Voter Turnout, Income Inequality, and Redistribution Henning Finseraas PhD student Norwegian Social Research hfi@nova.no
Introduction Motivation Robin Hood paradox No robust effect of voter turnout on redistribution Franzese: We need to interact inequality with turnout to capture the (politically) relevant measure of inequality. My contribution: Redistribution rather than social spending. Examine the micro-level foundation of why turnout should matter for redistribution.
Inequality and Redistribution Meltzer-Richard: Median voter model of redistribution Citizens vote over a flat tax-rate All tax income are redistributed as an equal transfer to all citizens Citizens know that there are disincentive effects of taxation Individual utility are solely determined by own income Individuals prefer the tax rate that maximizes own income Most important prediction: The preferred tax rate (level of redistribution) of the median income person (decisive person) increases with the distance between median income and average income (measure of inequality).
Inequality and Redistribution Thus, in a comparative perspective, the model predicts that the amount of redistribution should be highest in countries with the most unequal distribution of market income. Robin Hood paradox: Negative correlation between measures of market inequality and redistribution.
Figure 1: Redistribution and pre-transfer inequality (90/10 percentile), 13 OECDcountries. Averages based on observations from the 80s and 90s Average amount of redistribution 0.05.1.15 Belgium Finland Sweden Denmark Netherlands Australia Norway Germany Switzerland France UK Canada USA 2 2.5 3 3.5 4 4.5 Average pre-transfer inequality (90/10 percentile)
Voter Turnout and Redistribution Turnout is typically higher among the relatively rich the income of the median voter is above the median income: The politically relevant level of inequality is smaller than the actual level of inequality The more voting is biased in favour of the rich, the less redistribution.
Figure 2: Redistribution and voter turnout (90/10 percentile), 13 OECDcountries. Averages based on observations from the 80s and 90s Average amount of redistribution 0.05.1.15 Switzerland USA France Canada Finland UK Sweden Netherlands Denmark Norway Germany Australia Belgium 40 50 60 70 80 90 Average level of voter turnout
Voter Turnout and Redistribution No robust effect of voter turnout on redistribution in the empirical literature. Franzese: Need to introduce an interaction between inequality and turnout to capture the relevant measure of inequality Income inequality should matter for redistribution only if the poor actually vote
Empirical analysis: Macro-level Dependent variable: Redistribution = Gini (market) Gini (post taxes and transfers). Pre-transfer inequality: 90/10 percentile Turnout as a percentage of voting pop. Unemployment, net union density, GDPpc
Empirical results: Macro-level Random effects: The positive effect of inequality level reduces as turnout increases. Plümper and Troeger s fixed effects and Instrument Variable regression: No interaction effect.
Empirical results: Voter level Next, I examine voters and non-voters (using survey data) in an effort to understand why the theoretical expectations fail. Note that I analyze data from only one election (late 90s) in each country! A critical assumption in the theoretical framework is that voting is biased in favour of the rich This assumption is supported by my data (and previous research).
Empirical analysis: Voter-level Another important implication of the theory is that voting is biased in favour of those not supporting redistribution This is the case in 11 of 13 countries Finally, the theory assumes a strong correspondence between preference for redistribution and party choice. In 10 out of 13 countries, voting is biased in favour of the Left
Empirical analysis: Voter-level This finding has important implications: As voters vote for parties and not over single issues, it is not given that an increase in turnout will increase redistribution via more votes for (presumably more redistribution friendly) Leftist parties. One possible interpretation of my results (based on Roemer): An increase in turnout will increase redistribution only if redistribution is the most important issue in the election in this case, the extra votes are more likely to be cast according to preference for redistribution?
Figure 8: Leftist bias and the salience of welfare issues in 13 OECD countries. Leftist bias -8-6 -4-2 0 2 Denmark1998 Canada1997 US1996 Belgium1995 Germany1998 UK1997 Australia1996 Norway1997 Netherlands1998 France1995 Switzerland1995 Sweden1998 5 10 15 20 The salience of welfare issues Finland1995
Conclusion No support for the hypothesis that there is an important interaction between voter turnout and income inequality explaining redistribution. One possible interpretation: an increase in turnout matters for redistribution only if redistribution is the most important issue in the election. However, this interpretation immediately raises the question of what determines issue saliency, and how the salience of the redistribution issue is related to inequality...