Prof. Hiwatari, in a message distributed June 15, asks the right question for analyzing the impact of electoral system variables: how well can they explain past changes? By specifying the big changes in the japanese party system in the two decades from the 1960s to the 1980s, he seems to be implying that the electoral system can't be much of an explanation since it hadn't changed for years.
That makes good sense. But to get at what did cause change I think we need a better specification of the dependent variable. In particular the two decades from the 1960s to the 1980s could mean 1960-1980 or 1969-1979 or anything in-between, and it matters.
I don't remember Inoguchi's periodization of the convergance of party electoral platforms, but the change in the party system in terms of the shares of the vote (or the electorate, including non-voters) commanded by each party ended in the *early* 1970s in my view. The period from the late 50s to then was when the LDP share declined steadily and the two niche parties (three if the JCP is included) developed. Since then the shares of the LDP and the three niche parties have not shown much trend, though there has been some fluctuation; the only changes were the continued decline of the JSP (nichezation?) and the blip of the NLC in 1976, until the end of the 1980s.
So the interesting period is the 1960s, and there the obvious driver was urbanization (which itself slowed sharply around 1970). But I can't see what happened then as polarization. What it looks like is divisions of the growing urban electorate mainly by organizational means, with a pretty stable two-party division in rural areas. I thnk that is pretty consistent with a movement toward the equilibrium one would predict for this electoral system, takng the 1955 change as the starting point and urbanization as another independent force.
Incidentally, my thanks too for that essay by Gerry Curtis.
----------------