Abstract:
Tea industry is a strong pillar in Sri Lankan economy in terms of foreign earnings and employment. Millions of people are employed directly and indirectly in the tea industry. Therefore, it is important to study about the behavior of tea export earnings and forecast tea export earnings for several months. This study is an attempt to identify predictive models to forecast category wise tea export earnings namely Bulk tea, Tea bags, Tea packets and Total exports using monthly data obtained from Sri Lanka Tea Board covering the period January 2003 to October 2017. The study employed the conventional augmented dickey fuller (ADF) test to test for stationarity among the four variables and Johensen co-integration technique to determine the co-integrating equation. All the series were found to be I(1) and two co-integrating relationships among these series were evident. Hence Vector Error Correction (VEC) model was fitted. For the validation of the VEC model, residual analysis was carried out using Residual plot, Correlogram, Residual portmanteau test for autocorrelation and Serial Correlation LM Test. The results indicated that model was satisfactory. Finally, Impulse Response Function (IRF) and Variance Decomposition (VDC) were employed in order to illustrate the importance of each variable to tea export earnings when a shock is imposed to the system. The analysis revealed that bulk tea earnings positively relate to tea packets earnings but negatively relate to total export earnings whereas tea bags earnings negative relate to tea packets earnings in long run. The results showed that tea packets earnings significantly and directly affect both bulk tea and tea bags earnings in the short run. The study also generated an out-of-sample forecast to analyze and compare the statistical results in order to determine the accuracy of the fitted model. The accuracy of the forecasts was tested using MAPE. Therefore, it can be concluded that the developed VEC model can be used to forecast tea export earnings in Sri Lanka with considerable accuracy.
