Covid-Metrics - Econometrics, Time Series, and Risk Management of Covid-19
Abstract
The main goal of the project Covid-Metrics is to use the state of the art of econometric methods, time series analysis, and credit risk management knowledge, to model, estimate and predict the outcomes of Covid-19. The project has three parts. The first part deals with the impact of policy decisions like closing the schools on Covid-19’s outcomes, for instance the number of deaths. Such analysis is done by the medical literature. However, this literature ignores what econometricians and economists call endogeneity of policy decisions. Indeed, closing the schools is not an exogenous decision, it is related to the propagation of the epidemic and, hence, ignoring this endogeneity will create biases in the analysis of Covid-19’s outcomes. The second part of the projects uses the state of the art of time series modeling of economic and financial variables to improve the models used by the medical literature in order to study the dynamics of the epidemic events. An important limitation of these modes is that they assume that some key parameters that drive the dynamics, like the contagion parameters and the transition probabilities, are constant and deterministic, independent of time, sometimes independent of individual characteristics, and do not depend on common factors. These factors could be observed that like the state of the health system or unobserved like the fear of the population which could lead to the change of their decisions, for instance self-containment. The main goal of the second part is to relax these assumptions and provide dynamics of the outcomes that describe better the reality and provide better forecast of the outcomes. This part will also extract the true number of contaminated persons by using filtering techniques. The third part is aimed at characterizing the optimal public investment policy in intensive care units (respirators, stock of masks, …) to anticipate pandemic crises when their frequency and intensity are not known in advance. We will adapt the models of decision theory under uncertainty to risk aversion, ambiguity aversion, prudence and a scientifically-based interpretation of the precautionary principle.