Download to read offline
More Related Content
![[e-Government Program Action Plan : Guatemala City]](https://cdn.slidesharecdn.com/ss_thumbnails/03guatemalaactionplan-140728023606-phpapp02-thumbnail.jpg?width=640&height=640&fit=bounds)
Similar to Randomized Controlled Trials: Insights for Civic Tech
Randomized Controlled Trials: Insights for Civic Tech
- 1. Randomized controlled trials: insightsfor Civic Tech Andrew Westbury Center for Effective Global Action University of California, Berkeley [email protected] @andrewwestbury
- 2. Agenda • Who CEGAis • What we are doing on effectiveness in public institutions • Where we are headed (insights and open questions) • How you can get involved
- 4. “End poverty inall its forms everywhere”
- 5. ($130 Billion) Official developmentassistance. Public policy evaluations in low-income countries. (<$100 Million) Yearly average expenditure on...What’s the challenge?
- 6. 0 5 10 15 20 Intel Samsung GoogleFacebook Merck International Development Community What’s the challenge? Percentage of revenue spent on R+D in 2014 <1%
- 7. Identify the big problems Createproducts + solutions Rigorously field test ideas Iterate and adapt Support scale-up What we do.
- 8. Center for EffectiveGlobal Action • Network of 70+ faculty researchers across the West Coast, USA • 10 portfolio areas from agriculture to public health • Presence in 40 countries with government and NGO partnerships • 180 grants within the network to fund new products and solutions
- 9. Association between GDPper capita (Penn Tables, average 1990–1999, PPP) and democracy (Freedom House) Source: Acemoglu et al. (2008)
- 10. Economic Development &Institutions Goal: Provide insights into institutional changes and their impact on economic development Donor: This project is funded with UK Aid from the UK government Partners: Oxford Policy Management Limited University of Namur Paris School of Economics Aide a la decision economique
- 11. Randomized Evaluations Core component:Portfolio of randomized controlled trials (RCTs) to test institutional reforms designed to support sustainable economic development throughout the developing world. Cutting-edge RCTs: Inform their local institutional contexts and have high-potential for external validity. The Center for Effective Global Action (CEGA) at UC-Berkeley: Leading efforts related to EDI’s RCT portfolio.
- 12. What kinds ofprograms are we targeting? Thematic areas: Political institutions Legal institutions Information campaigns X X Selection and incentives X X Planning “linkages”: • To further important findings in current literature, include replication incentives • To improve synthesis across studies, share measurement approaches/tools
- 13.
- 14.
- 15.
- 16. The Counterfactual Problem: thecounterfactual cannot be observed Solution: We need to “mimic” or construct the counterfactual
- 17. Outcomes for bothgroups are measured Intervention Comparison Population is randomly split into 2 or more groups
- 18. 18 Key advantage ofexperiments Because members of the groups (treatment and control) do not differ systematically at the outset of the experiment Any difference that subsequently arises between them can be attributed to the program rather than to other factors Fewer assumptions, clearly explainable results 18
- 19. Impact evaluation canprovide insight into specific program activities and offer direction for major strategic decisions • Example: Microcredit • Throughout the 2000s microcredit was hailed as a “silver bullet” in international development • 2005 was designated as the “Year of Microcredit” by the United Nations, and in 2006 Muhammed Yunus won a Nobel prize for founding the Grameen Bank, one of the world’s first microcredit organizations • However, a series of rigorous RCTs proved in multiple geographies in multiple contexts that group lending strategies were not effective at reducing poverty • This results provided important insights into the investment strategies of donors to microcredit and changed the strategic approach Simple concept, profound impact
- 20. Where we areheaded
- 21. Landscape assessment • Hugegaps in evidence on governance issues • None of the questions really new, but some areas: • Have received virtually no empirical attention • Covered mostly by cross-country approaches • Others relied on micro, observational data • It’s an open question as to what state-of-the-art evaluation methods will find... • There are a number of organizations working to fill this gap, including the Evidence for Politics and Governance, The Abdul Latif Jameel Poverty Action Lab at MIT, CEGA, the World Bank, and IFC
- 22. An expanded rolefor technology? Existing and ongoing research • Surveyed past 5 years in the developing world • 175 RCTs (published and ongoing) • Leading supporters of impact research (J-PAL, EGAP, IGC, World Bank) • Identify geographic and thematic gaps Technology for governance Impact evaluations Tech Non-tech
- 23. “At the Intersection:A Review of Institutions in Economic Development” • Authors: Fred Finan and Ernesto Dal Bo (UC Berkeley) • 200 studies on a range of governance issues: • Administration of court and legal systems, monitoring of service delivery, political norms, electoral issues, and state capacity • Developed and developing world • Identified “40 open research questions”
- 24. Area of interest:information and transparency Thematic areas: Political institutions Legal institutions Information campaigns X X Selection and incentives X X
- 25. An asymmetrical relationship Politicians Citizens Stateof the world Competence Motives Effects of policy
- 26. Municipal audits inBrazil • Does access to information lead citizens to hold governments accountable? • Municipalities with mayors eligible for re-election in 2004 • Mayors with at least two corruption violations are significantly less likely to get re- elected (17% points) Ferraz & Finan (2008)
- 27. Voter Information inthe Philippines • Information to voters on spending programs, proposed allocations, promises of mayoral candidates • Changes in voter knowledge about the program and incumbent politicians • Incumbent politicians responded by increasing resources targeted at voters through vote buying Cruz et al (2015)
- 28. Political Debates: SierraLeone • Debates as a vehicle to provide information to voters • 112 of 224 randomly selected voting locations in the five weeks before the election • Politicians exposed to treatment: increased campaigning effort, as measured by gift giving, the monetary value of gifts, and the number of in-person visits Bidwell, Casey, and Glennerster (2015)
- 29. Open question: What determineswhether higher transparency leads to better political selection or a reinforcement of harmful political strategies by incumbents?
- 30. Where does civictechnology fit in?
- 31. Partnerships, matchmaking, andgrant making Partnerships • CEGA partners with practitioners to build research protocols into new initiatives to test results and generate insights • Expand in the civictech/govtech space • Liaison – understand your initiative, your goals, strategy and “match make” with researchers Grant making • Series of “research competitions” ongoing • Offering support for pilot and full-scale studies • Collaborations between faculty researchers and practitioners
- 32. Evidence to ActionSymposium: Smart Government Harnessing Technology for Public Good Top down • Smartphone monitoring for the public sector in Pakistan • Machine learning to fight corruption in Brazil • Sensors to improve public transport in Nairobi Bottom up • Social media and censorship evasion in China • Broadcasting political debates in Sierra Leone • Digital platforms and political participation in South Africa Orange Silicon Valley, San Francisco, May 10th
- 33.