Japan’s PM Naoto Kan made it official, and resigned today. His Democratic Party of Japan will choose a successor next week.

The successor will be Japan’s sixth since the departure of Junichiro Koizumi in 2006. That’s a lot of PMs in a short time.

Why does Japan have such short-lived PMs? In one regard, maybe Japan is “typical.” After all, unlike presidents, who are essentially never forced out by their own parties, prime ministers are by definition agents of their parties (as well as of the legislative majority). Indeed, Samuels and Shugart (2010: 96) report that 30% of all PMs in parliamentary systems leave office for “intraparty” reasons (N=354). So there is nothing unusual about parties “firing” a PM. (Kan was not formally fired; few are. But no one doubts that it was intraparty politics that has led him to this point. He survived an internal challenge just months ago, and promised then not to remain for long.)

Still, in Japan an “intraparty” termination of a prime minister happens rather more often than in most parliamentary systems. Moreover, while short-lived prime ministers logically follow from the inter-factional politics that used to characterize the Liberal Democratic Party, they make much less sense following the electoral reforms of 1994. Besides, this is not the LDP. It’s the DPJ, which won a resounding electoral victory in 2009–and is already about to have its third PM since that election!

Under the old LDP system, “back-room” factional politics was the kingmaker, and the electoral system (SNTV) made factions and individual politicians the agents of representation at least as much as the party itself. It thus also made them agents of inter-electoral bargaining, hence the PM-ship was always subject to renegotiation.

This was supposed to change with the electoral reform–to mixed-member majoritarian (MMM). And much has changed, with factions becoming less important to leadership selection within the LDP, according to various accounts. And the 2001-06 rule of Koizumi suggested a real turn in the position of the PM vis-a-vis the party. The instability of PMs following Koizumi (three of them from 2006 to 2009) perhaps could be written off as an exhausted and about-to-lose party casting about for what it stood for and who could represent it. But the DPJ presents a puzzle. Its first PM did not even last a year, and its second has made it just over a year. The disasters that struck Japan earlier could cut either way–they could have been an opportunity for the party to rally around the leader. In any case, he was already in trouble politically prior to those events.

Generally speaking, PMs in “Westminster” type systems with a single ruling party should be less vulnerable to internal party challenges. The logic is that such political systems maximize the alignment of incentives within the party, and give the party ample opportunity to vet potential PMs so that those they choose enjoy the backing of the party. Thus leaders who head their party when it wins election usually stick around for at least the term, and if they win again, usually for a second term. (Yes, it is stylized, which does not mean it is not generally accurate, Australia’s anomaly notwithstanding. And things often get sticky during third terms–see above point about exhaustion and casting-about.)

Japan does not have a Westminster system, by usual definition. Yet the reform of the electoral system, from SNTV to MMM, was supposed to move the country’s politics in that direction. In many respects, it clearly has done so: factions less important, policies more so, alternation in government, two clear blocs instead of one dominant party and a fragmented opposition, etc. So why not more stability in the top post?