Ideology and information in policymaking
| Author | Richard Van Weelden,Massimo Morelli |
| Published date | 01 July 2013 |
| Date | 01 July 2013 |
| DOI | http://doi.org/10.1177/0951629812473009 |
| Subject Matter | Articles |
Article
Ideology and information in
policymaking
Journal of Theoretical Politics
25(3) 412–439
©The Author(s) 2013
Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav
DOI:10.1177/0951629812473009
jtp.sagepub.com
Massimo Morelli
Department of Economics and Political Science, Columbia University, USA
Richard Van Weelden
Department of Economics, University of Chicago, USA
Abstract
We consider how the incentives for politicians to pander to public opinion depend on preference
heterogeneity and information. Politicians are more likely to pander on issues where politicians’
preferences are divided than on issues where there is a clear majority view. As pandering involves
ignoring socially valuable information that goes against the ex ante preferred action of the majority,
an increase in the ex ante probability that a politician may hold a minority view can then lead to
policy outcomes more biased towards the action ex ante preferred by the majority. In addition,
because the updating about the politician’s type is dampened when the voters are uncertain about
the state of the world, politicians are more likely to pander when voters are more informed about
which action is in their interest. It is then possible that increasing the information available to the
voters, by increasing the likelihood of pandering by politicians, can make all voters worse off.
Keywords
Information; pandering; preference heterogeneity; welfare
1. Introduction
In a representative democracy, voters delegate the authority to make decisions on their
behalf to electorally accountable policymakers. Howa policymaker exercises her author-
ity, and the resulting inferences that voters draw about whether she shares their interests,
determine whether or not she will be re-elected and tasked with making decisions in
the future. Inferring the policymaker’s preferences from her choices, however, is made
more difficult by uncertainty about the consequences of different actions. Voters’ ex ante
Corresponding author:
Richard Van Weelden, Department of Economics, 1126 E 59th Street, Chicago, Illinois 60637, USA.
Email: rvanweelden@uchicago.edu
Morelli and Van Weelden 413
preferred action may not coincide with their true interests when new information about
the true constraints and consequences for different policy choices in the current environ-
ment becomes available. Moreover, such information is likely to be available exclusively,
or at least disproportionately, to policymakers. In this paper we study how preference
heterogeneity and information asymmetry affect policymaking. In particular, we charac-
terize the incumbent policymaker’s incentive to act in conformity with public opinion on
a given issue, and how this relates to the degree of heterogeneity in initial preferences
and the sensitivity of voters’ preferred action to new information.
Taking the terminology from Canes-Wrone et al. (2001) and Maskin and Tirole
(2004), one of the most important potential distortions in representative democracy is
the possibility of ‘pandering’ by electorally accountable policymakers. Pandering refers
to the policymaker’s decision to take actions ex ante preferred by the majority in order
to enhance her own electoral prospects, even if this means ignoring potentially relevant
interim information. To study the incentives to pander, we consider an adaptation of the
model introduced in Maskin and Tirole (2004), in which politicians are more informed
than voters about the true state of the world, but may or may not share the policy pref-
erences of the majority of the voters. With this modeling choice, our analysis focuses on
distortions which are rooted in the politician’s incentive to signal ‘congruence’ with vot-
ers’ preferences (e.g. Maskin and Tirole, 2004; Morelli and Van Weelden, 2012; Morris,
2001) rather than the desire of policymakers to signal ‘competence’. Distortions induced
by the desire to signal competence have been studied in important parallel literature (e.g.
Ashworth and Shotts, 2010; Canes-Wrone et al., 2001; Fox and Stephenson, 2011), and
in Appendix B we show that the core insights wederive about information and pandering
also hold in such environments.
We extend the Maskin and Tirole (2004) model to allow for heterogeneous opinions
through the following modeling innovation: while the standard practice when modeling
asymmetric information is to assume that the agent (the incumbent politician in this case)
receives information about the state of nature that is binary, for example either a high or
a low signal, we believe that the interplay between preferences and information can be
best captured by three states of nature. In fact, even if we keep the possiblechoices down
to two, action aand action b, a natural way to integrate preferences and information is
to assume that the politician has an initial opinion and that in one state of nature (i.e. the
state in which no additional relevant information comes up) the politician maintains that
opinion. The addition of this third reasonable situation to the standard ones in which the
politician receives a strong signal in favor of aor bthat trump the initial opinion allows
us to extend the model to heterogeneous opinions while ensuring that policymakers’
informational advantage over voters is potentially valuable for policymaking.
Another modification of the Maskin and Tirole (2004) model that allows us to ana-
lyze how pandering incentives vary with the characteristics of a particular policy issue
is the introduction of stochastic valence. The addition of heterogeneity in the valence of
policymakers (e.g. Ashworth and Bueno de Mesquita, 2009; Stokes, 1963) means that,
while voters are more likely to prefer a candidate who is ideologically closer to them-
selves, if the valence advantage is large enough, they may prefer a different candidate
who is ideologically more distant. This serves to smooth the re-election probabilities:
in our model, the probability the incumbent will be re-elected varies continuously with
the voters’ belief about her type. Since re-election is non-deterministic, politicians have
To continue reading
Request your trial