Reputation and legislative leadership

ing away other aspects of legislative leadership, such as the nature of the collective action a leader tries to promote or the exact nature of the leader's own goals, we have constructed a stylized model of one important aspect of the leader's problem: the use of reputation. Reputation is really a problem of using the uncertainty of followers to the leader's best advantage, to make the most of whatever real resources the leader actually has. Although the leadership game has multiple equilibria when the leader is precisely indifferent between costless punishment and acquiescence, the slightest preference between those actions yields a unique equilibrium. When the preference is for costless punishment, this equilibrium involves creating a reputation, and we can use its properties to draw conclusions about (1) the nature and strategy of reputation-building and of the reactions of sophisticated followers; and (2) the manifestations of these strategic considerations in institutional and political change in Congress. Since nonuniqueness requires that h=0 exactly, we might say that equilibrium in the reputation game is generically unique across specifications of the punishment costs (or benefits). In addition, the various properties and comparative statics of equilibrium reputation-building will apply to each of the equilibria in the case h=0, except of course the extreme one in which no reputation-building takes place. Several technical extensions might give this model even more direct applicability to real-world legislative politics. Allowing followers to rebel in bunches, instead of one at a time, might cast more light on the strategy of being a member of the legislative rank and file. More could of course be learned about the dynamic nature of reputation-building if the model could be extended to an arbitrary finite number of periods, or to infinite length with discounting. However, because of the complexities of Bayesian updating, this apparently will require that the space of possible values of the w parameter be restricted to having only finitely many members rather than covering the unit interval. If such an extension is made, then it will be possible as well to examine the effects of random, exogenously generated "extra" information or random changes in the payoff parameters, another potentially important source of uncertainty in real life. Another interesting set of extensions involve explicitly modeling the context in which the reputation game is played. Rather than abstracting away the main collective action problem, it could be modeled directly, with the informational features of that game generating the kind of reputational considerations treated in this study. It would be helpful as well to directly consider the sources of the leader's punishment costs. This could be accomplished partly by modeling the leader's reciprocity relationships together with his more coercive "leadership" relations. Or, one might model the separate collective-action problem that followers face in banding together to depose a leader, and try to capture the effects of leadership sanctions on that problem. This would have the added attraction of presenting the leader's problem of coopting potential opposing coalitions through his distribution of rewards and punishments, a feature assumed away in the current model. Despite the apparent bare-bones nature of the model presented here, it serves to focus attention on a neglected aspect of political leadership. Since a successful leader cannot be constantly monitoring and punishing followers, it is necessary to establish in the followers a habit of obedience, a rule of thumb that the leader's wishes are to be followed. Habits and rules of thumb do not occur in a world of perfect and costless information. Thus the presence of uncertainty, and the artful manipulation of it, are crucial for successful leadership. Somewhat counterintuitively, this holds true regardless of whether the leader is a ruthless dictator or a benevolent provider of collective action. Copyright Martinus Nijhoff Publishers 1987