ratios to reflect less than full employment conditions. In general, the solutions suggested have been either to reduce the cost of the factors to reflect a lower i ce th early 1950s t ere has been a social opportunity cost, or alternatively, to increase the benefits of projects which employ otherwise unemployed resources. Sewell (1965), Harberger, et al. (1972), Khatkhatke (1961), and Marglin (1962) have argued that the adjustments should be made to costs, while McKean (1958) and the United States Army Corps of Engineers (1966) feel that the adjustment should be made to benefits. The Haveman and Krutilla study upon which this report is based follows the thinking of the former group. They have developed a method for evaluating a public expenditure project that adjusts the nominal factor prices to reflect any unemployment of resources and idle capacity in industry. Haveman and Krutilla studied the period between 1957 and 1964, a time of substantial nonfrictional unemployment that persisted nationally despite rising income and employment. They assessed the national and regional impact of public expenditures using models based on Leontief input-output systems. The models were designed to enable the analyst to decompose a dollar expended for a final good into its detailed sectoral
Read full abstract