Voting over Redistribution and the Size of the Welfare State: The Role of Turnout
| Author | Valentino Larcinese |
| Date | 01 October 2007 |
| Published date | 01 October 2007 |
| DOI | 10.1111/j.1467-9248.2007.00658.x |
| Subject Matter | Article |
Voting over Redistribution and the Size of
the Welfare State: The Role of Turnout
Valentino Larcinese
London School of Economics and Political Science
Theories of redistribution inspired by the Downsian model receive little support from empirical
investigation. In this article I argue that one of the possible explanations is that the standard Downsian
theory, and the empirical specifications derived from it, ignore electoral turnout. Empirical evidence
consistently shows that higher-income citizens are more likely to vote; office-seeking candidates should
therefore include this probability in their objectivefunction. As a consequence,the pivotal voter is not the
median in the income distribution, but is generally richer. Moreover, an increase in income inequality
does not unambiguously increase the political demand for redistribution, as most literature takes for
granted. Including turnout in the model restores the compatibility of the Downsian theory with current
empirical evidence.A regression analysis on panel data for 41 countries in the per iod 1972–98 confirms
the importance of turnout as an explanatory variable for social spending.
The introduction and evolution of welfare programmes can be explained by
many factors of both an economic and political nature.1Among them, it is
apparent that the expansion of the welfare state inWestern countries follows quite
closely the extension of democratic rights.A vast literature in political science and
economics has focused on this relation and linked the growth of welfare spending
to increased democratic representation of the poorer segments of the population.
The removal of wealth or literacy requirements for voting is thought to have
generally increased the support for popular and left-wing parties and for redis-
tributive electoral platforms.
Formal voting models have rigorously scrutinised this possibility. They have
provided firm foundations for an old idea of de Tocqueville’s that the size of a
government,measured by tax revenue and expenditure,depends essentially on the
spread of the franchise and the distribution of wealth in a society. Voting r ights
determine the extent of democratic representation of the various interests, while
wealth inequality generates the drive to use the gover nment to redistribute
resources in favour of the majorities.
Unfortunately this literature seems to have neglected the important role that
turnout plays in determining election outcomes. This is especially relevant as the
identity of voters is on average quite different from the identity of abstainers.
Tur nout can be strongly predicted by a number of social, economic and demo-
graphic characteristics. In different countries and elections, empirical research
consistently shows that the likelihood of voting is positively correlated with
income, age and education level, as well as with being a male citizen. It is quite
doi: 10.1111/j.1467-9248.2007.00658.x
POLITICAL STUDIES: 2007 VOL 55, 568–585
© 2007The Author.Jour nal compilation © 2007 Political StudiesAssociation
likely that such characteristics are correlated with policy preferences, especially
over redistributive issues. Thus, although the expansion of voting rights has
certainly changed the landscape of electoral competition and has probably
increased the political demand for welfare programmes, it should be recognised
that this has probably happened to an extent that is inferior to the predictions of
standard public choice models. As will be shown later, consider ing turnout
substantially alters some basic predictions of such models, restor ing a compatibil-
ity with empirical evidence that is currently in considerable doubt.
In the following I will first illustrate the logic of a basic rational choice model of
redistribution and derive its main predictions. I will then scrutinise the limited
empirical evidence available to conclude that this theory receives little support
from data.I will then introduce tur nout into the model and show how this affects
the predictions,making them compatible with the available evidence.I will finally
present new evidence from cross-countr y panel data and discuss the important
role that turnout can play as an explanatory variable of social spending.
Downsian Analysis of Redistribution
The median voter theorem is one of the most celebrated results in public choice
theory. It essentially states that, if the policy space is unidimensional and voters’
utility functions are single-peaked over the policy dimension, then the voter
whose ideal position is the median in the preference distribution will be pivotal
in majority decision-making.2Anthony Downs (1957) uses this result to analyse
electoral competition; he shows that two competing parties, able to pre-commit
to their proposed platforms, will converge on the policy preferred by the median
voter. Although under quite restrictive assumptions, this model delivers a clear
prediction about the impact of electoral competition on public policy.
Applications of the median voter theorem have been used to inter pret and explain
a very large number of issues in public policy-making.Among those, a vast body
of research has used the Downsian model to explain income redistribution
through fiscal policy. The basic models for this analysis have been developed by
Thomas Romer (1975) and Kevin Roberts (1977) and subsequently popularised
by Allan Meltzer and Scott Richard (1981).They assume that redistr ibution is
achieved by using a flat tax rate on all incomes and by using the tax revenue to pay
a lump sum benefit to all citizens such that the public budget is balanced.Although
redistribution assumes much more complex forms in the real world, this simpli-
fication tries to capture the essential features and has a very important advantage:
the policy space is unidimensional and equal to the tax rate. Given the condition
of a balanced budget, any change in the tax rate is reflected monotonically in the
benefit and therefore, by voting on the tax rate, a citizen is also implicitly voting
on the benefit to be distributed. Hence,the tax rate chosen by majority voting will
be that preferred by the citizen whose income is the median in the polity.
If tis the tax rate, Nis the number of citizens in the polity and mis the mean
income then the total revenue will clearly be tNmand the per capita benefit will
TURNOUT AND REDISTRIBUTION 569
© 2007The Author.Jour nal compilation © 2007 Political StudiesAssociation
POLITICAL STUDIES: 2007, 55(3)
To continue reading
Request your trial