So, Democrats are worried that a partial, V- or check-shaped recovery (e.g. unemployment goes from 3.5% pre-covid to 15% mid-covid to 9% by November) will help Trump. I tend to think this is somewhat misguided, empirically. https://www.politico.com/news/2020/05/26/2020-election-democrats-281470">https://www.politico.com/news/2020...

It& #39;s true that a partial recovery to 9% would be better for Trump than getting stuck at 15%. But 9% is still quite high, and voters could also compare to 3.5% or wherever it had been before. The literature on what time horizon voters use to evaluate the economy is not very clear.

And to an extent this is an intractable question because we don& #39;t have a large enough sample of presidential elections to know exactly (or even all that precisely) how voters evaluate the economy. Models that claim some particular indicator is supreme are bad/p-hacked/overfit.

The best strategy is therefore to use a broad basket of economic indicators over a broad range of time horizons. By that measure, voters& #39; assessments of the economy would likely still be fairly negative on balance by November even with a haflway recovery. https://fivethirtyeight.com/features/measuring-the-effect-of-the-economy-on-elections/">https://fivethirtyeight.com/features/...

If anything, being strictly empirical might push you toward assuming that voters& #39; perceptions of the economy are somewhat lagging, e.g. Bush didn& #39;t get any credit for the improving economy in 1992, but again these findings are based on small samples.

There& #39;s also a question of what this will look like in practice to voters if the improvement in the unemployment rate mostly comes from businesses hiring back after furloughs/layoffs but also a conspicuous number of other small businesses having been forced to close permanently.