[There is]... no evidence that tariff and non-tariff barriers rise systematically during cyclic downturns.
Conventional wisdom holds that protectionism is counter-cyclic, meaning that tariffs and quotas become more prevalent during recessions. In Protectionism Isn't Counter-Cyclic (Anymore) (NBER Working Paper No. 18062), Andrew Rose shows that while that may have been a valid description in the period before the Second World War, it is now inaccurate.
Rose motivates his study by presenting scatter plots that show no obvious correlation between business cycles and protectionism after World War II. For the United States, a positive relationship between the unemployment rate and tariffs is evident between 1906 and 1942, but this relationship is strikingly reversed between 1946 and 1982 when high U.S. unemployment seems to coincide with low U.S. tariffs. Moreover, there does not seem to be a correlation between the level of economic growth and the number of commercial disputes initiated under the GATT/WTO dispute settlement system. Interestingly, the "Great Recession" of 2009 was associated with a collapse of global growth but there was no coincident uptick in trade disputes.
Using a panel of data covering over 60 countries and 30 years, as well as eighteen measures of protectionism and seven measures of business cycles, Rose finds no evidence that tariff and non-tariff barriers rise systematically during cyclic downturns. In fact, the message seems to be that protectionism is essentially acyclic. The results are robust across all measures of protectionism and business cycles, with various controls and different estimation methods.
Rose also looks at the pre-war period by constructing a historical panel of data going back 140 years. He shows that protectionism was probably counter-cyclic before World War II, but it is hard to conclude that definitively because of the limited availability of reliable data for that period.