This paper proposes a simple mechanism for evaluating the relevance of credibility problems in politics. If parties are capable of making credible policy promises, we will not expect them to systematically adopt platforms that entail large probabilities of losing an election. This is because the adoption of very extreme platforms has the effect of shifting expected policies away from their ideal points. Parties who lack the capacity of making credible commitments, in turn, are unable to affect voters' expectations of the policies they will adopt upon reaching office. We test these predictions on a panel of US states by studying the relationship between the preferences of party constituents and enacted policies. We estimate this relationship using an econometric methodology that fully accounts for the possibility of multiple equilibria and find that its slope is in general not positive, a result that is inconsistent with the existence of a commitment technology.
Available at: http://works.bepress.com/ezambran/1/