Do Shifts in Late-Counted Votes Signal Fraud? Evidence From Bolivia

Abstract

Surprising trends in late-counted votes can spark conflict. When late-counted votes led to a narrow incumbent victory in Bolivia last year, fraud accusations followed—with dramatic political consequences. We study the pro-incumbent shift in vote share as the tally progressed, finding that we can explain it without invoking fraud. Two observable characteristics, rurality and region, account for most of the trend. And what looked like a late-breaking surge in the incumbent’s vote share—which electoral observers presented as evidence of foul play—was actually an artifact of methodological and coding errors. Our findings underscore the importance of documenting innocuous explanations for differences between early- and late-counted votes.