Instrumental Variables and Causal Mechanisms: Unpacking the Effect of Trade on Workers and Voters

Abstract

Instrumental variables (IV) are a common means to identify treatment effects. But standard IV methods do not allow us to unpack the complex treatment effects that arise when a treatment and its outcome together cause a second outcome of interest. For example, IV methods have been used to show that import exposure to low-wage countries has adversely affected Western labor markets. Similarly, they have been used to show that import exposure has increased voter polarization. However, standard IV cannot estimate to what extent the latter is a consequence of the former. This paper proposes a new identification framework that allows us to do so, appealing to one additional identifying assumption without requiring additional instruments. The added identifying assumption can be relaxed, and bounds instead of point estimates can be derived. Applying this framework, we estimate that labor market adjustments explain most to all of the effect of import exposure on voting, thereby providing rigorous evidence that the correct policy response to voter polarization has to be focused on labor markets.