Making Candidates Count: Strategic Coordination under Brazilian OLPR

"I have come to the conclusion that politics is too serious a matter to be left to politicians."
Charles de Gaulle, Brigade general and founder of the Fifth French Republic, circa 1950.

Abstract: Democracy is decided on numbers: we count heads to not decollate heads. Open-list proportional representation (OLPR) is often regarded as the most fragmented and personalistic of electoral systems — with Brazil as its most extreme case. Canonical accounts hold that coordination collapses in high-magnitude ($M$) districts, where weak party control, vote pooling, and informational constraints undermine strategic behaviour. This thesis challenges that view. I argue that coordination under OLPR is not only possible but observable — driven by viable candidates seeking to expurgate intra-list rivals, and by parties systematically allocating resources to those they prefer to see elected. District magnitude is cardinal: in low-$M$ districts, high thresholds sharpen intra-list hierarchies; in high-$M$ settings, rank volatility heightens incentives for pre-electoral filtering. Drawing on an original candidate-level panel covering all federal legislative races from 1998 to 2022, I combine descriptive analysis, fixed-effects specifications, and a regression discontinuity design to show that, despite enormous candidate pools, votes systematically concentrate around a small subset of viable contenders — those whom parties and elites support and fund before election day. From 2002 to 2014, barely elected candidates received up to R$98,000 more than their closest losing co-partisans. These patterns reveal how elites signal viability ex-ante, narrowing real competition to a select few. Brazil’s OLPR system, far from chaotic, reflects a hidden but decisive logic of partisan and elite coordination before voters even go to the polls.