Ferejohn: Incumbent performance and electoral control

Ferejohn. 1986. Incumbent performance and electoral control. Public Choice 30 (fall): 5-25..

In Brief

Voters can't trust anybody's campaign promises. Thus, challengers don't influence voters: voters simply have a referendum on the incumbent. They use retrospective voting: they vote based on what the incumbent has done for them lately. If they don't like the incumbent, they vote him out, effectively selecting a new representative at random from a pool of equally untrustable challengers. For this to work, voters must not vote selfishly--they most employ "sociotropic voting": "that is, voting based on an aggregate criterion." Elections are for sanctioning moral hazard, not for preventing adverse selection.

The Model

1. Rational, self-interested voters (and candidates) and an electorate with homogeneous preferences (i.e. sociotropic voting).
2. The officeholder is imperfectly monitored by the electorate.
3. Candidates cannot make credible campaign promises because voters assume that they will act self-interestedly once in office. Thus, the challenger can be anybody, and his appeal is merely his presence. Elections are just a referendum on the incumbent.
4. Given all this, voting is purely retrospective (i.e. strong economy, avoidance of major wars). However, since voters cannot perfectly monitor the incumbent, they do not observe actual policy; instead, they observe conditions (which combines policy and exogenous occurences) and judge incumbents accordingly.
5. Incumbents will work harder at reelection (i.e. responsiveness) if the office is more valuable.

Additional outcomes

1. If there are fewer parties in competition, the incumbent has a better chance of regaining office even if he loses office now (because the next election will be an incumbent on the other parties, with a new winner chosen randomly from the out-parties; fewer out-parties means better probability of regaining power). Thus, accountability is better served by a larger number of parties.
2. If voters aren't sociotropic (i.e. don't have homogenous preferences), incumbents can play one section of the electorate against another, dramatically reducing electoral accountability.