1. Impact Evaluation Impact Evaluation for Evidence-Based Policy Making Arianna Legovini Lead, Africa Impact Evaluation Initiative AFTRL

2. 2 Answer Three Questions  Why is evaluation valuable?  What makes a good impact evaluation?  How to implement evaluation?

3. 3 IE Answers: How do we turn this teacher…

4. 4 …into this teacher?

5. 5 Why Evaluate?  Need evidence on what works Allocate limited budget Fiscal accountability  Improve program/policy overtime Operational research Managing by results  Information key to sustainability Negotiating budgets Informing constituents and managing press Informing donors

6. 6 What is different between traditional M&E and Impact Evaluation?  monitoring to track implementation efficiency (input-output) INPUTSOUTCOMESOUTPUTS MONITOR EFFICIENCY EVALUATE EFFECTIVENESS $$$ BEHAVIOR  impact evaluation to measure effectiveness (output-outcome)

7. 7 Question types and methods  Process Evaluation / Monitoring: Descriptiveanalysis Causalanalysis ▫What was the effect of the program on outcomes? ▫How would outcomes change under alternative program designs? ▫Does the program impact people differently (e.g. females, poor, minorities)? ▫Is the program cost-effective? ▫Is program being implemented efficiently? ▫Is program targeting the right population? ▫Are outcomes moving in the right direction?  Impact Evaluation:

8. 8 Which can be answered by traditional M&E and which by IE?  Are ITNs being delivered as planned?  Does school-based delivery of malaria treatment increase school attendance?  What is the correlation between health coverage and under fives receiving treatment within 24 hr of fever outbreak?  Does the house-to-house approach lead to an increase in under fives sleeping under ITNs relative to level in communities with other community-based approaches?  M&E  IE  M&E  IE

9. 9 Types of Impact Evaluation  Efficacy: Proof of Concept Pilot under ideal conditions  Effectiveness: At scale Normal circumstances & capabilities Lower or higher impact? Higher or lower costs?

10. 10 So, use impact evaluation to….  Test innovations  Scale up what works (e.g. de-worming)  Cut/change what does not (e.g. HIV counseling)  Measure effectiveness of programs (e.g. JTPA )  Find best tactics to e.g. change people’s behavior (e.g. come to the clinic)  Manage expectations e.g. PROGRESA/OPORTUNIDADES (Mexico) Transition across presidential terms Expansion to 5 million households Change in benefits Battle with the press

11. 11 Next question please  Why is evaluation valuable?  What makes a good impact evaluation?  How to implement evaluation?

12. 12 Assessing impact  examples How much does an anti-malaria program lower under-five mortality? What is the beneficiary’s health status with program compared to without program?  Compare same individual with & without programs at the same point in time  Never observe same individual with and without program at same point in time

13. 13 Solving the evaluation problem  Counterfactual: what would have happened without the program  Need to estimate counterfactual i.e. find a control or comparison group  Counterfactual Criteria Treated & counterfactual groups have identical initial characteristics on average, Only reason for the difference in outcomes is due to the intervention

14. 14 2 “Counterfeit” Counterfactuals  Before and after: Same individual before the treatment  Non-Participants: Those who choose not to enroll in program Those who were not offered the program

15. 15 Before and After Example  Food Aid Compare mortality before and after Find increase in mortality Did the program fail? “Before” normal year, but “after” famine year Cannot separate (identify) effect of food aid from effect of drought  Epidemic

16. 16 Before and After  Compare Y before and after intervention B Before-after counterfactual A-BEstimated impact  Control for time varying factors C True counterfactual A-CTrue impact A-B is under-estimated Time Y AfterBefore A B C t-1t Treatment B

17. 17 Non-Participants….  Compare non-participants to participants  Counterfactual: non-participant outcomes  Problem: why did they not participate?

18. 18 Exercise: Why participants and non-participants might differ?  Mothers who came to the health unit for ORT and mothers who did not?  Communities that applied for funds for IRT and communities that did not?  Children who received ACT and children who did not? Child had diarrhea Access to clinic Costal and mountain Epidemic and non-epidemic Child had fever Access to clinic

19. 19 Health program example  Treatment offered  Who signs up? Those who are sick Areas with epidemics  Have lower health status that those who do not sign up  Healthy people/communities are a poor estimate of counterfactual

20. 20 Health insurance example  Health insurance offered  Who buys health insurance?  Who does not buy?  Compare health care utilization of those who got insurance to those who did not  Cannot separately identify impact of insurance and utilization on health

21. 21 What's wrong?  Selection bias: People choose to participate for specific reasons  Many times reasons are directly related to the outcome of interest Health Insurance: health status and medical expenditures  Cannot separately identify impact of the program from these other factors/reasons

22. 22 Program placement example  Government offers family planning program to villages with high fertility  Compare fertility in villages offered program to fertility in villages not offered  Program targeted based on fertility, so Treatments have high fertility Counterfactuals have low fertility  Cannot separately identify program impact from geographic targeting criteria

23. 23 Need to know…  Why some get program and others do not  How some get into treatment and other in control group  If reasons correlated with outcome cannot identify/separate program impact from other explanations of differences in outcomes  The process by which data is generated

24. 24 Possible Solutions…  Guarantee comparability of treatment and control groups  ONLY remaining difference is intervention  In this workshop we will consider Experimental design/randomization Quasi-experiments  Regression Discontinuity  Double differences Instrumental Variables

25. 25 These solutions all involve…  Randomization Give all equal chance of being in control or treatment groups Guarantees that all factors/characteristics will be on average equal between groups Only difference is the intervention  If not, need transparent & observable criteria for who is offered program

26. 26 The Last Question  Why is evaluation valuable?  What makes a good impact evaluation?  How to implement evaluation?

27. 27 Implementation Issues  Political economy  Policy context  Finding a good control Retrospective versus prospective designs Making the design compatible with operations Ethical Issues  Relationship to “results” monitoring

28. 28 Political Economy  What is the policy purpose? In USA derail from national policy, defend budget In RSA answer electorate In Mexico allocate budget to poverty programs In IDA country pressure to demonstrate aid effectiveness and scale up In poor country hard constraints and ambitious targets

29. 29 Political Economy  Cultural shift From retrospective evaluation  Look back and judge To prospective evaluation  Decide what need to learn  Experiment with alternatives  Measure and inform  Adopt better alternatives overtime  Change in incentives Rewards for changing programs that do not work Rewards for generating knowledge Separating job performance from knowledge generation

30. 30 The Policy Context  Address policy-relevant questions: What policy questions need to be answered? What outcomes answer those questions? What indicators measures outcomes? How much of a change in the outcomes would determine success?  Example: teacher performance-based pay Scale up pilot? Criteria: Need at least a 10% increase in test scores with no change in unit costs

31. 31 Prospective designs  Use opportunities to generate good control groups  Most programs cannot deliver benefits to all those eligible Budgetary limitations:  Eligible who get it are potential treatments  Eligible who do not are potential controls Logistical limitations:  Those who go first are potential treatments  Those who go later are potential controls

32. 32 Who gets the program?  Eligibility criteria Are benefits targeted? How are they targeted? Can we rank eligible's priority? Are measures good enough for fine rankings? Who goes first?  Roll out Equal chance to go first, second, third?

33. 33 Ethical Considerations  Do not delay benefits: Rollout based on budget/administrative constraints  Equity: equally deserving beneficiaries deserve an equal chance of going first  Transparent & accountable method Give everyone eligible an equal chance If rank based on some criteria, then criteria should be quantitative and public

34. 34 Retrospective Designs  Hard to find good control groups Must live with arbitrary or unobservable allocation rules  Administrative data good enough to reflect program was implemented as described  Need pre-intervention baseline survey On both controls and treatments With covariates to control for initial differences  Without baseline difficult to use quasi- experimental methods

35. 35 Manage for results  Retrospective evaluation cannot be used to manage for results  Use resources wisely: do prospective evaluation design Better methods More tailored policy questions Precise estimates Timely feedback and program changes Improve results on the ground

36. 36 Monitoring Systems  Projects/programs regularly collect data for management purposes  Typical content Lists of beneficiaries Distribution of benefits Expenditures Outcomes Ongoing process evaluation  Information is needed for impact evaluation

37. 37 Evaluation uses information to:  Verify who is beneficiary  When started  What benefits were actually delivered Necessary condition for program to have an impact:  benefits need to get to targeted beneficiaries

38. 38 Improve use of monitoring data for IE  Program monitoring data usually only collected in areas where active Collect baseline for control areas as well  Very cost-effective as little need for additional special surveys Add a couple of outcome indicators  Most IE’s use only monitoring data

39. 39 Overall Messages  Impact evaluation useful for Validating program design Adjusting program structure Communicating to finance ministry & civil society  A good evaluation design requires estimating the counterfactual What would have happened to beneficiaries if had not received the program Need to know all reasons why beneficiaries got program & others did not

40. 40 Design Messages  Address policy questions Interesting is what government needs and will use  Stakeholder buy-in  Easiest to use prospective designs  Good monitoring systems & administrative data can improve IE and lower costs