I really enjoyed reading Gerry Curtis's comments on what is going on in terms of campaign organizations and such since the decision to change the electoral rules. I, like Frances Rosenbluth, stick to my guns on the probable long-term effects of the electoral rules change. The reason for this contribution, however, is somewhat more theoretical in nature.

When Frances and I, and others, hypothesize about and look for evidence of the political effects of institutions and institutional change, we are always concerned with equilibirium behavior. I think most observers would agree that the new party system / campaign strategies / policy regime is not yet back in equilibrium since the change. People are changing parties and forming new ones left and right; big parties have not even decided how to decide on nominations, let alone instituted any rules; the realization of higher electoral thresholds has at this point only induced incumbents to seek out more support, in any way possible, not to identify and implement a new "optimal strategy." If politics really were a frictionless market, adjustment would be immediate. But as Curtis points out, things like extant parties, and individual incumbents, and koenkai-based campaigning are sticky. Change will take time.

That might sound like an implicit admission that "new institutional" scholars cannot explain the immediate effects of institutional change. Let me make it explicit: we cannot. Anyone who interprets new institutionalist hypotheses about the consequences of a change as basic as the rules by which politicains get elected as implying immediate adjustment is expecting too much; anyone who MAKES new institutionalist predictions about immediate adjustment is just plain silly.
So while current goings on are not a proper test of predictions about long-term effects, Curtis is dead right: there is no substitute for actually taking a look.

It might sound like a cop out: empirical tests do not corroborate our theoretically-derived predictions so we simply claim that the observed behavior is "out-of-equilibrium." I prefer to think about it as identifying the strengths and weaknesses of a particular approach to social science.

Let me give just one example. Curtis's description of the indistinguishability of LDP and Shinshinto policy platforms sounds right to me on most issues (although not on on issues such as deregulation). Bear in mind, however, that the prediction that the new electoral system will lead to party-based electoral competition centering on distinct policy platforms does NOT require that it be the same parties competing after the change as existed just before the change.
Why assume that the LDP and the Shinshinto will be around in 10 years, or that if parties with those same names are around, why should they be the "same"
parties as they were in 1993? The new institutionalism (or rat-cho, or whatever is the current favorite moniker) can't help much with this problem.

It's just like microeconomics can't tell us which specific firms will survive in a market -- it can only describe some characteristics (like profitability) that will describe all firms that do survive. As we learn more about how a firm operates in a particular market structure, we can be somewhat more specific than (only profitable enterprises will survive). WE can explain the superiority of certain managerial techniques or marketing strategies. But we can never make claims like "firms A,B, and C" will survive, but firm D will disintegrate in 36 months."

On the other hand, if these parties DO change their ways or DO dissolve in favor of different parties, we won't have to be surprised. (Indeed, I'm convinced that the 1993 LDP split would not have occurred had the defectors not expected some sort of electoral change -- they were jockeying for position under the new rules and for greater influence on what those rules would be.) If after a few elections the LDP and Shinshinto reamin the two largest parties in Japan, AND continue to advocate nearly identical policy platforms, AND continue to campaign in the same fashion as they did under SNTV, THEN I for one will be very surprised.

The answer in not to give up, but to do what Curtis suggests -- take a look! We as social scientists have NO models that predict the short-term effects of electoral change in Japan or generally, except the historical-cultural models that predict no effects whatsoever. We should not, however, throw the babies out with the bathwater. Theoretically-derived, general hypotheses are still a good goal to set, and help us to identify and explain patterns of political behavior across time and place. Obviously, the proof of the pudding is in the testing (to coin a phrase) but description and explanation and theorizing should be seen as compliments, not as competitors.

--Mike Thies, UCLA