Economy · June 23, 2022

“None of the models could identify this effect as other than 0”.

Key phrase buried in the appendix to the Badger Institute study titled: “Unemployment (Over) compensation: How the Federal Supplemental Unemployment Insurance Impacted Unemployment During the Pandemic” (April 2022).

From the executive summary:

Our analysis found that supplemental unemployment benefit (IU) seemed to delay people’s return to work and we estimate that unemployment was 3% to 6% higher due to the federal supplement in the states that kept it in place. until September 2021, which translates into an unemployment rate of 0.2 to 0.3 percentage points higher. If Wisconsin had given up on the supplement in June, total unemployment would have fallen faster than it has been. By September, we estimate, there would have been around 28,000 fewer unemployed.

The authors of this report (h / t Erik Gunn) find this result by exploiting the differences in the timing for the end of the supplemental unemployment benefit.

In general, econometrics are quite well done. The authors control the economic growth (although they do not define the variable), the temporal trend, the participation in the workforce.

However, they do not control public health measures. Furthermore, from a statistical point of view, the inclusion of endogenous variables such as labor force participation, which is likely also affected by the early termination of enhanced unemployment insurance.

But, from my point of view, the central issue is that it is problematic to draw political conclusions based on a crucial parameter estimate that is not statistically significant at conventional levels (whether we should use a lower significance level for political analysis is a legitimate topic of discussion).