Forecasting Global Reported Cases Of Coronavirus

Healthcare risk isn’t usually the leading factor in economic analysis, but the rise of coronavirus (Covid-19) has changed the focus and the calculus. Covid-19, in short, is on everyone’s short list for modeling the near-term macro outlook, in the US and around the world. At some point this, too, shall pass and something approximating a “normal” environment for economic analysis will return. Meantime, the uncertainty surrounding Covid-19 is shredding the usual macro modeling efforts. With that in mind, here begins a preliminary and evolving effort by for projecting the near-term path of cumulative reported cases of Covid-19 on a global basis, using a proprietary econometric forecasting methodology.

The basic procedure: generate forward estimates for the next 20 days based on the average of eight models. Academic and empirical research shows that using averages (or something comparable) via several models tends to provide a higher degree of accuracy (or smaller errors) through time compared with using any one model.

You can find a list of the models used in this project here. Note that one of the models (vector autoregression) isn’t used for forecasting Covad-19. The reason is that the dataset (published by Johns Hopkins) is univariate and vector autoregression is a multivariate application. The full set of nine models is normally used to forecast a range of key economic indicators—the 1-year change in US GDP, as shown here, for example.

The current projection points to an ongoing rise in cumulative Covad-19 cases over the next 20 days. When I ran the numbers a few hours ago, the last reported data point was for Feb. 26:  81,397. The model projects the total rising 29% to 105,310 by Mar. 17, based on the point forecast.

Note the band of uncertainty at the 95% confidence level that accompanies the point forecasts. Keep in mind, too, that the analysis presented represents a first approximation of what may lie ahead. In other words, all the usual caveats apply regarding forecasting–and then some for such a slippery subject as a new virus that the world is still struggling to understand and manage.

1 2
View single page >> |

Disclosure: None.

How did you like this article? Let us know so we can better customize your reading experience. Users' ratings are only visible to themselves.


Leave a comment to automatically be entered into our contest to win a free Echo Show.