118 MULTIPLE PURPOSE RIVER DEVELOPMENT 



defense and other programs. Our result is meant to apply only to 

 small tax cuts.*° 



The rates of return which we assumed are another possible 

 source of error. Through parts of the analysis, average rates were 

 used to approximate marginal rates of return. Households were 

 assumed to add to their assets in such proportion that the average 

 rate of return of their property incomes would remain constant. 

 For any one firm or household that is not likely to be a good 

 approximation; but for large aggregates the error will be smaller. 

 The typical household will not hold assets in the proportions of 

 the group average; some households will make investments in 

 their own business, others in common stocks, others in real estate — 

 depending upon the experience of the head of the household and 

 the opportunities to which he has access. Additional funds are 

 likely to be put into the household's major form of asset. With the 

 tax cut spread over all households in the income class, the money 

 is likely to be invested proportionately to total holdings. 



Similarly, business was assumed to make its additional invest- 

 ments at a rate equal to its average rate of return. In the case of 

 industry, the tax cut is diffused over successful and marginal firms 

 in many fields. This does not rule out a systematic bias between 

 marginal and average rates, but the direction of bias is not clear. 

 On the one hand, extra funds must go into investment opportunities 



*" We have made no allowance for effects originating in rounds of re-spending 

 subsequent to the economic units on whom the initial impact falls. The 

 increase in disposable income due to a tax reduction will have multiplier 

 effects on the incomes of others, of course, but we have assumed that a cor- 

 rectly managed fiscal and monetary policy offsets these multiplier effects. 

 Nevertheless, the multiplier effects and their policy offsets will lead to some 

 redistribution of income, and it is logically possible that there are systematic 

 differences between the time preferences of the gainers and the losers of this 

 redistribution. This would affect our estimate. But since both the initial reper- 

 cussions and their offsets are diffused more or less randomly through the 

 economy, it is most unlikely that there will be systematic differences between 

 the two groups in this regard. Further, the similarity of our estimates for 

 different taxes argues that if there are systematic differences in the two groups 

 with regard to important economic characteristics, the effect on the social cost 

 of capital will still be small. For a full discussion of these effects, see A. H. 

 Conrad, "The Multiplier Effects of Redistributive Public Budgets," Review of 

 Economics and Statistics, May 1955, pp. 160-73; also see M. Bronfenbrenner, 

 Taro Yamanc, and C. H. Lee, "A Study in Redistribution and Consumption." 

 ibid., pp. 149-59. 



