Our benchmark

The other day I had the chance to visit with Rohit Wanchoo, one of the three co-founders of Give Directly.

The idea behind Give Directly is so shockingly simple that you can’t help but feel, at first, that it can’t possible work.  The model is: donors give to GiveDirectly; GiveDirectly gives cash to poor households in northwestern Kenya.  Done.

Now here is some of the surprising information: this kind of giving (when you get fancy you call it either “conditional cash transfers” or “unconditional cash transfers”) is being practiced at a huge scale.  According to Rohit, it’s received by nearly 1 billion people globally.  In Mexico, it is called the Oportunidades program and it reaches about 5.8 million families, or 30% of the population.  In Brazil it’s called Bolsa Familia, it reaches 50 million Brazilians, and the program is much of the reason why, according to the NY Times, the incomes of the poor in Brazil grew seven times faster than the incomes of the rich from 2003 to 2009.

It’s not just that paying money increases incomes – how could it not?  Study after study has shown significant improvements in well-being, everything from improved childhood nutrition, increased birth weights, decreased HIV infection rates, increases in schooling and decrease in child labor. Also, the evidence shows that cash transfer programs do not increase spending on things like alcohol and tobacco.   And these are not just initial findings.  In fact, DfID, the UK’s main aid agency, recently published a comprehensive summary of the evidence to date, and DfID notes that cash transfers are “one of the more thoroughly researched forms of development intervention.”  Put another way: it’s widespread and has been studied a lot, and by and large the evidence is really positive.

So how does it work in practice?  At Give Directly – which is a new, small startup, though they just won a $2.4 million Google impact award – they go to a poor community in northwestern Kenya, survey households to determine which ones are the poorest (mostly by seeing whether houses have mud floors and thatched roofs – those that do qualify), and, for those that qualify, they give those households $1,000 over the course of one year, using cellphone-based money transfer powered by M-PESA.  There’s also a bunch of verification to check identity and protect against fraud.  And households can only receive funds for one year.  It sounds like a lot of steps, but all of this costs just a few cents on the dollar.  And from day 1, GiveDirectly has been opening up its to rigorous analysis to understand impact – not surprising since Rohit and his co-founders have Masters and PhDs in economics or development studies.

Just as a standalone idea, in terms of its direct impact, it’s interesting and important to understand this model.

But the big idea that really caught my attention was Rohit’s statement that what they really want to accomplish is to have direct cash transfer be the benchmark against which other poverty-alleviation interventions are measured.  Meaning, the development sector as a whole spends a huge amount of effort to demonstrate impact, and the benchmark is to show a positive impact relative to no outside intervention.  But couldn’t we, shouldn’t we, raise that bar?  Shouldn’t we be asking ourselves whether a given intervention (buying a cow, delivering food, or a vaccination or leadership training or anything) is more effective than the direct cash value of the program itself?

That sounds like a low hurdle to clear, but it’s higher than the hurdle the development sector is trying to clear today.

I haven’t studied the literature on cash transfer enough to really understand what is and isn’t known, what are the best ways to design the programs, what are the pitfalls.  Still, it seems like a very fair question to ask ourselves, to ask the big aid agencies, to ask the government: would your beneficiaries be better off if you just handed them the money?