It is gratifying to be the author of a paper has led to comments of the quality printed here. Though I differ with a number of the viewpoints expressed, I think on most of them the reader is best left to judge for himself. There is only one issue on which I want to offer a few remarks. In my paper, I addressed myself to the analytic difficulty posed by the possibility the subjective time discount rate is well below the opportunity cost rate of resources drawn from the corporate sector, and remarked the choice of social discount rate becomes a matter of the theory of the second best. Dan Usher takes up the challenge and tackles the difficult task of an explicit second best analysis, and argues with its help that . . . under certain quite general assumptions, the appropriate interest rate on government projects lies between the rate of time preference and the rate of opportunity cost between present and future consumption in the private sector. David Ramsey, confining himself to the opportunity cost approach, in a world in which the returns to resources differ among sectors of the economy, concludes . . . the social discount rate is a weighted average of observable pre-tax market rates of return. It is, in effect, an average of the opportunity costs associated with each of the sources of funds utilized by a government project, each opportunity cost weighted by the proportion of these funds drawn from the corresponding source. This suggests t . . . public projects which draw resources from low risk, low taxed areas will yield a lower [discount rate] than if the resources were diverted from [other sectors]. Alan Nichols, on the other hand, takes a diametrically opposite view. He argues . . . the correct [social discount] rate . . . is categorically the [highest] of the pertinent opportunity costs. The argument is, apparently, if resources are being employed in two private uses R and S where they yield respective returns of r and s percent, then (if r> s) funds should not flow to a government project unless it too offers at least r percent, even if the funds are derived from S, for in case it would be better to transfer resources from S to R than from S to the government project. Obviously neither of the authors is wrong. Nichols is a perfectionist who will have no truck with second best solutions, while Ramsey will accept any transfers to the public sector provided they constitute improvements. In this imperfect world, my own inclinations lie with Ramsey-I would be unhappy to see opportunities for a better use of resources passed up in an unwillingness to compromise with ideals [1]. Finally, I cannot resist thanking Estelle James for calling to our attention and analyzing an aspect of the problem I had overlooked completely-the case where governmental and private outputs are substitutes. In case the issue, essentially, is not which collection of services should be produced, but who should provide them. And here, clearly, the relevant considerations are not the same as those I discussed.