Panel Vector Auto-Regressive Model For COVID-19 Infected Cases and Deaths

Abstract

Abstract This study aims to model the dynamic relationships between the number of COVID-19 infected cases and deaths in all the districts of Kerala state, India, from January 2021 to December 2021 based on the panel vector auto-regressive model. The random effect panel vector auto-regressive model of order two was found suitable to model dynamic relationships. This model explains 62 % variations in the endogenous variable, deaths (number of deaths). The exogenous variable deaths (-1) are highly significant, whereas the exogenous variable cases (-1) are significant at a 5% level. Both of these exogenous variables positively influence the endogenous variable. The other exogenous variables, viz., deaths (-2) and cases (-2), are non-significant. The Durbin-Watson test statistic value confirms the independence of the residuals, and the Wald test confirms the validity of the significance of the estimated regression coefficients. JEL classification numbers: E18, HO, I1, J64, J88. Keywords: Fixed and Random Effect Models, Panel VAR model, Cointegration test, Levin-Lin-Chu unit root test, Granger causality test, Hausan test, Wald test.