ARE FOREIGN AID AND ECONOMIC GROWTH POSITIVELY RELATED? EMPIRICAL EVIDENCE FROM BANGLADESH

As an emerging country, the progress of Bangladesh is highly promising. Foreign aid may be one of the key players fueling such advancement. The note is an attempt to examine the effect of foreign aid on the economic growth of Bangladesh. With a view to fulfill our objective, the study employs annual time series data during the period of 1971 to 2019. It has used some econometric tools i.e., Unit Root Tests and OLS Methods to process the collected data. The dependent variable is Gross Domestic Product (GDP) while other independent variables i.e., Foreign Aid (ODA), Gross Capital Formation (GCF), Population (POP), and Education (EDU). The test results confirmed that GDP growth is positively related to foreign aid, gross capital formation, and education, but negatively related to population. In Bangladesh, if foreign aid increases by 1% then the GDP growth, gross capital formation, and education rate will accelerate to 0.1988%, 0.6015%, and 0.0652% respectively. So it is evident that foreign aid plays a propitious role to progress the economic growth of Bangladesh. It is a crying need to swell up effectiveness, transparency, proper accountability in allocation, and stronger management of aid inflows to speed up economic growth.


INTRODUCTION
country's policies? Employing the Burnside-Dollar policy index with two other broader measure, he noted that good policy has no effect on aid and alleviation of poverty in developing countries. A more formal attempt by Ghura et al. (1995) to detain the probable side effect of aid, i.e., "Dutch-Disease effect" and some policy variables to economic growth with the more sophisticated model sampling 41 countries from 1986-92 found the positive results with enormous evidence. A different conclusion of Burnside and Dollar (1997) was that aid effect policy variables strongly, but the ratio of aid-GDP does not significantly influence on the economic growth in the LDCs. Mosley (1980) reported a weedy inverse relation on aid-growth nexus developing a simultaneous equation model, but he got positive and significant statistical results from the poorest countries' sample. The test results of Boone (1996) reported that foreign aid has no effect on the investment growth as well as income to uplift the poor one rather than make healthier the political elites because they are receiving benefits continuously irrespective of liberal democratic, or highly repressive, suggesting the effectiveness of a short term aid targeted the program for LDCs. After having a wider sample of 120 countries, Lohani (2004) unveiled the relationship between foreign aid and development. Incorporating knowledge index, heath index and standard of living index for measuring development, this paper captured a negative link of aid-growth prescribing that the government should emphasize largely on foreign direct investment and domestic investment to boost up growth. Moreira (2005) addressed foreign aid enhances growth in the developing countries analyzing macroeconomic data for 48 developing countries from 1970-98 arguing the importance of time lags effective in short-run rather than long-run. Chheange (2009) documented aid triggers corruption rather than growth, even using panel data sampling 67 developing countries from 1986 to 2005 suggesting two lessons: avoid aid or utilize it for good governance by the recipient countries. Analyzing cross country data, Knack (2001) noted foreign aid corrodes the institutional quality of the public sector stimulating rent seeking behavior and corruption. Heavy aid dependency impedes good governance through raising bureaucratic complexity and misusing laws. Tait et al. (2015) studied aid-growth nexus on 25 Sub-Saharan African countries during 1970-2012. By testing the fixed effect model, they showed aid, in the form of grant, extend the economic growth positively only in the long run. Sahoo (2016) uncovered similar results studying the South Asian countries, especially for Sri-Lanka, India and Pakistan applying co-integration test and vector error correction model. By contrast, Fatima (2014) cast doubt to link up the issues for Pakistan both in aggregate and disaggregate level.
The above studies indicate that the aid-growth relationship is yet inconclusive that should welcome us to continue further research.

OBJECTIVES OF THE RESEARCH
 To identify the relationship of GDP growth and foreign aid.  To verify the effect of gross capital formation (GCF), population (POP) and education (EDU) on GDP of Bangladesh.  To recommend some policies for ODA, GCF. POP and EDU to strengthen economic growth.

Stationary Test Augmented Dickey-Fuller (ADF) Test
The unit root test has been applied to verify the data used in the study either stationary or not. The ADF test has been used for this purpose. If the data is stationary only in that case results are more reliable, but if time series data is not stationary, then the results will no longer be valid. Table 1 reports results of the Augmented Dickey-Fuller (ADF) test. The t-statistics are to be calculated for both intercept and, trend and intercept cases. The hypotheses are:H 0 : The variables have unit root, i.e., non-stationary, H 1 : The variables have no unit root, i.e., stationary. If t-statistics are greater than ADF critical values or p-values are greater than 5% level of significance, then the null hypothesis can't be rejected. That means the variables have unit root or the variables are non-stationary. On the other hand, if t-statistics are less than ADF critical values or p-values are less than 5% level of significance, then we can reject null hypothesis. Therefore, the unit root does not exist in the variables. So the variables will be stationary. Here computed ADF test statistics 0.114 is for GDP at level with intercept which is less than the critical values and p-value is not significant at the 5% level. That dictates the null hypothesis could not be rejected. So, it has the unit root problem and series are non-stationary at level. When the series turns into first difference, then the value (-7.333) becomes significant at 5% level and turns into stationary. Similarly at level with trend and intercept, the test statistics -3.277 is not significant but at first difference series becomes stationary.
In the same way, the computed ADF test (at level with intercept) statistics are -2.688, -0.164, -0.412 and 1.038 for ODA, GCF, POP and EDU respectively. All these values are insignificant at 5% level of significance. When they transformed into first difference, then all the variables become stationary. In case of the trend and intercept all the selected variables are stationary at the first difference but not level form.

Phillips-Perron (PP) Test
To testify unit root, an alternative test was suggested by Phillips in 1987 that was modified by Perron in 1988, and both Philips and Perron in 1988. Actually, it is a non-parametric statistical test that considered serial correlation of error terms without lagged difference terms. The asymptotic distribution of PP test is analogous to ADF test statistics.  Table 2 represents results of Phillips-Perron tests for five variables-Gross Domestic Product (GDP), Foreign Aid (ODA), Gross Capital Formation (GCF), Population (POP) and Education (EDU) in logarithmic form. The hypotheses are:H 0 : The variables have unit root, i.e., non-stationary, H 1 : The variables have no unit root, i.e., stationary. If t-statistics are greater than PP critical values, then we can't reject null hypothesis. That mean the variables have unit root or the variables are non-stationary. If t-statistics are less than PP critical values or p-values are less than 5% level of significance, then we can reject null hypothesis. So the variables become stationary. Computed PP test statistics are not statistically significant for all variables at level irrespective of both cases i.e., intercept or trend and intercept. The variables contained unit root at level form, but when they are transformed into the first difference then they become stationary. Table 3 represents the results of multiple regression analysis. It has been seen that the R 2 value 0.997110 indicates that there is more than 99% variation of the dependent variable is caused by independent variables. Again the p-values for the most of the variables are less than 5% that indicates the coefficients are statistically significant. In addition, the F-statistic 2587.646 is also significant at 5% significance level. All these characteristics indicate that the econometric model is fitted well. Using the value of the coefficients, the econometric model could be expressed as follows: The regression results indicate that GDP is positively related with ODA, GCF and EDU, but negatively related to POP. If foreign aid increases by 1% then the GDP growth will accelerate to 0.1988% in Bangladesh. In the same way, a 1% addition to gross capital formation and education rate will lead to change GDP growth by 0.6015% and 0.0652% respectively. In contrast, if the number of population increases to 1%, it will negatively affect the GDP growth by 0.2798% in Bangladesh.

Diagnostic Tests Normality Test
From the Figure 1 it is to be seen that the probability value of Jarque-Bara statistic is larger than 5% which is 0.6057. So the null hypothesis of residuals follows normal distribution could be accepted. Therefore, residuals of this model follow normal distribution.  Table 4 reported the results of Breusch-Godfrey serial correlation LM test. The observed Rsquared value is 1.381875 and p-value of Chi-Square 0.5011 is greater than 5%, indicating that there is no serial correlation.  Table 5 reported results of the Breusch-Pagan-Godfrey heteroskedasticity test. It can be noted that the observed R-squared value is 3.351028 and p-value is 0.5009>0.05 indicating the residuals have no heteroskedasticity problem.

Stability Test
To test out the stability of our model it is better to perform the cumulative sum of recursive residual (CUSUM) test. Following the Figure 2 it has been seen that the recursive error falls between the two critical lines that indicate the estimated model and parameters are stable during the sample period in our study.

CONCLUDING REMARKS AND RECOMMENDATIONS
The study explores the linkage of foreign aid and economic growth in Bangladesh using OLS method. Specifically, it attempts to find out whether the aid and growth are positively or negatively related. Both ADF and PP tests have been employed to verify the pattern of time series data either stationary or not. Data for all the variables i.e., Gross Domestic Product (GDP), Foreign Aid (ODA), Gross Capital Formation (GCF), Population (POP) and Education (EDU) are not stationary at level. After transforming into the first difference they changed into stationary which is a precondition of regression analysis. The regression model is fitted well passing normality, serial correlation, heteroskedasticity and stability tests and confirmed that GDP growth has positive association with foreign aid, gross capital formation and education, but negative association with the population. It is to be noted that a 1% increment in aid promotes the economic growth around 0.1988% based on sampling data. However, the study is suggesting that Government should focus on the proper allocation of aid funding through transparent and effective management. Additionally, an aid effectiveness program launched by UNDP could be strictly followed. However, how the foreign aid impacts in the development of specific sector could be a key concern for further research.