Rigorously evaluating cash transfer programs in the United States: Perspectives from other contexts
Cash transfer interventions—cash payments that can be one-time or recurring, conditional or unconditional, and direct or indirect—are gaining more attention worldwide. In September 2023, researchers conducting randomized evaluations of cash transfer programs in the United States gathered at Duke University to discuss their ongoing and completed research projects. J-PAL affiliate Paul Niehaus (UC San Diego) kicked off the conference with a keynote speech highlighting his work with GiveDirectly and research on cash transfers conducted in low- and middle-income countries. In this post, he shares his perspective on these studies and on how the United States can learn from other contexts.
You kicked off this convening with a keynote address about cash transfers in low- and middle-income contexts. Can you share what rigorous research on these programs has taught us so far?
There’s been an enormous amount—300 or so high-quality studies according to the 3ie Evidence Portal—so there is no quick synthesis. But in the talk, I emphasized what we’ve learned in low and middle income contexts on three broad questions that people often have about cash transfers.
- What have recipients done with the money? Across a wide range of contexts, across many studies, we see people spending money sensibly on things that improve their lives, often on things that matter for both the future (e.g., assets, education) and the the present (e.g.,. We haven’t seen evidence of much money being spent on things that could be self-harmful (e.g., alcohol, tobacco), which many policymakers are worried about. None of this proves that these are the “best” possible uses of money, but the overall picture has been positive.
- What do recipients do differently with their time? This question centers in particular on how much people work. There are some tensions there: funders often worry that recipients will work less (which they would view as a problem), while economists tend to take the view that leisure is a good thing. In any case, the pattern has been that,if anything, recipients typically work a bit more and/or shift some hours out of wage employment and into self-employment.
- How has all of this affected more subjective measures of well-being? Here we’ve seen reductions in many, though not all, studies on standard measures of depression or anxiety. I don’t think these are the end-all for measurement—speaking for myself, I’ve certainly felt more anxious after getting some big grants and that doesn’t mean it was a bad thing. There are also some results I find really striking. For example, some studies show reductions in the suicide rate due to transfers, capturing the effects of cash on extreme distress in a very poignant way.
You suggested three “more interesting” questions at the end of your presentation. Could you share what those are and why you think they’re important?
The questions above are common ones and totally understandable, but not actually the ones I would ask first as a policymaker trying to figure out how much of my budget to allocate to cash transfers and how to structure them.
First, I’d focus more attention on who benefits the most from cash. Part of this question is whether we can get better—and faster—at figuring out which households to give money to, and how much to give. Households themselves don’t always allocate resources equally, and it’s hard to measure who gets what within them. It would be useful to know how much individual people are impacted when their household gets a transfer, and whether we can target transfers based on those impacts.
Second is whether transfers are structured in ways that work well for the financial lives of people living in or near extreme poverty. Managing cash flow can be very tricky because their incomes and expenses tend to be volatile and the financial tools and services available to them are generally inadequate. So we want to structure transfers in ways that help. People often have a hard time saving up large lump sums, for example, so it could be valuable to build that option into public programs rather than requiring that everyone get small monthly payments.
Third is how transfers affect the people who don’t get them. We can be fairly sure there will be some indirect effects, if only because people who get money use it to transact, and that affects others involved in those transactions. Most of the evidence comes from evaluations that are too small to see these kinds of effects. There are a few notable exceptions: in one project I contributed to, we estimated that the local economy grew by US$2.5 for every US$1 transferred in. These effects are harder to value—it’s not as simple as saying “you got $1, which is worth $1,”—but are worth prioritizing.
After your keynote, you received a handful of questions about the place of governments and philanthropy in funding and administering cash transfers. Can you talk a bit about what roles you see for these and other stakeholders?
The usual maxim that philanthropy can afford to take risks that government or other public actors cannot certainly applies here. To take an example, a lot of people think that cash transfers won’t work well in more remote areas where people live further from markets. To test this theory, GiveDirectly (which I co-founded) identified several remote communities, gave people money, and evaluated the impacts. There’s value in testing conventional wisdom by trying things that people presume will not work.
That said, we sometimes overthink our philanthropy. My view is that simply giving money to low-income people and letting them decide how to use it is a great opportunity. It doesn’t always need to fit into a grand strategy for leveraging or influencing how other people deploy their money.
What advice would you give to researchers looking to study this topic?
The core advice I’d give is to try to get to conceptual clarity on what questions we want to answer and why. And to recognize that in doing that you’re unavoidably taking an ethical stance, which is something researchers often don’t like to own up to.
We live in this golden age of causal inference where we’ve become much better at measuring the effect of X on Y. There’s a lot to celebrate in that. But it doesn’t mean that it’s important that we measure the effect of every X on every Y. Cash transfers are a case in point because by definition they’re very flexible and so they tend to have some effect on pretty much everything. We could go on endlessly trying to estimate all those effects, trying to figure out why they are different in different contexts, how to deal with the fact that no one study is ever going to measure all the impacts that actually happened, and then trying to figure out how to communicate about all of that.
To me that misses the point. The main purpose of giving someone $1 is for them to have $1. Going after questions like, “are cash transfers a cost-effective way to improve food security,” say, can be problematic because the objective is narrow. I don’t think the ratio of food security to spending is an acceptable criterion for ethical decision-making. Ethically good decisions are going to depend on answers to some other questions like the ones above—what are the externalities, the effects on non-recipients, for example. So I try to focus on those.
In an accompanying post, Lisa Gennetian and J-PAL staff share key takeaways and reflections from researchers in the United States evaluating cash transfer programs.