Getting a handle on the COVID19 shock - Forecasts are sobering

We know the current pandemic/lockdown is bad, but using that words gives us little information. The question that needs to answered is how bad the shock will be and how long will this last. We know there is a lot of research from Wall Street firms, but we also want to highlight some of the work by academics to show how they approach this forecasting problem. While I may not completely agree with their methodology, their viewpoints provide a useful framework for consideration. 

In the new paper, "COVID19 and the Macroeconomic Effects of Costly Disasters" NBER working paper #26987, the authors use the times series from costly and deadly disaster to create a VAR model that can measure the impact of a shock on industrial production, services, and other economic variables. The data were collected from NOAA and the Insurance Information Institute. The assumption is that the economic response with all these shocks have some commonality.

Sizing the size of the shock as a number of standard deviations from normal can then be translated into macro variables. The authors make a reasonable assumption that the Great Lockdown will be a 3-month 60 standard deviation shock which translates into cumulative 10-month loss of 12.75% in industrial production and an employment loss of 17% or 24 million jobs before the recovery starts. The situation is even worse if a non-linear model is used.

This is the first shock where there is a shutdown of national production not associated with a shock to capital stock like a hurricane or terrorist attack. The global impact is not included in this model.

These are scary numbers that are not consistent with the behavior of financial markets. It is unclear whether the policy responses will be able to arrest this decline. While this is all in the realm of speculation, our attention should be focused on these possible scenarios.

See also Uncertainty and the COVID-19 recession - It is all about the uncertainty