Bangladesh Bank Training Academy Journal

Abstract

The study empirically examines the macroeconomic determinants of Foreign Exchange Reserve (FER) in Bangladesh using Auto Regressive Distributed Lag (ARDL) Model by using the data for the period of 1980 to 2024. Prior to estimation, the stationarity properties of the variables are examined using Augmented Dickey-Fuller (ADF), Phillips-Perron (PP) and KPSS tests. The results indicate that the variables are integrated of order I (0), I (1), or a mixture of both, with no evidence of I (2) behavior, thereby supporting the use of the ARDL framework. The bounds test confirms the presence of a long-run relationship among Foreign Exchange Reserves (FER), Personal Remittances (PR), Foreign Direct Investment (FDI), Foreign Loans and Grants (FLG), and the Current Account Balance (CAB). The long-run estimates show that foreign loans and grants and the current account balance are positively and significantly associated with Bangladesh’s foreign exchange reserves. By contrast, personal remittances and foreign direct investment show positive but statistically insignificant long-run coefficients in the preferred specification. The error correction term is negative and statistically significant, indicating adjustment toward long-run equilibrium after short-run deviations. Diagnostic and stability tests suggest that the model is free from serial correlation, heteroskedasticity and major functional-form problems, and remains stable over the sample period. The findings suggest that strengthening the current account position and reducing excessive reliance on debt-creating inflows are important for sustainable reserve management in Bangladesh.