In our paper, Smith and I investigate Sumner's claim empirically. We define an event of a recession wrought by severely restrictive monetary policy and an aggregate demand shortfall as occurring whenever the growth rate of Nominal Gross Domestic Product (NGDP) goes negative, meaning the overall level of nominal spending in the economy actually falls. This is very rare. It occurred in the United States during the Great Recession, but had not done so previously for a half century.
We establish a statistically significant fall in economic freedom five years, ten years, and fifteen years after the event, while including several controls. Due to our methodology, we had neither Nazi Germany nor the Great Recession in our sample, and given that these were the two primary motivating examples, it means that Sumner's hypothesis held true "out of sample" in some sense.
There's also some interesting discussion of how the Trump phenomenon fits in here. Of course Trump campaigned against free trade, but it's too soon to say how he'll actually govern.