ARE per-capita expenditures on foreign aid by different donor countries interrelated? In particular, do small countries tend to be free riders, exploiting the larger countries by spending relatively little themselves while deriving benefits from larger countries' expenditures? The participants in the 1960s debate over the allocation of defense expenditures among Western nations tended to assume that the answer to both questions was yes. Lumping foreign aid with defense, they argued that the smaller countries were assuming an insufficient share of the common burden of promoting world security. 1 In a provocative article, Olson and Zeckhauser (1966) proposed a theory of alliances to explain this phenomenon. The study developed the notion of a reaction function for each country, which determines its expenditures for any given level of the expenditure of other members of the alliance. Equilibrium for the alliance occurs when each country is in equilibrium given the spending of the other members. Since small countries benefit from the expenditures of large countries whatever they themselves do, they have little incentive to spend very much themselves. In the case of defense spending, this argument was supported by cross-section evidence of a positive correlation between defense spending as a percentage of GNP and GNP itself. However, in the case of foreign aid the evidence was less clear: for 1960 the authors found a positive relationship; for 1962 the relationship was not statistically significant. The Olson-Zeckhauser (O-Z) paper gives rise to a number of questions. The preferences of each country are expressed in terms of a social utility function, but no attempt is made to relate this function to the preferences of individuals within the country or to the domestic political process. Conceptually, one may distinguish three different reasons why foreign aid expenditures as a proportion of GNP might be related to the level of GNP. First, there is the influence of per-capita income. If foreign aid expenditures (or rather the returns from such expenditures) are a luxury good, aid as a percentage of per-capita income will vary positively with per-capita income. Second, there is the influence of population. It is reasonable to assume that what is generated by foreign spending is a public good consumed by residents of the donor country. Other things being equal, the price of that public good for the residents of different donor countries will be a decreasing function of their country's population. In other words, since a large country will have a greater number of taxpayers, the cost per taxpayer of a given total amount of aid spending will be lower than in a small country. Finally, there is the effect which O-Z had in mind; namely, the influence of spending by other members of the country's reference group or alliance. It would seem worthwhile to separate these three influences. Since the O-Z paper was primarily concerned with defense alliances, the implications for foreign aid spending were not studied in detail. However, there is evidence that multilateral aid, that is, funds from several countries channeled through a joint organism, is determined by a mechanism quite different from allocations of bilateral aid, which is transferred directly from donor to recipient.2 It would be desirable, therefore, to disaggregate total aid into these two components. This paper attempts to bridge the gap between individual and group preferences by means of the median-voter hypothesis, at the same time extending this approach to cases where group Received for publication February 24, 1978. Revision accepted for publication August 24, 1978. * Universite de Montreal. The author is greatly indebted to Claude Montmarquette for advice and encouragement offered in the preparation of this study. The suggestions of anonymous referees were also very helpful. Benoit Audet prepared the data and carried out the estimations. The research was supported by the Canada Council. I See, for example, Ypersele de Strihou (1967). However, Mason (1963) notes that in 1961 the United States was spending a smaller percentage of its GNP on aid than the other members of the OECD's development assistance committee. 2 Price (1967) examines the determinants of member shares for a number of international organizations. However, his capacity-to-pay model overlooks the relationship between member shares and the preferences of residents of member countries.

IT is generally recognized that the international monetary crises of late 1967 and early 1968 were precipitated by large, continuing deficits in the United States balance of payments. Underlying these deficits was a steady erosion of the United States surplus on current account, which fell from 8.5 billion dollars in 1964 to 4.8 billion dollars in 1967. Despite government programs restricting foreign investment by United States residents,1 therefore, the overall deficit on a liquidity basis reached a level of 3.6 billion dollars in 1967. The resulting outflow of dollars and gold greatly reduced the confidence of speculators and central banks in the ability of the United States to maintain the international value of its currency. It is of course no coincidence that this deterioration in the current account position of the United States accompanied a rapid expansion of its military commitments in Southeast Asia. But the amount of the deterioration which may be attributed to the War in Vietnam is a point at issue. The Administration has estimated the direct foreign exchange cost of the War to be about 1.5 billion dollars.2 Many have argued, however, that if secondary and indirect effects are taken into account, the total impact on the balance of payments has been much greater. In this paper, we shall attempt to reach a more satisfactory estimate -or range of estimates -of the total effect of the War on the balance of payments. In so doing, we shall try to answer the question: What would the United States balance on current account have been in 1967 if real expenditures had been lower by the amount channeled into the military effort in Southeast Asia? The wording of this question perhaps requires further explanation. We shall assume that the multiplier effects of this hypothetical difference in expenditures could have been completely and precisely offset by monetary and fiscal policy. Thus we suggest that in the absence of the military build-up, expenditures for consumption, investment and government purposes other than the War would have been identical with those that actually occurred. In addition, it must be recognized that part of the increase in military expenditures would have been spent even without the Vietnam War build-up. The men now serving in the armed forces would have eaten, for example, regardless of whether they were in military or civilian roles. For this reason, we exclude domestic expenditures on military pay, provisions and accommodation from our calculations, and consider only those expenditures which may clearly be assigned to the expanded campaign in Vietnam.3 One may note several possible effects which this military spending may have had on the United States current account balance: 1) Increased direct foreign purchases of food, services and finished goods to supply the military effort. 2) Greater purchases of foreign goods to be used as inputs in United States defense production. 3) Deterioration in the United States net exports, due to both war-stimulated inflation and to supply bottlenecks in those sectors of production most affected by the increased spending. In the following sections, we shall examine each of these effects in turn. By summing the three components we hope to provide a relatively accurate and comprehensive picture of the balance of payments impact of the War. * The authors are grateful to Robert Triffin and Richard Cooper for their helpful comments and suggestions. 1 Note that this practice of restricting capital cannot be continued indefinitely without worsening the balance on current account, because it will of course reduce United States earning assets abroad. 2 U.S. Treasury Department, Maintaining the Strength of the U.S. Dollar in a Strong Free World Economy (Washington, 1968), p. 103. 'As noted below, foreign military purchases of food, etc., by the United States will be included in the direct current account impact of the War.

Previous models of public spending fail to explain why high inflation rates are distributed non-randomly across countries. In the switching-regime specification proposed here, governments tend to resort to inflationary debt monetarization when the spending level chosen by voters exceeds actual revenue capacity. The voter-choice and fiscal-capacity models of earlier studies are special cases of this framework. Using cross country data for 1970, 1975 and 1980, the authors find that the voter-choice specification applies to most industrialized countries, whereas fiscal capacity appears to constrain government spending in most developing countries. High inflation appears to result from shocks that alter a country's fiscal capacity relative to voters' expectations. Copyright 1992 by MIT Press.