Daniel Graham and John Vernon's (G-V) comment on our paper raises four points: 1) that we have erred in the construction of the curve marginal to our of time schedule; 2) that we have mistakenly identified the area under the supply curve as the return which a network affiliate would have received had it remained an independent station; 3) that if the network regards the marginal cost of supplying a program to an additional affiliate as zero that this will lead to an oversupply of programming; and 4) that when our analIysis is extended to permiit variations in the distribution of advertising time between the network and the affiliate that one of our conclusions, that the ban on option time permits the affiliate to earn above normal profits, is vitiated. WVe are grateful to G-V for pointing out the technical slip in the construction of the curve marginal to the supply schedule and for indicating the inaccuracv in our description of area under the supply curve. We would point out only that while our description was inaccurate, our analysis was correct. G-V's third point is not correct. They are clearly right that the network must determine whether it should produce a program at all. And, clearly, for the network to be willing to do so the sum of the returns it gets from each of the stations that carry the program (the network share of advertising revenue minus the payment to the affiliate) must exceed the cost of carrying the program. But, contrary to their claim, the network need not have overinvested in even . . . if the marginal revenue product to the network of the last hour cleared by a representative affiliate (and thus by all affiliates) is . . . zero.... (p. 1034). To see whv this is so it is necessary only to recognize that the array of programs according to the sharing ratio which will induce the affiliate to accept programs can and in general will be different for different stations. The relative attractiveness of a given network program and its nonnetwork programming alternatives is likelys to differ from market to market because, for example, audiences tastes differ. If these differences exist, the return to the network from the carriage of the marginal program by each affiliate can be zero while all programs in the network schedule are profitable. A program which is marginal for one station will be inframnarginal for others and there will be contributions to network profits from the carriage of the program by all stations for which the program is inframarginal. G-V's analy-sis of this point is misleading in one important respect. It is generally not possible to write down an equation such as their profit-maximizing equation since it is not possible to know on the basis of data from a single station alone what share of the cost of a program it should bear if the network is to maximize its profits. For the case cited in G-V's footnote 4 in which all affiliates are identical, equal apportionment of cost gives the correct answer, but in general it will not be possible to employ this procedure. While G-V are correct that the network must earn positive profits on each program, their attempt to take production costs into account at the timne of establishing the terms of the affiliation contract runs the risk that the network will not produce a program which would increase its profits. The approach that we suggest, that the network first decide what its payment will be to each affiliate if a program is produtced and then decide whether the program should be produced at all, always yields the correct answer * Professors of economics, Rice University. We wish to acknowledge financial support from National Science Foundation, grant GI 38909.

The welfare implications of factor flows between countries, unlike those of flows of goods, have received very little theoretical treatment to date. It is widely accepted that emigration of highly skilled people constitutes a loss to a country (Oteiza 1965, Perkins 1966) and fairly generally agreed that emigration of unskilled labor can improve the lot of the remaining populace. Yet these propositions are not self-evident.Grubel and Scott (1966b) have challenged these views, concluding that, in general, if each person is paid his marginal product, no loss accrues to the remaining population from emigration. Although arguing in the context of a "brain drain," they have analyzed only the effects of a marginal outflow of human capital. When the effects of nonmarginal migration (which the word "drain” suggests) are considered, their conclusions do not hold. (Neither, however, do those generally accepted propositions referred to in the previous paragraph.) Yet the concern over the "brain drain" reflects the fact that nonmarginal flows have been and still are by no means rare. This is true even for unskilled labor. The analysis here is carried out under the limiting assumption of classical markets (for both goods and factors) within a country, but (at least for factors) not between countries. A gain (loss) is said to occur whenever the total income accruing to nonmigrants increases (decreases). This welfare criterion implicitly assumes that the marginal utility of income is the same for all of the persons in the nonmigrating groups. It is assumed, first, that the emigration is a once and for all affair and that the supply of resources to the domestic economy is perfectly inelastic. Because of the latter assumption, this case may be thought of as referring to the very short run in which resource supplies do not adjust to the impact of the migration. Given the assumptions used, to be spelled out below, loss occurs in all cases except where the emigrants own a relatively high proportion of the capital in the economy and do not take it with them. External effects related to the emigrants, increasing returns to scale, and other conditions, can affect the results, but since the direction of such effects is usually clear, the only interesting question is whether they are quantitatively important. The other cases discussed allow for the readjustment of factor supplies and factor proportions to the migration. If there are only two factors, capital and labor, the results depend on the relative propensity to hold wealth of emigrants and nonemigrants and on whether the emigrants take their capital with them or not. Loss occurs in all cases except where the emigrants have relatively high propensities to hold wealth but leave a large part of their capital behind, or (a special case) where they own the same amount of capital per person as the nonemigrants and take it all with them. If three factors are introduced, by distinguishing between skilled and unskilled labor, the result depends jointly on the relative propensities to hold wealth, the skill levels of migrants and nonmigrants, the ease of transforming unskilled into skilled labor, and the extent to which emigrants take their (nonhuman) capital with them. In general, the conclusion is that emigration leads to loss.