The Puzzle of Midterm Loss

This paper examines midterm election results since 1902, in order to account for the regularity with which the party controlling the presidency suffers an electoral decline at midterm. Standard interpretations account for midterm loss in terms of withdrawn presidential coattails (or regression-to-the-mean), surge-and-decline, or a negative referendum on presidential performance. None of these explanations is compatable with the historical data. A "presidential penalty" explanation, however, fits the data nicely. By this explanation, the midterm electorate penalized the president's party for being the party in power: holding constant the presidential year House vote, the president's party does much worse at midterm than it would if it did not control the presidency. Some implications of this finding are examined.