A new combined electoral method was adopted in the Japanese House of Representatives election, beginning with the 1996 general election. Three hundred seats are elected by the single-seat constituency system and the other by the proportional representation system. There was some argument that the adoption of the single-member constituency system gave a strong advantage to the most dominant party. However, no analysis has been made so far to confirm the electoral bias of the single-member constituency system in Japanese elections. Extensive research measuring electoral biases has been published in the US and UK contexts. The electoral process in those two countries is the single-member constituency and the two major party system, respectively. However, Japan has numerous parties. In the 1993 general election the largest party was divided and coalition governments have since appeared. To clarify the electoral biases under the multiparty system, the electoral bias of the 1996 general election was analyzed. Because some parties could not field candidates for some constituencies, votes for parties with no candidates did not exist in those constituencies. However, the conventional method (Gudign and Taylor, 1979) for measuring electoral bias requires a vote ratio for all parties in all constituencies. To overcome this problem, we used the voting data from the proportional representation system in each municipality zone as the unbiased votes for five major parties: the Liberal Democratic Party (LDP); New Frontier Party (NFP); Democratic Party of Japan (DPJ); Japan Communist Party (JCP); and Social Democratic Party (SDP). First, following the Gudign and Taylor (1979) method, the Japanese electoral biases were decomposed into the malapportionment and the waste-vote bias (Fig. 2). The former is clearly smaller than the latter. The magnitude of the latter is far bigger than the expected bias by the cube law, even if we take into account the variation of constituency zoning (Table 1). The power function model (equation 5) shows that the estimate of the bias-magnitude parameter is greater than four (Table 1). This indicates that the multiparty system tends to cause larger electoral bias than the two major party system. Second, another simulation is executed with the assumption that voters do not vote if there is no candidate from their preferred party for each constituency. As a result, we obtained a more realistic relationship between the ratios of votes and seats than in the first simulation. Because the sizes of the parties are relatively small under the multiparty situation, parties cannot field candidates for all constituencies. Parties tend to field candidates only for constituencies where there is significant support. This 'efficient' strategy reduces the number of wasted votes. At the same time, it does not reduce but hides the large electoral bias in the single-member constituency system. Finally, we investigated the regional effects of electoral bias. Shimizu (1959) previously demonstrated the malapportionment in the Japanese electoral system: the weight of votes in nonmetropolitan areas is unfairly heavy because the allocation of population for nonmetropolitan constituencies is small. The multiparty system is recognized in metropolitan areas that tend to have manifold party support of voters and more candidates than nonmetropolitan ones. Here we can consider a low probability of wasted votes as the strength of the vote. Thus it should be concluded that the regional difference in the degree of the multiparty situation expands the regional gap between the strength of one vote in metropolitan and nonmetropolitan areas: a metropolitan vote becomes weaker and a nonmetropolitan vote stronger. In other words, introducing the single-member constituency system in Japan expands the regional gap between the strength of one vote.
ASJC Scopus subject areas
- Geography, Planning and Development
- Earth-Surface Processes