Time series analysis for modelling earnings through tourism and workers' remittances in Sri Lanka

Abstract

Sri Lanka is considered as a middle-income country with a negative trade balance and the main foreign cashflows through workers’ remittances, exports, foreign direct investments, and tourism. These key sources play a major role in the Sri Lankan economy and the COVID-19 pandemic became a massive threat because Sri Lanka is hugely dependent on foreign reserves. This study was conducted to analyze the characteristics of workers’ remittances and earnings from tourism before and after the COVID-19 pandemic, then to forecast workers’ remittances and tourist earnings to get some idea of loss of foreign income and its relationship with economic growth. The official web database of the Central Bank of Sri Lanka was used as the data source for modeling workers’ remittances and tourist earnings (monthly 2010-2022). To identify the relationship between workers’ remittances and tourist earnings to economic growth, yearly data from 1998 to 2018 was extracted using annual reports of the Central Bank of Sri Lanka and Sri Lankan Tourism Development Authority statistical reports. EViews and Minitab software were used for the analysis and the ARIMA Box- Jenkins approach was followed to find the best fitted models. SARIMA(3,1,3)(1,1,1)12 - GARCH(1,0) model was found to be the best-fitted model for forecasting tourist earnings and SARIMA(0,1,1)(0,1,1)12 was for workers’ remittances. Based on the fitted models, estimated earnings loss of US$ 1277, 4973 and 5782 million in 2019, 2020 and 2021 years respectively from tourism. The best-fitted model for workers’ remittances was not enough for predictions after 2019 because the model only contained the moving average process part. It has to be remodeled using predicted values. When considering the relationship between workers’ remittances and tourist earnings to economic growth (GDP), no indication of causation between GDP, workers’ remittances and tourism earnings. To identify the long and short-run relationship and to check the stability of the long-run relationship, the Vector Error Correction Model was fitted. The long-run cointegration equation coefficients were significant and not stable according to the VECM results. Therefore, it was not suitable for policy-making decisions and needed to be tested using other variables.