Update (04/23/18): We’ve updated our take on the HS18 results, check out our latest here.

Back in 2011 we announced our first randomized controlled trial (RCT) in Rarieda, Kenya. A lot has changed over that time – we’ve moved to new areas, changed our transfer design ($1,000 lump-sum transfers are now GiveDirectly’s standard, whereas most Rarieda treatment households received $300, and a minority $1,000 spread over 9 payments), and changed eligibility rules, most recently to enrolling entire villages in some areas. We now have several other RCTs underway testing the impact of these designs.

We embark on these studies not because they’re easy (they’re not), produce results quickly (they don’t), or always tell a straightforward story (more on that later), but because changing the lives of those living in extreme poverty is too important an endeavour to leave to hunches. One important question is about the duration of impacts: while there is more evidence about the long-term impacts of cash than just about any other intervention, there is still a lot to learn about how long impacts persist, and for whom.

We’re excited that first draft results measuring impacts 3 years after transfers began have been released for the Rarieda study, providing new evidence about the longer-term impact of GD transfers, albeit for a different transfer design to what we use today.

Overall the findings are encouraging. The treatment effects on all the main outcomes (assets, earnings, expenditure, food security, and psychological wellbeing) were sustained after 3 years. Gains on an education index that were not significant at 9 months also becomes significant at 3 years, driven by increased spending on school fees, uniforms, books and supplies. The size of these impacts at 3 years are broadly similar to those at 9 months – in fact, the impact on assets increases significantly, even though the value of assets owned by control households doubled over that time.

The main caveat is that the study generally is not able to estimate “spillover” effects precisely. Spillovers refer to impacts on households that did not receive transfers themselves, but lived near other households that did. There are fewer such households today (particularly where we’ve switched to enrolling entire villages), but in the early days of the Rarieda study, the researchers randomized which eligible households received money initially try to measure spillovers and maximize statistical power to measure the impact of cash. (We later gave money to this control group as well on ethical grounds.)

To measure spillovers, the study compared control households in ‘cash villages’ to households in ‘pure control’ villages (where nobody received cash). There was no baseline survey for the pure control sample, and a range of analytical approaches were deployed to address other methodological challenges, including the late determination of ‘eligibility’ in pure control villages and higher study dropout rate in pure control villages. This means that multiple estimates are reported for every potential spillover effect, and most spillover impacts were not significant after these approaches to bounding the uncertainty were applied (the exception being a negative spillover for consumption).

The paper concludes that “a larger sample is needed to robustly assess treatment effects across villages and to validate spillover effects”. We couldn’t agree more, and that’s why we can’t wait for the results of the large study measuring the 18 month impacts of $1,000 lump sum cash transfers across 650 villages that’s in the works.

Michael Cooke is GiveDirectly’s Research Director.

Back to List