A comparative analysis of the effect of aid and microfinance on growth1 Maricruz Lacalle-Calderón Applied Economics Department Faculty of Economics and Business Universidad Autónoma de Madrid Avda. Francisco Tomás y Valiente 5 28049 Madrid (Spain) Telephone: +34 914972895 Email: maicu.lacalle@uam.es Coro Chasco Applied Economics Universidad Autónoma de Madrid Facultad de Ciencias Económicas y Empresariales, Módulo 14 Avda. Francisco Tomás y Valiente 5 28049 Madrid (Spain) Telephone: +34 914975266 Email: coro.chasco@uam.es Javier Alfonso-Gil Applied Economics Department Faculty of Economics and Business Universidad Autónoma de Madrid Avda. Francisco Tomás y Valiente 5 28049 Madrid (Spain) Telephone: +34 914973434 Email: javier.alfonso@uam.es Isabel Neira Econometrics Department Faculty of Economics and Business Universidad de Santiago de Compostela Av.Xoan XXIII s/n 15704 Santiago de Compostela (Spain) Telephone: +34 981 566287 Email: isabel.neira@usc.es Maricruz Lacalle‐Calderón and Coro Chasco acknowledge financial support from the 8th UAM‐Santander Bank Project. Coro Chasco also thanks the Spanish Ministry of Education and Science (Grant No. ECO2012‐36032‐C03‐ 01). 1 1 A comparative analysis of the effect of aid and microfinance on growth Abstract The objective of this paper is to study the effect of official development aid (ODA) and microfinance (MF) on economic growth and compare the results. To achieve this objective, we identify and analyze the transmission mechanisms from ODA and microfinance to growth. We then study the dynamics of ODA and microfinance on growth, using an unbalanced panel of 67 countries for the period 2001-2011. After country and time effects are accounted for in a dynamic panel data model, our results show that microfinance has a positive and statistically significant effect on economic growth through private investment, while ODA has no effect on growth. This conclusion suggests that deploying resources in the microfinance industry will foster economic development in poor countries. Key Words: Microfinance; Official Development Aid; Economic Growth; Panel Data. JEL classification codes: O20; O40; G23; C23 2 I. Introduction Foreign aid in the form of Official Development Aid (hereafter ODA) reached a record amount of 180.1 billion USD in 2010 from all donor countries, totaling 4.2 trillion USD over the past 50 years (OECD 2013a). There is still substantial controversy, however, concerning ODA’s expected impact on growth in receiving countries (Easterly 2008 and Gibson et al. 2009). In recent decades, microfinance (MF) has emerged as a complementary approach to financing development. MF is a bottom-up financial tool developed in a businesslike way and with a sustainable focus. The gross loan portfolio of the worldwide microfinance industry reached 64.769 billion USD in 2011 (MIX Market 2011). As in the case of ODA, there is some uncertainty as to the impact of microcredit on growth (Hulme 2007, Dichter 2007, Bateman 2010, Roodman 2012 and Duflo et al. 2013). The objective of this paper is to study the effect of ODA and MF on economic growth and compare the results. There is an extensive empirical literature on the effect of aidi on growth, but this is not the case for microfinance. Although both ODA and MF are funding tools for development and have very similar objectives, their characteristics and dynamics are not the same. The channel by which ODA affects growth may, therefore, be different than that of microfinance. Empirical analysis must explain and address both transmission mechanisms. We therefore test the indirect impact of ODA on growth through a two-step panel data model following applications similar to those used in Gomanee et al. (2005). We repeat this procedure for microfinance to ensure proper comparison of the funding tools. To compare the effects of these two instruments on economic growth, we conduct the analysis for an unbalanced panel of 67 countries, all of which are ODA recipients and have MF activity (and can therefore provide data on both tools). The results show a positive and statistically significant association between MF and economic growth through private investment, but not between ODA and growth. This paper contributes to the literature by: a) exploring the macroeconomic effects of microfinance, b) focusing on the comparative effects of aid and microfinance on growth, and c) determining whether the transmission mechanisms from these tools to growth are the same. The rest of the paper is organized as follows. Section 2 analyzes theoretically how and why ODA may impact economic growth (i.e., transmission mechanisms) and summarizes the most important literature on the effectiveness of ODA. Section 3 analyzes how and why microfinance may affect growth. In Section 4, a panel data model is used to explore the dynamics of ODA and MF on economic growth. We also perform a battery of sensitivity checks to confirm the robustness of our results with different specifications and estimation strategies. Section 5 presents the results 3 derived from comparing the impacts of ODA and MF on economic growth. Section 6 presents the conclusions. II. Foreign aid and economic growth According to the definition provided by the OECD’s Development Assistance Committee (DAC), ODA represents those flows to recipient countries and multilateral institutions provided by official agencies (including state and local governments or their executive agencies) and administered fundamentally for the promotion of the economic development and welfare of developing countries. To constitute ODA, these flows must be concessional in character and include a grant element of at least 25% (OECD 2013b). ODA is considered a top-down financing tool because it flows from governments or international agencies to governments. It is basically a one-wayii financial flow that operates in large-scale development projects to address the shortages in capital goods in poor countries, where the insufficiency of domestic savings leads to weak investmentsiii. In economic theory, several authors have identified capital formation shortages as a significant impediment to development, arguing that aid could be a significant determinant for these investments (Lewis 1954, Burnside and Dollar 2000, Hansen and Tarp 2001, among others). The basic transmission mechanisms from ODA to growth were presented by Chenery and Strout (1966) in their two-gap model. First, aid was needed to close the savings-investment gap, known as the internal gap. Second, they also argued that aid was needed to fill the external gap, that is, the difference between imports and exports. Poor countries not only lack sufficient internal resources but often have low and volatile export earnings. They thus lack the resources to finance importation of the capital goods, technology, and intermediate goods (primarily fuel) needed for growth. Aid, in the form of hard currency, could also fill the foreign exchange gap (Gomanee et al. 2005). Finally, since ODA flows fundamentally to governments, it could encourage public spending and consumption and thus affect growth negatively. Several authors have argued that, rather than fostering growth-promoting investment or improving human development indicators, large free aid flows increase the size of government, resulting in excessive and unsustainable government consumption and public deficit (Boone 1996, Burnside and Dollar 2000, Moss et al. 2008). Over the past 60 years, the question of whether aid has a macroeconomic impact on growth has been studied from a variety of methodological approaches. We can define two main but opposing positions on the effectiveness of aid, although the evidence is still ambiguous and the debate continues (Easterly 2008 and Sachs 2011). Some studies have obtained positive results for ODA’s influence on growth (Levy 1988), while others show negative results (Mosley 1980, Mosley 4 et al. 1987, Boone 1996). In the new century, Burnside and Dollar (2000) found that aid works in “good policy environments”, an argument that has influenced the World Bank’s view that aid boosts economic growth, reduces poverty, and improves social indicators in good policy environments (Easterly 2003). Hansen and Tarp (2001), Collier and Dollar (2002), and Sachs (2004) obtain similar results supporting the effectiveness of aid. Instead, Easterly et al. (2004), Lensink and White (2001), Banerjee et al. (2006), and Rajan and Subramanian (2008) determine that aid does not affect growth at all. Moreover, Djankov et al. (2006 and 2008) finds that foreign aid has a negative impact on democracy and economic growth in developing countries. It must be said that these studies have significant problems, such as the erratic nature of the concept of aid, poor data quality, the low ratio of aid to GDP in most recipient countries, the problem of endogeneity, and the use of weak instruments in econometric models (Tarp 2006). Recently, two meta-analyses have been performed with contrary results. Doucouliagos and Paldam (2011) find that aid is ineffective in promoting growth, while Mekasha and Tarp (2013) find that the effect of aid on growth is positive and statistically significant. Based on an extensive review of the literature, one could conclude that, although the expected impact of aid on growth is dubious, the entire development community, from Sachs to Easterly, believes foreign aid is needed to stimulate progress. ODA should be rethought (Easterly 2008) but not rejected. Indeed, foreign aid has been effective where projects have had narrow and monitorable goals that meet the needs of the poor directly (Schultz 2004, Kremer and Miguel 2007, Banerjee and He 2008, Banerjee and Duflo 2011, Karlan and Appel 2011). An appropriate framework to study empirically whether aid has any impact on growth should thus consider explicitly the three transmission mechanisms shown above –investment, imports, and government consumption– treating the data and econometric models properly (Chenery and Strout 1966 and Gomanee et al. 2005). The following sections of this paper will perform such a study. III. Microfinance and economic growth Microfinance is a recent bottom-up approach to financing development that focuses mainly on the individual. It was born in response to the lack of access to financial services experienced by millions of people worldwide who are excluded from the formal financial system (Helms 2006). The World Bank estimates that 2.7 billion people globally have no access to formal financial services (Chaia et al. 2009 and World Bank 2011) and must rely on informal financial services that may be more costly and less reliable (CGAP 2010). 5 Microfinance offers small-scale loans, savings accounts, insurance, housing loans, and other financial services to the poor (CGAP 2009), providing the possibility of overcoming these constraints. It helps the poor to improve their financial security, allows them to take advantage of new business opportunities, and enables many to expand and diversify their economic activities and increase their incomes (Robinson 2001 and Armendariz and Morduch 2010). Further, the potential of MF not only opens new possibilities to those excluded but provides them with the social network and institutional capital created in the process of providing MF (Matin et al. 2007). According to the most recent report of the Microcredit Summit Campaign (Reed 2013), microcredit increased exponentially from 1997 to 2010, from 7.6 million poorest clientsiv receiving microcredit at the end of 1997 to 137 million at the end of 2010. In 2010, the total numberv of clients reached by the 3652 microfinance institutions reporting to the Summit was 205 million. According to data from the MIX Market (2011), the gross loan portfolio of the 1115 worldwide microfinance institutions reporting data to this institution totaled 64.769 billion USD. Based on all of these characteristics, how might microfinance affect growth? We begin by recognizing that economic activity or economic growth is impossible without financing (Hermes and Meesters 2011 and Roodman 2012). By analogy the argument by Schumpeter (1934) that financial intermediation spurs growth by identifying and funding productive investments, microfinance funds productive microenterprises for the poor through funding of investment in physical capital (Okurut et al. 2005). Microfinance is not limited to small scale loans. It also includes access to saving accounts for the poor and unbanked, mobilizing savings and channeling these savings into microentrepreneurs’ productive investments. Traditionally, private banks thought poor people could not save, but several case studies have reported this is not true. Poor people can and do save (Robinson 2001). Microfinance institutions thus help to achieve financial deepening. They fulfill an unmet need for financial intermediation by receiving savings from poor people and using those savings to fund productive microbusiness that would otherwise be unfeasible. Without microfinance services, thousands of unbankable entrepreneurs could not invest and create or maintain labor, income, and consumption. Khandker (2005) and Mahjaeen (2008) report that microfinance appears to play a very significant role in increasing income and consumption. Based on these arguments, the transmission mechanism from microfinance to growth may occur primarily through productive private investment and secondarily through private consumption (Okurut et al. 2005, Khandker 2005 and Mahjaeen 2008). Caution should be exercised, however, in accepting this conclusion, since microfinance is still an immature and unproven tool in some areas (Dichter 2007). Cases of over-indebtedness and 6 repayment problems with microcredit in India in 2010—and previously in other countries, such as Bolivia, Bosnia-Herzegovina, and Morocco—have proven that microcredit can also be a debt trap (Maes and Reed 2012). Access to financial services may improve the well-being of many poor people, but the process is not automatic (Roodman 2012). After thorough research, Roodman (2012) finds little evidence that the microfinance movement has lived up to its claims of achieving development or reducing poverty in the last thirty years. In recent decades, several studies have reported that well-managed microfinance institutions have been successful in smoothing consumption, creating jobs and income, and improving women’s empowerment and education (Pitt and Khandker 1998, Khandker 2005, Khandker et al. 2013, Karlan and Ziman 2010, Karlan and Apple 2011, among others). Microfinance increases freedom, since it gives the poor a new way to navigate tricky financial currents (Sen 1999). Moreover, five randomized control trials (RCTs) published recently confirm some positive microeconomic outcomes of microfinance (Crépon et al. 2011, Attanasio et al. 2011, Augsburg et al. 2012, Angelucci et al. 2013 and Duflo et al. 2013).vi At the macro level, some recently published papers study how microfinance affects macroeconomic activity. Among them, Buera et al. (2012) and Ahlin and Jiang (2008) propose two theoretical models to study the potential of MF to generate a gradual process of development (the former) and its potential to create a redistributive effect on the overall economy in a country (the latter). Imai et al. (2012) study the impact of MF on national poverty levels. Finally, Ahlin et al. (2011) and Hermes and Meesters (2011) investigate the relationship between macroeconomic conditions and the performance of microfinance institutions. Still, almost no empirical research studies the macroeconomic effect of MF on economic growth. To our knowledge, Sodokin and Donou-Aeonsou (2010) is the only empirical study. It measures the impact of the joint effort between banks and MF to promote economic growth in the West African Monetary Union region. To sum up, the transmission mechanisms from microfinance to growth may occur primarily through productive private investment and secondarily through private consumption, since microcredit activity is basically credit to create or maintain microenterprises, as well as to increase or maintain income and consumption. IV. Empirical analysis Our empirical analysis focuses on the effect of ODA and MF on growth, taking into account the transmission mechanisms explained in the previous sections. ODA may affect growth through public investment, imports, and government consumption spending. The effect of ODA on growth is thus considered to be indirect. Exactly the same dynamism occurs with MF, which may affect 7 growth indirectly via private investment and/or private consumption. All of these indirect impacts of ODA and microfinance on growth will be tested in the following pages. Data The dataset used in this paper to analyze the relationship of ODA and MF to economic growth was obtained by combining a variety of databases, as shown in Table 1. The variable records were matched by country and year. After the exclusion of some records that could not be matched, the final dataset is composed of an unbalanced panel of 67 countriesvii for the period 2001-2011. The availability of data for MF limits the analysis to this period. ------- TABLE 1 ------Main model The base specification is a dynamic panel data model of the logarithm of real GDP PPP (gdpit), as in Sodokin and Donou-Adonsou (2010). The model is specified as follows: gdpit   gdpi,t 1   pit  tit  cit  i  t  vit (1) Initial GDP growth (gdpi,t1) is used to control for persistence of economic development over time. The measures of the principal variables, ODA and MF, are designated by p. Two GDP panel data models will thus be specified from Equation (1), depending on the main variable considered: total net ODA or microcredit gross loan portfolio. We incorporate two vectors of other groups of variables. The vector of transmission variables (t) includes private and public investment, private and public consumption, and total imports. Following Gomanee et al. (2005), Acemoglu (2009), and Compton et al. (2011), the vector of the control variables (c) considers a wide array of socioeconomic variables conditioning economic growth (e.g., labor force, malnutrition prevalence, credits, inflation, interest, and real exchange rates), as well as other political indicators (democracy, role of law, and control of corruption). All variables are specified in log form, except the rule of law and control of corruption indexes and the inflation and interest rate variables. The term  i captures the unobservable time-invariant country heterogeneity, since we allow for the presence of country effects;  t represents time effects; α, β, , and  are parameters to be estimated; and vit is the error term, capturing all other omitted factors, with E[vit]=0 for all i and t. We estimate dynamic panel data model (1) using the consistent General Method of Moments (GMM) estimator proposed by Arellano and Bond (1991), because this method allows for unobserved country-specific effects, measurement errors, and endogeneity problems, not only of the lagged dependent variable but also of any other regressor (Arndt et al. 2010). Causality from ODA 8 and MF to economic growth could run both ways. MF could affect economic growth, but economic growth could also have a positive impact on microfinance institutions’ performance (Ahlin et al. 2011 and Hermes and Meesters 2011). We must test for the same effect in the case of ODA, as in Asiedu and Nandwa (2007) and Hansen and Tarp (2001). Thus, neither ODA nor MF can be considered a priori as exogenous to economic growth in Equation (1). Arellano and Bond’s “difference GMM” estimator uses lagged levels of first difference of variables as instruments to treat endogeneity accurately in both the lagged dependent variable and the other explanatory variables. This method identifies how many lags of the dependent variable, the predetermined variables, and the endogenous variables are valid instruments and how to combine these lagged levels with first differences of the strictly exogenous variables into a potentially large instrument matrix. We employ the so-called “two-step” difference GMM estimator, which allows for heteroskedasticity in the error terms. Additionally, the Windmeijer biascorrected robust VCE has been computed, since non-robust results for inference on the coefficients tend to be severely biased downward (see Arellano and Bond 1991 and Windmeijer 2005 for details). Hence, if we time difference Equation (1), we obtain the following specification: gdpit  gdpi,t 1  pit  tit  cit   t  vit (2) where the fixed country effectsi are removed by time differencing. This estimator requires that there be no autocorrelation in the idiosyncratic errors (vit). When the idiosyncratic errors are independently and identically distributed (i.i.d.), the first-differenced errors (vit) are first-order serially correlated. Nevertheless, serial correlation in the first-differenced errors at an order higher than 1 is not allowed, since this would imply that the moment conditions used in the model are not valid. For this reason, we have calculated the Arellano–Bond test for firstand second-order autocorrelation in the first-differenced errors of the specified models. We have also computed the Sargan test derived by Arellano and Bond (1991), which contrasts the null hypothesis that the overidentifying restrictions are valid. Rejecting this null hypothesis would indicate that we need to reconsider either our model or our instruments. Estimation of the transmission mechanisms for ODA and MF As discussed above, three potential transmission variables are considered for aid: public investment (gcfpub), government consumption (cpubl), and imports (impor). For MF, these variables are private investment (gcfpriv) and private consumption (cpriv). In a first step, we test whether these variables are indeed transmission mechanisms for the effect of ODA/MF on economic growth. We do this simply to determine whether the coefficients of the ODA/MF variable are significant in a 9 dynamic panel data regression of ODA/MF–and other explanatory variables–for each particular transmission variable. These models are estimated using a two-step GMM, as in Equation (2). The formal models are the following: a) Transmission mechanisms for ODA: gcfpubit  gcfpubi,t 1  oda it   1credit it   2tot i,t 1  t  vit  cpublit  cpubli,t 1  oda it   1taxit   2densit  t  vit impor  impor  oda   expor   tot    inf    v it i,t 1 it 1 it 2 i,t 1 3 i,t 1 t it  (3) To account for the dependence of public investment (gcfpub), we include its own one-period lag, bank private credit (credit), one-period lagged net barter terms of trade index (tot), as well as ODA. Government consumption spending (cpubl) is explained by its own time lag and by both domestic and foreign sources of government revenue, such as tax revenue (tax), population density (dens), and ODA. Finally, imports (impor) are considered to be determined by their own time lag, one-period lagged inflation rate (inf), one-period lagged net barter terms of trade index (tot), exports (expor), and ODA. b) Transmission mechanisms for MF: gcfpriv it  gcfpriv i,t 1  mfit   1credit it   2 infi,t 1   3demoit  t  v it  cpriv it  cpriv i,t 1  mfit   1credit it   2interi,t 1   3infi,t 1  t  vit (4) Private investment (gcfpriv) is explained by its own time lag, bank private credits (credit), one-period lagged inflation rate (inf), a democracy index (demo), and MF. Private expenditure (cpriv) is determined by its own time lag, bank private credits (credit), one-period lagged interest rate (inter), one-period lagged inflation rate (inf), and MF. V. Econometric results: the impact of ODA and MF on economic growth The estimation results of the models in Equations (3) and (4) are presented in Table 2. Recall that our aim is to determine whether ODA and MF are significant determinants of cross-country variations in the level of the transmission variables under consideration. First, we perform the Durbin and Wu-Hausman tests to investigate whether ODA and MF are exogenous (null hypothesis) in each of the models in which they appear. This involves comparing the results of OLS and IV regressions using the Sargan test for validity of instruments.viii As shown in Table 2, these tests clearly fail to reject the null hypothesis that the exogenous regressors (ODA/MF) and error terms are uncorrelated. Consequently, our sample provides no evidence of the need to use instruments for either ODA or MF. 10 The Arellano-Bond serial autocorrelation tests also obtain the expected outcomes in all cases: they are always significant for a first order (t–1) but no longer significant for the second order. The Sargan specification test always accepts the null hypothesis of validity of the overidentifying restrictions, indicating good specification of both model and instruments. The results displayed in Table 2 show coefficient estimates with the expected signs, but the results are not always statistically significant. In fact, the first important result is that we cannot obtain evidence of a relevant effect of ODA on either of its potential transmission variables, while MF is found to be highly significant in explaining both private investment and private consumption. On average, an increase of 1 percentage point in MF raises private investment and private expenditure by about 0.09% and 0.06%, respectively. ------- TABLE 2 ------The results suggest that there are no mediating mechanisms for ODA to impact on economic growth, while private investment and consumption are significant transmission mechanisms for MF Therefore in this last case, it is necessary to consider the ‘double-counting’ problem to avoid MF being counted more than once. In a second step, we contrast the role of the transmission variables in a GDP growth equation in which ODA and MF have not yet been considered. Although aid is not found to be a significant determinant of public investment (gcfpubl), government consumption (cpubl), and imports (impor) in our sample, we include these key variables in the growth equation. The results are presented in the second column of Table 3. Of the full set of explanatory variables included in Table 1, only private investment (gcfpriv), imports (impor), the net barter terms of trade index (tot), and the democracy index (demo), together with the one-period lag of the dependent variable, are found to be statistically significant for GDP growth and to show the expected signs. Thus, neither the variables (private/public) consumption nor public investment can be considered relevant determinants of economic growth for our country sample, although imports are. Steps 1 and 2 reach the following conclusions: 1) ODA has no significant transmission mediations on income growth; 2) MF has two significant transmission mechanisms that impact economic growth, private investment, and private consumption; 3) of these last two significant mediations, only private investment has been found to have a relevant impact on growth. ------- TABLE 3 ------In order to quantify the impact of ODA and MF on income growth, we thus include these variables in the previously estimated growth model. In the third column of Table 3, we contrast to what extent ODA has –at least– a direct impact on GDP, what must also be rejected due to the nonsignificant level of its corresponding coefficient. In the fourth column, we have estimated Equation 11 (2) with MF and the original private investment variable (gcfpriv). The coefficient of MF is low (0.0032) and only significant at the 80% confidence level. Hence it is not possible to prove a direct effect of MF on GDP growth. Nevertheless, since MF does affect private investment, which, in turn, has a relevant impact on growth, we are not taking into account the aforementioned doublecounting problem. Specifically, the total effect of MF in growth is spread out across the coefficients on the transmission variable (gcfpriv) and MF, and the net effect of MF cannot be disentangled. Double counting problems can be solved as in Gomanee et al. (2005), who recommend the use of the generated residual variable procedure proposed by Pagan (1984) to build the part of each transmission variable that is not attributed to ODA or MF. This method consists of constructing a new regressor for each transmission variable from the residuals of a supplementary bivariate regression of the original transmission variable on ODA or MF, as is appropriate. This method captures transmission from ODA/MF to their corresponding transmission variables. It also enables us to identify that portion of the effect on growth produced by the relevant transmission mechanism that is not directly due to ODA/MF, avoiding problems of double counting and omitted variables. In our case, since only MF has a significant transmission mechanism to growth through private investment, we generate a residual variable for this mediator from a bivariate regression of private investment on MF (gcfprires). This new variable corresponds to the portion of private investment not explained by MF.ix Inclusion of both regressors (gcfprires–instead of the original gcfpriv– and MF) in Equation (2), as well as the other significant explanatory variables, leads to the results presented in the fifth column of Table 3. Both the private investment generated and the MF variables show very significant estimators in this equation, as do the other regressors. This result demonstrates that the MF coefficient in a regression including the original investment term will be an underestimate of the true effect of MF on growth. Namely, there is a significant transmission mechanism from MF to economic growth via private investment. To test for robustness, a dynamic panel data model is estimated for ODA and MF as the only explanatory variables (with the one-period lagged dependent variable) of growth. Again, only MF obtains a highly significant estimator for this sample, while ODA remains unconnected to growth. We conclude this section by affirming that ODA is not found to be a significant determinant of public investment, public consumption, or imports, since none of the coefficients of these three variables is statistically significant (see Figure 1). Although the import variable is a relevant determinant of economic growth, ODA has no effect on economic growth since it does not have any previous effect on imports. This result differs from that of Gomanee et al. (2005), who find the transmission mechanism working from aid to investment, and from investment to growth. ------- FIGURE 1 ------12 The results for microfinance (columns 4 to 5 of Table 2) provide strong evidence that microfinance is associated with its transmission mechanisms. Microfinance is a significant determinant of private investment and private consumption. As we see in the fourth column of Table 2, all coefficients are significant and have the expected sign. Private investment depends basically on access to financial resources. Following Okurut et al. (2005), credit may increase production capacities through funding of investment on capital. Financial resources come primarily from the banking sector in developed countries, but in poor and developing countries (like those in our sample), financial resources for financing private investment come from both formal and informal financial intermediaries. The results in Table 2 show the importance of these two variables (credit and MF). The microcredit gross loan portfolio coefficient is positive and highly significant. Credit is also positive and significant. This means that microfinance increases private investment in the countries in the sample. As one might expect, inflation affects private investment negatively. Its coefficient is negative and significant. Finally, private investment may occur in a stable and democratic environment. As to private consumption (column 5, Table 2), again, microfinance and credit has positive and statistically significant coefficients. This result is similar to that of Khandker (2005) and Mahjabeen (2008), who find that microfinance is highly significant in increasing household consumption. The other coefficients in this column are as expected. An increase in the inflation rate or in the interest rate reduces private consumption. All of these estimation results are as expected and fit economic and microfinance theory quite well. Although MF is a relevant determinant of both private investment and private consumption, as mentioned above, only the former is found to be statistically significant for GDP growth (see Figure 1). Again, our primary and most important result is that the effect of ODA on growth is not statistically significant, while microfinance presents a positive and statistically significant association with economic growth (Table 3). This finding agrees with Buera et al. (2012) and Ahlin and Jiang (2008), who also detect positive effects of MF on economic activity, as well as with the most recent aid-growth literature, which does not find a significant effect of aid on growth (Mosley et al. 1987, Boone 1996, Djankov et al. 2006 and 2008, Rajan and Subramanian 2008 and Doucouliagos and Paldam 2011). VI. Conclusions The impact of ODA on growth has been studied for a long time, but there is still substantial controversy concerning its effects on growth. Microfinance, in contrast, is a newer tool, and almost no empirical studies have been performed to test its impact on growth. Our paper has studied and 13 compared the effect of these two tools on economic growth and finds that there may be different transmission mechanisms from these tools to growth. ODA would affect growth indirectly through public investment, government consumption, and imports, whereas microfinance would impact growth through private investment and private consumption. We performed an empirical analysis to test these hypotheses completely using dynamic panel data models of 67 countries for the period 2001-2011. The results of the empirical analysis show that ODA has no effect on any of its potential transmission variables: public investment, public consumption, and imports. There are no mediating mechanisms for ODA to impact growth. In contrast, microfinance shows two significant transmission mechanisms to impact economic growth: private investment and private consumption, and, of these two significant mediations, only private investment has been found to have a relevant impact on growth. Therefore, ODA has no impact on growth, and microfinance has a significant impact on growth, but only through private investment. Based on the empirical results obtained, the policy recommendation for the international community is to allocate resources to the microfinance industry in order to provide more options for the economically active poor and to enable them to choose their own path. The terms Aid and ODA will be used interchangeably in this paper. ODA flows include grants and concessional loans whose grant component is at least 25% (OECD 2013). The grant component of ODA has increased over the years. At present, almost all aid flows are grants. The average repayment rate of ODA (the difference between gross and net ODA) from 1960 to 2010 has been only 9.42%. iii Investment should be understood not only as fixed capital but also as the improvement of human skills and institutional change. iv According to the Summit Report, “poorest clients” are those people living on less than $1.25 a day, adjusted for PPP. v All microcredit clients, not only the poorest. vi These RCTs obtain mixed results but prove that microfinance does not cause harm. Duflo et al. (2013) concludes that some favorable impacts result from the poor having direct access to credit but that these impacts are not especially strong. The other four RCTs (Crépon et al. 2011, Attanasio et al. 2011, Augsburg et.al. 2012, and Angelucci et al. 2013) obtain more positive results on decrease in expenditure on “temptation goods,” growth of self-employment activities, and increase in business revenues and profits. vii The sample comprises all countries that are ODA recipients and that have an MF sector, that is, all countries for which there are data on ODA and microfinance. Other countries for which there is no data for ODA, microfinance, or even GDP have been eliminated. Zimbabwe, Democratic Republic of the Congo, Lao P.D.R., and Micronesia Fed. State have been excluded because their data is unreliable. Kosovo and Thailand have also been excluded because they have not received ODA in net terms in last five years. Any other exclusion is due to missing observations on microfinance data. The base sample is composed of 67 countries (see Appendix 1). viii ODA has been instrumented by its own time lag, rule of law (rol), and malnutrition prevalence (malnut) as a proxy for poverty, plus the purely exogenous variables in each regression. MF has been instrumented by its own time lag plus private credits, as in Sodokin and Donou-Adonsou (2010), and the purely exogenous variables in each regression. The determination coefficient values for the regressions of ODA/MF on their corresponding instruments are all above 0.87 and 0.92, respectively. ix Although ODA is not statistically significant in its corresponding first-step equations, we have also generated new residual variables for ODA’s potential transmission mechanisms: public investment (gcfpures), government expenditure (cpubres), and imports (impores). Nonetheless, as stated above, only import is a significant regressor in the growth equation, although it was not found to be relevant as a mediator for ODA. The generated residual variables have been i ii 14 estimated from the following bivariate regressions: variable gfpures from gfpubl=13.85+0.38·oda, R2=0.04; variable cpubres from cpubl=17.70+0.24·oda, R2=0.02; variable impores from import=19.68+0.20·oda, R2=0.02; variable gfprires from gfpriv=18.00+0.25·mf, R2=0.10; and variable cprires from cpriv=20.34+0.22·mf, R2=0.11. 15 References  Acemoglu, D., S. Johnson, J.A. Robinson and P. Yared (2008) "Income and Democracy" American Economic Review, 98(3): 808-42.  Ahlin, C. and N. Jiang (2008): “Can micro-credit bring development?” Journal of Development Economics, 86, 1 – 21.  Ahlin, C., J. Lin and M. Maio (2011) “Where does microfinance flourish? Microfinance institution performance in macroeconomic context”, Journal of Development Economics, 95(2), 105–120.  Angelucci, M., D. Karlan and J. Zinman (2013) “Win some lose some? Evidence from a randomized microcredit program placement experiment by Compartamos Banco”. NBER Working Paper No. 19119, Cambridge, M.A.  Arellano, M. and S. Bond (1991) “Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application of Employment Equations” Review of Economic Studies, 58 (2), 277-97.  Armendáriz, B. and J. Morduch (2010) The economics of Microfinance, 2nd Edition. Cambridge: The MIT Press.  Arndt, C., S. Jonesand and F. Tarp (2010) “Aid, Growth and Development: Have we come full circle?” Journal of Globalization and Development, 1(2), 1-29.  Asiedu, E. and B. Nandwa (2007) “On the Impact of Foreign Aid in Education on Growth: How Relevant Is the Heterogeneity of Aid Flows and the Heterogeneity of Aid Recipients?” Review of World Economics, 143(4), 631-649.  Attanasio, O., B. Augsburg, R. De Haas, E. Fitzsimons and H. Harmgart (2011) “Group lending or individual lending? Evidence from a randomized field experiment in Mongolia”. Pub ref: MPRA Paper No. 35439.  Augsburg, B., R. D. Haas, H. Harmgart and C. Meghir (2012) “Microfinance, poverty and education”. IFS working paper.  Banerjee, A. and E. Duflo (2011) Poor Economics: A Radical Rethinking of the Way to Fight Global Poverty, Random House India.  Banerjee, A. and R. He (2008) Making Aid work. In: Easterly, W. (Ed.) Reinventing foreign aid, pp. 47-92.Cambridge: The MIT Press.  Banerjee, A., A. Deaton, N. Lustig and K. Rogoff (2006) An evaluation of the World Bank research: 1998-2005.Washington: The World Bank.  Bateman, M. (2010) Why doesn’t microfinance work: The destructive rise of local neo liberalism. London: Zed Books.  Boone, P. (1996) “Politics and the effectiveness of foreign aid” European Economic Review, 40, 289-329.  Buera, F., J.P. Kaboski and Y. Shin (2012) “The Macroeconomics of Microfinance”, NBER Working Paper No. 17905, Cambridge, M.A.  Burnside, C. and D. Dollar (2000) “Aid, policies and growth” American Economic Review, 90 (4), 847-868.  CGAP (Consultative Group to Assit the Poor) (2009)“About Microfinance”, www.cgap.org  CGAP (Consultative Group to Assit the Poor) (2010) Access to finance for the poor, Annual Report 2010.Washington D.C: CGAP.  Chaia A., A. Dalal, T. Goland, M.J. Gonzalez, J. Morduch, and R. Schiff (2009) “Half the World is Unbanked”, Financial Access Initiative Framing Note, New York, Cambridge, New Haven.  Chenery H.B. and A. Strout (1966) “Foreign assistance and economic development” American Economic Review, 56: 679–733. 16                       Collier, P. and D. Dollar (2002) “Aid allocation and poverty reduction” European Economic Review, 45 (1), 1-26. Compton, R.A., D.C. Giedeman and G.A. Hoover (2011) “Panel evidence on economic freedom and growth in the United States” European Journal of Political Economy, 27, 432-435. Crépon, B., E. Duflo, F. Devoto and W. Pariente (2011) “Impact of microcredit in rural areas of Morocco: Evidence from a randomized evaluation”. J-PAL working paper. Dichter, T. (2007) Can microcredit make an already slippery slope more slippery? In Dichter & Harper (Ed.) What´s wrong with microfinance? London: Practical Action Publishing. Djankov, S., J.G. Montalvo and M. Reynal-Querol (2006) “Does foreign aid help?” Cato Journal, 26 (1), 1-28. Djankov, S., J.G. Montalvo and M. Reynal-Querol (2008) “The curse of aid” Journal Economic Growth, 13, 169-194. Doucouliagos, H. and M. Paldam (2011) “The ineffectiveness of development aid on growth: an update” European journal of political economy, 27(2), 399-404. Duflo, E., Banerjee, A., Glennerster, R. and C.G. Kinnan (2013).The Miracle of Microfinance? Evidence from a Randomized Evaluation. NBER working paper No.18950, Cambridge, M.A. Easterly, W. (2003) “Can Foreign Aid buy growth?” Journal of Economic Perspectives, 17 (3), 23-48. Easterly, W. (2008) Reinventing Foreign Aid. Cambridge: The MIT Press. Easterly, W., R. Levine and D. Roodman (2004) “Aid, policies and growth” American Economic Review, 94 (3), 774-780. Gibson, C.C., K. Andersson, E. Ostrom and S. Shivakumar (2009) The Samaritan´s Dilema. Oxford: Oxford University Press. Gomanee, K., S. Girma and O. Morrissey (2005) “Aid and Growth in Sub-Saharan Africa: Accounting for Transmission Mechanisms” Journal of International Development, 17, 10551075. Hansen, H. and F. Tarp (2001) “Aid and growth regressions” Journal of Development Economics, 64, 547-570. Helms, B. (2006) Access for all, Building inclusive financial systems, Washington, D.C.: CGAP, The World Bank. Hermes, N. and A. Meesters (2011) The performance of microfinance institutions: Do macro conditions matter? In: Armendariz, B. and M. Labie (Eds.) The Handbook of Microfinance, World Scientific, Singapore. Pp: 173-201. Heston, A., R. Summers and B. Aten (2012) Penn World Table. Version 7.1. Pennsylvania: Center for International Comparisons of Production, Income and Prices at the University of Pennsylvania (CICUP). Hulme, D. (2007) Is microdebt good for poor people? A note on the dark side of microfinance. In Dichter & Harper (Eds.) What´s wrong with microfinance?, London: Practical Action Publishing. Imai, K.S., R. Gaiha, G. Thapa and S.K. Annim (2012) “Microfinance and Poverty—A Macro Perspective”, World Development, 40(8): 1675-1689. Karlan, D. and J. Apple (2011) More than good intentions, How a New Economics Is Helping to Solve Global Poverty, Dutton Press. Karlan, D. and J. Zinman (2010) “Expanding Credit Access: Using Randomized Supply Decisions to Estimate the Impacts” Review of Financial Studies, 23 (1), 433-464. Khandker, S.R. (2005) “Microfinance and Poverty: evidence using panel data from Bangladesh” The World Bank Economic Review, 19(2):263–286. 17                         Khandker, S.R., S. Hussain and A. Rubaba (2013) Does access to finance matter in microenterprise growth? evidence from Bangladesh.World Bank Policy Research Working Paper No.6333, Washington D.C.: World Bank. Kremer, M. and D. Miguel (2007) “Illusion of sustainability” Quarterly Journal of Economics, 112(3):1007-1065. Lensink, R. and H. White (2001) “Are there negative returns to aid?” The journal of Development Studies, 37 (6), 42-65. Levy, V. (1988) “Aid and growth in Sub Saharan Africa: The recent experience”, European Economic Review, 32: 1777-95. Lewis, W.A. (1954) “Economic Development with Unlimited Supplies of Labor”, The Manchester School of Economic and Social Studies, 22(2): 139-191. Maes, J. and L. Reed (2012) State of the Microcredit Summit Campaign Report 2012.Washington D.C.: Microcredit Summit Camping. Mahjabeen, R. (2008) “Microfinancing in Bangladesh: Impact on Households, Consumption and Welfare”, Journal of Policy Modeling, 30: 1083-1092. Matin, I., M. Sulaiman and M.A. Saleque (2007) Imagining microfinance more boldly: Unleashing the true potencial of microfinance. In: Dichter & Harper (Eds.) What´s wrong with microfinance? London: Practical Action Publishing. Mekasha, T.J. and F. Tarp (2013) “Aid and Growth: What Meta-Analysis Reveals” The Journal of Development Studies, 49 (4), 564-583. MIX Market (2011) Mix Market database. Available at: http://www.mixmarket.org Morduch. J. (1999) “The microfinance promise” Journal of economic literature, 37, 1569-1614. Mosley, P. (1980) “Aid, savings and growth revisited” Bulletin of the Oxford University Institute of Economics and Statistics, 42, 79-95. Mosley, P., J. Hudson and S. Horrell (1987) “Aid, the public sector and the market in less developed countries” The Economic Journal, 97, 616-641. Moss, T., G. Pettersson and N. van de Walle (2008) An aid-institutions paradox? A review essay on aid dependency and State building in Sub-Saharan Africa. In Easterly (Ed.) Reinventing Foreign Aid. Cambridge: The MIT Press. OECD (2013a) DAC database. Available at: http://www.oecd.org/dac/stats/internationaldevelopment-statistics.htm. OECD (2013b) Official Development Assistance: definition and coverage. Available at: http://www.oecd.org/dac/stats/officialdevelopmentassistancedefinitionandcoverage.htm Okurut, N., A. Schoombee and S. Berg (2005) “Credit demand and credit rationing in the informal financial sector in Uganda” South African Journal of Economics, 73(3): 482–497. Pagan, A. (1984) “Econometric issues in the analysis of regressions with generated regressors”, International Economic Review, 25:221-247. Pitt, M. and S.R. Khandker (1998) “The Impact of Group-Based Credit Programs on Poor Households in Bangladesh: Does the Gender of Participants Matter?” The Journal of Political Economy,106(5), 958-996. Polity IV (2012) Available at: http://www.systemicpeace.org/inscr/inscr.htm Rajan, R. and A. Subramanian (2008) “Aid and growth: What does the cross country really show?” The Review of Economics and Statistics, 90(4), 643-665. Reed, L. R. (2013) State of the Microcredit Summit Campaign Report 2011, Washington D.C.: Microcredit Summit Camping. Robinson, M. (2001) The microfinance revolution: Sustainable finance for the poor. Washington D.C.: The World Bank. Roodman, D. (2012) Due Diligence: An impertinent inquiry into microfinance. Washington D.C.: Center for Global Development. 18          Sachs, J. (2004) “Ending Africa´s poverty trap” Brookings Papers on Economic Activity, 1, 117240. Sachs, J. (2011) The end of poverty: How can make it happen in our lifetime. London: Penguin. Schultz, P. (2004) “School subsidies for the poor: Evaluating the Mexican PROGRESA Poverty Program” Journal of Development Economics, 74, 199-250. Schumpeter J.A. (1934) The theory of economic development: An inquiry into profits, capital, credit, interest, and the business cycle. Oxford: Oxford University Press. Sen, A. (1999) Development as freedom, Anchor Books. Sodokin, K. and C. Donou-Adonsou (2010) “Banks, microfinance institutions and economic growth in the West African Economic and Monetary Union” African Development Review, 22(4): 495-510. Tarp, F. (2006) “Aid and Development” Swedish Economic Policy Review, 13(2), 9-61. World Bank (2011) News and Broadcast, www.worldbank.org. WDI (World Development Indicators) (2012) WDI Database, available at: http://data.worldbank.org/data-catalog/world-development-indicators 19 APPENDIX 1 Countries in the sample Albania Argentina Armenia Azerbaijan Bangladesh Benin Bolivia Bosnia and Herzegovina Brazil Cambodia Cameroon Colombia Congo, Rep. Costa Rica Chile China Dominican Republic Ecuador Egypt, Arab Rep. El Salvador Ethiopia Georgia Ghana Guatemala Guinea Haiti Honduras India Indonesia Jordan Kazakhstan Kenya Kyrgyz Republic Lebanon Madagascar Malawi Mali Mexico Moldova Mongolia Morocco Mozambique Nepal Nicaragua Niger Nigeria Paraguay Peru Philippines Rwanda Senegal Sierra Leone South Africa Sri Lanka Swaziland Syrian Arab Republic Tajikistan Tanzania Togo Tunisia Turkey Uganda Uzbekistan Venezuela, RB Vietnam Yemen, Rep. Zambia 20 TABLES: Table 1 Variable list specification, 67 countries, and period 2001-2011 Variable Description Dependent variable: gdp Real GDP, PPP (constant 2005) Units Source Internat. $ World Development Indicators a) Persistence in economic development over time gdpl1 GDP (constant 2005) in the preceding year Internat. $ World Development Indicators b) Main variables oda Total Net ODA mf Gross Loan Portfolio on Microfinance Internat. $ Internat. $ OECD (DAC database) MIX Market Internat. $ Internat. $ Internat. $ Internat. $ Internat. $ World Development Indicators World Development Indicators World Development Indicators World Development Indicators World Development Indicators Internat. $ Internat. $ % Internat. $ % Index Index Internat. $ % of GDP Years # persons # persons Persons/Km2 % 010 2.5 to 2.5 2.5 to 2.5 IFSIMF IFSIMF World Development Indicators World Development Indicators World Development Indicators World Development Indicators World Development Indicators World Development Indicators World Development Indicators Barro and Lee database International Labor Organization Penn Word Tables Penn Word Tables and GIS* World Development Indicators Polity IV World Development Indicators World Development Indicators Explanatory variables: c) Transmission variables gcfpriv gcfpubl cpriv cpubl impor Private investment Public investment Private final consumption expenditure Govern final consumption expenditure Imports of goods and services (constant 2005) d) Control variables credit depos inf tax inter er tot expor open edu lf popu dens malnut demo rol corrupt Bank private credit Deposit money bank assets Inflation rate (consumer price index) Tax revenue Real interest rate Real effective exchange rate index (2005=100) Net barter terms of trade index (2000=100) Exports Openness at 2005 constant prices Total years of schooling Labor force (economically active population) Total population Population density Malnutrition prevalence (% of children under 5) Democracy index Rule of law Control of corruption * Developed by the authors with a Geographical Information System (GIS) 21 Table 2 Estimation results of the transmission mechanisms for ODA and MF(1), 67 countries, and period 2001-2011 Dependent variables: oda mf lagged dependent variable credit tax inf (lagged) inter(lagged) tot (lagged) expor lf dens demo Arell.-Bond 1storder autocor. test 2ndorder Sargan specification test Endogeneity Durbin tests for W-Haus. oda and mf Sargan ODA transmission equations (Eq. 3) Public Government investment consumption Imports (gcfpubl) (cpubl) (impor) 0.0570 0.0193 0.0083 0.3416*** 0.6270*** 0.3200*** 0.3888*** 0.2010*** 0.0016*** 0.2850*** 0.1795*** 0.4668*** 0.6549** * 0.6077 0.0123* 2.0679** 2.4951*** 1.7120* 0.2331 0.5245 0.42681 46.8469 46.1080 49.1945 0.6264 2.0501 0.2875 0.1496 1.5084 0.1697 5.3637 0.1544 0.7722 MF transmission equations (Eq. 4) Private Private investment consumption (gcfpriv) (cpriv) 0.0878*** 0.0614*** 0.6244*** 0.5843*** * 0.2157 0.1226** 0.0055* 0.0019** 0.0018*** 0.0170** 2.4995** 1.7698* -0.67261 0.87741 48.8144 47.9938 0.1433 0.8416 0.1410 0.8289 0.5516 0.9010 (1) All variables are expressed in log form, except the inflation and interest rate variables. *** Significant at pvalue<0.01; ** significant at p<0.5; * significant at p<0.1. Table 3 Estimation results of the economic growth models for ODA and MF(1), 67 countries, and period 2001-2011 Dependent variable: GDP growth (gdp) oda mf gdpl1 gcfpriv gcfprires impor tot demo Arellano-Bond 1st order autocorr. test 2nd order Sargan specification test Endogeneity Durbin test for W-Hausman ODA/MF Sargan Initial growth model ODAgrowth model - 0.8310*** 0.0189** 0.1111*** 0.0422** 0.0045** 2.3080** 0.4934 51.7735 - 0.0019 0.8420*** 0.0197** 0.1114*** 0.0042* 2.3473** 0.4473 51.1756 2.2029 1.0202 4.5673 MFgrowth model 0.0032 0.8091*** 0.0189** 0.1123*** 0.0428** 0.0043* 2.2630** 0.5207 53.3160 - MF correctedgrowth model 0.0078** 0.8091*** 0.0185** 0.1123*** 0.0427** 0.0042* 2.2630** 0.5207 0.6813 0.6696 0.9665 Robust checks ODAgrowth 0.0058 0.9942*** - MFgrowth 0.0115*** 0.9280*** - *** significant at p-value < 0.01; ** significant at p < 0.5; * significant at p < 0.1. (1) All variables are expressed in log form, except the inflation and interest rate variables. 22 FIGURES: Figure 1. Model estimation results of transmission mechanisms of ODA and MF on GDP growth Public investment Government expenditure ODA ˆ2  0.1123  Imports ˆ1  0.0185 ˆ  0.0878  Private investment MF ˆ  0.0614 GDP growth  ˆ  0.0078 **  Private consumption Direct effect Indirect effect Developed by the authors 23 V i e w p u b l i c a t i o n s t a t s