Factors Affecting Corona Deaths in Sri Lanka: Time Series Modeling Approach

Abstract

Whole world has been affected by COVID-19 Pandemic which kills people on a large scale. Identifying, controlling and taking preventive actions for the factors that cause such deaths is crucial. This work intends to investigate the factors affecting COVID-19 deaths reported in Sri Lanka, during the period of 2020 to 2021 by using Vector Auto Regressive model. The empirical results of the model indicated the factors that significantly affected COVID-19 deaths short term as well as long term. Short term, factors such as increase in reported new cases in the previous day, positive number of test results, additional hours per day spent at residence compared to the median value of duration stayed at residence from 3rd January to 6th February 2020(difference between the actual hours and median hours spent at residence has been considered), number of new visitors to outdoor places and a decrease in previous day’s deaths. In a long term forecast, variables such as reproduction rate, new vaccination doses, stringency index, additional time spent at residence, new users of public transport, new users of retail and recreation and new visitors to outdoor spaces significantly influence on the mortality. The Granger Causality test confirmed the past values of new cases and positive number of tests have a predictive ability in determining the present values of deaths. On the other hand, the Variance Decomposition method indicated that the variation in deaths in short term is due to deaths itself.