Matthews: Veto threats

Matthews. 1989. Veto threats: Rhetoric in a bargaining game. Quarterly Journal of Economics.

Puzzle

Presidents say all kinds of rhetoric to try to influence Congressional appropriations bills. (1) Does the rhetoric influence the bills? (2) Is the rhetoric only a bluff? (3) If so, will his bluff be called, and will he renege on his promises to veto?

Place in the Literature

Criticizes the two-stage proposal-veto game (e.g. Kiewiet and McCubbins 1986) in which the Congress proposes a bill just close enough to its ideal point that the president will be indifferent to the status quo. (1) This model predicts no vetoes, yet vetoes occur regularly. (2) This model assumes that rhetoric has no meaning, despite the common belief that presidential veto threats have some credibility.

The Model

Overview

1. Only the president knows what bills he would prefer to the status quo (uncertainty). The president is of one of four types:
   • Accomodator: Wants more reform than Congress, therefore lets Congress move policy to its ideal point.
   • Moderate: Wants somewhat less reform than Congress
   • Extreme: Wants far less reform than Congress
   • Recalcitrant: Wants reform in the opposite direction as Congress wants.
2. Thus, Congress does not know which bills will be vetoed.
3. Thus, the president's rhetoric is a signal to Congress.

NOTE that each type doesn't require that policy move to its ideal point. It requires that Congress move policy only far enough that the new policy is closer to the ideal point than the status quo. But if the Moderate type could convince Congress that he were Extreme, than Congress would propose a policy just good enough that an Extreme president wouldn't sign it--and this would probably closer to the Moderate president's ideal point than what Congress would propose if it knew the president were Moderate.

• If rhetoric didn't matter, than Congress would estimate the probabilities that the president is A, M, E, or R. Using these probabilities, it would propose a policy that it thought was (1) as close to its ideal point as possible (2) given the probability that the president will pass it. But in its effort to preempt a possible veto, Congress would make concessions that Accommodating types wouldn't want.
• Thus, the president wishes to signal that he is an Accomodator (if he is). An accomodating speech is always credible, since no other type of president would have an incentive to make an accomodating speech. Any of the other types might threaten a veto, however. Thus, the presence of a veto threat does convey information. After a threat, Congress can update its beliefs slightly ("he's definitely not an accomodator") and make a slightly less drastic proposal than it might otherwise make.
• THUS: Congress makes a larger concession to the president's preferences following a veto threat.
• Moreover, a Moderate might issue a threat even if he prefers Congress's ideal point to the status quo. See above: by issuing a threat, the Moderate might extract concessions making the proposal even better for him.
• THUS: A president may issue a veto threat even though he would accept Congress's ideal point.

Implications

1. Rhetoric matters
2. Bills will sometimes be vetoed

Sequence of Play

1. The "chooser" (president), who alone knows his type, uses rhetoric to send a message. This rhetoric is "cheap talk", not a costly signal.
2. The "proposer" responds with a proposal.
3. The choose accepts or vetoes the proposal.
4. Outcomes: Acceptance results in the proposal, veto results in status quo.

Assumptions

1. Unidimensionality. [Why? When funding multiple agencies, there can be multiple dimensions. But see Cameron, p 87.]

(Perfect) Equilibria

1. "Size one equilibria." Rhetoric doesn't matter. The proposer always makes a compromise proposal. Might be accepted or vetoed.
2. "Size two equilibria." Chooser sends either an 'accomodating' or 'threat' message.
   • Proposer responds to 'accomodating' message by proposing his ideal outcome, which is 'surely' accepted by the types of choosers willing to signal accomodation.
   • Proposer responds to 'threat' message by proposing a compromise. Proposer might accept or reject this compromise.

Criticisms and Comments

Doesn't see vetoes as costly enough. Moreover, the game (in real life) wouldn't end with a veto--a veto would simply lead to the next stage of the game.