TWO CENTURIES OF PRESIDENTIAL PUZZLES

Abstract

Contrary to an extensive literature, I show that the presidential puzzle, whereby the market performs better under Democratic presidents, reverses prior to 1927. The political cycle, whereby the market performs significantly better in the second half of presidential terms, does not hold in sample or in the pre-1927 period. Both patterns appear to be artifacts of small samples. In line with Pastor and Veronesi (2020), I show that the presidential puzzle instead reflects when a president is elected: presidents elected under high risk aversion experience higher subsequent returns, a pattern associated with Republican presidents before 1927 and Democratic presidents thereafter.

Keywords: Presidential puzzle, political cycles, out-of-sample tests, long-run asset pricing. JEL Codes: G12, G14, D72