• rmacculloch

The Covid Modelers: some more humility, perhaps?

In the economics profession, we're "experts" at making forecasts that turn out to be wrong. And talking of forecasts, today the total number of virus cases arising from the present outbreak exceeded 1,000. And we're still counting.

Yet on 24 August one of the lead official government Covid modelers told Newshub, "I would say a best case scenario would be something like what we saw in August last year - probably, higher than that. Maybe 200 [cases] might be the best case scenario but it could go as high as 1,000 [cases]. That's still a possibility". And on 8 September, as case numbers started to soar, the modelers said, "We're still forecasting around about 1,000 cases in total".

Hang on. "Still"? But their original forecast wasn't given as a point estimate of 1,000. It was given as a range, from 200 lowest to 1,000 highest, the mid-point of that range being 600.

Meanwhile, back in September 2020, the modelers called the Productivity Commission's cost-benefit-analysis which it did to help work out whether staying at Level 4 was justified on social welfare grounds, a "toy model". Yet this evening, the modelers are quoted as saying, "You could persist at level 4 & stamp this out, I'm pretty sure that'd be the case ... On the other hand, everybody's a bit over level 4, people need a bit of light at the end of the tunnel .. It's probably worth taking a calculated risk."

Oh, I see. So now the modelers are doing their own informal, back-of-the-envelope "cost benefit analysis", based on what appears to be little of the expert knowledge one needs to actually do a proper one, and instead basing their judgement on a hunch that maybe the people are "a bit over level 4" and need "a bit of light".