Under the plausible assumption that binding policy commitments cannot be made, pre-electoral announcements influence post-election choices only if they are self-enforcing, due to reputational concerns of some form. But whose reputation is more important? Generally speaking, we believe that the collective reputation of political parties is more important than the individual reputation of single politicians. But this raises another difficult question: how should the collective choices of political parties be modeled fully?
In our analysis, as in virtually all of the literature, there is no meaningful distinction between a party and a politician. On a final point, our models were designed to shed light on the stark cross-section variation in the data on government spending. Would similar models be useful in shedding light also on the stark time-series variation we observe in the same data, particularly the well-documented growth of government and the expansion of transfer payments in the last 30 years? This is far from obvious.
We intended this lecture to illustrate how we—as economists—might embark upon new research on comparative politics. We believe such research should rely on solid theoretical foundations and aim at strong empirical content. In our view, the theory is challenging, but doable. Indeed, researchers have recently made progress in understanding the consequences of different rules for allocating decision-making authority over legislation and government formation.
Empirical content is essential, in more than one way. Constitutions across the world provide a great deal of observable variation that can provide precise empirical guidance when formulating extensive-form game theoretic models. This may help avoid the ”with-the-right-assumptions-you-can-prove-anything-critique”, sometimes launched against game-theoretic research in Industrial Organization. Moreover, the theory can, and should, be formulated to yield predictions over observable policy variables. New or better data on political institutions or measures of government performance may be necessary; but a strong theoretical backing would greatly facilitate primary data collection. Collaboration between economists and political scientists on the boundary of our disciplines is also essential. Fortunately, such collaboration has become more frequent in the last few years.
To us, what lies ahead is a wide-open research agenda. In this lecture, we suggested a possible approach, by deriving testable predictions from some simple theoretical models. Even though the empirical results we presented are preliminary, we think they are encouraging enough to proceed. We hope to have convinced other economists that more research on comparative politics is both worthwhile and exciting.