Join us for an ERIM BOM seminar.
Abstract
The incumbency advantage, the electoral benefit from holding office, is widely estimated using regression discontinuity designs (RDDs) in close elections. Existing estimates typically pool first-term and multiple-term incumbents. We show that this pooling attenuates RDD estimates when candidate characteristics contain persistent shocks, such as quality: a multiple-term incumbent who barely wins has already revealed low quality, fully absorbing the incumbency premium. Using U.S. House (1942-2008) and Portuguese municipal elections (1976-2025), estimates for one-term incumbents are 13.5 and 13.8 percentage points versus 7.8 and 2.8 pp for multiple-term incumbents. Standard pooled RDD estimates understate the causal incumbency advantage from first-time officeholding by 20-30 percent.
