Christian Dippel, Robert Gold, Stephan Heblich, Rodrigo Pinto, 12 April 2017

Finding exogenous variables to establish control mechanisms is difficult outside of randomised control trials. This column shows that under certain circumstances, it is possible to separate out the causal effect of an unknown variable on the observed and unobserved variables. When applied to trade exposure and voter sentiment for populist parties, the model is largely accurate and gives the surprising finding that 170% of the total effect of trade exposure on populist voting is explained by labour markets, meaning that trade exposure’s other effects on voting – those that do not run through labour markets – are politically moderating.