Legislative Voting Patterns and Party Defection in the Japanese National Diet

Ryo C. Kato

(will not be doing this in 2^nd module)

Background

In 2009, the Liberal Democratic Party (LDP), which had been the governing party since 1955 (except for 11 months in 1993), was handed the worst defeat of a sitting government in the history of Japanese electoral politics. The opposition party, the Democratic Party of Japan (DPJ), captured 308 seats out of 480 in a landslide, and secured the office of Prime Minister. This event was touted as a major landmark for Japanese democracy and was interpreted as a repudiation of the LDP for its domination over post-war politics. However, after the 2012 elections, only three years after utter defeat, the LDP was back in power. In both elections, there were a number of legislators, some of who were instrumental to the outcome of the elections, who had defected from their parties. The kingmaker of the DPJ, Ozawa Ichiro, for instance, was a powerful chief secretary of the LDP before defecting in 1993 to form a series of opposition parties with other former members of the LDP, one of which was the DPJ. In 2012, he once again defected and formed a new party, the Life Party, taking with him several former DPJ members.

A recent paper showed that there is a correlation between seniority within a party and switching affiliations. It found that very junior or very senior legislators were more likely to switch. What are other characteristics of the politicians who switch party affiliations? Do they hail from the same prefectures, do they represent similar industrial and commercial interests in their constituencies, do they sit in the same committees, have they spent time in the same or similar companies, or are they old-boys of the same universities? Or is it voting patterns that correlates most with party switching?

Primary Question

Can patterns of voting on a selection of bills be used to predict the cohesiveness of political parties in Japan? Restated: when legislators switch political parties or create new ones, do they do so with legislators with whom they had voted together on important bills?

Hypothesis

How politicians switch their party affiliations correlates with patterns of voting on legislation and connection to powerful kingmaker politicians.  

Data

I will need two sets of networks for the members of the National Diet, one from immediately before and one from immediately after the 2012 elections. Current legislators are listed on websites maintained by the two-houses, the House of Representatives and House of Councilors. Although historical compositions of the legislature are not kept on these websites, it is relatively easy to find this information from government databases and academic sources online. While the information of specific legislation at various points in the law making process are easy enough to find, the record on how legislators voted on them is a different matter. While the House of Councilors website publishes their members’ voting records on bills, the House of Representatives’ website does not. From what I have gathered, finding voting records on bills introduced to the House of Representatives will require requests for data from a government office. 

Attributes for the legislators will include, party affiliation, age, prefecture where their district is located, alma mater, category of industry they worked in prior to political life,

I will manipulate the two 2-mode network of legislators and bills, one each for before and after the 2012 elections, into a pair of 1-mode networks of legislators, where the ties represent common votes on legislation. The thickness of the line will represent the number of common votes.

Important Network Measures

A large part of this assessment will follow looking for visual patterns and comparing the pre- and post-election networks. The first, the 2-mode network of legislators and legislation, will look for clusters of nodes that vote similarly on legislation. There will be on visualization showing edges representing ‘nay votes,’ and another where the edges are ‘yay votes.’ I will compare the clusters of the pre- and post-election networks.

The second visual will be used to look for patterns in 2-to-1 mode networks of the pre- and post-election. The nodes represent legislators and the edges represent shared votes, and the thickness will show the number of shared votes. Do clusters in the pre-election networks resemble clusters in the post-election network? Do clusters based on party affiliation in the post-election network resemble clusters in the pre-election network based on other attributes (shared votes, shared committees, etc).

I would also look at measures of degree centrality in the pre- and post-election 2-to-1 mode networks of legislators with ties indicating shared votes. If shared votes do in fact correlate with party affiliation, then power politicians that carry other legislators to other parties, such as Ozawa may exhibit high degree centrality in this network.

Conclusion

Social network analysis can help narrow the study of Japanese legislator’s switching political affiliations in the National Diet. What are other attributes other than age that correlate with this behavior? Do similar voting patterns between legislators in a party or in the factions within that party correlate with party defection? Or are the connections that share many attributes with influential politicians, such as Ozawa, more correlated with party defection?

Additionally…

Given the historical strength of certain political families, some of which trace their lineage to the leaders of the Meiji Revolution, it may be of interest to conduct an intertemporal and intergenerational examination of political cooperation and party cohesion. Are the network characteristics of the Meiji oligarchs or post-war political elites reflected by their grandsons and great-grandsons?