1 of 18

Information and Women’s Participation: Experimental Evidence from Kenya

​

Solomon Zena Walelign

ICED and World Bank

​

Soazic Elise Wang Sonne and Ganesh Seshan

World Bank

​

Africa Summit

June 29 - July 01

​

2 of 18

  • Increase in community driven development projects
    • Implemented in more than 105 countries (Saguin 2018)
    • As of June 2021, the World Bank supported 374 active CDD projects in 93 countries

​

  • Women’s participation in CDDs is lower than men

​

Motivation

Women’s participation in community driven development (CDD) projects

Barron et al. (2009)

3 of 18

  • Barriers for women’s participation
    • Cultural (Fongchingong 2006; White et al. 2018)
    • Poor information (Kongolo et al. 2009, White et al. 2018)
      • Women’s awareness about CDD programs is 0.5 times less than men

​

  • We study the impact of information provision on women’s participation in community driven development project.

​

Motivation

Poor information could be one reason for the low participation

4 of 18

  • Two strands of literature
    • Impact of CDDs on different outcomes (White et al. 2018, Casey 2012)
      • Infrastructure development
      • Social cohesion
      • Women participation
      • Institution formation

​

    • Information experiments on various other domains of participation and outcomes
      • Voting (Banerjee et al. 2011)

​

      • Educational outcomes (Jensen 2010)
      • Health outcomes (Dupas 2011)

​

  • Evidence on the effect of information on women participation is limited

Literature

Focus on the impact of CDDs

5 of 18

  • Most CDDs have not been effective (White et al. 2018, Casey et al. 2012)

​

  • Participation of women in leadership roles has linked with improved outcomes
    • Lower incidence of corruption and bribery (Jha and Saranji 2018; Bauhr et al. 2018)
    • Improved social services, e.g., childcare (Bratton and Ray 2002)
    • Better aspiration levels and educational attainments for children (Beaman et al. 2012)
    • Higher investment in public goods and infrastructure benefitting women and the community at large (Chattopadhyay and Duflo 2004)

​

  • Hence, women participation could bring a positive outcome of CDDs (Onwuaroh 2020). How?
    • Improve the design of CDDs
    • Reducing corruption in managing the CDD funding
    • Invest more in the needs of the community and women themselves

​

​

Motivation

Why do we care about women participation in CDDs?

6 of 18

  • Research questions
    • Does providing information increase women’s participation in community driven projects?

​

​

​

​

​

​

Research questions

7 of 18

  • Funds community driven projects in refugee host communities in Kenya

​

  • Encourage the participation of disadvantaged peoples and communities
    • Encourage participation of women (aged between 25 and 60 years) and the poor

​

  • So far, about 470 investments has been made or community groups has been formed.

​

​

​

KDRDIP

​

​

​

​

​

​

​

8 of 18

​

​

​

​

​

​

​

​

​

​

​

​

​

Treatment and placebo scripts

9 of 18

  • Treatment variable
    • Information treatment
    • Information treatment take-up

​

​

​

​

Treatment variables

Sample woman

Control

Information treatment assignment

Treatment

Did not understand the script

Understand the script

Control

Information treatment take-up

10 of 18

Presentation of experiment instrument for the treatment and placebo group

11 of 18

  • Sampling
    • of a total of 830 eligible households

​

  • Data from a total of 709 households (women) from 50 sublocations
    • 341 assigned to treatment
    • 368 assigned to control

​

  • Data
    • Likelihood of participation (prior and participation)
    • Perception on the effectiveness of the information script
    • Prior knowledge of KDRDIP
    • A bunch of socio-economic variables

​

Sampling and data

12 of 18

 

Empirical strategy

13 of 18

​

​

​

Results:

OLS estimation – information treatment assignment

14 of 18

​

​

​

Results:

OLS estimation – information treatment take-up

15 of 18

​

​

​

Results:

IV estimation – information treatment take-up

16 of 18

  • Information treatment (assignment to the information)
    • Did not increase women’s likelihood of participation

​

  • Information treatment take-up
    • Had a non-robust positive effect on women’s likelihood of participation

​

  • Prior likelihood of participation was a strong predictor of posterior likelihood of participation

​

  • Why null effect?
    • Did not affect (change) woman’s perception to change their prior likelihood of participation
    • The information experiment does not overcome cultural barriers
    • Providing information just about the project has the same effect as providing information on the benefits of participating in the project

​

​

​

Conclusions

17 of 18

  • Conduct follow-up survey (end of July)
    • Actual participation
    • Spillovers to female and male household members
    • Spillover to other households (male and female)

​

  • Spillovers
    • Network questions (Carter et al. 2021; Conley and Udry 2010)

​

​

  • If treatment does not affect actual participation,
    • Explore why?

​

​

​

​

Next steps

18 of 18

​

​

​

​

​

Thank You

​

​