So proud of everyone involved in this study, 2nd of the @USAID #cashbenchmarking initiative. @AndrewZeitlin and Craig McIntosh, @EDCtweets, @GiveDirectly, @USAIDRwanda, @poverty_action, @CEGA_UC , @LizBrow29385570 and the incomparable @DanielHandel17
BIG REFLECTIONS THREAD https://twitter.com/poverty_action/status/1301542920940589059">https://twitter.com/poverty_a...
BIG REFLECTIONS THREAD https://twitter.com/poverty_action/status/1301542920940589059">https://twitter.com/poverty_a...
Setting up & executing high-quality RCTs is really HARD. @deankarlan has a book about the things that can go wrong. If N = number of things that can go wrong in a standard RCT, evaluating 2 orgs with distinct interventions is not 2N, its N^2. H/T to PIs & IPs managing complexity!
@USAID relies on both hierarchical AND consensus-driven decision-making. Table stakes for any new program: bring all of the relevant stakeholders into the fold, BUT ALSO buy-in from leaders. These studies couldnt have happened w/o @DanielHandel17 and an ARMY of other advocates.
USAID also relies heavily on precedent. Costs of innovation are high to innovators, but once you& #39;ve done something new like #cashbenchmarking or #impactbond with @village_ent & @DFID, its easier for the rest of the agency to replicate the innovation. https://twitter.com/village_ent/status/917811982509551616?s=20">https://twitter.com/village_e...
This is why programs like #DIVatUSAID in the @GlobalDevLab are essential: they take risks for the rest of the Agency. As a 1st-mover in the case of the VE impact bond, in generating evidence that wouldnt be produced w/cash benchmarking. And it pays off! https://scholar.harvard.edu/kremer/publications/development-innovation-good-investment-which-innovations-scale-evidence-social">https://scholar.harvard.edu/kremer/pu...
But the key questions are: do the innovations scale? Does the agency change its approach in response to evidence?
Well.... these are the questions we should be asking after the 2 benchmarking studies in Rwanda!
So what have we learned?
Well.... these are the questions we should be asking after the 2 benchmarking studies in Rwanda!
So what have we learned?
Newest study has 2 main findings:
https://abs.twimg.com/emoji/v2/... draggable="false" alt="1⃣" title="Tastenkappe Ziffer 1" aria-label="Emoji: Tastenkappe Ziffer 1"> Neither cash nor the job training program (Huguka Dukore or HD)
https://abs.twimg.com/emoji/v2/... draggable="false" alt="⬆️" title="Pfeil nach oben" aria-label="Emoji: Pfeil nach oben"> employment *main outcome of interest*
https://abs.twimg.com/emoji/v2/... draggable="false" alt="2⃣" title="Tastenkappe Ziffer 2" aria-label="Emoji: Tastenkappe Ziffer 2"> Cash *outperforms* HD on all other primary and secondary outcomes except business knowledge
So HD does not clear the cash benchmark
So HD does not clear the cash benchmark
Put this evaluation in context: a recent performance evaluation by @DexisConsulting concluded "the majority of youth find new or improved employment after the HDAK project and therefore have higher incomes now than before training"
Is the perf eval wrong?
https://pdf.usaid.gov/pdf_docs/PA00WGVG.pdf">https://pdf.usaid.gov/pdf_docs/...
Is the perf eval wrong?
https://pdf.usaid.gov/pdf_docs/PA00WGVG.pdf">https://pdf.usaid.gov/pdf_docs/...
NO, but it does not evaluate HD against a COUNTERFACTUAL (what happens in the absence of the program). If you look at Andy/Craig& #39;s study, you& #39;ll see control employment rt. @ baseline = 33%, endline = 48%.
People tend to go from unemployed to employed over time... yeah!
People tend to go from unemployed to employed over time... yeah!
The timeline of the performance eval, btw, partially overlapped with the study period. So roughly same point in time.
Therefore a majority of HD participants can find employment and increase income relative to pre-program... AND HD CAN STILL HAVE NO IMPACT on income/employment.
Therefore a majority of HD participants can find employment and increase income relative to pre-program... AND HD CAN STILL HAVE NO IMPACT on income/employment.
Zooming out a bit: that Dexis performance eval - historically that& #39;s the main type of eval the agency conducts on its activities. ~90+% of all evaluations ever conducted at USAID are performance evals! There& #39;s useful info in there... but it cant tell us if we& #39;re really
https://abs.twimg.com/emoji/v2/... draggable="false" alt="⬆️" title="Pfeil nach oben" aria-label="Emoji: Pfeil nach oben">outcomes
But a revolution is brewing at USAID: an impact evaluation revolution! @BrookingsInst
https://www.brookings.edu/blog/future-development/2018/10/08/a-quiet-revolution-in-impact-evaluation-at-usaid/
This">https://www.brookings.edu/blog/futu... recent study is part of a new wave of more rigorous research being supported by @USAID - we should celebrate this while acknowledging how far we still have to go!
https://www.brookings.edu/blog/future-development/2018/10/08/a-quiet-revolution-in-impact-evaluation-at-usaid/
This">https://www.brookings.edu/blog/futu... recent study is part of a new wave of more rigorous research being supported by @USAID - we should celebrate this while acknowledging how far we still have to go!
The 1st benchmarking study had similar findings:
https://abs.twimg.com/emoji/v2/... draggable="false" alt="1⃣" title="Tastenkappe Ziffer 1" aria-label="Emoji: Tastenkappe Ziffer 1"> Neither the child nutrition program (CNP) nor cost-equivalent cash improved anthropometrics or diet. diversity
https://abs.twimg.com/emoji/v2/... draggable="false" alt="2⃣" title="Tastenkappe Ziffer 2" aria-label="Emoji: Tastenkappe Ziffer 2"> Cash
https://abs.twimg.com/emoji/v2/... draggable="false" alt="⬆️" title="Pfeil nach oben" aria-label="Emoji: Pfeil nach oben">a range of other outcomes more than the CNP
CNP did not clear cash benchmark. https://www.poverty-action.org/study/benchmarking-wash-and-nutrition-program-cash-rwanda">https://www.poverty-action.org/study/ben...
CNP did not clear cash benchmark. https://www.poverty-action.org/study/benchmarking-wash-and-nutrition-program-cash-rwanda">https://www.poverty-action.org/study/ben...
Where does this leave us? Are we just peering into the abyss?
I am reminded of this great piece by @80000Hours. Two big takeawys:
https://abs.twimg.com/emoji/v2/... draggable="false" alt="1⃣" title="Tastenkappe Ziffer 1" aria-label="Emoji: Tastenkappe Ziffer 1"> Most (70-90%) social programs don& #39;t hold up under rigorous evaluation
https://abs.twimg.com/emoji/v2/... draggable="false" alt="2⃣" title="Tastenkappe Ziffer 2" aria-label="Emoji: Tastenkappe Ziffer 2"> There are still interventions we know WORK WELL https://80000hours.org/articles/effective-social-program/">https://80000hours.org/articles/...
I am reminded of this great piece by @80000Hours. Two big takeawys:
So lets do more of the stuff that works and continue rigorously evaluating so we can LEARN. and huzzah to USAID for this bold research agenda - onward!
And ask people like @AnneHHealy @MichaelEddy @NormaAltshuler @sasha_gallant @KareninKenya @DanielHandel17 about what works!
And ask people like @AnneHHealy @MichaelEddy @NormaAltshuler @sasha_gallant @KareninKenya @DanielHandel17 about what works!