Electoral Turnout and State Redistribution
| Published date | 01 June 2014 |
| Author | Piotr R. Paradowski,Vincent A. Mahler,David K. Jesuit |
| Date | 01 June 2014 |
| DOI | 10.1177/1065912913509306 |
| Subject Matter | Articles |
research-article2013
Article
Political Research Quarterly
2014, Vol. 67(2) 361 –373
Electoral Turnout and State
© 2013 University of Utah
Reprints and permissions:
sagepub.com/journalsPermissions.nav
Redistribution: A Cross-National Study of DOI: 10.1177/1065912913509306
prq.sagepub.com
Fourteen Developed Countries
Vincent A. Mahler1, David K. Jesuit2, and Piotr R. Paradowski3
Abstract
This study explores the relationship between electoral participation and income redistribution by way of social transfers,
using data from the European Social Survey, the Comparative Study of Electoral Systems, and the Luxembourg Income
Study. It extends previous research by measuring the income skew of turnout rather than using average turnout as
a proxy for its income bias. We find that, controlling for a number of other variables, the income skew of turnout is
negatively related to transfer redistribution and that electoral participation by those in poverty is positively associated
with redistribution in their favor.
Keywords
electoral turnout, income redistribution, developed democracies
Few political phenomena have attracted as much schol-
have suggested that it is—although many other variables
arly attention for as long a time as electoral turnout. This
also play a role. At the heart of these analyses, however,
abiding interest is hardly surprising. The right to partici-
there is almost always a missing step. Nearly all broad
pate in competitive elections is a defining feature of
cross-national studies have measured average electoral
democracy, and the fact that widely varying proportions
turnout in a country when what they have really been
of all eligible citizens actually exercise that right is one of
interested in is the degree to which turnout is skewed in
the most striking political differences among contempo-
favor of high-income groups. Is it, though, actually the
rary democratic regimes.
case that the average level of electoral turnout is directly
The variation in electoral participation across democ-
related to its income skew? Certainly, when turnout is
racies is so large that a substantial cross-national litera-
very high, there is little room for participation to vary
ture has considered its implications for political outcomes.
systematically by income group: if 90 percent of eligible
Political scientists studying the developed world have
persons vote, all income groups will necessarily partici-
devoted much of their attention to the relationship
pate at similar rates. However, what of the difference
between turnout and the extent of government redistribu-
between an average turnout of 75 percent and one of
tion by way of social transfers, which itself varies widely
50 percent? Does income skew systematically increase as
across the affluent democracies. The basic intuition is
average turnout declines? Similarly, can countries with
that higher turnout reflects a more equal representation of
the same average level of electoral turnout safely be
low-income groups in the political process, which in turn
assumed to manifest the same degree of income skew?
results in a greater effort to redistribute market income in
These are questions that cannot be addressed by the usual
favor of disadvantaged groups. In the words of Lijphart
practice of using average national turnout as a proxy for
(1997, 4), “who votes, and who doesn’t, has important
its income skew.
consequences for who gets elected and for the content of
public policies,” including, especially, redistributive
policies.
1Loyola University Chicago, IL, USA
2
Is the expectation that electoral turnout is positively
Central Michigan University, Mount Pleasant, USA
3Luxembourg Income Study, LIS Cross-National Data Center,
related to the size and redistributive effect of social trans-
Luxembourg
fers borne out by the cross-national evidence? In the last
decade, a growing number of empirical studies (e.g.,
Corresponding Author:
Vincent A. Mahler, Department of Political Science, Loyola University
Brady 2009, 117; Iversen 2005, 154; Kenworthy and
Chicago, Chicago IL 60660, USA.
Pontusson 2005, 459–60; Lupu and Pontusson 2011, 325)
E-mail: vmahler@luc.edu
362
Political Research Quarterly 67(2)
Most cross-national studies fail even to mention these
from the Luxembourg Income Study and then tests multi-
issues, implicitly assuming that average turnout is a direct
level models that include both individual- and country-
proxy for the income skew of turnout. One recent excep-
level variables. Together, these analyses offer a more
tion is a careful empirical study by Pontusson and Rueda
detailed and direct assessment of the redistributive effect
(2010) that focuses on the political mobilization of low-
of electoral participation than is typical in the literature,
income voters, particularly its effect on the willingness of
which, as has been indicated, ordinarily simply assumes
left parties to ameliorate inequality by government action.
that average national turnout is a direct proxy for the
The authors of this study are clearly aware of the limita-
actual variable of interest, the degree to which turnout is
tions of using average national turnout as a proxy for its
skewed by income.
income skew. However, in the end, practical consider-
More broadly, we aim to contribute to the literature on
ations compel them to do so; as they suggest, while
class bias in political “voice” and its effect on democratic
“aggregate turnout is, of course, only a rough proxy for
accountability. As described by Schlozman, Verba, and
relative turnout by income . . . it has the advantage of
Brady (2012, 3), political voice performs two essential
being readily available and comparable across countries
functions in a democracy: communicating information to
and elections” (Pontusson and Rueda 2010, 681). In sum,
policy makers and providing them with incentives to act.
even as careful a study as this runs up against the hard
If the actual electorate differs systematically from the eli-
fact that fully comparable data disaggregating turnout by
gible electorate in so important and policy-relevant a
income group have heretofore been available for very
characteristic as income, it is instructive to ask whether
few countries and elections.
this has consequences for one of the core functions of
The aim of this article is to address this limitation by
contemporary governments, redistributing market income
assembling comparative data for a number of developed
by way of social transfers (Gelman et al. 2008, 211).
countries that disaggregate electoral turnout by income
group—quintiles, to be exact. In so doing, we will follow
Individual-level Turnout
several steps. First, self-reported individual-level data on
turnout have been gathered from two major compendia of
One basic way of measuring turnout is at the individual
electoral surveys, the European Social Survey (ESS)
level. Individual-level data on electoral participation are
(2013) and the Comparative Study of Electoral Systems
derived from the surveys that are conducted in many
(CSES) (2013). These studies ask respondents not only
countries at the time of national elections, which invari-
whether they voted in a given election but also the income
ably ask respondents whether they voted in the election in
group within which their household falls.
question. Again, the most important advantage of election
One problem with individual-level election studies is
studies is that they offer information about respondents’
that respondents commonly overreport electoral partici-
household income—information that is obviously not
pation by a nontrivial amount, often indicating turnout
available in national-level aggregate data on turnout.
levels ten or more points higher than those reported in
Despite this advantage, electoral studies have a disad-
aggregate election statistics. In adjusting for this overre-
vantage that limits their usefulness in comparative work:
porting, the second step of our analysis is to adjust self-
self-reported data almost always indicate a higher rate of
reported turnout figures with reference to aggregate
turnout than data from national electoral rolls, which are
figures available from national election authorities. Here,
based on actual election results. To make matters worse,
an additional complication arises. While aggregate-level
the extent of overreporting varies considerably from
figures for turnout are clearly more accurate than those
country to country. A good deal of attention has been paid
based on self-reported totals, there is a debate in the lit-
to the overreporting problem in the scholarly literature,
erature (summarized in Dettrey and Schwindt-Bayer
particularly with reference to the American National
2009) as to whether the denominator of turnout rates
Election Studies (ANES, 2013). The consensus is that the
should be registered voters or the voting-age population.
most important reason for overreporting is a tendency for
We will make a case for focusing on votes/registered in
some respondents to seek social approval by telling inter-
most, but not all, countries.
viewers that they voted in a given election even if they
Finally, we will explore the relationship between the
did not actually do so (Karp and Brockington 2005; Katz
income skew...
To continue reading
Request your trial