Adding Structure to Statistics: A Study on COVID-19 Dynamics in India

This study pertains to COVID-19 in India, and begins by uncovering the statistical relationship between three time series-number of cases, number of deaths, and number of tests each day, using structural vector autoregression. Further, impulse responses of the before-mentioned series are studied. Effect of temperature and humidity on number of cases is analysed using the fixed effects model on city-level panel data. The next model utilises exponential smoothing for forecasting and conjecture for identifying peak specific to this data is presented. Lastly, multiple iterations of compartmental modelling, possible scenarios, and effect of underlying assumptions is analysed. The models are used to forecast number of cases (regression for short term and epidemiological for long term). In the end, policy implications of different modelling exercises and ways to check the implication for policy planning are discussed.

Download paper here

Recommended citation: Arundhati Dixit, Sarthak Vishnoi, Sourabh Bikas Paul (2020). Adding Structure to Statistics: A Study on COVID-19 Dynamics in India; medRxiv.