Panacea, placebo, or poison? The impact of development aid on growth

Abstract

Abstract This article holistically evaluates the impact of developmental aid on economic growth. Although most scholars find that aid has no direct impact on economic development (for example, Easterly 2001) and some even question the value of aid altogether (Moyo 2009), we highlight that these claims are premature and unsupported by empirical evidence. While our structural equation model confirms that there is no direct impact of aid on growth, we find that aid has an indirect positive effect on economic development. Developmental aid leads to a decrease in a country's level of corruption, which then positively affects growth. The implications of this finding are that most aid is effective in that it leads to better governing practices and more transparency. These improved governing practices then positively impact a country's level of development.

Résumé Cet article évalue globalement l'impact de l'aide au développement sur la croissance économique. Bien que la plupart des spécialistes (par exemple Easterly, 2001) montrent que l'aide n'a pas d'impact direct sur le développement économique et certains (par exemple, Moyo. 2009), en questionnent même sa valeur, nous devons souligner que ces allégations sont sous-développées et non basées sur des preuves empiriques. Bien que notre modèle d'équation structurelle confirme qu'il n'y a pas d'impact direct de l'aide sur la croissance, nous constatons que l'aide a un effet positif indirect sur le développement économique. L'aide au développement mène à une diminution du niveau de la corruption d'un pays, ce qui a ensuite une influence positive sur la croissance. Les implications de ces résultats sont que l'aide, dans la plupart des cas, est efficace dans le cas où elle conduit à de meilleures pratiques de gouvernance et à une plus grande transparence en général. L'amélioration de ces pratiques de gouvernance a ensuite un impact positif sur le niveau du développement d'un pays.

Keywords:

• Development aid
• Corruption
• Economic growth

1. Introduction

In September 2000, 189 states of the world spoke in one voice, declaring their commitment to eradicate extreme poverty and help better the lives of the approximately 1 billion people living on less than a dollar per day. Documenting this resolution, the Millennium Declaration was signed by 147 heads of state and passed unanimously by the UN General Assembly (World Bank 2010). It seemed that most of the world was ‘on board’ when it came to development. Yet, while policymakers, institutions – and academics for that matter – have maintained their commitment for growth promotion in principle, once solutions to this challenge are proposed, consensus crumbles. At the center of this debate is the issue of aid: does development aid hinder, help, or have no effect on developing countries?

Indeed, the question of whether development aid spurs growth has been a contentious one in both academic and policymaking circles for decades. Whereas early aid ‘optimists’ (for example, Chenery and Strout 1966, White 1992) argued that aid caulks the investment and export gaps that erode growth, more recent aid ‘pessimists’ point out that far from a panacea for the developing world, aid only promotes growth under beneficial economic or political conditions (Burnside and Dollar 2000, Easterly 2001, 2003, 2006). Others have even boldly charged that aid's positive impact on growth is nothing more than a ‘myth’ (Moyo 2009). The raft of empirical tests of the aid–growth relationship have only added further ambiguity, as numerous scholars report no significant impact of aid on growth in either direction (Rajan and Subramanian 2005, Doucouliagos and Paldam 2006, Paldam and Herbertsson 2007). Indeed the aid–growth link has become a proverbial football tossed from one side to another (Easterly 2003).

What, then, is the relationship between aid and growth? Is aid a panacea, placebo, poison, or none of the above? In this paper we argue, in line with Easterly, that the answer to this question can be found not in a mere retesting of the empirical relationship, but in a refined focus on theoretical modeling and empirical specification (Easterly 2003, p. 30). In particular, we posit that while there is no direct relationship between aid and growth, there is a positive and indirect relationship between the variables, one that flows through corruption. Bridging both the aid–corruption and the corruption–growth bodies of literature, we test this tripartite relationship between aid, corruption and growth through a simultaneous equation model and find that as aid increases, it weakens incentives for corruption (Van Rijckeghem and Weder 1997, Tavares 2003). This decreased corruption then subsequently opens the door for economic growth (Mauro 1995, Ades and Di Tella 1996, Gupta et al. 2002, Podobnik et al. 2008). While supporting the aid–growth studies that have included corruption (Burnside and Dollar 2000, Morissey 2001, Kaufmann 2009), our findings challenge their treatment of corruption as a mere control variable to be included in a long list of aid determinants. Rather, corruption acts as a mediating factor – one that is both a dependent variable impacted by aid and an independent variable impacting growth.

To demonstrate this relationship, we proceed as follows. After detailing the highly debated aid–growth link in Section 2, we then retest this relationship through regression analysis in Section 3. Revealing that aid and growth have no direct impact, we then show how aid impacts growth indirectly through corruption. To do so we begin with our theoretical argument, which we construct in Section 4 by fusing two bodies of major scholarship: (1) the aid–corruption literature; and (2) the corruption–growth literature. We then test this theoretical argument empirically through a simultaneous equation model in Section 5, and reveal our conclusions in the final section.

2. The impact of aid on growth

For 30 years, the effect of aid on growth has been the object of interest in academia given both the structure of the world economy, the interests of aid dispensing organisations and the draw of altruism. Despite hundreds of studies there appears to be little consensus as to the effects of aid on various developmental indicators. This holds true for older studies that examined the effect of aid on the finance gap, projects that tried to gauge the affect of aid on economic growth in the aggregate as well as more recent work which looks at the effect of aid on growth under conditionality (Morrissey 2001, Radelet 2006, Doucouliagos and Paldam 2007).

Initially, the Harrod–Domar growth model theorised that aid channelled to lesser developed states could fill the twin gaps of savings and export earnings. This would allow the developing country to buy the necessary imports crucial to climbing the industrial ladder and allow its population to begin long-term savings accumulation, which at the time were seen as necessary conditions for sustained economic growth. This approach eventually evolved into the two-gap model advocated by Chenery and Strout (1966) and White (1992), and was further buttressed by the theoretical success of financial liberalisation as a means to build economic growth as highlighted by McKinnon (1973), Shaw (1973) and the seminal empirical work of King and Levine (1993) (see also Ang 2008a, 2008b for a broad survey of the financing literature).

Despite the success of the finance model, the belief that aid creates economic growth was eventually challenged by economists. Rather than bridge the gaps in domestic savings or financing for export industrial development, foreign aid was found to increase public consumption without producing commensurate savings or financing, to say nothing of aid's role in lining the pockets of recipient country corrupt elites (Boone 1996). The amount of aid given was found to be generally equivalent to target country expenditure (Griffin and Enos 1970, Mosley et al. 1987, White 1992, Boone 1996, Easterly 2003). In particular, Barro and Sala-i-Martin (2004) found that this caused more damage considering the negative effect of public consumption on economic growth.

In the mid-1990s various regression techniques were applied to study the aid–growth nexus. For example, by squaring aid, Hadjmichael et al. (1995), Durbarry et al. (1998) and Lensink and White (2001) found that aid shows a positive effect on growth. Another perspective suggests that there is a non-linear effect in the aid–growth relationship due to diminishing returns to aid (see Lensink and White 2001, Gomanee et al. 2002, Dalgaard et al. 2004). These regression-based studies have featured over 60 papers and thousands of regressions, nearly half of which found small positive effects for aid on growth directly, with the other half finding no statistical significant results. However, these models are not only extremely sensitive to sample size, the significance of the effect of aid on growth diminished to the point of non-significance as the data for more countries became available (Doucouliagos and Paldam 2005).

Finally, in the late 1990s and early 2000s a fourth array of studies – the conditionality literature – emerged. Specifically, there are three separate arguments within the conditionality literature. The first is ‘the good policy model’ suggested at the World Bank, which holds that aid works if the recipient government pursues good policies. This perspective began in 1995, but is best typified by the work of Burnside and Dollar (2000, 2004) and Collier and Dollar (2002). The second is ‘the medicine model’ promoted by the Copenhagen Group, which holds that aid, like medicine, is beneficial in the correct dosage but harmful or ineffective in incorrect dosages (Collier and Hoeffler 2004, Dalgaard et al. 2004, Roodman 2004, Jensen and Paldam 2006). The third category, which can broadly be called ‘institution models’ (Doucouliagos and Paldam, 2006), states that aid works beneficially if the right governmental institutions are in place.

Despite the rather positive undertone of many studies in the conditionality literature, the current consensus is nevertheless that there is no relationship between development aid and growth (Rajan and Subramanian 2005, Doucouliagos and Paldam 2006, Paldam and Herbertsson 2007). Excluding those who are extremely sceptical regarding the effect of aid (see Easterly et al. 2004, Clemens et al. 2004, Easterly 2006), the typical finding from meta-analysis portrays a slight positive linkage or no association between development aid and economic growth. For example, a recent study by Doucouliagos and Paldam (2009), using a meta-analysis covering 68 published studies across the last 30 years, found that the effect of aid on growth estimates scatter considerably and add up to a small positive, but insignificant, effect on growth. Our ensuing regression confirms this small and substantively negligible positive effect. However, before presenting and discussing these results, we will first present our control variables and our modeling procedures.

3. Re-establishing the non-relationship between aid and economic growth

To stand on firm ground, our preliminary model retests the relationship between aid on the right-hand side and economic growth on the left-hand side. Growth is operationalised in the standard manner through the economic growth rate as of 2008 (IMF 2008a). Aid, the independent variable of interest, is operationalised by the amount of aid per capita a country received in 2008 from all DAC countries of the OECD, divided by the total population (IMF 2007). While it is the main goal of this preliminary regression analysis to test the impact of aid on economic growth, this first model controls for six independent variables commonly found in the literature. We expect a country's inflation rate, unemployment rate, amount of debt, as well as a state's involvement in armed conflict to have a negative impact on economic growth.1 In contrast, we assume the amount of foreign direct investment a country receives and its GDP per capita to positively impact growth.

The first two indicators, inflation and unemployment, are expected to hamper growth. Pertaining to inflation, most economists would agree that at high levels, inflation has distortional effects on economic growth. High inflation interferes with foreign and domestic capital and renders the realisation of long investment projects difficult, if not impossible (Goncalves and Salles 2008, Lin and Ye 2009). A similar logic should apply to unemployment. Not only does high unemployment drain a state's resources because governments must disburse increased unemployment benefits or social assistance, but it also prevents a country from reaching a high level of productivity (Muscatelli and Tirelli 2001). Both indicators are measured as the net inflation and unemployment rates respectively and are gathered from the IMF (2008b) and the United Nations Statistical Division (UNSD 2009), respectively. The third macroeconomic indicator – a country's amount of foreign debt – is also expected to be negatively correlated with economic growth. For example, analyses have shown that doubling a country's foreign debt reduces its growth by 0.5 percentage points (Patillo et al. 2002). Debt is measured as the percentage of the GDP that a country owes to foreign lenders. Data for total debt were collected from the World Bank (2005) and divided by the total GDP, gathered from the IMF (2008c).

Following Barro (1991, 1996), we also assume that civil and/or interstate conflict inhibits economic growth. Not only are financial resources diverted into the military complex but, even more importantly, conflict creates an atmosphere of fear and distrust, damages the infrastructure of a country and hinders trade. We measure conflict as an ordinal variable. Countries that did not experience war in 2008 are coded zero, countries with minor wars (25 to 1,000 battle deaths) one and countries facing major conflicts (states where more than 1,000 people are killed due to fighting) two. These data were collected from the UCDP (2008) dataset.

In contrast, we expect foreign direct investment to have a positive impact on economic growth. We assume that financial and technological transfers not only profit recipient firms, but also the economy as a whole (De Mello 1999). FDI is gauged as foreign direct investment inflows as a percentage of GDP. The data were gathered from the UNCTAD (2008) website. The final factor for which we expect a positive relationship is a country's GDP per capita. Richer countries have a better infrastructure, a higher educated workforce and more technological innovations, characteristics which should all foster growth (Demetriades and Hussein 1996). GDP per capita data (calculated as for Purchasing Power Parity [PPP]) were gathered from IMF (2008d).

We use OLS regression analysis to retest the impact of developmental aid on economic growth. Our data yields 110 cases and is organised in a cross-sectional design. In fact, our dataset includes all aid recipient countries, for which data are available. The regression equation for this study is as follows:

This results of the first regression model confirm that the variable aid per capita has no impact on economic growth (Table 1).

Table 1. Regression results for the interactive models.

Economic growth	Beta	Std error	Sig.
Aid per capita	.003	.002	.258
Inflation rate	–.066	.022	.003
Unemployment rate	–.039	.023	.096
Debt as % of GDP	–.409	.362	.260
D_Conflict	–.260	.434	.559
FDI as % of GDP	–6.086	5.484	.261
GDP per capita	.00012	.00004	.003
Constant	2.521	.464	.000
Notes: N = 110. Rsquared: .429.

As shown by most recent works (Rajan and Subramanian 2005, Paldam and Herbertsson 2007), the variable is not statistically significant, nor is its coefficient substantively relevant. Rather, aid per capita has a negligible positive impact on economic growth. The model predicts that per $1.00 increase in per capita aid, growth will increase by 0.002 points, which is so miniscule that it does not impact the economic performance of countries. Of the control variables we find that inflation rates are negatively correlated with economic growth, whereas GDP per capita is positively correlated with economic growth. All the other controls do not seem to impact growth rates. While they all exhibit the predicted signs, their impacts are neither statistically nor substantively relevant.

Having established that there is, in fact, no direct impact between economic aid and economic growth, we now take the literature in a new direction, positing a possible indirect relationship between aid and growth. In particular, we argue that aid directly impacts corruption, which then impacts economic growth. To do so, we begin theoretically, discussing the causal mechanisms that link aid, growth and corruption. Based on the subsequent discussion, we then build a structural equation model, which confirms the indirect effect between aid and growth.

4. The indirect effect of aid on growth: corruption as mediating variable

Rather than having a direct impact on economic growth, we contend that aid may have an indirect effect on economic growth through the mediating variable of corruption, or the ‘misuse of public office for private gain’ (Rose-Ackerman 2008, p. 551). Previous studies have included corruption control and institutions (Burnside and Dollar 2000, Morissey 2001, Kaufmann 2009) in their aid–growth studies; however, they treat corruption as essentially a control variable to be included alongside an array of other factors. Such modeling assumes that corruption is an exogenously given condition that simply feeds independently into growth outcomes. Our study, by contrast, argues that corruption is not merely another factor impacting growth alongside aid, but is actually affected by aid itself. The subsequent level of corruption then drives economic growth. We support this ‘mediator’ assumption by linking two established bodies of literature: first, that aid affects corruption; and second, that corruption impacts aid. This section will explore these two areas of scholarship to better understand the aid–corruption–growth link.

4.1. The link between aid and corruption

A burgeoning literature has underlined the potential relationship between aid and corruption. Interestingly, however, the direction of this relationship (positive or negative) is disputed as policymakers and scholars alike have displayed mixed views on the relationship. Some assert a positive view of aid. The World Bank, for instance, sees aid as a harbinger of ‘good governance’ and ‘corruption control’ (World Bank 2009).2 Aid and aid conditionality are viewed as economic incentives (for example, ‘carrots and sticks’) for governments to alter their corrupt habits (Alesina and Weder 2002, p. 3). Assuming that donor governments are able to monitor recipient governments, they can either allot or increase aid to reward governments for good governance or remove it to punish governments for corrupt acts.3

Scholars such as Van Rijckeghem and Weder (1997) and Tavares (2003) have supported the rosy assertions of these agencies, positing that aid might help the survival of reform-seeking governments by providing them with much needed assistance. Indeed, the construction of better institutions of governance comes with a monetary cost. Development aid, especially aid that is directed toward good governance, may provide proactive governments the opportunity and resources to build better institutions (Knack 2000). Outside of this governance support, aid may assist in other ways. By increasing liquidity, aid can lessen budget constraints and provide higher salaries to civil servants (Fritz and Kolstad 2008). This increase in salary might then reduce incentives for bribe solicitation (Van Rijckeghem and Weder 1997). In this vein, Tavares (2003) finds a negative relationship (increased aid decreases corruption), bolstering support for the claim that aid helps depress corruption and pull countries out of a potentially growth-inhibiting pattern.

However, this positive view has been widely disputed in recent years (see Svensson 2000, Knack 2000, Alesina and Weder 2002, Ali and Isse 2003). Instead of serving to decrease corruption, aid may simply become fodder for corrupt officials (Collier and Dollar 2004, p. 263), who either dip directly into public coffers, or engage in bribery, graft, nepotism, patronage and extortion to pocket a portion of collective goods (Easterly 2006). This is largely a product of a principal–agent problem, in which aid agencies are unable to monitor the use of funds (Milner 2004, p. 5). Others (see Hope and Chikulo 1999, Rimmer 2000) point out that aid is not merely consumed by corruption, but that it actively fuels a higher level of corruption. Outside of simply pocketing funds, corrupt officials may use these funds to widen patron–client relations and patrimonial hierarchy (Hope and Chikulo 1999, Rimmer 2000, p. 123). In the process, the public may be cut out of the equation entirely. In an aid-driven ‘paradox of plenty’ (Karl 1997) aid then may allow government officials to short circuit the process of taxation, which is directly linked to representation and accountability (Knack 2000). When they are less responsive to the population, politicians may be freer to hide and increase their corrupt actions, not only of aid money but all other revenues as well.4

In addition, a moral hazard problem may ensue from aid donations, in which ‘having … access to [these] resources … actually induces riskier … behaviour’ (Brautigam and Knack 2004, p. 263). If donors plan to sever funding when economic or political reforms are achieved, then (self-interested) politicians have few incentives to reach these goals (Brautigam and Knack 2004). Instead, they may allow wasteful corruption and relax ‘credible constraints’ on overspending. Especially in the absence of ‘appropriate contracts’ (for instance, aid conditionalities) (Isopi and Mattesini 2008)5, aid can become a ‘narcotic, fostering addictive behaviour among states that receive it’ (Goldsmith 2001, p. 411).

The aid–corruption spiral may not be restricted to intentionally predatory states, however, but can afflict reform-minded governments as well. First, aid may exacerbate collective action problems that impede improved governance and decreased corruption. Whereas resource-poor governments could have a higher incentive to work together to resolve pressing governance problems, high injections of aid, which may satisfy the immediate cravings of politicians, could make leaders less willing to engage in campaigns of reform. Second, instead of acting as a ‘safety net’, omnipresent aid may become a ‘hammock’ – giving governments little incentive to track their balance rigorously or control corruption. Providing a soft budget constraint in which ‘expenditures bear little relationship to revenues’ (Brautigam and Knack 2004, p. 264), aid could lead to a ‘deceptive mirage’ about how much money is truly needed for investment. This then may only encourage a lackadaisical environment in which corruption of present funds and even dips into future budgets are tolerated (Campos and Pradhan 1996). Bloating the government, aid increase may increase government consumption, ‘creat[ing] opportunities’ for increased corruption (Ali and Isse 2003).

As this section has pointed out, while there appears to be a recent trend toward a positive aid–corruption relationship (aid increases corruption), this puzzle is hardly solved. There is still conflicting evidence on the link between development aid and corruption. On the one hand, scholars and policymakers suggest that aid can act as a financial asset to assist reform-minded governments or an economic carrot and stick that can forcibly undermine corrupt practices. This downward pressure on corruption can ultimately help fuel growth. On the other hand, scholars point out that aid actually assists corruption by creating a principal–agent problem between donors and recipients, an aid-based paradox of plenty, a moral hazard problem of poor budgetary constraints and a collective action problem that impedes group efforts to combat corruption. Whether aid increases corruption or decreases it is of high importance given that corruption may then influence the level of growth a country experiences. The second part of this mediating effect will be explored the subsequent section.

4.2. The corruption and growth link

The impact of aid on corruption is extremely important because corruption has been a known contributor to depressed economic growth. Though early scholarship (see Merton 1957, Leff 1964, Nye 1967, Huntington 1968) argued that corruption helps grease the wheels of slow or inefficient bureaucracies and economies (Kawata 2006),6 the vast majority of recent scholarship (North 1990, Mauro 1995, Ades and Di Tella 1996, Gupta et al. 2002, Podobnik et al. 2008) counters these claims. Corruption not only allows governments to lose revenue as officials are unable to tax money hidden as bribes (Podobnik et al. 2008, p. 409), but dissipates resources that could have been used productively. Bribery in particular makes economic transactions ‘irrational’ when governmental jobs end up going to the highest bribe payer instead of the lowest paying, best-quality provider (Pobodnik et al. 2008, p. 410) and the higher risk of bad contracts means that corruption may also provide disincentives to invest (North 1990). Mauro's (1995) empirical tests bolster the thesis that corruption lowers investment and thus corruption, and further points out that while ‘speed money’ may grease the wheels of business for a single individual, it actually ‘increases red tape for the economy as a whole’ (p. 687). Thus, even viewed in the best light possible, ‘financial schemes … [create] a vicious circle: they compensate for the deficiencies in formal institutions … but by the same token, they undermine formal institutions and retard their effectiveness’ (Ledeneva 2006, p. 161). By undermining economic institutions and the ‘rational’ logic of the market, corruption is likely to hinder economic growth.

5. Testing the indirect link between aid and growth

In the preceding sections we theoretically explored the potential mediating effect of corruption: how aid might act on corruption to then affect growth. We have sought to understand whether aid either increases or decreases corruption, and whether the resulting level of corruption affects growth. We now move to empirically test this potential mediating effect. If aid does, in fact, heighten corruption, ultimately depressing growth, then policymakers may need to rethink the form of their assistance. If, however, aid serves to depress corruption, ultimately raising growth, then the case for increased aid is strengthened. In the following section we lay out the methodology and results of our empirical test of the mediating effect of corruption.

5.1. Methodology

To test whether and how corruption mediates the impact of growth and corruption, we run a recursive simultaneous equation model by including corruption as a mediating variable. Corruption is measured through the World Bank's (2008) Control of Corruption Index. For Equation (1) and (2), in which corruption is the dependent variable, we include the predictors: a country's level of development; its degree of democracy; as well as its oil-producing status aside from developmental aid. We expect democracies will be less corrupt. Not only do free states allow citizens to oust corrupt officials through elections, they also promote accountability as well as an active civil society; both of which should foster less corruption (Cheibub and Limongi 1999). A country's level of democracy is measured by the Freedom House Index, a two dimensional numerical index that measures both political rights and civil liberties.

The socio-economic variable – material wealth – should positively impact the governance practices of states. Richer countries have a more sophisticated bureaucracy and a better infrastructure, which should both facilitate the implementation of government decisions and the functioning of the state apparatus (Sklar, 1987). Finally, we assume that oil-producing countries – those countries for which the oil trade makes up at least a third of their GDP – have higher corruption. The logic behind this argument is that high rents, in the absence of competition, give bureaucrats and those that control these rents a high likelihood of malfeasant behaviour (Ades and Di Tella 1999). Data for this variable were gathered from Fearon and Laitin's (2003) replication dataset.

For Equation (3), we assume that economic growth is a function of the three variables, which we found significant in our previous regression model – a country's inflation rates, its level of development and its degree of corruption. The two equations to be estimated are:

Both equations are over-identified. The variables ‘development aid’ and ‘degree of democracy’ help in the identification of the corruption equation. ‘Inflation’ helps identify the economic growth equations. The variable ‘GDP per capita’ loads on both the corruption and the economic growth equation. Figure 1 graphically illustrates the complete simultaneous equation model. As suggested by various modification indexes, we correlate the independent variables (for example, GDP per capita and democracy). The loadings in the model refer to the standardised coefficients.

Figure 1. The recursive simultaneous equation model.

Our recursive model fits the data well as indicated by various fit indices (see Table 2). Not only is the chi-squared value of 12.7 with eight degrees of freedom non significant (p = 0.123), but other commonly used fit indices also highlight a well fitted model. The Comparative Fit Index is 0.978 and the RMSEA is 0.073, which is not statistically different from 0.05. (The PCLOSE is 0.266).Table 2 presents the unstandardised coefficients, standard errors and significance level of the maximum likelihood equation. First, and most importantly, our study finds that corruption mediates the impact of development aid on economic growth. Mediation is present because both the path from developmental aid to corruption and the path from corruption to economic growth are statistically significant. Since both causal paths are positive, we can establish the following causal pattern: first, developmental aid negatively impacts corruption, which then positively impacts economic growth. Except for the loading of GDP per capita on economic growth, all the other loadings are statistically significant and in the expected direction. Interestingly, the previously significant variable, GDP per capita, is no longer statistically significant once we introduce the variable corruption. In fact, it is not wealth per se that increases growth. Rather, wealthy countries are normally less corrupt, which then bolsters economic growth.

Table 2. The unstandardised loadings of the simultaneous equation model.

Corruption	Beta	Std error	Sig.
Aid per capita	.001	.000	.047
Democracy	–.076	.014	.000
GDP per capita	.0001	.00001	.000
Oil	–.376	.133	.008
Economic growth
Corruption	.769	.365	.035
Inflation	–.091	.012	.000
GDP per capita	.0001	.0001	.147
Notes: Chi-squared = 12.7 with 8 DF, CFI = .978, RMSEA = .073. N = 110.

The finding that developmental aid has an indirect positive impact on economic growth is reassuring. On the one hand, the study provides evidence that the aid international organisations and developed countries offer is not a waste. Rather, it leads governments of recipient countries to be more transparent and less corrupt. This latter point also highlights that donor organisations' and countries' focus on good governance has been effective. Not only is some of the more recent aid directly channelled toward a reform of institutions, but recipient countries also realise that in order to receive aid they have to be transparent in the allocation of services and resources. In a nutshell, this paper argues that aid is beneficial for the development of industrialising countries. While the overall amount is probably too little to directly and significantly impact economic growth, it is sufficient to render countries more accountable, which then directly and positively impacts growth.

6. Conclusion

This study has sought to provide a more thorough investigation of the relationship between development aid and economic growth. While an investigatory regression revealed no direct impact of aid on growth, the subsequent simultaneous equation model showed a statistically significant mediating effect of corruption between these two variables. Aid decreases corruption, which subsequently increases economic growth.

These results bring important evidence into a burning debate on aid and growth that pits aid optimists and supporters against sceptics. In particular, it provides support for the view that aid can positively affect a country's economic growth. Yet, our results simultaneously cast doubt on the nature of this relationship, seeing as the aid–growth link is not direct, but runs indirectly through corruption.

Aid works through the lowering of corruption to influence economic growth, a finding that simultaneously informs the two bodies of literatures from which it derived the mediating relationship hypothesis (the aid–corruption and corruption–growth literatures). In terms of the aid–corruption literature, our study throws into question the growing consensus that aid promotes or allows massive corruption through a principal–agent problem, an aid-based paradox of plenty, a moral hazard problem or poor budgetary constraints (see Hope and Chikulo 1999, Knack 2000, Rimmer 2000, Svensson 2000, Goldsmith 2001, Alesina and Weder 2002, Ali and Isse 2003, Brautigam and Knack 2004). Rather, we find that aid can erode corruption. While future studies should investigate the exact mechanisms of this relationship, we suggest that aid could theoretically help increase economic incentives for good governance, bolster reform-seeking governments and reduce bribe solicitation and the wasting of funds by providing civil servants with higher salaries (see van Rijckeghem and Weder 1997, Tavares 2003). On the corruption–growth link, this study supports the well-documented idea that better governance and less corruption mean less financial scheming, bribery and misuse of aid and domestic funds, ultimately boosting a country's opportunity and desire to invest in a growth-producing manner. Given this relationship among aid, corruption and growth, our study also suggests that corruption should be treated in future research not simply as one more control to be included in a list of growth determinants, but as a mediating variable that is influenced by aid and influences growth.

In sum, our results would appear to suggest that donor countries should boost economic donations to developing countries. While on the surface an innocuous and beneficial proposal, this advice comes with at least two caveats. The first is that not all aid may have an equivalent impact on corruption or economic growth (Mavrotas 2003). Aid that is made conditional upon good governance or that is explicitly targeted toward improving institutions may have a stronger impact on corruption control, and thus the creation of growth. In addition, the degree to which aid impacts corruption and thus growth may depend in part on the ‘modality of aid,’ or how the aid is given to the recipient country (Fritz and Kolstad 2008). Scholars such as Fritz and Kolstad (2008) have suggested that aid given as ‘budget support’ (aid transferred directly into the partner government's national treasury) may be riskier in allowing for corruption than ‘project support’ (aid earmarked to particular projects). While it is beyond the scope of this paper, future scholarship should disaggregate development aid so as to see how various aid types and modalities affect the aid–corruption–growth nexus.

Biographical Notes

Daniel Stockemer is an Assistant Professor in the School of Political Studies at the University of Ottawa. His research interests are comparative politics and political methodology. He has published in International Politics, Comparative European Politics and the Social Science Journal, among others.

Bernadette La Montagne is a PhD candidate in the Political Science Department at the University of Connecticut. Bernadette`s main research interests are welfare state policies. She has published in Perspectives on European Politics and Society and Contemporary Politics, among others.

Jason Charrette is a PhD student in the Political Science Department at the University of Connecticut. He is interested in international relations theory, political economy and governance.

Notes

The variables inflation and debt rates are operationalised by the net.

The World Bank has recently identified governance and anticorruption issues as ‘critical to improving development outcomes, such as better delivery of services in health, education, roads, water, and electricity’ (World Bank 2009).

The United States' Millennium Challenge Corporation perhaps lays this out most clearly, asserting that it ‘forms partnerships with some of the world's poorest countries, but only those committed to good governance, economic freedom, and investments in their citizens’ (MCC 2010).

Some (see Ear 2007) have even posited that aid undermines not only civil society, but also the domestic economy in general. The only industry left standing is thought to be the aid industry. By creating a situation rife with poverty, low wages, and high inflation, the ‘aid curse’ or ‘Dutch Disease’ creates economic incentives for those numerous civil servants who flock towards the aid sector to dip into funds.

Isopi and Mattesini (2008) argue that not only does aid always fail to reduce endemic corruption, but it can even create its own aid-related corruption if appropriate contracts are not in place.

These authors assert that corruption quickens monetary transactions by allowing politicians to circumvent poorly functioning government and bribes provide incentives for civil servants to work harder. Additionally, the ‘black market’ created by corruption deals would further promote the survival and growth of more efficient firms by allowing richer resourceful firms to outbid poorer weaker companies.