Is too much foreign aid a curse or blessing to developing countries?

3.1. Data

This article examines the nexus of aid and economic growth; and, the role of institutional quality and economic freedom in the relationship between them. For this purpose, 48 developing countries given in Table 2 were selected and the necessary data was collected from available databases like World Development Indicators (WDI), World Economic Outlook (WOE), Worldwide Governance Indicators (WGI), Heritage Foundation, and United Nations development program (UNDP) over the period covering 2006–2019. The sample of countries considered in this study is made up of low income and lower middle-income countries according to the World Bank (WB) country classification. Obviously, these countries are not the only developing countries of the world. Developing countries not included in this study are excluded purely because of unavailability of data. The study period (2006–2019) is decided in an attempt to obtain strongly balanced panel data set as panel threshold regression cannot be implemented with unbalanced. Variables of the study were selected on the basis of theory of economic growth. These are: real GDP per capita (the dependent variable), institutional quality index, economic freedom index, aid, population growth, inflation, government expenditure, investment, financial development and schooling.

Table 3 provides description of the variables adopted by the study. In this study, aid is represented by official development assistance and expressed in its share of GNI and sourced from WDI. Consumer price index has been adopted by various contemporary empirical growth studies to measure cost of living (see Asongu et al., 2018; Bekere and Bersisa, 2018; Ruiz, 2018; Hayat, 2019; Osei and Kim, 2020; Krasniqi and Demukaj, 2021). However, GDP deflator is preferred in this study due to its ability to capture prices of wider goods and services. While some recent studies show inflation is harmful to the economy only after some threshold level (López-Villavicencio and Mignon, 2011; Kremer et al., 2013), there is consensus among different schools of economic thought that high growth rate of inflation is unfavorable at all (Thanh, 2015). The data is obtained from WDI. Government expenditure is expressed in percent of GDP and collected from WOE. The same proxy for the variable was used by Hayat (2019); Jain et al. (2021). Government expenditure is an important economic variable to stimulate the economy during deficiency of demand in Keynesian economics. Government expenditure directed towards development enhances growth (Wahab, 2011). Total investment as percentage of GDP is considered in this study as another important control variable and sourced from WEO. Investment plays a pivotal role in the neoclassical growth theories and often considered as an indispensable source of growth. Feeny (2005) and Alshammary et al. (2020) used the same proxy variable in their study. Mean years of schooling is used to proxy human capital (Law and Singh, 2014; Chattopadhyay et al., 2021), which is obtained from UNDP. While the annual population growth rate is collected from WDI, financial development is provided by global financial development database.

In order to examine if the effect of aid on economic growth hinges on the level of institutional quality, the study relies on the six variables designed by world governance indicators (WGI) of the World Bank. The six variables are namely control of corruption, political stability and absence of violence, regulatory quality, and government effectiveness, rule of law and voice and accountability. The variables are scaled from – 2.5 (indicates weak quality of institutions) to +2.5 (indicates strong quality of institutions). For the purpose of the current study, the average of the six indicators is used because there is no variation in relative importance attached to each indicator. In the study by Yahyaoui and Bouchoucha (2020), it is found that aid performs better under better institutional quality while institutional quality does not play significant role in shaping the link between aid and economic growth in the paper by Adusei (2020).

In the same vein, economic freedom is represented in this study by index of different indicators and gathered from the heritage foundation. It measures the extent of governments' control over domestic economy and graded on scale of 0–100. The higher the scale the more liberal the economy and vise versa. According to Burnside and Dollar (2000), aid promotes economy if given to countries with good economic policy environments. However, the study by Vasquez (1998) and Powell and Ryan (2011) fails to confirm this as improvement in economic freedom index is not rewarded with more aid. Some western donors like World Bank and the International Monetary Fund (IMF) prefers to support more free economies. If huge amount of aid stimulates recipient's economy and if huge amount of aid is more likely to flow to ward countries with better economic freedom, it can be said economic freedom do matter in the relationship between aid and economic growth.

3.2. Empirical model

This study relies on panel threshold regression developed by Seo and Shin (2016) to achieve its objectives. Panel threshold regression (PTR for short) was first proposed by Hansen (1999). It is a great innovation in the field as linear models are becoming incapable of detecting the relationship between various economic variables accurately. PTR is a widely used method in research areas like inflation (Kremer et al., 2013; Thanh, 2015; Khan and Hanif, 2020), financial development (Law et al., 2018), foreign direct investment (Osei and Kim, 2020), public debt (Alshammary et al., 2020), mobile penetration (Asongu et al., 2018), etc.

The method of Hansen (1999), though notable contribution in the field, is static in nature and cannot capture dynamics. Some scholars have tried to extend the method of Hansen (1999) by introducing its dynamic version (Caner and Hansen, 2004; Dang et al., 2012; Ramírez-Rondán, 2015; Kremer et al., 2013). Even though all of these models move one-step beyond Hansen's model, they are still defective since they impose too restrictive requirement of exogeneity of either the predictors or the threshold variable or even both (Seo and Shin, 2016). Seo and Shin (2016) developed a model that relaxes the restriction imposed on properties of predictors and threshold variables and allow them to be endogenous. The proposed estimator is implemented by the method of Arellano and Bond (1991). Therefore, the current study employs modified versions of the original equation developed by Seo and Shin (2016) as given in Eqs. 1a, 1b, and 1c. According to this panel threshold model, the threshold effect is identified using the nonparametric i. i.d. bootstrapping proposed by Seo and Shin (2016). The null hypothesis of no threshold will be rejected if the bootstrap probability value is less than the conventional critical values (1%, 5%, and 10%).

Where $y_{it}$ is the logarithm of per capita real GDP, $X{\prime }_{it}$ is a $K_1\times 1$ is vector of covariates that vary with time and it includes lag of aid, aid represents amount of aid received, ${\text{ins}}_{it}$ is institutional quality, ${\text{freed}}_{it}$, is economic freedom, $1\{\ast \}$ is an indicator function, ${\text{q}}_{it}$ is the threshold variable, $\delta$ is the threshold parameter, ${\phi }_1$ and ${\phi }_2$ are respectively coefficients of the lower and upper regime, ${\mu }_i$ is country fixed effect and ${\epsilon }_{it}$ is random error, t = 1,…, T indexes time and i = 1,…,N indexes country.

4.1. Descriptive statistics and correlation analysis

The result of descriptive analysis is exhibited in table. As depicted in the table, the log of GDP per capita is 7.275 on average with standard deviation of 0.719, which is an indication of low variability in the data series. Over the study period, the net aid inflow to the sampled developing countries amounts to 4.97 percent of GNI on average. Since there is a big gap between the minimum and maximum values of the series, the data points constituting the series have high variability. The highest value of 24.734 is observed for Haiti in 2010 while the lowest value of -0.288 is observed for Sri Lanka in 2018. Standard deviation of financial development (29.877) which shows that high variability in the spread of data characterizes the variable. As reported in the table, the mean value of institutional quality index is -0.581 with lower standard deviation of 0.399, indicating poor institutional quality. Among the sampled countries, the lowest institutional quality (−1.75) is recorded in Myanmar in 2009 while the highest (0.572) is in Cape Verde in 2010 while the mean value for the whole of the sampled countries is 55.284, which, according to the heritage foundation's country classification, would classify the sample countries as “mostly unfree”.

The result of correlation analysis is reported in Table 4. It is indicated in the table that the independent variables other than inflation, population growth, aid, and square of aid have positive correlation with economic growth. Coefficient of correlation between economic growth and aid (−0.61) shows strong and negative correlation between the variables, indicating that they move in opposite direction. However, though still negative, the correlation between aid and economic growth becomes moderate as shown by coefficient of correlation of -0.47 between square of aid and economic growth.

4.2. The relationship between aid and economic growth

The current study has examined the relationship between aid and economic growth using the method of Seo and Shin (2016). As indicated in Table 5, the result reveals that the relationship between aid and economic growth is nonlinear because the bootstrap p-value based on 1000 replication is significant at 1% critical value indicating that there is a threshold effect of aid. As reported in the upper part of the table, the estimated threshold level of aid is 7.03, such that about 74% of the observations fall below the threshold level.

In the middle part of the table, the impact of aid on economic growth is given. Accordingly.

$\widehat{{\phi }_1}$ and $\widehat{{\phi }_2}$ represents the coefficient of aid below and above the threshold level respectively. Below the threshold level, the impact of aid on economic growth is negative while it turns out positive above it. Specifically, a 1-percentage point increase in aid to GNI ratio decreases economic growth by 1.4% for any amount of aid no more than 7.03 percent of GNI and increases economic growth by 1.3% for aid to GNI ratio exceeding 7.03. Thus, aid is nonlinearly related to growth taking U – shape, which means that the impact of aid on economic growth is favorable if and only if it is above certain threshold level. In other words, too much aid to developing nations benefits, not harm their economy. Similar result was obtained by Gyimah-Brempong et al. (2012) and Yiew and Lau (2018). Furthermore, the threshold estimate recorded in this study is closer to the minimum 6.6 percent documented by Gyimah-Brempong et al. (2012). On the other hand, the result obtained in this study is in stark contrast with what was found by Ali and Isse (2005); Fiodendji and Evlo (2013). These authors found inverted U shape relationship between aid and economic growth. The finding also differs from what was recorded by the study conducted by Khan and Ahmeed (2007). The authors concluded that foreign aid is not blessing both at aggregated and disaggregated level from their linear model specification.

Certain amounts of aid received by recipient countries will be added to their capital stock, and hence, bolster investment activities. When viewed from this angle, aid will stimulate economic growth (Yiew and Lau, 2018; Ovaska, 2003). This provides theoretical base for the observed positive effect of aid on economic growth above the threshold level. However, it is also possible that aid requires good policy to be effective (Burnside and Dollar, 2000; Niyonkuru, 2016) and deteriorate balance of payment by causing national currency of the recipient country to appreciate (Liew et al., 2012). According to Borjas (2015), an increase in non-labor income reduces the number of hours people choose to work, which reduces the total labor supply in the economy. Since foreign aid tends to increase non-labor income, therefore, it reduces labor supply and this harms the productive capacity of the country (Gyimah-Brempong et al., 2012). Thus, the claim of the existence of some conditions for effectiveness of aid and the tipping point is not unfounded.

Generally, as the results of this article reveal, the impact of foreign aid received by developing countries on their economic growth is contingent on the amount received. Since it is established that the impact of aid on economic growth is negative (positive) when amount of aid received is below (above) the threshold level respectively, it implies that foreign aid received by the developing countries inevitably entails both favorable and unfavorable effects. Therefore, below the threshold level (7.03% of GNI), the positive effect of aid on economic growth cannot completely swamp the negative effect and foreign aid remains detrimental to the economy. When it is sufficiently above the threshold level, however, it starts to stimulate economic growth. Significance and sign of the control variables are expected except financial development in this study.

The role of institutional quality in the relationship between aid and economic growth.

The role of institutional quality and economic freedom can also be examined using panel threshold regression by designating quality of institutions and economic freedom as threshold variables. This time, however, the threshold variables are institutional quality and economic freedom. In order for institutional quality and economic freedom to matter, the threshold estimate of institutional quality and economic freedom should be statistically different from zero. As shown in Table 6, quality of institutions and economic freedom matter in the nexus of aid and economic growth as confirmed by the bootstrap p-value of less than 1%.

As indicated in Table 6, the institutional quality threshold is -0.320, such that about 72% of the total observations fall below this. Below the threshold level, the effect of aid on economic growth is positive but insignificant. Above the threshold level, however, aid contributes to the economic growth positively and significantly. Accordingly, it is found in this study that a 1% increase in aid to GNI ratio is associated with 0.9% increase in economic growth in countries with average institutional quality greater than -0.320. Similar results concerning the positive role of good quality institutions in the impact of aid on economic growth is obtained by prior studies (Burnside and Dollar, 2000; Tang and Bundhoo, 2017). However, the argument that aid undermines growth under poor in institutional quality is not established in this study since the coefficient of aid below the threshold level, though insignificant, is positive.

As regards to the role of economic freedom, the threshold level is 53.481 and it is significant at 1% significance level. As shown in Table 6, the effect of aid on economic growth varies below and above this threshold level. Below the threshold level, the impact of aid on economic growth is negative, but it becomes insignificant when economic freedom index exceeds the threshold level. Specifically, a 1-percentage point increase in economic freedom below the threshold level costs 0.3% in economic growth. The observed negative effect of aid on economic growth when economic freedom is below the threshold level of 53.481 might be for two reasons.

Firstly, the low level of economic freedom index is an indicative of huge involvement of government in the overall economy than private sectors. In situations where government officials are opportunistic and/or corrupt, however, there is little chance for funds obtained through aid to be directed toward productive activities, and thereby promote growth.

Secondly, as found in this study a low level of aid received (below 7.03% of GNI), by developing countries is detrimental to their economy. Some western donors like IMF and WB requires the aid receiving economies to reform their institutions in line of neo liberalism. Countries that show improvement in economic freedom measures will be rewarded with more aid from these donors. Thus, countries with low economic freedom faces difficulties in attracting more aid and thus forced to experience the negative impact of low level of aid received. Out of the 48 sampled countries in this study, 32 have the overall economic freedom index above the threshold level on average. Aid flows to these countries amounts to 5.18% on average while the remaining 16 countries with economic freedom index less than the threshold level managed to attract 4.5% of GNI as aid. Therefore, it implies that countries with better economic freedom are rewarded with more aid than those with poor economic freedom.

Author contribution statement

Chala Amante Abate: Conceived and designed the experiments; Performed the experiments; Analyzed and interpreted the data; Contributed reagents, materials, analysis tools or data; Wrote the paper.

Funding statement

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Data availability statement

Data will be made available on request.

Declaration of interest's statement

The authors declare no conflict of interest.

Additional information

No additional information is available for this paper.