Speaker
Description
A key identification assumption required for causal claims of instrumental variables (IVs) is the exclusion restriction. This paper assesses the validity of this assumption for topographic variation as a widely used IV in empirical economics. A systematic review of leading economics journals identifies 56 different variables that the reviewed literature causally links to topographic variation. We interpret this as strong indication for a problematic prevalence of confounding variables in most topography-based IV studies. While we find that it is common in the literature to control for some of these variables, we point out that this leads to violations of another key identification assumption, the unconfoundedness of the instrument. As an example, we assess in more depth the interwined causal relationship of roads and electrification for Dinkelman (2011), a seminal study that popularized the use of topographic variation IVs. We apply tests to gauge the strength of potential exclusion-restriction violations and discuss common issues in the literature, including IV weakness in the presence of potential exclusion-restriction violations. Our findings suggest a higher standard in transparently discussing, systematically testing, and ultimately choosing to use topographic variation as an IV.
