WHO CARES ABOUT GLOBALIZATION? MATCHED SURVEY OF LEGISLATORS & VOTERS DURING THE FINANCIAL CRISIS IPES 2010 @ Harvard University Megumi Naoi, Assistant Professor, Dept. of PoliMcal Science, UCSD Acknowledgement: Surveys used in this paper were funded by the JSPS Grant in Aid for ScienMfic Research and were conducted collaboramvely with Ikuo Kume (PI), Arata Kuno, Yoshitaka Nishizawa, Masahiko Tatebayashi, and Yomiuri Shimbun.
In a Nutshell Research Ques+on Is globalizamon a poli7cally salient issue for voters and legislators during the world recession? Mixed views & evidence IPE scholars: We hope or assume Yes. Guisinger (2009) shows No in the American polimcal context. Public opinion researchers: No. DomesMc issues more important, valence more important than policy based compemmon. LegislaMve studies: No, globalizamon has become less parmsan due to the cross cu_ng nature of issues. Yes, to the extent that organized interests contribute polimcal finance to legislators. Broz (2005). This paper Provides some of the first systemamc evidence of the hidden but high salience of globalizamon issues using an original, matched survey of legislators and voters during the current financial crisis.
Three Components of Poli7cal Salience (i) Issue a>en+on: voters interests and knowledge & how much effort legislators invest during campaigns & gaining policy expermse (e.g., commidee affiliamon). (ii) Congruence: do voters actually vote for a candidate based on his/her issue posimon on globalizamon? (iii) Divisiveness/polariza+on: issue posimons diverge and polarize & do they map onto the parmsan divide? Data limitamons, especially, the lack of coordinated data collecmon at the levels of voters and legislators, have been a problem.
ContribuMons of this Paper I. New data & original matched surveys in Japan Issue AFen7on Congruence Divisiveness/ Polariza7on Measurem ent (a) Media coverage of issues during elecmon. (b) Voters selfassessment. Ideal point distance b/w a legislator & voters. Ideal point dispersion for voters & legislators. Method (a) Key word count & content analysis (b) Survey Scaling survey responses using MDS (mulmdimensi onal scaling) Scaling survey responses using MDS (mulmdimensi onal scaling) II. New Findings & a Further Puzzle Issue AFen7on Congruence Divisiveness Very Low MaFers Very High III. Test hypotheses on why low issue attention is associated w/ more divisiveness & congruent voting.
Matched Survey of Legislators & Voters in Japan during the Financial Crisis Asked legislators and voters the exact same 23 ques7ons on policy issues related to domesmc polimcal economy, society, foreign affairs, and globalizamon. Nov Dec, 2009 All Lower House legislator survey. Teamed up with Yomiuri Shimbun (the News daily with largest subscripmon). Obtained 58.7% response rate (282/480, 167/300 SMD). February, 2010 On line voter survey at Yahoo! Research. Opt in monitors whose demographics are close to those of namonal census in each municipality. 100 respondents per a singlemember district for 160 districts=16,000 voters.
23 Policy Issues Economy: big vs. small gov t, consumpmon tax, highway tax, redistribumon, public work, pension, fiscal health. Society: gender, namonalismc educamon, polimcal inheritance, regulamon of polimcal donamon, vomng age. Globaliza+on: general trade, agricultural protecmonism, Japan U.S. FTA, inward foreign investment, foreign workers. Security and na+onalism: defense, consmtumonal revision, foreigners right to vote & North Korea etc. Non globaliza+on issues were chosen by (i) Vote Match program which uses coding of party manifestos (amount of adenmon & parmsan divisiveness) and (ii) the polimcal desk of Yomiuri Daily. Thus, these should be high salience issues.
MulMdimensional Scaling Ideal point esmmates (like NOMINATE) for survey responses using Keith Poole s blackbox program. Similar to principal component analysis but uses responses as direct input and deals w/ missing data. Maps legislators and voters issue posimons in a common two dimensional policy space. Dimensions of policy debates: (i) what is the underlying structural brief or ideology that predict respondents issue posimons? (ii) what policy issues divide/polarize respondents (i.e., cleavages & coalimons)? 1st Dimension: Picks up issues that are the most predicmve & divisive. 2nd Dimension: Picks up issues that account for the residuals from the first dimension. Policy posi7ons: proximity on the map represents proximity of issue posimons. Useful to describe congruence and distance of issue posimons.
Fit StaMsMcs by Issues Voters & Legislators (pooled) 1 st Dimension (R-square=.416): Globalization. Picks up import, agri protectionism, foreign workers, U.S- Japan FTA, voting age (not picked up: inward FDI). 2 nd Dimension (R-square=.071): Govt. intervention. Picks up tax, highway fee, gender, nationalism, foreigners right to vote at local elections
The Puzzle: GlobalizaMon issues are low adenmon & very divisive inverse relamonship.
Mapping policy posimons: DPJ, LDP, CGP, Leqist Voters Protectionist Pro-Globalization big vs. small govt- Legislators big vs. small govt- Partisan divide Protectionist Pro-Globalization
Plo_ng Distance in Policy PosiMoning b/w Voters and a Legislator xxyy xx: prefectural id yy: district id M: Median Voter D: Median voter of DPJ supporters L: Median voter of LDP supporters DPJ Legislator LDP Legislator Left Legislator Protectionist Pro-Globalization
The Effect of Ideal Point Distance on the Probability of VoMng for a Legislator Congruence matters for voting more so for globalization issues (except for party voters in #4) 1 2 3 4 Distance 1 (GlobalizaMon) 0.233 (2.26)* 0.234 (2.19)* 0.420 (3.61)** 0.157 (0.43) Distance 2 (Govt. Role) 0.257 (2.90)** 0.146 (1.61) 0.301 (3.06)** 0.169 (0.55) DPJ Support (ParMsan ID) No 1.090 1.032 (39.03)** (33.75)** 0.985 (31.13)** Party vote index*dis1 Party vote index*dis2 No No No 0.439 (2.55)* No No No 1.080 (7.44)** Party vote index 0.384 (10.05)** Organized Vote No No No Yes Only Voted No No Yes Yes Obs 14799 14799 11809 11809 Pseudo R 0.0007 0.082 0.077 0.083 Note: Absolute value of z-statistics in parentheses. *significant at 5%, ** significant at 1%.
Q: Is globalizamon a polimcally salient issue? A: YES, but it is hidden salience. Issue A>en+on: No. Low adenmon. Congruence: Yes, the greater an ideal point distance on the globalizamon dimension, the less likely that a voter voted for a given legislator. The distance on globalizamon (1 st dimension), moreover, has a larger impact on vomng than the distance on issues related to government intervenmon (2 nd dimension). Party based voters are more sensimve to the distance on 2 nd dimension (govt s role). Divisiveness/Polariza+on Yes, globalizamon issues (trade & migramon) are the most divisive at individual levels, but this division does not map onto the parmsan divide.
Next Step: Endogenizing Issue AdenMon, Congruence, and Divisiveness Possible hypotheses. Nature of globalizamon issues: technical, distant & elitest low adenmon & uninformed voters & legislators express their gut hunch reacmons that are more likely to be extreme than moderated toward median voters. Elite s and Party manipulamon: legislators do not want to focus on a divisive issue, especially if it cross cuts parmsan lines focus on less divisive issues and moderate their posimons to median voters (e.g., the second dimension).
Supplementary Slides
ParMsan distribumon of respondents on two dimensions
Ideal point distance in 143 Single-member districts (Globalization dimension) Protectionist Pro-Global xxyy xx: prefectural id yy: district id M: Median Voter D: Median voter of DPJ supporters L: Median voter of LDP supporters DPJ Legislator LDP Legislator
Ideal point distance in 143 Single-member districts (Govt. intervention dimension) Big vs. Small Govt xxyy xx: prefectural id yy: district id M: Median Voter D: Median voter of DPJ supporters L: Median voter of LDP supporters DPJ Legislator LDP Legislator