AEA PAPERS AND PROCEEDINGS AY990 estimates for the 1909-59 period. in the tion and the preceding one United States, output power man-hour grew correcting for human capita 如m tials re- about one percentage point faster than capi duces the predicted return tal per man-hour. Denison estimates a very rich and very poor countries from about growth rate of h, attributed entirely to 58 at least to about 5, and possibly, if knowl- growth in schooling, of 009. With the tech- edge spillovers are local enough, to unity nology(3), this implies that(1-B+y)time he growth rate. 009 of human capital equals Ill. Capital Market Imperfections 01. With a capital's share B= 25, these numbers imply y=36. That is to say, a 10 I have been discussing capital flows in percent increase in the average quality of static terms, taking it for granted that dif- those with whom I work increases my pro- ferences in marginal products of capital at a based on the assumption that the total stock through time. In the one -good context I am of human capital grows at the same rate using, such flows are simply borrowing con- 009, as that part of the stock that is accumu- tracts: the poor country acquires capital from lated through formal schooling. I do not the rich now, in return for promised good have any idea how accurate an assumption flows in the opposite direction later on this is Suppose countries A and B are engaged Now taking the Krueger estimate that five in such a transaction, and that the capital Indians equals one American, the predicted stocks in the two countries are growing on rate of return ratio between India and the paths that will eventually converge to a com- United States becomes(3).5-=1.04. That mon value. If we look at goods flows through is, taking the external effects of human capi time between these two countries, we al into account in the way I have done phase in which goods fow from advanced A entirely eliminates the predicted return dif- to backward B, followed by a phase(which ferential. Notice that this result is in no way lasts forever)in which goods flow from B to built into my estimation procedure. The value A in the form of interest payments or repa of y estimated from the 1909-58 U.S. com- triated profits. This sort of pattern was im- parison exactly eliminates the return differ- plicit in my statement of the capital fiow ential in a 1959 India-US comparison problem. For such a pattern to be a competi- One might accept this calculation as a tive equilibrium, it is evident that there must resolution of the question I posed in my be an effective mechanism for enforcing in- title. This was the argument in my earlier ternational borrowing agreements. Other paper, based on U.S. data only, and I am wise, country B will gain by terminating its surprised how well it works in a cross-coun- relationship with A at the point where the try comparison. But it portant epayment p blesome I think, to note that the cross-coun- this, country A will never lend in the first comparison is based on the assumption place. A capital market imperfection of this that the external benefits of a countrys stock type is often summarized by the term" polit of human capital accrue entirely to produc- ical risk ers within that country. Knowledge spillovers A serious difficulty with political risk as across national boundaries are assumed to an explanation for the inadequacy of capital be zero. Ordinary experience suggests that flows lies in the novelty of the current politi while some of the external benefits of in- cal arrangements between rich and poor na creases in individual knowledge are local, tions. Until around 1945, much of the Third confined to single cities or even small neigh- World was subject to European-imposed le- orhoods of cities, others are worldwide in gal and economic arrangements, and had scope. But, without some real evidence on been so for decades or even centuries. A the scope of these external effects, I do not European lending to a borrower in India or see how to advance this quantitative discus- the Dutch East Indies could expect his con- sion any further. The argument of this sec- tract to be enforced with exactly the same