As Michael Thies has mentioned, the new electoral system is a modified version of the "personalized PR" borrowed from Germany. (More precisely, it resembles the "trench system", a variety which would have favored the larger parties but which fell into disrepute after Adenauer attempted - in vain - to implement it when he wanted to get rid of the parliamentary representatives of his smaller coalition partners). German legislators invented this system in order to combine the alleged virtues of the British single member district system with the traditional German preference for PR. One might therefore look into the German experience to try to answer some of the questions raised in the present debate.
I am of course aware that certain ceteris paribus assumptions are unavoidable to do this, and I have to neglect the impact of the "trench system" element.

1. Different from the experience of the Weimar Republic, where a rigid PR list systems was credited with strengthening party headquarters, the present system has weakened them in the Federal Republic. Not only is it (legally, but still more politically) impossible to impose any outside candidate on constituency party organizations. Moreover, a candidate will not find an advantageous position on the Landesliste (the PR list on the basis of federal member states) if he does not also run for a single-member district. That means, he has first to be adopted by a constituency party organization as their candidate if he wants a chance to be elected by PR. The reason obviously is that constituency party organizations want to have their own MP even when they are minoritarian in the district, and these local candidates (unsuccessful in the SMD but elected by PR) are nevertheless considered as representing the district. As far as I understand the Japanese law does not permit this double candidature.
Nevertheless, the German case does not support the hypothesis that "personalized PR" would have a centralizing effect by strengthening party headquarters (neither on the federal nor on the state level). Also, under personalized PR party headquarters do not lead the party. Neglecting the impact of the federal structure, German parties are rather led by the top leadership of their parliamentary party groups, and these are too much dependent on the loyalty of individual MPs to meddle actively into the process of candidate selection.
One has furthermore to keep in mind that Germany does not have Japanese-style koenkai. Only in some extreme cases will constituency parties resemble a personal clientelistic machine of the candidate - in spite of their localism they normally have a strong identification with the national party organization (as distinct from headquarters), and on the other hand they are also dependent on party headquarters for services in the centralized campaign organization. I do hence not see why koenkai should suffer from the new system.

2. From what I said does not follow that elections under personalized PR are strongly candidate-centered. Localism in German elections is mainly due to the organizational self-interest of constituency parties. Voters largely follow national party appeals.

3. I strongly doubt that the longtime tendency of the German party system toward bipolarization was a consequence of personalized PR. Already in the first state elections in 1946 CDU and SPD led far ahead before the rest, and this was probably a consequence of the realigment of ideological camps (notably, the fusion of the protestant conservative camp with "political catholicism") that had taken place in reaction to the nazi experience. Personalized PR served more to stabilize this bipolarity - for several decades, to be precise. In recent elections, the system after all has not prevented the emergence of new parties and the relative decline of the two large parties. Given the German experience,
I have some doubts that the adoption of personalized PR in Japan will favor a trend toward a two-party system. (I must admit, however, that the PR element is clearly more important in the German system than in the modified Japanese version).

4. Similar to the scenario developed by Gerald Curtis, under personalized PR German parties tend to cater for the median voter and therefore to blur differences, in particular during electoral campaigns. Insofar as the German system produces clear policy differences this is not a result of the electoral process but of the needs felt by party activists to promote the internal cohesion of parties. Given the lack of such an activist program-oriented party organization in Japan it is still less likely that policy differences will be accentuated as a result of the new system.

5. There is one element in the German version of personalized PR - of which I do not know whether it has also been adopted in Japan - which has so far favored coalition government over a two party system, namely strategic vote-splitting.
This is well illustrated by the outcome of the last federal elections when the liberal FDP, in danger to fall below the 5 per cent threshold and hence to lose its representation in parliament, was apparently saved by strategic vote-splitting from CDU voters. These gave their SMD vote to the CDU but their PR vote to the FDP because they were afraid that the FDP might disappear, but the CDU not win an absolute majority of votes. In this case their party might have been rejected into the opposition because it would probably not find another coalition partner. I wonder whether the Japanese law permits such vote-splitting, and how this opportunity would eventually be exploited by sophisticated parts of the electorate. To be sure, Japanese voters were accustumed to one-party government while the German electorate has normally tended to avoid this. (There were only four years with a one-party majority of the CDU, 1957-1961). Apparently, when one of the large parties was widely considered as the likely winner of an absolute majority of seats, a significant number of voters switched their PR vote toward smaller parties (like the liberal FDP) because they wished to uphold the "checks and balances" considered as an important virtue of coalition government. Again I wonder to what degree the new Japanese electoral system might favor such strategic voting.

Gerhard Lehmbruch

(office:) Fakultaet fuer Verwaltungswissenschaft
Universitaet Konstanz
Postfach 5560 D 81 Tel. +49-7531-882318
D-78434 Konstanz (Germany) Fax +49-7531-882601
Email: Gerhard.Lehmbruch[atx]uni-konstanz.de