W ’ A F A ? HO S FRAID OF OREIGN ID T D ’ P HE ONORS ERSPECTIVE ALBERTO CHONG MARK GRADSTEIN CESIFO WORKING PAPER NO. 1833 CATEGORY 1: PUBLIC FINANCE OCTOBER 2006 An electronic version of the paper may be downloaded • from the SSRN website: www.SSRN.com • from the RePEc website: www.RePEc.org • from the CESifo website: www.CESifo-group.de T T CESifo Working Paper No. 1833 W ’ A F A ? HO S FRAID OF OREIGN ID T D ’ P HE ONORS ERSPECTIVE Abstract With efforts across the industrial countries to increase the amount of foreign aid mounting, it is important to understand its determinants. This paper examines the factors affecting the support for foreign aid among voters in donor countries. A simple theoretical model, which considers an endogenous determination of official and private aid flows, relates individual income to aid support through the elasticity of substitution and also suggests that government efficiency is an important factor in this regard. The empirical analysis of individual attitudes, based on the World Values Surveys, reveals that satisfaction with own government performance and individual relative income are positively related to the willingness to provide foreign aid. Furthermore, when using donor country data we find that aid is adversely affected by government inefficiency and income inequality. JEL Code: O10, H41. Keywords: foreign aid, donors, perceptions, development, corruption. Alberto Chong Mark Gradstein Research Department Department of Economics Inter-American Development Bank Ben Gurion University Stop B-602 Beer Sheva 84105 1300 New York Ave, NW Israel Washington, DC 20577 [email protected] USA [email protected] Gianmarco Leon provided excellent research assistance. The findings and interpretations are those of the authors and do not necessarily represent the views of the Inter-American Development Bank or its corresponding executive directors. 1. Introduction For many developing countries foreign aid constitutes a significant part, often in excess of ten percent or more, of their national income, see the World Bank Development Report, 2005a. In contrast, it is a relatively miniscule item in the donor countries' national accounts. Despite the recognition that donations to poor countries may serve the rich countries' interests – either because of strategic, economic or humanitarian concerns – and many pledges by rich countries’ governments to significantly increase their amounts, flows of foreign aid from any given country do not typically exceed one percent of the national product.1 Because of the belief that foreign aid may be essential for maintaining minimal living standards in poor countries, providing immediate humanitarian relief, or even spur investment and growth, international agencies, NGOs, and other organizations have been conducting for years vigorous campaigns to enhance its flows – with very limited success. It appears, therefore, that the relatively small amount of aid is a reflection of deeply rooted priorities, ideologies or political forces in donor countries. Indeed, the calibrations in Kopczuk et al., 2003, show that the actual levels of foreign aid relative to domestic income transfers correspond to citizens in rich countries attaching the welfare of citizens in poor countries only 1/2000 of the weight assigned to the welfare of their own poor. Additionally, the amount of foreign aid differs significantly across the donor countries. For example, official development assistance constitutes around one percent of the GDP in Denmark, Sweden, and the Netherlands, but only one tenth of a percent in the United States; the total amount of aid (including, in particular, private flows) reaches almost two percent of the GDP in Sweden and 2 the Netherlands, but only a quarter of a percent in the United States, see Table 6.3 in USAID, 2004, and the World Bank Development Report, 2005a. It is, therefore, important to have a better understanding of the forces that shape the overall levels of aid flows and their differences across countries. To this end, this paper offers a political economy model of international giving, where politically determined amount of official aid interacts with private donations.2 While the weight assigned to foreign aid by the citizens in rich countries obviously plays a crucial role in the analysis as one would expect from Kopczuk et al., 2003, we identify additional potentially important elements. Thus, the perceived inefficiency of the recipient countries' use of aid as well as domestic government's inefficiency can potentially shape the political support for foreign aid. Yet another element is incomes. While generally ambiguous, under empirically plausible assumptions on preferences, the income effect implies that more affluent individuals should be no less supportive of aid – and likely more so – than poorer individuals. This also implies that in the resulting equilibrium, richer countries are more generous and that the lower is the degree of income inequality, hence, the more well off is the politically decisive voter, the higher is the level of political support for more generous giving. We also consider the case of multiple donors identifying a free riding effect, which implies an inverse relationship between the number of donors and the amount of aid provided by an individual donor country. Our empirical findings are consistent with the model's predictions. The empirical analysis of individual attitudes, based on the World Values Surveys, reveals that satisfaction with the government performance and individual income are positively related to the willingness to provide foreign aid. This, in particular, implies that more efficient and less corrupt governments would be more conducive to public support for foreign aid. Further, the analysis of actual foreign aid flows 1 At recent international forums, calls were made for each donor country to increase its aid commitment up to 0.7 percent of its national income, but the plan is uet to be implemented in full. 3 reveals that they are linked with domestic government corruption and income inequality, as predicted by the model. Earlier theoretical work Dudley and Montmarquette, 1976, Mosley, 1985, and Hatzipanayotou and Michael, 1995, contains valuable insights as to the modeling of foreign aid. In particular, some of these papers emphasize its public good aspect from the donors' perspective. The recent paper by Lahiri and Schweinberger, 2006, focuses on the interaction between government and private aid. We extend their political economy analysis to the standard majority voting framework while incorporating the "warm glow" motive for giving. More importantly, perhaps, we also incorporate factors neglected in the above literature, which nevertheless may have empirical significance such as government inefficiencies described above. The existing empirical literature on foreign aid has been almost entirely devoted to the issues of aid allocation and the effects of aid in receiving countries, see e.g., Alesina and Dollar, 2000, Alesina and Weder, 2002, Boone, 1996, Burnside and Dollar, 2000, Collier and Dollar, 2002, and a critical survey in Easterly, 2003. Our empirical focus is different in considering the attitudes toward foreign aid in donor countries and in this sense, it should be viewed as complementary to the existing line of research. Knack and Rahman, 2006, also consider donor incentives to provide aid, focusing on interactions among multiple donors; here, instead, the focus is on the donor country's characteristics that are more or less conducive to aid. Other work has previously used survey data to study views on international flows, in particular, migration, see e.g., Mayda, 2005; the empirical part of this paper is complementary to that work. The paper is organized as follows. The next section describes the basic theoretical framework, which is then followed by its analysis in Section 3. The empirical part begins in Section 2 In the United States, at least, an argument is often made that, whereas the amount of official aid is relatively small, private flows more than compensate for this, see USAID, 2004. Private flows of aid often constitute a significant aid component, about two thirds of the total aid in the United States, but only around a quarter in Denmark and France. 4 4 with the description of the data, followed by their analysis in Section 5; finally, Section 6 concludes. 2. Basic Framework Consider an economy populated by a continuum of individuals indexed i, whose measure is normalized to one.3 The individuals differ only in terms of their initial income endowment, y, i whose distribution is exogenously given, and are assumed to derive utility from private consumption c; from the aggregate amount of donated foreign aid, A; and from their private donation, z, which i i captures the warm glow motive for giving. The modeling approach that perceives foreign aid as a public good is quite standard, see, for example, Dudley and Montmarquette, 1976, Hatzipanayotou and Michael, 1995, and, Lahiri and Schweinberger, 2006, who similarly distinguish between government-provided and private aid. It also reflects the idea that, whatever the motivation is (strategic or humanitarian), foreign aid is often perceived as a national interest – see, e.g., USAID, 2004, which outlines the basic philosophy of aid from the US perspective. The incorporation of the warm glow has become standard in the related literature on public goods provision, see Andreoni, 1989, 1990. Thus, the individual preferences are U(c) + V(A) + W(z) (1) i i where U, V and W satisfy the standard assumptions and, in particular, are monotonic and concave.4 Income is allocated between consumption, paying taxes to finance the official or publicly provided aid, as well as private aid donations. We assume that a proportional income tax is used to 3 While the main analysis considers one donor-one recipient framework, an extension in the appendix exhibits a richer environment. 4 The separability assumption, while substantially simplifying the analysis, is not essential. 5 finance official aid and let T, 0 < T < 1, denote the tax rate. Then, normalizing prices to unity, the individual budget constraints can be written as: y = c + Ty + z (2) i i i i Letting Y denote the aggregate income and Z – the aggregate amount of private aid, the total amount of donation is A = TY + Z (3) The decision on the amount of official aid, or alternatively, on the tax rate used to finance this aid, is made collectively, by a majority vote among the individual voters; it is then followed by the individually made decisions on the allocation of the disposable income between consumption and private donations of aid. The political economic equilibrium consists of the majority-supported tax rate and individual optimal donations which are mutually consistent. The assumption that a democratic vote determines the amount of official aid is especially apt in our context where the donor countries are all exclusively democratic. Indeed, the assumed responsiveness of collective decisions to individual attitudes also constitutes one of the rationales behind the empirical focus on the attitudes’ data below. 3. Analysis The analysis proceeds backwards. Suppose first that the tax rate is exogenously given and consider the individual allocations of disposable income between consumption and private aid. The optimal individual allocation maximizes the utility (1) subject to (2), (3). The first order condition is 6 -U'(c) + W'(z) = 0 (4) i i whose solution can be written as c = h(y(1-T)), z = y(1-T) - h(y(1-T)) (5) i i i i i where 0 < h' < 1 ; so that, integrating across the individuals, Z = Y(1-T) - ∫ h(y(1-T))di (6) i and A = TY + Z = TY + Y(1-T) - ∫ h(y(1-T))di (7) i Differentiating we obtain: dA/dT = ∫ h'(y(1-T)) y di > 0 (8) i i This result implies that the crowding out of private donations by official aid is incomplete, so that a marginal increase in the official aid is not fully offset by a corresponding decrease in private aid, thus causing an increase in the overall amount of aid. Suppose now that, in anticipation of the private donation equilibrium as described above, the individuals vote on the tax rate. The preferred tax rate for individual i satisfies the first order condition: -U'(c) y + V'(A) dA/dT < 0 (9) i i 7 where dA/dT is given by (8); and (9) is strictly negative if T = 0. In general, the relationship between individual income and foreign aid preferences is complex. To facilitate its understanding, suppose that the all sub-utilities are isoelastic functions, so that (1) has the form c1-σ/(1-σ) + αA1-σ/(1-σ) + βz1-σ/(1-σ), α, β > 0 (1') i i The parameter α reflects the taste for foreign aid as such, whereas β represents the warm glow motive from making a donation. When σ > (<) 1, the elasticity of substitution is smaller (larger) than unity; σ =1 corresponds to the case of logarithmic preferences. Straightforward calculations reveal then that, for a given tax rate, c = β-1/σ y(1-T) / (1 + β-1/σ), z = y(1-T) / (1 + β-1/σ), Z = Y(1-T) / (1 + β-1/σ), i i i i A = TY + Y(1-T) / (1 + β-1/σ) (10) The first order condition for the tax rate (9) in this case takes the form -c-σ y + αA-σ(Y + dZ/dT) = -c-σ y + αA-σY(1 - 1/ (1 + β-1/σ)) = i i i i -[β-1/σ y(1-T) / (1 + β-1/σ)]-σ y + α[TY + Y(1-T) / (1 + β-1/σ)]-σY(1 - 1/ (1 + β-1/σ)) < 0 (11) i i where the second line is obtained by substituting from (10); this condition can be rewritten as follows: -(1-T)-σ (y/Y)1-σ + α[T + (1-T) / (1 + β-1/σ)]-σ < 0 (12) i 8 Differentiation of the left hand side in (12) reveals that the second order condition holds, implying that the individual preferences are single peaked, implying that the political economic equilibrium exists, and the median voter is decisive in determining the political outcome. It follows from differentiating (12) that the preferred tax rate increases (decreases) with the relative individual income depending on whether the elasticity of substitution is greater (is smaller) than unity. Because in any case, tax preferences depend monotonically on individual income, in the political equilibrium, the median income voter is decisive, and the equilibrium condition is -(1-T)-σ (y /Y)1-σ + α[T + (1-T) / (1 + β-1/σ)]-σ < 0 (13) m where y is the median income. The ratio y /Y will be interpreted as being inversely related to m m income inequality (as would, for example, be the case with lognormal income distribution). The above analysis bears implications for the resulting amounts of private aid. The relative share of private aid out of the total amount of aid is Z/A = 1/[1 + T(1+β-1/σ)/(1-T)], which clearly decreases in T. Since total aid is an increasing function of the tax rate, this implies a negative relationship between the amount of total aid and the share of private aid in it. To summarize, Proposition 1. Support for government-provided foreign aid increases (decreases) with income depending on whether the elasticity of substitution is greater (smaller) than one. Correspondingly, in the resulting political economic equilibrium income inequality reduces the political support for official aid when the elasticity of substitution is greater than one; enhances it when the elasticity of substitution is smaller than one; and is irrelevant in the case of logarithmic preferences. The share of private aid out of total is a decreasing function of the tax rate used to provide aid financing. 9