Improving Electoral Engagement: A Narrative on the Evidence Tavneet Suri November 5 th 2015
Democracy Expanding Rapidly
Across the World
Since 1800
In Africa
Governance Remains a Challenge
Corruption
Safety
Poor Public Service Delivery
Electoral Process Sometimes Flawed Clientelism and vote-buying Voting based on ethnicity Parties not distinguished by policy differences or platforms Ballot fraud Elections violence
Why Aren t Elections Delivering Results Voters often lack political knowledge Access: Voters illiterate, do not speak national language, live in remote areas Supply: Limited media dissemination of political information Even with greater knowledge, will voters change beliefs and behavior Vote along ethnic, regional, or caste lines or for candidate with the best gifts How do we enable more engaged and informed voters, support policy-based elections, and promote accountable politicians?
Improving the Electoral Process
Research: Improving the Electoral Process Basic information on electoral process Date of election, offices to be voted on, etc. Encouragement, support for registration Passive information about candidates SMS, newspapers, flyers, radio Active engagement of candidates and voters Debates, town hall meetings, door-to-door canvassing Appropriate medium depends on the local context
Information on Candidates Changes Behavior Voters may vote for incumbent/their ethnic group/based on gifts But several examples where information changed behavior Benin Brazil India Kenya Mexico Mozambique Philippines Sierra Leone
Basic Information can Matter Mozambique (Aker et al 2013): flyers, SMS and newspaper All given mobile phone hotline number to report electoral problems All three interventions increased turnout, on average by 5 pp (from base of 44%) Newspapers increased demand for political accountability (voters could send SMS to president-elect with policy priorities) by 9 pp (from base of 15%)
Information on Candidates Matters Brazil (Ferraz & Finan 2008) Releasing audit results decreased (increased) reelection rate for corrupt (clean) incumbents Effects larger in municipalities with radio stations Delhi, India (Banerjee et al 2011) Report cards on candidate qualifications and legislator performance in slums Increased turnout by 2pp (from base of 57.5%) Reduced vote share for worse-performers, incidence of cash bribes Voters only reacted to information that mattered to them
But Increased Knowledge Can Reduce Turnout Negative information may de-motivate: Mexico (Chong et al 2010) Flyers reporting portion of resources mayor spent in corrupt manner Reduced voter turnout by 1.3 pp (from base of 52%) Effects larger when high levels of corruption exposed Bad news discouraged and de-motivated voters, though many already perceived the government to be dishonest
Information Must be Correct and Credible Kenya: 2013 National Elections (Suri et al forthcoming) Electoral Commission sent 11m SMS to registered voters Message 1: Reminders about election, encouragement to vote Message 2: + Information about each position to be voted for Message 3: + Transparency, neutrality, and capabilities of electoral commission Increased turnout by about 0.5-1 pp (already high) Messages reduced trust in the electoral commission by 4pp (80% in control) Concentrated among voters of ethnic group affiliated with losing candidate Messages created expectations need to fulfil these to get the gains
Information Must be Correct and Credible Uganda (Humphreys & Weinstein 2012) Scorecards on performance of all MPs Informed a random set of MPs Some MPs tried to undermine the info Info changed voters beliefs and attitudes No effect on electoral outcomes Information interacts with the context
Involving Politicians Makes Dissemination Easy Ability to inform voters when little data are available Activity generates information on candidates qualifications, policy stances Organizations do not have to take on costly task of gathering data Greater credibility Difficult to take back what is said publicly More holistic information: not just one measure of performance Might help address multitasking problem
Active Dissemination Promotes Involvement Changes how people vote Generates more positive views of politicians Studies presented today Debates in Sierra Leone (Bidwell et al 2015) Town hall meetings in Benin and the Philippines (Wantchekon and Fujiwara 2013, Wantchekon et al 2015)
Does Engagement Create Good Governance Evidence suggests that voters can change behavior Ultimately, this should lead to better governance outcomes Two pathways Selection: Different candidates are elected to office Incentives: Same candidates behave differently Impact Information dissemination increases voters political knowledge Gains in knowledge affect voting behavior Different candidates are elected to office Same candidates are more engaged and responsive Higher quality governance
Conclusions It is not impossible to change voters beliefs and behaviors Voters discern and act on relevant information But providing information is not unambiguously good Backlash to negative or inaccurate information Backlash if information creates expectations that are not met Must create programs that provide credible information More research needed on the impact on quality of governance