The Relationship Between Foreign Aid and Economic Growth: Empirical Evidence from Bangladesh

Abstract

Bangladesh's growing foreign aid has sparked controversy over whether it affects the country's economic performance. This review assesses foreign aid's influence on the country's economic growth with annual data covering the 1989-2018 period. The Autoregressive Distributed Lag (ARDL) model is applied to achieve the research objective, and the empirical results indicate a substantial and robust impact of foreign assistance on economic growth. The outcome further reveal that domestic investment also contributes significantly to the country's economic evolution. However, trade openness plays a substantial positive role in the short run, although the impact is immaterial in the long run. The empirical findings indicate that the association between aid, domestic investment, and growth has a confident meaningful effect at 1 per cent level in the long run, whereas aid influences more than domestic investment. However, in the short run, aid, domestic investment, trade openness, and growth show positive and noteworthy response also at 1 percent level. This review undertakes a detailed analysis about the country's economic growth, and grounded on its outcome, this work suggests that focus should be placed more on creating domestic investment, promoting more export, and allocation of aid should be determined by the relative needs of the country.

1. Introduction

Bangladesh, formerly known as East Pakistan, is the world’s largest delta on the Bay of Bengal’s shore, liberated from nearly 200 years of British colonial rule and becoming part of Pakistan’s newly formed kingdom in 1947. In 1971, an independent state called Bangladesh was born from East Pakistan through a bloody war. In the post-independence period, poverty and various natural calamities were constant companions of poverty-stricken Bangladesh; besides, long-term political instability and some military coups have repeatedly damaged the country’s economic structure. Many policymakers, at that time, were pessimistic about the potential economy of Bangladesh. However, the black cloud accumulated in the sky of Bangladesh’s economy is slowly beginning to dissipate. At present, Bangladesh has achieved unprecedented progress in poverty alleviation and economic development. According to World Bank data, in 2018, it became the world’s 8^th most populated nation, 94^th in terms of total area, with a total population of 161.36 million, per capita GDP of 1698.35 US$, and a poverty rate of 26.20 per cent. Besides, in the post-independence period, various social pointers, such as child mortality, adult literacy, primary school enrollment, and life expectancy, have improved significantly in Bangladesh (World Bank, 2019a).

Just as Bangladesh has used its domestic resources for long-term economic development, so too have friendly countries sided with her since independence. The Government of Bangladesh’s long-term development plan aims to position the country as a middle-income and high-income nation by 2021 and 2041, respectively, and has already been recognised as a lower-middle-income nation in 2015. Since independence, various donor agencies and other friendly countries have extended a helping hand on the country’s economic progress, and by 2018, Bangladesh gathered approximately US$64.13 billion in total as foreign aid. Statistics show that the flow of foreign aid to Bangladesh stood at US$13.01 billion from 1981–1990, US$11.48 billion from 1991–2000, US$12.21 billion from 2001–2010, and US$20.60 billion for the last eight years of 2011–2018 (World Bank, 2019b).

However, the nature and trend of foreign aid have changed somewhat in recent times. The flow of foreign aid as a scale of GDP was massive at first, it has declined since 1996, when the economy expanded further. Statistics show that the contribution of domestic resources to ADP funding was 35.83% and 43.03% in FY1993–94 and FY1994–95, respectively, while in FY2016–17 and FY2017–18, it was 55.86% and 64.92%, respectively, and therefore, it can be assumed that the economic dependence on foreign assistance, as a whole, is continually decreasing (Bangladesh Economic Review 2005, 2005; Bangladesh Economic Review 2019, 2019). It is a testament to the government’s success in using resources from internal sources and the non-public field’s progress. At present, the constructive aspect of reducing the proportion of foreign aid to GDP is that Bangladesh wants to become self-reliant and moving toward it; in other words, Bangladesh is now gradually establishing the necessary funds for its development. The reduction in the proportion of overseas assistance to GDP, and simultaneously the achievement of economic growth of the country, may indeed give a fresh measure of the relationship of foreign aid with economic progress.

Moreover, studies conducted in different countries have also yielded results of divergent characteristics. Many have manifested a credible positive link between aid and growth (Gounder, 2001; Hansen & Tarp, 2001; Loxley & Sackey, 2008; Tra, 2014), while others have observed a significant negative association (Mallik, 2008; MM, 2016; Tendongho, 2016). Besides, a group of researchers did not find any connection between the two, or if there was a relationship, it was very little or not very worthy (Burnside & Dollar, 2000; Eregha & Oziegbe, 2016; Khan & Ahmed, 2007). Early studies in Bangladesh are not without controversy. For example, Islam (1992) listed some causes restricting the findings’ effectiveness, such as low data quality, poor model specification, collinearity problem, and limited sample period. Later, Hossain et al. (2018) used VECM, where error terms from different periods might be correlated as no serial correlation test was found. Many previous studies have either used OLS (Hossain, 2014; Islam, 1992) or 2SLS method (Ahmed, 1992) and did not confirm the stationarity of the data. So, there is a high chance of counterfeit results, as time-series data, most of the time, suffer from the non-stationary problem. Quazi (2005) also studied the aid-growth nexus and used the Engle-Granger cointegration approach. Still, this study did not present any strong argument to prove whether the residuals of the error correction model are normally distributed.

Thus, contrary to previous studies on Bangladesh, this work is designed to examine the cohesion between overseas assistance and economic growth, adopting the ARDL model. This review employs unit root tests to verify data stationarity. It does not give any spurious regression results and generates correct estimates of the coefficient even when endogenous regressors exist. It is followed by applying the ARDL method to ascertain the association between aid and growth and confirmed the absenteeism of any serial correlation and heteroskedasticity backing the notion of normally distributed errors.

This review examines this association intending to contribute to the richness of pertinent studies, primarily because, as mentioned above, the foreign aid-to-GDP ratio is slowly declining. At the same time, the use of domestic funds and the GDP growth rate is increasing. The findings are useful especially for policymakers, economic relations division (ERD), and Bangladesh’s government. It will help them implement a proper strategy for the country’s economic development.

2. Literature Review

What contribution does foreign aid bring to a nation’s economy is a widely discussed and debated issue. The Keynesian economic growth model (Domar, 1946; Harrod, 1939) analyzes foreign aid’s significance in forming capital and domestic investment by closing the savings gap, thereby accelerating the economic growth of a developing nation. The prominent big-push concept of Rosenstein-Rodan (1943) specifies the requirement of a minimum level of high investment figure for tackling growth obstacles in an underdeveloped economy. Besides, the economy should not be instructed way-by-bit; instead, the government should accentuate each economy segment altogether. Chenery and Strout (1966) claimed that, due to the scarcity of savings and the lack of foreign exchange, the underdeveloped countries are caught in a vicious deprivation loop. In this case, foreign aid can help bridge two sensitive lacunae, namely, (i) the savings-investment gaps and (ii) the export-import gaps. Besides, Gupta (1975); Levy (1988), and Papanek (1973) supported Chenery and Strout (1966) and concluded that aid helps to raise capital indirectly through increasing domestic income and savings, and thereby increases direct investment in non-industrialized countries. Additionally, aid in technical assistant shape also enhances technical knowledge and human skill, indirectly serving a constructive role in emerging nations’ economies.

However, a group of economists (Easterly, 2003; Griffin, 1970; Harms & Lutz, 2004; Heller, 1975; Leff, 1969) criticized Chenery and Strout (1966). They claimed that aid might not work well in solving the two-gap model’s problem; instead, look forward to developing a sound tax system. The three-gap model introduces the fiscal gap along with the previous two gaps (e.g., savings-investment and export-import), where it is argued that most of the emerging countries are struggling to manage the funding they need for government-planned investments. Foreign aid is acknowledged as a vital instrument to fill that gap (Bacha, 1990). However, when foreign aid comes as a loan rather than a grant, it could have some hostile consequences on the long-run’s overall economic system. Although the inflow of loans in the name of foreign aid can minimize the trade gap in the short-run, a country needs to progress foreign trade to become self-sufficient in the future without the help of foreign aid (Anthony et al., 2019).

Moreover, Griffin and Enos (1970) claimed that aid upsurges public consumption and does not go about as an addition to investment funds. A group of researchers (Burnside & Dollar, 2000; Collier & Dollar, 1999) blames the country’s poor strategy on the failure of aid and argue that foreign assistance is a blessing for those countries which can adopt the right policy, else it is a squander. Also, the contributors’ strategic preference performs a more substantial role in providing aid than its recipients.

The empirical literature of foreign assistance on growth is mixed. Islam (1972) acknowledges the cohesion between foreign assistance and economic expansion in the Pakistani context and revealed that aid helps build domestic savings and investment. Although, at some level, it increased the rate of inflation, it played a title role in the enrichment of GDP. Gounder (2001) analyzed the function of aid in Fiji’s growth, and the outcome represents a significant relationship between the bilateral, grant, and technical aid with growth. However, this study also concludes about the ineffectiveness of domestic resources on growth. Hansen and Tarp (2001) have acknowledged foreign aid as a driving force behind the growth using domestic investment. The effectiveness of aid has been cited as a response to the choice of control variables. Later, Dalgaard et al. (2004) have observed the efficacy of aid, citing the inflow of aid and good policies as crucial to its impact. They show the positive aspects of foreign aid on production, but simultaneously it is said that aid cannot be the only remedy to poverty alleviation. In Africa, Loxley and Sackey (2008) identified an essential positive connection between aid and growth. In unison, the researchers have suggested reducing aid dependence rather than relying on it in the long-run.

Although Pigka-Balanika (2013), in SSA, revealed a negative consequence of trade openness on growth, from 1984 to 2007, Sakyi (2011) demonstrated the association between trade openness and economic growth with foreign aid, where both have exposed a constructive link between growth and assistance. Hye et al. (2016) and Nantharath and Kang (2019) agree with Sakyi (2011) and posit that trade openness is positively linked with growth. Fasanya and Onakoya (2012) have sought to determine foreign aid’s incentive on economic growth, indicating a meaningful positive association between the two. Tra (2014) inspected the connection of aid on economic advancement from 1993 to 2012 and found a significant positive association between the two. Some of the policies recommended include coordination between donors and recipients, evaluation of completed projects, strengthening institutional and managerial capacity, development of good fiscal system and above all development of the capacity for sustainable economic development. A recent survey (Nguyen, 2020) also revealed the fact that foreign aid give rise to economic development, and at the same time, exports contribute meaningfully to the overall growth of the country.

However, some studies show a negative relationship between aid and growth. Ahmed (1992) looked into the aspect of foreign aid in Bangladesh’s economic growth from 1972 to 1991, where the inverse connection between assistance and growth was found. Mallik (2008) studied the six most impoverished countries in the SSA, citing that despite receiving foreign aid, those countries’ real per capita income either decline or stay the same, and five out of six countries found negative growth relation with aid. Later, Appiah-Konadu et al. (2016) studied the consequence of overseas assistance on Ghana’s progress and found a negative relation between the two. This study suggested that capital and human skills should be built through foreign aid to get positive outcomes from it. In a study from 1960 to 2013, Tendongho (2016) found that Cameroon was increasingly dependent on foreign aid despite having ample natural and other resources. Even with a large amount of aid received, it could not minimize the foreign exchange and savings gaps. The study uncovered a credible negative linkage between foreign assistance and growth. However, there was a constructive affiliation between domestic investment and growth.

A recent study, MM (2016), applied the DOLS model to reveal the influence of aid on progress from 1976 to 2014. The empirical outcome identifies a negative relation between aid and growth. In another study, Sothan (2018) assessed the growth-aid nexus adopting ARDL approach where, results of the analysis showed that aid and its distinctive factors have made a noteworthy stimulus to the economic progress of Cambodia. The findings disclose that both in short and long run, trade openness gear up the growth; domestic investment leads to economic progress only in the long run; however, aid has a supportive influence on growth only in short-run, besides in long-run growth is negatively affected by aid.

3. Methodology

3.1. Data and Model

This review utilizes data for Bangladesh on four variables – growth proportion of gross domestic product (GDP), foreign aid (FAID), domestic investment (DI), and trade openness (TO) – with an annual scale for the period 1989–2018. Data series are collected from the World Bank and OECD National Accounts. This study adopts the model by involving foreign aid and a few control variables to diagnose the association between economic growth and aid. The main equation, to explore the affiliation between economic evolution and overseas aid with some additional control variables, is expressed as follows:

\(\mathrm{GDP}_{t}=\beta_{0}+\beta_{1} \mathrm{FAID}_{t}+\beta_{2} \mathrm{DI}_{t}+\beta_{3} \mathrm{TO}_{t}+\varepsilon_{t}\)       (1)

Here, GDP_t, FAID_t, DI_t and TO_t represent the gross domestic product, foreign aid, domestic investment, and trade openness, respectively. β₀ is constant and ε_t represents error term. β₁, β₂ and β₃ are coefficients of FAID, DI, and TO, respectively. The real GDP growth rate surrogates economic growth; foreign aid is surrogated by net official development assistance received as a proportion of GDP; the gross fixed form of capital surrogates domestic investment as a proportion of GDP and, trade openness adding exports with imports as a ratio of GDP.

To explore the influence of external assistance on Bangladesh’s economic evolution, this work applies the Autoregressive Distributed Lag model. This method is preferred to many other traditional methods, e.g., Johansen co-integration tests (Johansen, 1991) and Engle and Granger approach (Engle & Granger, 1987) because it gives relaxation to some assumptions that other traditional methods do not. This method does not provide any restriction that all defined variables have to be integrated in identical order; instead, it can be involved in the order in zero, one, or in a combined order of the two. This model also reveals the short-run and long-run coefficients of the explanatory variables at the same time. Besides, the ARDL is equally effective for a small sample size. However, if any variable is integrated into the second-order or higher, this model cannot be extended.

This work applies two widely accepted unit root screen methods, namely, Augmented Dickey-Fuller and Phillips- Perron unit root tests, to reveal the integration sequence and authenticate the data’s stationarity. Aimed at exploring the affiliation between foreign assistance and economic progress, this review derives the subsequent ARDL model in equation (2), where the main concern is the magnitude and instruction of the coefficient of foreign aid, even though the results of the other control variables still give some valuable viewpoints.

\(\begin{aligned} \Delta \mathrm{GDP}_{t} &=\beta_{0}+\sum_{r-1}^{m_{1}} \beta_{1 r} \Delta \mathrm{GDP}_{t-r}+\sum_{r-0}^{m_{1}} \beta_{2 r} \Delta \mathrm{FAID}_{t-r} \\ &+\sum_{r-0}^{m_{3}} \beta_{3 r} \Delta \mathrm{DI}_{t-r}+\sum_{r-0}^{m 4} \beta_{4 r} \Delta \mathrm{TO}_{t-r}+\delta_{0} \mathrm{GDP}_{t-r} \\ &+\delta_{1} \mathrm{FAID}_{t-r}+\delta_{2} \mathrm{DI}_{t-r}+\delta_{3} \mathrm{TO}_{t-r}+\mu_{t} \end{aligned}\)       (2)

Here,

\(\begin{array}{l} \boldsymbol{H} \boldsymbol{\theta}: \delta_{0}=\delta_{1}=\delta_{2}=\delta_{3}=0 \text { (no cointegration) } \\ \boldsymbol{H} \boldsymbol{A}: \delta_{0} \neq \delta_{1} \neq \delta_{2} \neq \delta_{3} \neq 0 \text { (cointegration exists) } \end{array}\)

Where ∆ denotes is the first difference operator, β₀ is constant, β_1r, β_2r, β_3r and β_4r are the short-run coefficient of GDP, FAID, DI, and TO, respectively, in equation (2). m₁–m₄ represents the optimum lag length, and μ_t denotes the residual term.

H0 specifies the nonappearance of cointegration between the variables, whereas the HA takes the form of a long-run association between the variables. F statistics is needed to figure out the long-run connection inside the variables, wherein Pesaran et al. (2001) advanced two classes of key standards, namely, upper bound I(1) and lower bound I(0), need to be compared. If the measurement of F statistics >I(1), the H0 of no cointegration cannot be accepted signposts the occurrence of long-run affiliation between variables. If the measurement of F statistics < I(0), the H0 of no cointegration is confirmed, implies omission of long-run affiliation between the variables. Contrariwise, an inconclusive outcome is derived when the measure of F statistics resides between the critical values of I(1) and I(0).

If the long-run affiliation between variables is ensured, the subsequent is to determine the short-run relationship with error correction term. The equation (3) is the representation of the ECT with the short-run relationship of the variables:

\(\begin{aligned} \Delta \mathrm{GDP}_{t}=& \beta_{0}+\sum_{r=1}^{m_{1}} \beta_{\mathrm{lr}} \Delta \mathrm{GDP}_{t-r}+\sum_{r=0}^{m_{2}} \beta_{2 r} \Delta \mathrm{FAID}_{t-r} \\ &+\sum_{r=0}^{m_{3}} \beta_{3 r} \Delta \mathrm{DI}_{s-r}+\sum_{r=0}^{m_{4}} \beta_{4 r} \Delta \mathrm{TO}_{t-r} \\ &+\eta_{1} \mathrm{ECT}_{r-r}+\mu_{t} \end{aligned}\)       (3)

Where, η₁ denotes the speed of adjustment, and ECT_t−r signifies lagged error correction term. ECT’s value must have to be minus value and statistically meaningful, which indicates the time that it will take to get back in the long-run equilibrium if there is any shock in the short-run. Finally, some diagnostic and stability checks are completed to determine whether the model fits correctly. The tests include serial correlation, heteroscedasticity, the CUSUM and CUSUM-SQ test.

4. Results and Discussion

Although the ARDL bound testing method does not require unit root testing, this work checks for the data’s stationarity. It is done to confirm whether the data of any variable is integrated at more than level 1. Table 1 shows the outcomes of the unit root check. This inquiry applies two methods of unit root test, e.g., ADF and PP for cross-checking the outcomes, where ADF indicates all the variables are stationary at the I(1) for constant; in contrast, on condition of constant and trend, except GDP, which is stationary at I(0), the other three variables (FAID, DI and TO) are appeared to be stationary at I(1). However, the PP test confirms the stationarity of all variables at the I(1), excluding FAID that is stationary at I(0) in constant, and in constant and trend all the variables are stationary at I(1) expect GDP. The mixed integrating nature of variables and the absence of integration in the I(2) confirms the feasibility of estimating the long-run ARDL model.

Table 1: Unit Root Check Results

Notes: (i) ***, ** and * indicate the p-value is statistically substantial at 1%, 5% and 10% level, respectively. (ii) “−” Symbolizes “not applicable.” (iii) I(0) and I(1) refers at level and first difference, respectively. (iv) C and CT indicate constant, and constant with trend, respectively.

Before approaching the ARDL bound testing model, it estimates optimum lag based on LR, FPE, AIC, SC, and HQ for the respective equation. Finally, it selects lag 1 as an optimum that is accessible in Table 2.

Table 2: Best Lag Choice

After recognizing the sequence of integration of the variables and estimating the maximum lag, the following phase is to verify if there is any long-run linkage between variables using ARDL bound testing approach. Table 3 demonstrates the outcomes of ARDL bound testing where the calculated value of F statistics (8.175) for model F_GDP(GDP|FAID, DI, TO) is more than the computed critical value of I(1) (5.61) at a 1 percent level of significance. It permits to refuse the of no co-integration among variables and allows accepting HA of long-run association among GDP, foreign aid, domestic investment, and trade openness.

Table 3: Results of Bound Testing

Notes: (i) The first variable outside the bracket is the dependent variable. (ii) *** indicates the p-value is statistically substantial at 1% level.

Having confirmed the presence of long-run association in the model of F_GDP(GDP|FAID, DI, TO) this study estimates and divulges ARDL long-run and short-run outcomes in Table 4; the GDP growth rate is the dependent variable. The foremost explanatory variable is foreign aid, where domestic investment and trade openness are control variables. The long-run results of Table 4(I) indicate that FAID has a positive credible (p < 0.01) consequence on GDP. It suggests that the higher the foreign aid, the better the GDP growth rate, which leads to Bangladesh’s economic progress. More precisely, an upsurge in the foreign aid of 1 percent promotes economic growth, via increasing the GDP growth rate by 0.660894. This finding also empirically confirms that from several earlier works, for example, (Fasanya & Onakoya, 2012; Gounder, 2001; Hansen & Tarp, 2001; Islam, 1972; Loxley & Sackey, 2008; Tra, 2014). Another control variable is DI, which has a positive, and significant (p < 0.01) consequence on GDP, indicating a 1 percent gain in domestic investment will upsurge GDP by 0.441599, and with which similar work has been found close to that of Tendongho (2016). Finally, the empirical outcome also shows that TO be positive, but statistically inconsequential. The reasons could be internal blockers that increase trade costs, weak overland connectivity, low human capital index, technology, and institutional inefficiency.

Table 4: ARDL of GDP Model (1, 0, 0, 1)

Notes: (i) ∆ signifies the first difference operator. (ii) *** indicates the p-value is statistically substantial at 1% leve

Table 4(II) divulges the regression outcomes of the short-run ARDL model of F_GDP(GDP|FAID, DI, TO). All explanatory variables demonstrate a statistically significant (p < 0.01) positive association with GDP. One important thing is that, unlike in long-run, TO appears to perform a noteworthy character in the nation’s economic progress in the short-run. This study uses ECT to observe the short-run affiliation between the variables and confirm the reliability of long-run results. The coefficient of ECT is negative and statistically meaningful (p < 0.01), signposts that if there is any change in GDP from the short-run to the long-run, it is adjusted by 92.28% percent every year.

This study confirms the non-existence of serial correlation (Table 5), as the p-value of the x² of LM Test is 0.9021, which is greater than 0.05. Besides, the heteroskedasticity test also supports the absence of heteroskedasticity and backing the notion of normally distributed errors as here, the probability value of x² (0.1411) similarly exceeds 0.05.

Table 5: Results of Diagnostic Test

Lastly, the constancy of the model is confirmed through the CUSUM and CUSUM-SQ assessments. Figure 1 shows the CUSUM and CUSUM-SQ test plot, where, in both cases, the blue lines reside between the two red lines. As they do not cross the critical value of a 5 percent level of significance, it confirms the model’s firmness.

Figure 1: Plot of the CUSUM Test and CUSUM-SQ Test

5. Conclusion

This work intends to examine foreign assistance’s influence on Bangladesh’s economic progress from 1989 to 2018 using the ARDL method. Foreign aid was the independent variable that was proxied by the net ODA. Additionally, two other control variables (viz. domestic investment and trade openness) were also involved in the model. Economic growth, the dependent variable in this review, was surrogated by the GDP growth rate. The study’s emphatical evidence finds that aid and domestic investment are two substantial determining factors, both in the long and short run, of Bangladesh’s economic advancement. However, while trade openness may not yield a weighty character in the country’s economy in the long run, plays a vital role in the short run.

The outcomes of this investigation have consequences for strategy. Such results are, for example, beneficial to the government of Bangladesh, and especially to the policymakers in economic relations division (ERD) of the country as they are responsible for determining the need for foreign aid and signing of grants and loan agreements with different foreign donors. It may be mentioned that this study is conducted at an aggregate level, and the relationship between foreign loans and grants separately on economic growth was not considered. Therefore, further studies could be undertaken to overcome these limitations. However, it is especially recommended that as foreign aid is playing a significant positive role in the economy, it can be accepted. Nevertheless, long-term aid dependency cannot be acceptable in any sense, and it hinders the achievement of sustainable economic growth. So, policymakers should boost export and investment and develop a fair tax system to cut aid dependence.