Impact of Foreign Reserve in Economic Growth: An Empirical Study on Bangladesh

Bangladesh maintained its robust performance in development. There have been upbeat exports and remittances. Overall inflation is slow & the current account deficit was reduced by higher exports and lower import inflation, but the decrease in the financial account surplus diluted the effect of the decline in the current account deficit on the overall balance of payments deficit. In the context of the Error Correction Process, this research studied the impact of economic development on Bangladesh foreign reserve assets using data for the period 1980- 2014. The findings show that economic growth is extremely important. The model's estimate, that economic growth and foreign reserves have a positive long-term relationship. The model checked by error correction estimates for error correction is negative and statistically important. In addition, the model showed that economic growth has short-term relationship too. The adjustment speed is more than 60 percent, suggesting that the term for error correction corrects the imbalance of the previous year. Granger causality test confirm that there is long run & short run causality among the variables. The question is whether the accumulation of foreign currency reserves is a required prerequisite for economic development. The empirical findings in this paper indicate that the rise in foreign exchange reserves induces GDP growth, although causality has not been demonstrated in the opposite direction.


Introduction
Bangladesh's economic freedom score is 56.4, making its economy the 122nd freest in the 2020 Index. It has increased by 0.8 point overall, led by a higher score for property rights. In the Asia-Pacific region, Bangladesh is ranked 29th out of 42 nations, and its overall score is far below the regional and world averages. During the last five years, Bangladesh has made steady, although incremental, progress towards greater economic independence. Although its economy has remained stuck in the largely unfree category, its GDP growth has been robust over the same period. A welcoming attitude to foreign investment and restraint on government growth can partially explain the discrepancy. Bangladesh is ranked 29th among 42 countries in the Asia-Pacific region, and its overall score is well below the regional and world averages. Bangladesh has made steady albeit incremental progress toward greater economic freedom during the past five years. Although its economy has remained stuck in the mostly unfree category, its GDP growth during the same period has been robust. The difference can be partly explained by a friendly approach towards foreign investment and restriction on government growth. In order to eventually break into the ranks of relatively free Bangladesh, the government will have to make a concerted, multi-year effort to strengthen the three indices of the rule of law and enable more foreign banks and the best practices they will carry with them to join the country. 25 percent is the top federal tax rate, and 45 percent is the top corporate tax rate. A value-added tax is used with some rates. 9.1% of gross domestic revenue is equal to the aggregate tax burden. Over the last three years, government investment has contributed to 13.7 percent of the country's output (GDP) and budget. The combined number of products and services exported and manufactured is equivalent to 38.2 percent of GDP. The total tariff rate added is 10.7 percent. Layers of non-tariff barriers hinder the complex flow of trade. Government tolerance to foreign investment is smaller than the average. The financial market is also underdeveloped and vulnerable to government intervention. The extent of state ownership and banking regulation is considerable, and the deficits have averaged 3.6 percent of GDP. Public debt is equal to 34,8% of GDP.

Literature Review 2.1 Review of Foreign Direct Investment and Economic Growth:
Foreign direct investment boosts a country for better economic growth and development. Mawugnon and Fang (2020) found by using the Granger Causality Test FDI exhibited a positive relationship, its contribution to economic growth. Understanding the direction of causality between the two variables is crucial for formulating policies that would encourage more private investors. Afzalur Rahman (2015) argued that growth in FDI has not been associated with positive economic growth, FDI was linked with increases in the inflation rate and a negative trade balance. FDI appears to have a small positive, but not statistically significant, impact on GDP growth.

Review of Foreign Reserve and Economic Growth:
In recent decades foreign reserve reached an extended level in many developing countries. Fukuda and Yoshifumi (2007) found that, an increase in foreign exchange reserves raises both liquid and total debt, while shortening debt maturity. It also leads to a decline in consumption, although investment and economic growth may improve when the tradable sector is capital intensive. Kashif1 P. Sridharan2 S. Thiyagarajan (2017) argue that, economic growth has positive impact and dynamic relationship between economic growth and international reserves. Their results reveal that, 1 per cent increase in economic growth will lead 0.16 per cent upsurge in international reserves. A statement from Polterovich and Vladimir (2002), The accumulation of foreign exchange reserves is neither a necessary nor a sufficient condition of economic growth. It may well be that countries that do not accumulate reserves grow faster than others because of better investment climate, better institutions, and greater involvement in international trade achieved through greater openness of their economies even though their exchange rate is at equilibrium level.

Review of Trade Balance and Economic Growth:
To measure a country strength trade balance plays a vital role in economy. Trade balance on economic growth as it is stated that trade is one of the key factors promoting economic growth by facilitating capital accumulation, modernization of industrial structure, technological progress, and institutional progress (Blavasciunaite, Lina Garsviene and Kristina, 2020). Trong, Thu, Van and Thai (2020) evident that, the effect of economic growth on trade balance is negative but insignificant. More specifically, an increase in economic performance will not impact on trade balance.

Review of Inflation Rate and Economic Growth:
High Inflation indicates a country strong economic growth but too much high inflation may not bring good for economy. Because it always reduces the value of money except interest rate is higher. Shapan Chandra (2016) and Behera (2014) said that, there is positive relationship between inflation and economic growth in long run. Adversely Khayroollo (2015) argued that, when inflation exceeds 4 percent, inflation will start to significantly reduce the economic growth rate. Fikirte Tsegaye (2012) found that Inflation was negatively and significantly related to economic growth. It means that inflation has an adverse effect on economic growth.

Methodology 3.1 Sampling Technique
At present, the study allots for intensive review which takes in the area of foreign reserve, economic growth and its impact. Study showcase is chosen for the country of Bangladesh. 29 years' time series data ranging from 1990 to 2019 were collected. There are five variables which have been taken where economic growth is dependent variable and foreign reserve is independent variable proxied Trade balance, FDI and Inflation rate. All the variables are taken in their real value with log transformation.

Model Specification:
The analysis variables chosen include economic growth, foreign reserve, foreign direct investment, inflation rate, Trade balance. Details notation and variables identification sign are shows in table1. Here, Log economic growth is dependent variable, Foreign Reserve, Foreign direct investment, Inflation and Trade Balance are the independent variables. Foreign currency reserve greatly influences the economic growth. FDI increase which reflect positively towards economic growth. In diagnostic Test the unit root has been demeanor to Test stationarity either data has stationary or nonstationary then followed by those result auto correlation Test will placed and developing a regression model to remove the problem of heteroskedasticity meaning that it would be homoscedastic. If all the variables showed in the same order process, they might be co-integrated then we can apply Johansen Test to check either the model is long run or short run process. In error correction will use the most advantage model which is VECM. This model shows cointegrated VAR and granger's representation and lastly robust Test, which is a kind of regression to detect influential observation & which is reported below. D(ECONOMIC GROWTH) = C(1)*( ECONOMIC GROWTH(-1) -0.38*FOREIGN RESERVE(-1) -0.15*FOREIGN_DIRECT INVESTMENT(-1) + 0.07*INFLATION RATE (-1) -0.08*TRADE BALANCE(-1) -4.25) + C(2)*D(ECONOMIC GROWTH(-1)) + C(3)*D(ECONOMIC GROWTH(-2)) + C(4)*D(FOREIGN RESERVE(-1)) + C(5)*D(FOREIGN RESERVE(-2)) + C(6)*D(FOREIGN_DIRECT INVESTMENT(-1)) + C(7)*D(FOREIGN_DIRECT INVESTMENT(-2)) + C(8)*D(INFLATION RATE (-1)) + C(9)*D(INFLATION RATE (-2)) + C(10)*D(TRADE BALANCE(-1)) + C(11)*D(TRADE BALANCE(-2)) + C(12)

Data sources
Both primary and secondary data used in this practice is representative sample of Bangladesh. The main source is form world bank website and secondary data are taken from different publications, journals etc.

Dependent Variable
Economic growth: Economic growth is determined by production of economic goods which is increased in extreme manner and supporting by services in a specific time period. The most popular way to measure the economic growth is GDP and entire economic output of a country take into account and in this study it is figure out as dependent variable. GDP Stands with Gross domestic product simply mean GDP, is the total output of a country including goods and other services that produced by a country. Everything around us is part of GDP. It declares as economic indicator which shows the health structure of a country. Most phenomenon way to calculate the GDP = C+G+I+NX (Consumer spending, Government spending, Investments, Net export).

Independent Variables
Foreign reserve: It simply mean to reserve foreign currency includes with banknotes, bonds, T-bill and other government securities. Authority of this reserve is taking by central bank of its country. There are many reasons to reserve foreign currency but mostly uses to manage their currency value. It builds a country confidence and provide necessary support in the national monetary and exchange rate management policies. Foreign reserve has been considered as independent variable. Foreign Direct Investment: Another factor that influence economic growth which is foreign direct investment. An increasing level of FDI recommended for higher growth of its economy. A country structure can be change vastly by FDI. It influences economic stimulation but mostly deals with development of human capital. Inflation: When it come on mind, firstly we think it discuss about declining of price level but precisely focus on declining purchase power of individuals. It contrasts between currency and goods while one is falling and another is rising. Higher inflation is good but extensive level of inflation can cause damages and it should be control through its contractionary monetary policy. Trade Balance: Without considering all financial transfers, investments and other components of goods and difference between the monetary value of a country's imports and exports over a given time period. Trade balance is very import and beneficial for an economy, where it involved trade policy decisions in the time of positive or negative also the size of the trade imbalance. Economists generally agree that neither trade surpluses or trade deficits are inherently "bad" or "good" for the economy.
Trade Balance is positive (+) > imports and is referred to as a trade surplus. Trade Balance is positive (-) < imports and is referred to as a trade deficit.

Unit root Test
Initial process of this analysis to check whether the data series is stationary or non-stationary. To getting this result ADF (Augmented Dicky Fuller) and PP (Phillis and Perron) Test have been done to verify the co-integration and stationarity of those variables which represent in this model. It is important here to determine the stationary property because if the time series has no stationary the regression result will fictious. Here multiple regression process is done and assumption result shows that all the variable has unit root. Tested at level and at 1 st difference with constant, with constant and trend lastly without constant and trend to evaluate the EG, FDI, FR, CPI and TB. Below table 2 is representation of ADF & PP Test. Prob. 0.1644 0.0000 0.0004 0.0000 0.0000 n0 *** *** *** *** Notes: (*) Significant at the 10%; (**) Significant at the 5%; (***) Significant at the 1%. and (no) Not Significant UNIT ROOT TEST Prob. 0.1644 0.0000 0.0004 0.0000 0.0000 n0 *** *** *** *** *MacKinnon (1996) one-sided p-values. Notes: (*) Significant at the 10%; (**) Significant at the 5%; (***) Significant at the 1%. and (no) Not Significant

Result of Stationary Test
From above Test result there are many ways to check stationarity, here constant at 1 st difference which is intercept has been shown. Variable will be rejected if it is null hypothesis and choose the alternative hypothesis because null hypothesis means that, there is no stationarity among those variables. This Tests the null hypothesis that demand follows a unit root process. Intercept at 1 st difference P < 5%, Data is stationary Here approximate p-value of Economic growth is 0.0142 foreign reserve is 0.0023, foreign direct investment is 0.0003, inflation rate is 0.0000 and trade balance is 0.0001. All the variable shows less than 5%, so reject the null in all these cases and choose the alternative hypothesis. Also, all the observation of Test statistic is smaller than 10% critical value. This confirms that, all the variables are stationary at 1 st difference while doing intercept.

Auto correlation
It suggests when two error terms related to each other meaning that correlation of the same variables between two continuous time intervals. Measurement of this value is determined by lag criteria which is related to time series and observations occur at different points in time. Most phenomenon way to Test autocorrelation is the Durbin-Watson for the 1 st order and for higher order correlation LM Test to be shown. Table 4: Durbin-Watson stat Test From the regression model dependent variable is Economic Growth and FG, FDI, CPI and TB is independent. Normally when data is time series, serial correlation is appearing. Here, Durbin-Watson stat value is 0.978659 (table 4) and R-squared value is negative. which is less than 2, according to DW guideline there is positive auto correlation.  (2) 0.0150 Will check it by second confirmation that higher order of Breusch-Godfrey Serial Correlation LM Test suggest auto correlation or not. Here the probability of chi-square value is 0.0150 (table 5) which is less than 5%, it means this model has serial correlation. That is not good for regression. Reject the null and choose the alternative hypothesis. Now, the process is to remove the serial correlation from the model. 0.000000 Durbin-Watson stat 1.693761 Here Economic-growth is dependent variable and leg 1 of dependent variable shows (table 6) that it is highly significant because probability value 0 which is less than 0.05 and Durbin-Watson stat value increase. Where Rsquared is have 99% meaning that data is nicely fitted because it is more than 60%. Corresponding probability value of f-stat is also highly significant, it is less than 5%. SO, R-squared value is high and probability value is significant concluded that Economic-growth of lag (-1) and other 4 independent variable can explain economic growth jointly because out of these 5 variables 50% of the variables is significant. For good regression model it is required, at least 50% of the independent variable should be significant. This model can be affected by serial correlation and normally heteroscedasticity problem does not appear specially when the time series is natural and it appears when the data is cross sectional. Prob. Chi-Square (2) 0.1707 By creating a new variable and that variable would be lag of dependent variable that is Economic-growth (-1). After having lag of Economic-growth, new model Breusch-Godfrey Serial Correlation LM Test suggest that there is not serial correlation and null hypothesis cannot be rejected according to (table 7). Pro value is 0.1707, meaning that it is more than 5% and this is desirable. Serial correlation or auto correlation now remove from this model.

Heteroskedasticity
Firstly, did the heteroscedasticity Test with absolute value and the estimated variance of the residuals from a regression are dependent on the values of the independent variables that means heteroscedasticity on the model and picture shows that. Heteroskedasticity appears when the error variance has non-constant variance. The variance of the observed value of the economic growth around the regression line is non-constant and now work for removals. Most popular way to remove heteroscedasticity from the model, is to convert all the variables into log, which is known as log transformation and did the transformation in excel from the beginning. R-squared -0.003705 Adjusted R-squared -0.119517 Durbin-Watson stat 0.978659 Here probability value of chi square 0.1260 (table 9), which is more than 5% and couldn't be reject the null hypothesis rather accept null. Meaning that residuals are not heteroscedastic & Economic growth is influenced by all the variables. its homoscedastic after converting the variables into log transformation. Now the model if completely free from heteroscedasticity and can use this model for further Test.

Johansen Cointegration Test
The series transformed non-stationary to stationary after first difference which is done in unit root process. Now it is same order of integration. Under this continuity, the variables are unpredictable and entirely not useless. To verify further the relevance of the model, there is need to Test for cointegration. Here, the output releases two statistics, Trace Statistic and Max-Eigen Statistic.

.05 level **MacKinnon-Haug-Michelis (1999) p-values
The Maximum Eigen value Test (table 11) suggest two cointegrating equations at the 5% level, which confirming the Trace Test. Therefore, these two Tests confirm a cointegrating relationship over sample period. Hence, it is concluded that a long-run relationship exists among those variables. The long-run relationship means that those variables will move together in the long-run while showing divergence in the short-run. In other words, those variables are related to each other in some way. Since the Johansen Test shows presence of cointegration can run the VECM in next step.

Error Correction Test 2.1 Vector Error correction Model
If a set of variables are found to have one or more cointegrating vectors then a suitable estimation technique is a VECM (Vector Error Correction Model) which adjusts to both short run changes in variables and deviations from equilibrium. From Johansen Cointegration Test the result shows that the variables are cointegrated. So, the VECM model can be run. Lag length criteria also suggest the chosen of one lag for estimating VECM. Coefficient of error correction term is the most typical parameter in the estimation of the VECM, which measures the speed of adjustment of economic growth to its equilibrium level. Thus, the expectation is positive and variables on evolution of FDI, Foreign reserve and Trade balance. From Johannsen Test it shows that variables are cointegrated and main objective would be focus on dependent variable which is economic growth. The vector error correction result with long and short run causality under ordinary least square model is showed below. Here C determine Error correction term mean speed of adjustment towards equilibrium. If C is negative in sign and significant, it means there is a long run causality running from foreign reserve, FDI and trade balance to Economic growth.
Here the result of C(1) to C(5) and C(8) shows that, Coefficient is not negative and also not significant because probability value is more that 5% meaning that there is no long run causality running form foreign reserve, FDI and trade balance to Economic growth. Alternatively C(6) has negative coefficient and significant where (7) and C(9) also negative but not significant which provides a mixed indication. The negative speed of the adjustment parameter, it shows how fast the variables go to the equilibrium. -0.001655 0.6649 The error correction C (4), C (5), C (8) and C (9) together they become zero by Wald Test and probability of Chi Square value is more than 5%, it means there is no short run causality also which running from foreign reserve to economic growth. The probability of Chi Square value C (6) and C (7) is 0.0114 (table 12) which is less than 5% and that particular variable becomes significant. So, here the error correction term becomes significant. When the ECT is significant and its sign is negative it means that Trade Balance have short-run causality in economic growth. That means Trade Balance causes economic growth rate in the short-run. Short run effects are captured through individual coefficients of the differentiated terms. Rest of the others are more than 5% which means among those variables there are no short run causality. Value of R-squared is 0.732260 (table 12) which is more than 60% and its well fitted model to use. F-stat and probability value is significant which is less than 5%. Finally, Histogram Normality Test declare that. The residual is normally distributed which is tested by Jarque-Bera result. Note that the kurtosis is more than 3, and the Jarque-Bera statistic has a p-value of 0.8329 (table 13), implying that the data are consistent with a Normal distribution (and constant variance) Table 13: Histogram Normality Test

Granger Causality Test
In stark contrast to the Johannsen test, which is concerned with long-run relationship, the Granger causality test is concerned with short-run relationship between variables. Cointegration and vector error correction have driven both the long and short-term connection of these variables. The path of causality shows whether the FDI influenced the growth of GDP or the FDI affected economic growth, and so on. In the direction of the variable, at least one Granger causality may have occurred. 0.0158** 0.0115** Notes: (*) Significant at the 10%; (**) Significant at the 5%; (***) Significant at the 1%. And (no) Not Significant Between Trade Balance and Foreign Reserve, they have advance lag relationship. Granger causality between them implies that Trade Balance can influence the Foreign Reserve in a unidirectional causality movement. There is also unidirectional causality comes from Economic Growth to Trade Balance at 5% level of significance which means Economic growth has a predictive capacity to assess the present value or ranking of the Trade balance. At 5% level of significance Foreign reserve cause the Economic Growth to change based on previous movement in the variable. FDI causes Economic growth that mean FDI can predict the future movement in Economic growth & has short run relationship.

Robust standard errors
Robust standard errors or White-Huber standard errors techniques are called (heteroskedasticity) standard errors. This method is accurate but completely empirical. On the heteroskedasticity system, couldn't impose any conclusions. The robust standard errors would become only traditional OLS standard errors, even though there is no heteroskedasticity. Particularly under homoscedasticity, the robust standard errors are also adequate. Here coefficient of foreign reserve is strongly negative. High percentage of foreign reserve, the smaller proportion will be the economic growth. The probability of foreign reserve 0.7324 (table 15) which is more than 5% and suggested Foreign reserve is negative impact on Economic growth. FDI seems negative also but its marginal & highly significant. Prediction was negative for Inflation & Trade Balance but the of Robust test shows the positivity of these two variables toward Economic Growth significantly.

Findings and Conclusion
The effect of foreign reserves on economic development has been the subject of relatively few researches. In the case of a particular area in Asia, this research is intended to add to this methodological analysis. Usage of sample data from Bangladesh during the period 1990 -2019. Statistical results revealed an insignificant high positive correlation with economic growth and the Foreign Reserve, a marginal positive correlation with economic growth and inflation and a modest correlation with FDI. The study found there is no negative correlation among them. Therefore, the results obtained in this research suggest that growth in all are the variables have been associated with positive economic growth in Bangladesh. FDI was rarely linked with increases in the inflation rate and a positive trade balance. A higher rate of inflation lowers people's buying power and has been blamed for economic deflation. Additionally, when a country's imports are larger than exports, a negative trade balance exists and this creates a resource deficit for developing nations. FDI, on the other hand, tends to have a strong positive and statistically effect on economic development. Therefore, in order to inspire a trade surplus, Bangladesh should concentrate on exports and exchange rate devaluation policies. Another purpose of this paper was to econometrically analyses and examined the causal relationship among certain economic indicators in Bangladesh by using VECM model. Which contributes to making understood cointegrating and causal relationship between foreign direct investment, economic growth, FDI & Inflation in Bangladesh case. Which is the Johansen approach to cointegration and the procedure for Granger non-causality test. VECM was used for regression model and run it in order to test for the presence of a long-run relationship between variables. Form Johansen Test it is found that all the variables are cointegrated & there is no long run & short run causality running form foreign reserve, FDI and trade balance to Economic growth While, in the case of economic growth by Granger Causality Test, there is unidirectional linkage running from economic growth to trade balance with positive sign, which means that economic growth positively affected by trade balance. In comparison, a higher level of trade balance can definitely be reached by a nation with higher levels of domestic income. The accumulation of foreign reserves is neither an imperative nor an adequate condition for economic development. It may well be that countries that do not accumulate reserves grow faster than others because of better investment climate, better institutions, and greater involvement in international trade achieved through greater openness of their economies even though their exchange rate is at equilibrium level. It could also be the case that countries accumulating reserves are not able to increase their investment/GDP ratios since the export externality is not strong enough or due to poor investment climate. A significant macroeconomic mechanism for rising long-term economic growth rates is the accumulation of foreign exchange reserves. This would lead to a reduction of the exchange rate, a rise in export sector revenues and profits, an increase in investment and economic activity dependent on higher exports. Other macroeconomic indicators (such as unemployment rate, gross national product, parity of purchasing power, poverty level, and foreign exchange rate) are recommended for future analysis, which can help to better understand the effects of foreign reserves on Bangladesh's economic development.