WPS6956 Policy Research Working Paper 6956 Containing Volatility Windfall Revenues for Resource-Rich Low-Income Countries Anton Dobronogov Octave Keutiben The World Bank Africa Region Poverty Reduction and Economic Management Unit June 2014 Policy Research Working Paper 6956 Abstract An abundance of natural resources is both an opportunity resource revenues. This paper develops a growth model and a challenge for developing countries. Several à la Barro in which the government receives windfalls resource-rich, low-income countries receive amounts of (from natural resources and foreign aid) and rent-seeking foreign aid that are similar to or larger than their actual agents contest for public funds. The key conclusion or potential revenues from natural resources. In such is that making aid countercyclical helps to achieve countries, the donors may have an opportunity to help higher economic growth, and so does conditioning a government to use its resource revenues productively disbursements on enhancement of public capital. and minimize the magnitude of risks created by resource Introducing elements of insurance in the design of both rents. Development of aid instruments tailored for such aid products financing investments in infrastructure and purposes might be helped by model-based analysis of social services and supporting policy and institutional the effects of foreign aid on resource-rich, low-income reforms may help to achieve both of these objectives. economies and its interactions with the flows of natural This paper is a product of the Poverty Reduction and Economic Management Unit, Africa Region. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be contacted at adobronogov@worldbank.org and o.keutiben.njopmouo@umontreal.ca. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Containing Volatility: Windfall Revenues for Resource-Rich Low-Income Countries∗ Anton Dobronogov† Octave Keutiben‡ June 2014 Keywords: Economic growth; Foreign aid; Natural resources; Rent-seeking; Volatility JEL classification: D72; F35; 043; Q33 ∗ We thank Phil Keefer, Fulbert Tchana Tchana, Nelnan Koumtingue, Kevin Carey, and Sebastien Dessus for helpful comments. All errors are ours. † World Bank, 1818 H Street, NW, Washington, DC 20433, USA. adobronogov@worldbank.org ‡ Department of Economics, Université de Montréal, C.P. 6128, succursale Centre-ville, Montréal, Québec, H3C 3J7, Canada. Phone: +1-514-686-0264; Email: o.keutiben.njopmouo@umontreal.ca 1 Introduction Abundance of natural resources is both an opportunity and a challenge for de- veloping countries. One may expect a positive impact if the country’s policies and institutions are conducive for accelerating accumulation of physical and human capital. The discovery and development of natural resources can help to increase the economy’s investment rate, bringing about faster economic growth in the short run. Further, revenues generated from these resources can be used to enhance so- cial outcomes related to health and education that induce economic growth in the long run. These revenues may also help growth by allowing the governments to finance their expenditures with lower levels of taxation of economic activity. However, if not handled well, natural resources adversely affect economic growth , particularly in low-income countries.1 Natural resource revenues are in- trinsically temporary and highly volatile (for example, the price of Brent oil var- ied between $9 and $157 during the 1999-2009 period). Cash flows from natural resources cause an appreciation of the real exchange rate, thus shifting factors of production out of technologically sophisticated activities with positive spillovers to other sectors of the economy such as manufacturing, and producing a phenomenon known as Dutch Disease. An increase in oil revenues owing to a rise in oil prices or increased production drives up the real exchange rate, causing domestic production of exportables to drop and imports to rise. Problems occur if the increased revenues prove temporary. Reversing the above scenario through depreciation often proves difficult, and repeated boom–bust cycles will lower investment in domestic pro- duction of tradable goods. Because most innovation and technological progress 1 See Frankel (2012) for a more detailed review of these effects. 2 take place in the tradable sector, shifting resources into the less dynamic nontrad- able sector can have long-term consequences for economic growth. Resource-rich economies vulnerable to Dutch Disease remain highly dependent on revenues de- rived from those resources and therefore exhibit boom–bust cycles in their growth performance. In economies where a large portion of revenues accrues to the public sector, public expenditure initiates the cycle, as government spending increases and de- creases with the rise and fall of revenue. An obvious solution would be to smooth government expenditure, with the government saving all or part of the windfall during upswings and drawing down assets to maintain expenditure during the bust. If the government is to accumulate assets during booms, it must establish a stabi- lization fund. The objective of the policy is not to prevent real appreciation per se, which would be impossible if any benefit were to be derived from the oil, but rather to prevent the volatility associated with the boom–bust cycle.2 Unfortunately, the political economy of most resource-rich countries is not con- ducive to strengthening the policies and institutions needed to avoid procyclical spending, and high price volatility further challenges these policies and institu- tions. A country rich in natural resources may feel a false sense of security that leads the government to focus on distribution of the resource rents and to delay im- plementation of necessary economic reforms and other growth-enhancing policies. Among other things, the government may blunt demands for reform by adopting economically inefficient but politically effective patronage policies. Abundance of natural resource rents may cause wasteful rent-seeking. Exten- 2 In principle, the deleterious consequences of real appreciation can be avoided by basing spending on a stable stream of income rather than on volatile current revenue. 3 sive state intervention, usually accompanied by meager accountability and trans- parency at the implementation stage, invites resource-wasting competition for fa- vorable bureaucratic treatment. A windfall gain, such as higher oil revenues, may lead to intensified rent-seeking, a more-than-proportional rise in fiscal redistribu- tion, and lower growth—a phenomenon known as the voracity effect (Lane and Tornell, 1996). Rent-seeking may take many forms, from corruption to civil war. In the resource-rich countries of Latin America, central governments have permit- ted extensive rent-seeking by subnational governments, state-owned companies, and the military, institutions that have limited incentives to minimize costs and in fact may have incentives to overspend so as to qualify for increased transfers. The result has been political instability and short-sighted policies, both of which make it very difficult to carry out economic reforms (Dornbusch and Edwards, 1991). As a result, stabilization funds in low-income environments are at risks of be- coming institutions for fiscal transfers from fiscally responsible to fiscally irrespon- sible governments if deterioration of governance over time is not avoided (Collier, 2007). Resources accumulated in the fund represent a rent which can be fought for, unless the quality of the country’s institutions is sufficiently high to prevent that from happening, so while a stabilization fund might become a useful tool en- hancing macroeconomic stability, in certain circumstances it may also increase incentives for rent-seeking and thus contribute to institutional deterioration. Large-scale foreign aid is another factor which may affect the impact of re- source revenues on economic development in low-income resource-rich countries. A number of low-income countries receive amounts of foreign aid (typically, from a large number of donors and in a variety of forms) which are similar to or larger than their actual or potential revenues from natural resources. For example, three 4 countries from the Great Lakes region, Burundi, Democratic Republic of Congo, and Uganda, all of which have substantial discovered reserves of oil and miner- als, are also reported by OECD’s Development Assistance Committee to receive official development assistance of the magnitude equivalent to double-digit per- centages of their respective GDPs. In a resource-rich low-income country, the donors may have an opportunity to help a government to use its resource revenues productively and minimize the magnitude of risks created by resource rents. As their natural resource revenues increase, these countries will be facing an opportu- nity of transition from aid dependency to greater fiscal self-sufficiency. During this transition period, macroeconomic and institutional impact of the oil revenues on the economies of low-income countries will much depend on the amounts and the structure of foreign aid being delivered after the oil revenues start flowing in. There are two principal channels through which foreign aid may interact with resource revenues. First, foreign aid may potentially help to stabilize the government revenues, making it easier for a country with limited institutional quality to stabilize expen- ditures. This will happen if amounts of aid are systematically countercyclical with respect to oil revenues (or with respect to oil prices if production remains relatively stable). In principle, this can be achieved, but is by no means automatic. On av- erage, foreign aid is in fact mildly pro-cyclical.3 Since the donor agencies have incentives to disburse aid according to pre-determined targets, during the periods when oil revenues are rapidly increasing they may supply amount of resources which may exceed what the government can spend with reasonable efficiency and 3 See Perry (2009) for a review of the relevant literature. 5 effectiveness. Second, a trajectory for substitution of oil revenues for aid will determine whether the countries will receive support with resolving the problems they are not yet institutionally equipped to handle. Collier (2006) has shown that on av- erage aid has been more effective at promoting development than oil revenues, predominantly because of complementary inputs by the donors, such as technical assistance and scrutiny of public expenditures. Major change in the scale and structure of the government’s own revenues calls for a corresponding change in design of the foreign aid programs. Typically, donors gear such efforts towards strengthening institutions for oil wealth management by means of conditionality and technical assistance – that is, towards improving gov- ernment’s capacity to absorb financial resources productively. These undoubtedly useful efforts need to be complemented by changes in design of aid instruments helping to stabilize the government’s overall revenues. If such changes are made only ad-hoc in response to deterioration in the quality of the country’s institutions and economic performance after it occurs, critical time will be lost. Development of such instruments might be helped by model-based analysis of the effects of for- eign aid instruments on resource-rich low-income economies and their interactions with the flows of natural resource revenues. Economic effects of aid and natural resource revenues have been examined separately, though frequently compared, in the literature.4 Our paper contributes to this body of work by developing a model which offers a coherent analytical framework for looking at these effects jointly.5 4 See Morrison (2012) for a review of this literature. 5 To the best of our knowledge no model has looked at these effects jointly before 6 We develop a growth model à la Barro in which the government simultaneously receives windfalls from natural resources and foreign aid, and rent-seeking agents contest for public funds. By jointly consider foreign aid and natural resource rev- enues, we show that making aid countercyclical helps to achieve higher economic growth, and so does conditioning disbursements on enhancement of public capital. The remainder of the paper is organized as follows. Section 2 outlines the analytical framework and its key characteristics. Section 3 characterizes the solu- tion of the model and discuss the main results. Section 4 highlights some policy implications and concludes. 2 The model Our baseline model is an extension of Hodler (2007) accounting explicitly for natural resource revenues. We consider a resource-rich country who produces and consumes a single commodity. There are private agents and government. Govern- ment revenue can come from foreign aid, resource rents, and income tax. Private agents can allocate labor to output production or to rent-seeking activities.6 2.1 Private agents There is a continuum of private agents of measure 1. Each agent i is endowed with one time unit at each date t , a fraction lit of which can be devoted to labor supply in the productive sector, and the balance, 1 − lit , to rent seeking. At each period t , agent i can generate income by producing output yit and by appropriating 6 By considering a one sector growth economy, our model mainly focus on resource curse causes by the volatility of natural resource revenues or weak institutions rather than on Dutch Disease. 7 rents bit from government revenue. The production technology is given by α yit = Akit (lit gt )1−α , (1) where α ∈ (0, 1) and A > 0. kit denotes the physical capital input of agent i at date t and gt is the stock of productive public capital (public infrastructure). The economy is initially endowed with a stock of public capital g0 > 0 considered as given.7 The produced income can be used for current private consumption cit and to accumulate private capital kit , whereas the usurped income contributes directly to the utility of agents.8 Hence, given the initial endowment in the stock of private capital ki0 = k0 , the capital accumulation constraint for each individual i is described by kit +1 = (1 − τt )yit − cit + (1 − δk )kit , (2) where τt ∈ [0, 1) is a proportional tax rate on income and δk ∈ [0, 1] is the rate of depreciation of private capital. The appropriation technology is such that agent i 7 Public capital should be broadly understood to include not only physical infrastructure (railroads, roads, airports, etc.), but also social services such as education and health. 8 It is assumed that domestic households do not have access to world capital markets, so private foreign borrowing is zero. The assumption of no external financial market access approximates the non-concessional borrowing ceilings agreed between the government and the donors as well as major constraints faced by domestic private sectors of the low-income countries in their access to international capital markets. Given this, the appropriated income is introduced directly in the utility function (as in Svensson, 2000) because it does not contribute to the domestic economy. This consideration seems plausible since most of the misappropriated public funds are either consumed or secured abroad. 8 rent appropriation is given by   0   for eit = 0, bit = (3) eit et bt for eit > 0,    1 where eit = 1 − lit denotes the rent seeking effort of agent i, et = 0 e jt d j the ag- gregate rent-seeking effort, and bt the total amount of public funds appropriated by rent-seekers. At each date t , each agent i must take two different decisions. First, she must allocate her time between rent seeking and productive activities; second, she must decide her consumption-saving plan. The objective of each agent i is therefore to maximize her expected lifetime utility ∞ E0 (Ui ) = E0 ∑ β t u(cit , bit ), (4) t =0 where E0 is the expected value operator, β ∈ (0, 1) is the discount factor. The instantaneous utility function, u(·) is assumed to satisfy the Inada conditions. 2.2 Government At each date t , the government invests gt +1 in public infrastructure.9 To fi- nance its expenses, the government levies a proportional tax on output at the rate 9 Productive government infrastructure is modeled as a stock such that the aggregate stock of public capital at t , gt , is not provided out of current output but results from past public investments (Irmen and Kuehnel, 2008). g g Normally, the stock of public capital evolves according to gt +1 = It + (1 − δg )gt , where It denotes current public expenditures on infrastructure. For reasons of tractability, we assume full depreciation of public capital (i.e., δg = 1), such that gt +1 denotes public investment as well. 9 τt and receives windfall revenues from natural resource, Zt , and from foreign aid, At , which can be a pure (untied) transfer or is tied to investment in public infras- tructure. Following Chatterjee et al. (2003), Neanidis and Varvarigos (2009), and Borensztein et al. (2013) we assume that foreign aid disbursements and windfall from natural resources are measured in proportion of domestic income, such that: At = at yt ≡ ay ¯ t , Zt = zt yt , (5) 1 where yt = 0 yit di is the aggregate output. The windfall revenues from natural resources is highly volatile such that {zt }t 0 is assumed to be a sequence of identically distributed random variables. More pre- cisely, the exogenously determined resource revenue process {zt }t 0 is assumed to be a linear Markov process given by 2 ¯ + ρ zt −1 + εt , εt ∼ i.i.d.(0, σε zt = (1 − ρ )z ). (6) ¯ is the average resource revenue (the steady state level of zt ) and ρ ≤ 1 where z captures the persistence of the revenue process. For this specification, the variance of the process {zt }t 0 2 /(1 − (using the invariant distribution) is given by σz2 = σε ρ 2 ). This dynamics of resource windfalls reflects both a possible upward trend (an ¯ if natural resource revenues increase) and volatility (an increase in increase in z σz ).10 To capture the donor disbursement policy in a more realistic way, it is further assumed that aid flow at time t , at , is indexed to the flow of resource revenue, zt : 10 The volatility in resource revenues can be due to exogenous changes in either the price of the resource or the endowment of the resource. 10 when resource revenue falls (goes up) below (above) its average level, aid flow to the recipient country is adjusted upwards (downwards), proportionately. More specifically, aid flows to the recipient country are given by: at = a ¯ − zt ). ¯ + µz¯ (z (7) ¯ is the average level of resource revenue and µz¯ ∈ ¯ is the average level of aid flow, z a (0, 1) is the adjustment coefficient that determines the degree of aid indexation.11 The government cannot issue debt claims and therefore is required to maintain a balanced budget each period. Thus, public capital evolves according to12 gt +1 = [1 − φ (et )] pt + λ at yt , (8) where pt = τt yt + zt yt + (1 − λ )at yt is the government funds targeted by rent- seekers, φ (e) is the share of government funds appropriated by agents engaging in rent seeking activities and λ ∈ [0, 1] represents the degree to which foreign aid contributes to the formation of public capital.13 Thus, the aggregate amount appro- 11 The aim of indexing aid flows to some macroeconomic indicator of the recipient country’s in- come is to reduce its macroeconomic volatility thereby stabilizing its income. Aid flows can be indexed to various indicators (terms of trade, GDP, etc.), each having advantages and inconveniences (see for example Dhasmana, 2008). Resource windfall indexed aid is chosen because resource rev- enue is mostly the main determinant of aggregate income in low-income resource rich countries. It should be expected that µz¯ increases with z¯. 12 We abstract from considerations of international borrowing to keep the analysis tightly focused on resource volatility. Nonetheless, this could be justified by evidence provided by Prati and Tressel (2006) who argue that the vast majority of aid recipient countries have accumulated very high levels of debt that severely restricts their capacity to borrow in international markets (see also Neanidis and Varvarigos, 2009). 13 λ = 0 means Program aid (also referred to as budget or “untied” aid) which generally takes the form of a cash disbursement and is perfectly fungible. By contrast, λ = 1 means project aid (or “tied” aid), consisting of transfers for investment projects agreed between the donor and the recipient country; whether it is fungible or not depends on whether, prior to the aid commitment, the recipient 11 priated from government coffer is given by bt = φ (et ) pt = φ (et )[τt + zt + (1 − λ )at ]yt . (9) The function φ (·) is such that φ (·) > 0 and φ (e) ∈ [0, 1) for all e.14 φ (·) is assumed to be an increasing function of e, the aggregate rent seeking activity, to capture the idea that the quality of institutions in low-income resource-rich coun- tries tends to decline during the periods of rapidly increasing government revenues. This is because in equilibrium, the aggregate rent-seeking effort may be positively related to government revenues. Therefore, an increase in government revenues would increase the aggregate rent-seeking effort e, thereby increasing φ (et ), mean- ing a decrease in the quality of institution.15 To sustain an equilibrium of on-going growth (Barro, 1990), it is assumed that every period the government adjusts its policy such as to keep the rate of public investment constant at gt +1 = gy ¯ t (10) Substituting (10) in the government budget constraint (8) yields the following tax country intended to finance the project itself (Agénor and Aizenman, 2010). Tied aid is assumed to be secured from appropriation. 14 Following Hodler (2007), φ may be interpreted as an inverse measure of institutional quality: The higher φ is, the poorer are the institutions that restrict appropriation of public funds by rent seekers. 15 It is worth emphasizing that this feature of rent-seeking ignores the congestion effects in appro- priation technology, whereby the marginal product of rent seeking ought to be lower, the higher the total amount of rent-seeking activity, as the amount of public funds available for each rent seeker to steal falls when everybody is stealing from it (see Mauro, 2004). 12 rule ¯ − λ at g τt = − zt − (1 − λ )at 1 − φ (et ) (11) ¯ g φ (et ) = − zt − at − λ at . 1 − φ (et ) 1 − φ (et ) This equation shows that windfall revenues may allow the government to substitute for domestic tax financing. In particular, the tax rate may be very low when aid is tied and used to co-finance public capital (i.e., λ > 0). 3 Private sector optimal decisions At each period t , agent i considers public infrastructure gt , aggregate rent- seeking effort et , aggregate public funds appropriated bt and her current capital stock kit as given. Then she allocates her time and chooses her consumption such as to maximize her expected utility, given in equation (4), subject to her initial capital endowment ki0 and the capital accumulation constraint (2). The Lagrangian for each individual i maximization problem is: ∞ Li = E0 ∑ β t {u(cit , bit ) + νit [(1 − τt ) f (kit , gt , lit ) − cit − kit +1 + (1 − δk )kit ]}, t =0 where νit is the shadow value of wealth and f (kt , gt , lt ) = Aktα (lt gt )1−α = yt . Her optimal choice is characterized by the following first-order conditions: ∂ Li ∂ u : (t ) = νit , (12a) ∂ ct ∂ c ∂ Li bt ∂ u ∂f : (t ) = νit (1 − τt ) (t ), (12b) ∂ lt et ∂ b ∂l 13 ∂ Li : β Et (νit +1 Rit +1 ) = νit , (12c) ∂ kit +1 ∂ Li α : kit +1 = (1 − τt )Akit (lit gt )1−α − cit + (1 − δk )kit . (12d) ∂ νit where Rt = (1 − τt )α Aktα −1 (lt gt )1−α + 1 − δk ≡ (1 − τt ) ∂ f ∂ k (t ) + 1 − δk and Et is the conditional expectation operator given the information available at time t . In addition, the transversality condition on capital, lim j→∞ Et (β j νt + j kt + j+1 ) = 0 must be satisfied. Equation (12a) is the traditional condition equating the marginal utility of con- sumption to the shadow value of wealth. Equation (12b) equates the marginal ben- efit (marginal product of rent seeking times its marginal utility) with the shadow value of the after-tax marginal product of labor, which is the productive income foregone. Equation (12c) is the dynamic optimality condition, equating the marginal cost with the expected marginal benefit of an additional unit of private capital in- vestment. Equations (12a) and (12c) can be summarized by equation (13) which is the usual Euler equation characterizing the intertemporal efficiency condition. ∂u ∂u (t ) = β Et (t + 1)Rt +1 . (13) ∂c ∂c For analytical tractability and to derive closed-form solutions , we will use the momentary utility u(·) = ln(·) and assume a full depreciation of physical capital, (i.e., δk = 1). We further assume that utility is separable so that u(ct , bt ) = ln(ct ) + θ ln(bt ), where θ > 0 represents the relative importance of misappropriated funds in utility. Using the logarithmic utility, the first order conditions are rewritten as: 14 ∂ Li 1 : = νit , (14a) ∂ ct cit ∂ Li θ yit : = νit (1 − τt )(1 − α ) , (14b) ∂ lt 1 − lit lit ∂ Li : β Et (νit +1 Rit +1 ) = νit , (14c) ∂ kit +1 ∂ Li α : kit +1 = (1 − τt )Akit (lit gt )1−α − cit . (14d) ∂ νit The Euler equation (13) can also be rewritten as 1 1 α −1 1−α = β Et (1 − τt +1 )α Akit +1 (lit +1 gt +1 ) . (15) cit cit +1 Multiplying both sides of equation (15) by kit +1 and using equation (14d) yields kit +1 kit +2 = αβ + αβ Et . (16) cit cit +1 Equation (16) is a first order stochastic difference equation which can be solved by the method of repeated substitution. Given the transversality condition on capital, lim j→∞ Et (β j kt + j+1 /ct + j ) = 0, one can infer that the solution to this first order stochastic difference equation is a constant, say H , such that kit +1 /cit = H for all t . Therefore, expression (16) can be rewritten as H = αβ + αβ H . (17) 15 Expression (17) can be solved for H = kit +1 /cit to yield16 kit +1 αβ = . (18) cit 1 − αβ Substituting equation (18) in (2), one can solve for kit +1 and derive the optimal private investment (saving) and consumption rates as kit +1 = αβ (1 − τt )yit = sit (·)yit . (19a) cit = (1 − αβ )(1 − τt )yit . (19b) Combining equations (19b) and (14a), and substituting in (14b) determines the optimal level of labor and rent-seeking effort θ (1 − αβ ) 1−α e∗ it = ≡ e∗ ; 1 − e∗ ∗ it = lit = ≡ l∗. (20) θ (1 − αβ ) + 1 − α θ (1 − αβ ) + 1 − α Equation (20) shows that each agent i chooses the same rent-seeking effort e∗ over- time. Given that ki0 = k0 , it follows that all agents produce the same output yi0 = y0 , obtain the same rent bi0 = b0 and accumulate the same amount of capital at date 0. Therefore, they choose again the same rent-seeking effort in the ensuing period. Repeating this reasoning infinitely, one can conclude that e∗ ∗ it = e , kit = kt , cit = ct , bit = bt and yit = yt for all i and t . Using the tax rule (11), optimum saving and 16 This solution satisfies the transversality condition on capital and can be checked by direct sub- stitution in the stochastic difference equation (16). 16 consumption rates (19) can be rewritten as g¯ φ (e∗ ) kt +1 = αβ 1 − + zt + at + λ at yt = st (·)yt . (21a) 1 − φ (e∗ ) 1 − φ (e∗ ) g¯ φ (e∗ ) ct = (1 − αβ ) 1 − + zt + at + λ at yt . (21b) 1 − φ (e∗ ) 1 − φ (e∗ ) From (21a), it immediately follows that ∂ st (·)/∂ zt > 0; hence, a temporary rise in the resource windfalls increases private capital accumulation. Intuitively, a tem- porary increase in resource windfalls provides more resources to the government budget that may allow a reduction in income tax rate (thereby leaving additional resources to individuals for both consumption and saving) or an increase in the provision of public infrastructure (thereby increasing the marginal productivity of private capital, again leading to a positive accumulation of private capital). Using the fiscal rule (10) along with the saving rule (21a) in the t + 1 variant of equation (1), we can derive the stochastic, endogenous growth rate of output as yt +1 g¯ φ (e∗ ) α ¯ ∗ )1−α 1 − = A(αβ )α (gl + zt + at + λ at , yt 1 − φ (e∗ ) 1 − φ (e∗ ) which, after substituting for at from (7), can be rewritten as yt +1 α ¯ − zt ) + wa = Ω v + zt + µz¯ w(z ¯ , (22) yt φ (e ) ∗ ¯ ¯ ∗ )1−α , v = 1 − 1−φg where Ω = A(αβ )α (gl (e∗ ) and w = 1 + λ 1−φ (e∗ ) . The temporary growth rate of output is a random variable with different realiza- tions each period depending on different realizations of resource revenues shocks, zt . The long-run growth rate of output, γy , is the mean value obtaining by taking 17 expectations on (22). This ultimately yields yt +1 1 + γy ≡ Mean = ΩE0 [∆(zt )], (23) yt where, the function ∆ is defined by α ¯ − zt ) + wa ∆(zt ) = v + zt + µz¯ w(z ¯ . (24) g¯−λ a¯ ∂ γy From (24), ∂∆ ∂φ = − [1− φ (e∗ )]2 < 0.17 Then, ∂φ < 0, implying that a decrease in institutional quality, i.e., an increase in φ dampens growth. This is because low institutional quality makes rent-seeking relatively more attractive, entailing a de- crease in labor inputs. Notice that the negative impact of weak institutional quality on growth is larger when aid is untied (i.e., λ = 0). Accordingly, the negative im- pact of weak institutional quality may be mitigated if aid is tied and the contribution of aid to the formation of public capital is large, that is if λ is highly positive. ∂∆ ∂∆ It also follows from (24) that ∂z > 0 and ∂a > 0. Therefore, all else equal, an increase in windfall revenues (from natural resources or foreign aid) enhances growth. This is because abundance of windfall may induce the government to lower the tax rate thereby leaving private agents with higher disposable income allowing them to consume and to save more. The increase in saving translates to an increase in private investment and ultimately has a positive effect on output growth. Essentially, equation (22) shows that the actual growth rate is a random vari- able. Hence, rather than growing at a constant rate, the economy exhibits over time temporary growth rates that depend on different realizations of resource windfall 17 This is because g ¯−λa ¯ > 0, which follows from equation (11), giving that the tax rate is required to be positive. 18 shocks in different periods. From equations (23) and (24), it is straightforward to show that the long-run effects of the variability of windfall revenues on trend growth are negative. Reducing overall volatility of revenues by making aid countercyclical with re- spect to the resource revenues helps to achieve higher trend growth. Indeed, it can be easily shown that higher volatility in resource revenues implies a lower expected growth rate in the long run. The intuition of these results goes as follows: recall from (21a) that a temporary increase in resource windfall has positive effect on investment and growth. However, noting that the specification in (24) is concave in zt , the temporary rise in growth resulting from an increase in resource revenue is not as pronounced as the the reduction in growth resulting from a decrease in resource revenue of equal magnitude. Consequently, higher volatility in resource revenue on average has a negative impact on growth.18 In effect, the economy would be better off getting a constant amount of resource revenue with certainty than getting the same amount on average, but with a non-zero variance. The index- ation of aid inflows to natural resource revenues may help to mitigate the negative impact of resource volatility. 4 Policy implications What can we learn from this analysis for the purposes of aid instruments de- sign? The key conclusion is that making aid countercyclical helps to achieve higher 18 This can be better seen through the function ∆. Indeed, α < 1 implies that the production technology exhibits a diminishing marginal productivity of both private and public capital. Therefore ∆ is concave with respect to zt , implying that volatile resource windfall reduces expected growth. This follows immediately from Jensen’s inequality. 19 economic growth, and so does conditioning disbursements on enhancement of pub- lic capital. We start from discussing how the latter objective can be achieved, and then look at how it can be combined with the former one. Aid can enhance the stock of public capital in a number of ways, most ef- fectively when multiple ways are applied simultaneously as there are important complementarities among them. First, public capital in this context must not be thought of as only physical cap- ital, and foreign aid which can enhance its stock is not limited to financing infras- tructure projects. Investments in human capital through public spending on social services such as primary health care and primary education may have positive ex- ternalities in the developing countries, benefiting not only their direct recipients of these services. Hence, a portion of human capital produced by such investments can be thought of as public capital. Second, aid may enhance public capital if it makes public spending on infras- tructure and social services more effective by means of supporting policy reforms and institutional changes, such as strengthening fiscal rules and improvements in the budgeting process or in public financial management. Implementation of in- vestment projects financed by the donors might be more efficient compared to the projects implemented by the recipient government on its own, for example because the donor’s involvement may help to transfer new technologies to the economy or because use of the donor’s procurement rules may reduce the costs of the project. However, the investment projects approach to foreign aid has its limitations. Since money is fungible, the government may offer the donors to finance the projects which it would have implemented anyway, and reallocate its own resources to other, less efficient or less benign uses. It may also fail to finance maintenance 20 of the completed pieces of infrastructure. Donors’ support to improvements in policies and institutions governing public finance helps to reduce the magnitude of such problems. In “budget support” op- erations serving this purpose, disbursements are conditioned on implementation of pre-agreed changes and become a part of the country’s budget. One way then to en- sure that foreign aid is both countercyclical and helps to enhance the stock of public capital is to introduce elements of insurance in the design of both aid products fi- nancing investments in infrastructure and social services and supporting policy and institutional reforms. For the investment projects, this can be done through their co-financing by the government and the donors at a variable rate: if oil prices are low, the donors contribute more and the government contributes less; and the other way around if the oil prices are high. For the budget support operations, the fact of disbursement conditioned on changes in policies and institutions might be delayed for certain period (say, a year), and its amount might be conditioned on the average price of the relevant commodity during that period. In addition to reducing the risk of macroeconomic instability, such designs would help to tie-in some of the oil rents into more productive spending, thus also making unnecessary excessive ac- cumulation of resources in the stabilization fund and reducing amount of resources otherwise available for rent-seeking. Yet no design of aid products can ensure that the governance in a low-income resource-rich country does not deteriorate to the point when it substantially affects amounts of foreign aid which can be absorbed by the country’s economy with reasonable efficiency. This is particularly true about the budget support operations where use of resources is subject to the country’s own public financial management rules and institutions. To mitigate this problem, Dobronogov, Gelb, and Saldanha (2014) propose another approach would be to adjust the level of program 21 disbursements in response to resource shocks so that countercyclical aid flows provide a degree of insurance to the development program, which could include a range of budget support, project support and other instruments. This would complicate project agreements but it offers some advantages, especially in situations when the donor is not comfortable in providing the equivalent of budget support. To conclude, if design of foreign aid programs and projects is tailored to spe- cific circumstances of low-income resource-rich countries, the donors may have an opportunity to help their governments to use their resource revenues productively and to increase their overall envelop of fiscal resources without amplifying the risks of rent-seeking. The analysis in this paper offers some theoretical underpinnings for developing such products and programs. 22 References Agénor, P.-R., Aizenman, J., 2010. Aid volatility and poverty traps. Journal of Development Economics 91, 1–7. Barro, R., 1990. Government spending in a simple model of endogenous growth. Journal of Political Economy 98 (5), S103–S125. Borensztein, E., Jeanne, O., Sandri, D., 2013. Macro-hedging for commodity ex- porters. Journal of Development Economics 101, 105–116. Chatterjee, S., Sakoulisb, G., Turnovsky, S. J., 2003. Unilateral capital transfers, public investment, and economic growth. European Economic Review 47, 1077 – 1103. Collier, P., 2006. Is aid oil? an analysis of whether africa can absorb more aid. World Development 34 (9), 1482–1497. Collier, P., 2007. The Bottom Billion: Why the Poorest Countries are Failing and What Can Be Done About It. Oxford University Press. Dhasmana, A., April 2008. Welfare gains of aid indexation in small open economies. IMF Working Paper 08/101. Dobronogov, A., Gelb, A., Saldanha, F., 2014. How should donors respond to resource windfalls in poor countries? From aid to insurance. Policy Research Working Paper 6952, The World Bank. Dornbusch, R., Edwards, S. (Eds.), 1991. The Macroeconomics of Populism in Latin America. University of Chicago Press. Frankel, J., 2012. The natural resource curse: A survey of diagnoses and some pre- scriptions. In: Arezki, R., Pattillo, C., Quintyn, M., Zhu, M. (Eds.), Commodity Price Volatility and Inclusive Growth in Low-Income Countries. International Monetary Fund. Hodler, R., 2007. Rent seeking and aid effectiveness. International Tax and Public Finance 14, 525–541. Irmen, A., Kuehnel, J., May 2008. Productive government expenditure and eco- nomic growth. Discussion Paper Series No. 464, University of Heidelberg. Lane, P., Tornell, A., 1996. Power, growth and the voracity effect. Journal of Eco- nomic Growth 1 (2), 213–241. 23 Mauro, P., 2004. The persistence of corruption and slow economic growth. IMF Staff Papers 51 (1), 1–18. Morrison, K., 2012. What can we learn about the “resource curse” from foreign aid? The World Bank Research Observer 27 (1), 52–73. Neanidis, K. C., Varvarigos, D., 2009. The allocation of volatile aid and eco- nomic growth: Theory and evidence. European Journal of Political Economy 25, 447–462. Perry, G., 2009. Beyond Lending: How Multilateral Development Banks Can Help Developing Countries Manage Volatility. Washington, DC: Center for Global Development. Prati, A., Tressel, T., 2006. What is the most effective monetary policy for aid- receiving countries? Working Paper no. 12, United Nations Department of Eco- nomic and Social Affairs. Svensson, J., 2000. Foreign aid and rent-seeking. Journal of International Eco- nomics 51, 437–461. 24