Headline

• Design: Individual level random assignment to take part in extended municipality decision making
• Findings from RIE:
  • No evidence for effects on a survey based measure of trust
  • Quite strong evidence in a behavioral task: allocations to the municipality
• Mixed methods:
  • Qualitative and quantitative analyses agree on the major moving parts of the model
  • Qualitative findings add important nuance to how citizens interpret participation and the roles of the municipality

Motivation

• Much development funding goes to participation: bringing residents into local decisions
• Donors expect this to pay off socially, for example in trust in government
• But the evidence is mixed (Wong 2012; Casey 2018; King and Samii 2014; Mansuri and Rao 2013)
• Most studies so far:
  • compare whole communities, not individual residents
  • test programmes that run alongside government, not inside it
  • do not find that participation builds social cohesion (horizontal or vertical)

Motivation

• Much development funding goes to participation: bringing residents into local decisions
• Donors expect this to pay off socially, for example in trust in government
• But the evidence is mixed (Wong 2012; Casey 2018; King and Samii 2014; Mansuri and Rao 2013)
• Most studies so far:
  • compare whole communities, not individual residents
  • test programmes that run alongside government, not inside it

Interest in Participation

• 28 percent say they are definitely not interested
• 30 percent say they certainly are

Interest in Participation

• The willing set is somewhat more male and somewhat younger
• Those who say corruption is a problem are marginally more likely to volunteer
• Those who allocated more in the hypothetical allocation task at baseline are more likely to volunteer
• Differences are not huge

Actual attendance

• Among those invited, 37 percent attend no meetings
• 43 percent attend at least two

Reported experience

Subjective experiences appear very positive

• (About 80% of) invited participants report an increase in trust due to participation in meeting
• Invited participants are much more likely to be satisfied with initiative selection

Trust index

• Trust index: average of three survey questions, in standard-deviation units (trust in local government, fairness of the board, corruption)
• Difference between invited and not invited: 0.01 (95 percent CI −0.09 to 0.12), i.e. no detectable effect

Behavioral task

• Task: residents split a real sum (about $20) between the municipal board and the Red Cross/Crescent
• Impact: Being invited raised the share allocating any funds to the municipality by 10 percentage points
• Average amount to the municipality: \(+65,277\) LBP higher (\(p < 0.05\)); \(+162,399\) LBP among those who attended at least two meetings

Estimated effects

• We do not see much heterogeneity by gender or local marginalization.

Theory of change (before)

Theory of change (updated)

Qualitative Findings

• The quality of participation cannot be attributed to taking part in working groups
• Residents credited the NGO, not the municipality for openness / innovations: felt as peers with officials in a joint process
• Many of the marginalized did not feel included
• Some suggestions of effects on horizontal inclusion

Conclusion

• We evaluate the effect of inclusion (individuals engaging in municipal decision-making) on trust in local authorities, not the effect of the program itself.

• Our results are not designed to answer questions about the overall effectiveness of the program. But they do speak to how to implement.

• Along the way we show that learning is possible in unstable contexts. We ran this through war and municipal elections.

Summary of findings

• RIE
  • Practitioners favor participation but actual demand for participation is quite low
  • Trust index does not move; allocation to the municipality rises
  • Suggests that citizens believe municipality has important functions and can be kept in line, but do not necessarily trust it to perform well
• Interviews
  • Participation improved perceptions, but the NGO, not the municipality, got the credit
  • Marginalized residents were least reached, and their trust was unaffected
• Both methods agree on the major moving parts of the model. Survey DAG

Conclusion

Looking forward:

• Participation as part of design. Where a municipality already plans to include people, this evidence suggests that is probably a good thing. But our results give no grounds to set up programs specifically to pursue this goal.
• Put the municipality in the lead. If the aim is trust in local government, municipal staff should facilitate and oversee the process. Otherwise the NGO gets the credit.

Timeline

• February 2024. Baseline begins (first 11 municipalities)
• August 2024. Working groups begin, then suspend in September
• September–November 2024. War; programming interrupted
• March 2025. Second baseline wave (13 further municipalities). Meetings resume
• May 2025. First municipal elections in nearly a decade
• February 2026. Endline survey
• June 2026. 48 structured interviews

Survey DAG

Same layout as the interview graph. Blue is positive, red is negative, and line thickness is the size of the survey parent effect. The number on a node is the share coded 1.

Back to the interview graph

Outcome Measures

Survey based outcome plus behavioral outcomes

Index for levels of trust is a simple mean of standardized responses to the following questions:

• “I trust that my local government acts in the interests of its citizens.”
• “I believe that the resource-distribution by the municipal board is fair.”
• “Corruption is widespread in local government.”

Behavioral money allocation task (approx.$20):

• “You have the opportunity to decide how the development funds to help poor people should be divided between your municipal board and the Red Cross/Crescent. Please choose one of the options below that shows how you would like the funds to be allocated.”

Invited residents are more satisfied

Score from 1 (not at all) to 5 (very satisfied); 3 is not knowing about the project. Not invited on the top row.

Being invited raises satisfaction with initiative selection

Municipality fixed effects and inverse-probability weights. Score from 1 (not at all) to 5 (very satisfied).

Participants certainly reported an effect

Most invited residents say confidence rose

Treatment group only. Score from 1 (much less confidence) to 4 (much more). Means are unweighted.

Attributed confidence lines up with allocation and trust

Treatment group only.