Seminal studies show that naive lab participants accurately predict who wins real world elections based solely on candidate photos. It is unclear what this implies for the health of democracy without knowing whether candidates who look more electable or competent in photos actually behave more competently in office. Combining novel data on politician performance with lab-in -the- field experiments, I show that voters can identify which politicians divert less public money and communicate more persuasively based solely on their headshots. Such inferences do not predict politician effort convening public meetings, but neither do other readily available metrics like professional qualifications. I run these experiments in a low income country where candidate photos appear on the ballot and weak institutional checks raise the stakes for selecting innately competent leaders.
