This document discusses approaches to improving governance through better information and citizen participation, and proposes a new approach.
[1] Current approaches of providing information and encouraging citizen participation have shown mixed results in improving outcomes. Information is processed through social and political institutions, limiting its impact. Participatory programs also have design flaws.
[2] A better approach is to focus on how information shapes social capital and institutions. The document describes experiments in Benin evaluating the impact of deliberative town halls on voter behavior, clientelism, and support for candidates. Initial results found increases in voter turnout, support for some candidates, and reductions in clientelism.
[3] Future work will evaluate how information presented in
2. An Observation
The experimental literature on governance
has focused on two main approaches:
1. Better information, and
2. Encouraging participatory or community-driven
development
However, there is no consistent evidence
that these approaches improve outcomes
we care about
2
3. Outline
I. As social scientists, what is our
current approach to improving
political decision-making and
governance?
II. Why isn’t this approach working?
III. What should we do instead?
3
4. Key Concepts
• Governance
– Service delivery, security, infrastructure, etc.
• Constructive social capital
– Social capital -> Connections among
individuals and the norms of reciprocity and
trustworthiness that arise from them (Putnam,
2000)
– Constructive -> Beneficial to the whole
community
4
5. Providing Information: Theory
• Information is power
• Information is key to democratic
governance
– Better informed voters will vote more often and
base their choices on important issues
• Information will lead to better service
delivery and less corruptions
– “Sunlight is the best disinfectant” – Justice Brandeis
I. What we are doing now? II. Why isn’t it working? III. What should we do?
5
6. Engaging Citizens: Theory
• More engaged citizens should lead to
improvements in governance
– Participatory democracy
– Decentralization
– Community-driven development
• The World Bank has spent nearly $28 billion
on CDD over the last decade1
I. What we are doing now? II. Why isn’t it working? III. What should we do?
1. http://www.worldbank.org/en/topic/communitydrivendevelopment
6
7. Providing Information: Results
Table 1: Selected Information Campaign Evaluations
I. What we are doing now? II. Why isn’t it working? III. What should we do?
Study Outcome(s) Results
Banerjee, Banerjee, Duflo, Glennerster, & Khemani
(2010)
Learning ✗
Chong, De La, Karlan, & Wantchekon (2015)
Corruption, Voter
participation
?
Ferrez & Finan (2007) Corruption ✓
Lieberman, Posner, & Tsai (2014)
Private action,
Collective action ✗
Peisakhin & Pinto (2010)
Access to public
services
✓
Reinikka & Svensson (2005) Enrollment, learning ✓
Ravallion, van de Walle, Dutta, & Murgau (2013) Employment ✗
7
8. Engaging Citizens: Results
Table 1: Selected Participatory Governance Evaluations
I. What we are doing now? II. Why isn’t it working? III. What should we do?
Study Intervention Outcome(s) Results
Casey,
Glennerster, &
Miguel (2011)
Block grants for communities in
Sierra Leona that promote
democratic-decision making
and inclusion
Material well-being,
Institutions, Power
dynamics
?
Chaudhury &
Parajuli (2010)
Transferred local school mgmt
to the communities in Nepal
School access,
Learning
?
Duflo, Dupas, &
Kremer (2008)
Community monitoring of
teachers in Kenya
Teacher attendance,
Student performance
✓
Humphreys, de la
Sierra, & van de
Walle (2012)
Organized elected village
committees in the DRC that
chose development projects
Governance, Gender
inclusion, economic
impacts
✗
Olken (2008)
Community participation in the
monitoring of road projects in
Indonesia
Corruption ✗
8
9. Key Takeaway
• The empirical evidence on information
campaigns and participatory governance
strategies is mixed
– Why don’t these approaches
consistently have the hypothesized
outcomes?
I. What we are doing now? II. Why isn’t it working? III. What should we do?
9
10. Providing Information:
Why it doesn’t always work
• Limits to the exogeneity assumption
– When people engage with new information
they automatically use different frames,
methods of acquisition, and beliefs on the
credibility of the source
– Information also has spillovers
• As a result, the information given is rarely
the information acted upon
I. What we are doing now? II. Why isn’t it working? III. What should we do?
10
11. Engaging Citizens:
Why it doesn’t always work
• Design issues
– They implicitly assume all forms of participation
are equal and that all participatory institutions
are automatically good
• But the“rules of the game” matter
– For example, the possible rules for a town hall
include
• Deliberation or no deliberation?
• With or without voting?
• Who sets the agenda?
• What is the agenda?
I. What we are doing now? II. Why isn’t it working? III. What should we do?
11
12. A different approach?
• Typically, we “jump” straight from
information to governance
• However, before governance, information is
“processed” in organizations/social
networks, under specific rules.
• Information -> Social capital -> Governance
I. What we are doing now? II. Why isn’t it working? III. What should we do?
12
13. A different approach?
• Instead of only focusing on information or
some “vaguely” defined notion of
participation, we should focus more on
how (together) these concepts shape
social capital and institutions.
– Estimating intrinsic causal effects of institutions
– Separate out the effect of information from
that of institutions
I. What we are doing now? II. Why isn’t it working? III. What should we do?
13
14. Evaluating Institutions: the
Problem
• Instead of randomly assigning a treatment (e.g.
information), assign decision-making process.
• By deliberation?
• Institution is assigned to groups of individuals, who
collectively and endogenously generate the treatment.
• The question: How to best identify and separate out the
effect of the institution from the effect of the policy
(information)?
15. Example 1: Democracy Experiment in
Indonesia (Olken 2010)
• 49 villages randomly-assigned to choose development projects
via
1. Representative-based meetings, or
2. Direct election-based plebiscites
• Findings
– Direct participation in decision-making increased satisfaction
and legitimacy compared to representative-based meetings
✓ much higher satisfaction
✓ increased knowledge about project
✓ greater perceived benefits
✓ higher reported willingness to contribute
– Unclear the difference between total effect and the
"democracy effect.”
– Ignore selection effect in the treatment group.
– Policy outcomes not always the same in treatment and
control groups.
I. What we are doing now? II. Why isn’t it working? III. What should we do?
15
17. The Institutional Intervention
• Tested a voter engagement intervention in Benin and
the Philippines: Party-endorsed town hall meetings
where people deliberate over platforms proposed by
candidates
• Deliberation might outperform rallies because
– Of its effect on attendees:
• Voter coordination: Learn about other voters’
preferences and beliefs
• Platform transparency: Better understand a candidate’s
platform
• Platform customization: adapt platform to local
conditions
– Deliberative campaigns may affect those who do not attend:
• Information sharing: Learn about the candidate’s
platform from others in a social network
I. What we are doing now? II. Why isn’t it working? III. What should we do?
17
18. Experimental Design
Treatment Villages:
Town hall meeting
• Meetings followed standard
format:
– Introduction to candidate’s
platform
– Deliberation about
platform, participants
invited to propose
amendments
– Summary of meeting and
commitment to transmit
feedback to party
leadership
Comparison Villages
Business-as-usual
– Festive political rallies,
sometimes with gift
distribution
– Speech by candidate or
representative about policy
agenda
– One-way communication
– Mobile propaganda teams
– Posters in public spaces
I. What we are doing now? II. Why isn’t it working? III. What should we do?
18
20. Effects on Voter Behavior, Vote Share &
Clientelist Practices
• Town hall meetings increased turnout by 3.3%
in Benin and had no impact on turnout in
Philippines
• Increased electoral support for some
candidates or parties that engaged in
deliberation:
– Benin: 8.6% increase in vote share for opposition; no
increase in incumbent vote share
– Philippines: Vote share for treatment parties
increased 50%
• Deliberative campaign reduced index of
clientelism 0.23 standard deviations in Benin.
I. What we are doing now? II. Why isn’t it working? III. What should we do?
20
21. What’s next?
• Bureaucracies in Benin
– Providing audit reports during closed town hall
meetings twice a year
• No mayor present, only chief of staff
• Can talk freely without fear of retribution
• How is information processed in this
setting?
• Does it affect bureaucratic cohesion and
motivation?
21
22. Conclusion
• Information, generating constructive
social capital can improve governance
through horizontal accountability and
cooperation among individuals.
22
23. Credits for Illustrations
All illustrations came from The Noun Project
1. Illustrations of people, Gan Khoon Lay
2. Newspaper, unlimicon
3. Radio, AFY Studio
4. Brochure, Xinh Studio
23