Trade imbalance and currency devaluation in Uganda.(1989-2019)

Abstract

The main objective of the study was to analyse trade imbalance and currency devaluation in Uganda, using time series data from 1989 to 2019. The study sought to achieve three study objectives; to examine short-run relationship of currency devaluation in Uganda, to ascertain long-run relationship of currency devaluation in Uganda, and to assess causal relationship of currency devaluation in Uganda. Secondary data was employed in the study, got from reliable sources; Bank of Uganda and World Development Indicators. First and foremost, univariate analysis showed that the mean and median of all variables were good measures of central tendency since these values lied in between the minimum and maximum values, for example, the mean of currency devaluation (1927.174), and the median (1781.13) both lied mid-way the lowest (223.09) and highest (3713.35), and as well the standard deviation (961.6953) was a good measure of dispersion since its value was small, implying that the data points in the variables do not over deviate from the central value/mean. Bivariate analysis also showed that correlations for all variables were statistically significant since their respective p-values were smaller than 0.05. At bivariate stage, the natural log of imports and currency devaluation had a high positive significant relationship (R-value = 0.8769). The natural log of exports and currency devaluation had a very high positive significant relationship (R-value = 0.8815) and government debt and currency devaluation had a poor negative significant correlation relationship (R = -0.4256). The Augmented Dickey-Fuller test was used to test forstationarity of the time series data, and all variables; currency devaluation, natural log of imports, natural log of exports and government debt had MacKinnon approximate p-values for Z(t) for equations 1 and 2 as (0.8962 and 0.6961), (0.9359 and 0.6912), (0.892 and 0.7041) and (0.7627 and 0.9215) respectively which were greater than 0.05 level of significance, proving non-stationarity at level I (0). And after first differencing, all variables currency devaluation, natural log of imports, natural log of exports and government debt had MacKinnon approximate p-values for Z (t) for equations 1 and 2 as (0.0002 and 0.0023), (0.0061 and 0.0391), (0.0004 and 0.0021) and (0.0422 and 0.1704) which were less than 0.05 level of significance making all variables stationary at I (1) apart from equation 2 of government debt. The data was normally skewed, neither heteroscedasticity nor autocorrelation existed in the time series. The Johansen test of co-integration showed existence of a long-run relationship. And so proceeded with the Vector Error Correction model (VECM) which was used to determine if there existed a short-run and long-run relationship, and Granger causality if there existed a causal relationship on currency devaluation. The findings showed that; there existed no short-run relationship on currency devaluation and a long-term relationship existed but couldn’t be explained by the error correction term (-0.7027408), since its probability value (0.277) exceeded the 0.05 level of significance. The study recommended that policy makers should be considerate while implementing currency devaluation policy to ensure it does not lead to galloping inflation.