Innovations for Poverty Action, the most influential charity you’ve never heard of, tackles poverty in less developed countries by rigorously investigating the effectiveness of poverty-reduction programmes, and supporting the expansion of the best. Its main investigative tool is the randomised control trial, developed in medical research (where it may well have quietly saved your life several times) now increasingly used elsewhere. Randomised control trials (RCTs) are powerful because they, uniquely, demonstrate what happened which would not have happened otherwise, i.e., the true impact of a programme. The sample size, duration and rigour of IPA’s studies makes them a great deal more robust than most charity/development evaluations.
IPA has run over 350 studies in 40 countries, across many aspects of poverty, including health, education, agriculture, governance, micro-finance and environment. Its research is normally led by tenured academics, many at universities such as Harvard, University College London, the London School of Economics Massachusetts Institute of Technology, Yale, and most of it is published in respected peer-reviewed journals. The evaluations are normally of programmes run by NGOs and/or governments. IPA now has 600 staff in 14 offices. Typical research questions include:
– Do better cook-stoves really save all the carbon, fuel, time and health problems which their advocates claim? (Answer: sadly not.)
– Which is a better way of getting teachers to turn up to school in India: giving them financial incentives to attend, or putting cameras in class rooms? (Answer: cameras are miles better, though obviously affect the power dynamic.)
– To dissuade poor parents from taking their children out of school, you might give them cash when their children come to school (‘a conditional cash transfer programme). The conditionality is quite expensive to administer, so does it matter if you ditch it? (Answer: not in Colombia, but yes in Malawi.)
Caroline Fiennes, Director of Giving Evidence, is now working with IPA in Europe, aiming to raise awareness of its method, usage of its research findings, and resources to enable further study and further uptake of the findings. I’m delighted: given my passion for giving based on the evidence of what works, it’s natural to support more which generates, uses and shares that evidence.
That extreme poverty persists, and now only a few hours flight from our comfortable lives, is one of the biggest shames of our age; and IPA deploys against it the scientific method, perhaps mankind’s greatest achievement. As a result, it has a cupboard full of surprising insights which need to be heard and used.
IPA sounds a lot like J-PAL (the Abdul Latif Jameel Poverty Action Lab at MIT). Is it the same?
IPA and J-PAL are sister organisations, both doing RCTs in development, and they collaborate more closely than any charities I’ve ever seen. For instance, their names – which share the ‘poverty action’ bit – are deliberately similar.
J-PAL is a network of academics, based at MIT and now with centres in four other universities (in India, South Africa, Latin America, Indonesia). Hence J-PAL’s studies are all undertaken by academics. IPA is a free-standing charity. Its studies are often led by academics (many of whom are affiliates of both IPA and J-PAL) but sometimes led by IPA staff. This provides an important freedom. Academics are incentivised to do studies which will get published, which encourages innovative studies. But suppose somebody’s studied cook-stoves in northern Kenya and now we need to see if the surprising findings are also true in southern Kenya: despite being important, that study is pretty unattractive to an academic, but IPA might well do it.
IPA and J-PAL deliberately avoid having offices in the same countries as each other.
By a weird quirk of fate, the leaders of J-PAL and IPA are actually sisters. J-PAL was co-founded by Esther Duflo (MIT professor, and co-author of the prize-winning book Poor Economics), and IPA’s CEO is Annie Duflo. IPA was founded in 2002 by Professor Dean Karlan at Yale University: Dean did his PhD with Abhijit Banerjee, the other co-founder of J-PAL.
Are RCTs really better than other forms of evaluation?
Yes, a lot (in the right circumstances). Explained here, about Goldman Sachs, and here about evaluations in general.
But RCTs can’t be used in all circumstances.
Correct. Nobody said they could. Their limitations, and what to do about them, is discussed in my book.
Doesn’t helping IPA raise money conflict with Giving Evidence’s work advising donors?
Not really. First, it’s hardly a secret – and I discuss it explicitly with donors to which it pertains. Second, it’s no secret that I advocate giving based on evidence, and hence is hardly surprising that I’m working with the people who generate it. Third, in any instance where a donor is interested in development work, I’m ‘acting for’ the donor rather than IPA and IPA knows that. Fourth, I have no performance incentive with IPA. Etc etc.
IPA sounds great. How do I…
– find out more? Here
– find its research? (IPA’s research is all summarised here (and here are summaries of J-PAL’s research.)
– give it money? Either here, or we’re setting up a UK mechanism: coming soon.
– support good programmes it’s discovered? Either here, or we’re setting up a UK mechanism: coming soon
– contact you for advice about my/ my company’s giving? At enquiries [at] giving-evidence [dot] com
The FT on how (as result of J-PAL’s work) development is becoming a science—>
The Economist on findings of a J-PAL study and IPA replication about nailing poverty—>
Bloomberg on how J-PAL, IPA and RCTs “speed our way toward a better world”–>
RCTs find how to buy one, get 24 for free!-