This is not another post about the evidence for cash transfers (we have plenty of those, and you can read them all here and here). Rather, it’s about how cash transfers offer a useful lens to analyze the effectiveness of other development programs.
Yesterday I attended a seminar with impact evaluation results presented on four different public works (PW) programs around Africa: one from Malawi, one from Sierra Leone, one from Côte d’Ivoire (paper not out yet), and one from Ethiopia (here and here). Public works, for those unfamiliar with the concept, are programs that provide public employment to the poor, often working on road repair or some other community project. With the exception of the Malawian program, all these programs had at least some positive impacts. Yay!
But in each case, I found myself thinking of a simple unconditional cash transfer program as a reference. This isn’t because unconditional cash transfers solve all problems. It’s because they are simple and transparent. They solve one problem (liquidity constraints). The costs are relatively straightforward: In addition to the actual transfers, the administrative costs are unlikely to be higher than those for other, more complex developing programs, like a package of services for the ultra-poor, or a training program.
So then, rather than asking whether a development program has a positive impact, we can ask how it stands up next to a transfer of cash. Does it cost more than simply transferring cash? If so, does it offer commensurate returns?
This is a pretty basic economic concept, but I repeatedly find impact evaluations asking whether programs find a positive impact – the equivalent of saying, Is giving people a bunch of stuff better than absolutely nothing? – rather than thinking about a simple plausible alternative. Here is how this thinking could play out in the context of a public works program.
How are public works programs different from cash?
1. How do public works programs stand up next to cash in terms of offering direct resources to the poor? In short, they offer less money to the poor than simply transferring cash. In the Sierra Leone project, household income rose only about two-thirds the amount of the income due to the households. In Côte d’Ivoire, the earnings gains were much less than the value of the transfer, in part because of substituting from regular work into the public works program.
So PW programs offer less money to households than cash. But at the same time, public works programs offer a job! (Sort of. A short-term job, within the context of the project.) A job is better than cash, right?
Why might we like jobs better than cash?
2. By encouraging people to work for the money they receive, perhaps PW programs reduce dependency. Last year, Banerjee et al. showed – across 7 government-implemented cash transfer programs – that there is “no systematic evidence that cash transfer programs discourage work.” So even if PW programs do discourage dependency, it’s not an advantage relative to cash transfers.
3. By providing people with work experience, perhaps PW programs increase future employability. People are working. Maybe they’re gaining some basic skills. Of the four programs yesterday, only the Côte d’Ivoire program examined employment after the program. They observe some post-program impacts on self-employment earnings, particularly for participants who did PW combined with self-employment training. Of course, cash transfers have also shown increased earnings several years later in Uganda (Blattman et al.).
4. By providing people with work, perhaps PW programs reduce unrest. We have all these unemployed youth; perhaps PW programs make them less likely to cause social problems. None of the studies examined this directly. The Côte d’Ivoire study showed positive impacts on psychological well-being. We have evidence that cash transfers can also have positive psychological effects in the short-run, in Malawi (Baird et al), although the story is nuanced. They also improved happiness and reduced stress in Kenya (Haushofer & Shapiro). It would be very interesting to see results of PW programs on actual unrest or engagement in disruptive activities.
5. Public works programs may reduce the need for detailed targeting systems. The idea is that by offering a relatively low wage, the only people who will want the public works program are those who really need it. That would remove the need for complicated proxy means tests or community targeted processes, often used in cash transfer programs. However, in Sierra Leone, Malawi, and Côte d’Ivoire, the evidence suggests that the wage isn’t low enough to accomplish that kind of sorting, so the programs don’t end up actually targeting the poorest; and lower wages run into political and potential ethical problems, being lower than national minimum wages. So while this argument may work in theory, it doesn’t seem to be a realized advantage of PW programs in practice.
6. Public works programs actually build something, in addition to giving people resources. Of course, the workers in the PW program are actually building something to improve their community. In Sierra Leone, Malawi, and Côte d’Ivoire, participants did road maintenance. In Sierra Leone and Malawi, they also worked on other community projects. In Ethiopia, participants worked on watershed development (conservation activities that later allow better irrigation). This could be a big plus over a simple cash transfer program: Maybe these communities have great roads at the end of the project, and that has additional benefits on commerce. Only the Ethiopia evaluation showed results on this (although it was simple before-after), but it looked like water access was up and waterborne diseases were down, among other outcomes. Identifying these impacts well is more difficult, especially if randomization of the intervention is within-community, but I’d propose it’s worth trying.
Take away. How do public works programs hold up? On the areas that have actually been measured, it’s not clear that they outperform a simple cash transfer. I hope that these and future evaluations can make sure to report on outcomes that may make PW programs a preferable alternative.
Ultimately, the proposal here is to scrutinize each development program – whether it’s providing training or giving away shoes or a package of services targeted to the ultra-poor or a merit-based scholarship – not simply on whether it has a positive impact. Positive impacts don’t always show up, but they’re still too low a bar. Rather, we can consistently ask whether the program would likely do better than cold, hard cash.
Bonus reading
Yesterday I attended a seminar with impact evaluation results presented on four different public works (PW) programs around Africa: one from Malawi, one from Sierra Leone, one from Côte d’Ivoire (paper not out yet), and one from Ethiopia (here and here). Public works, for those unfamiliar with the concept, are programs that provide public employment to the poor, often working on road repair or some other community project. With the exception of the Malawian program, all these programs had at least some positive impacts. Yay!
But in each case, I found myself thinking of a simple unconditional cash transfer program as a reference. This isn’t because unconditional cash transfers solve all problems. It’s because they are simple and transparent. They solve one problem (liquidity constraints). The costs are relatively straightforward: In addition to the actual transfers, the administrative costs are unlikely to be higher than those for other, more complex developing programs, like a package of services for the ultra-poor, or a training program.
So then, rather than asking whether a development program has a positive impact, we can ask how it stands up next to a transfer of cash. Does it cost more than simply transferring cash? If so, does it offer commensurate returns?
This is a pretty basic economic concept, but I repeatedly find impact evaluations asking whether programs find a positive impact – the equivalent of saying, Is giving people a bunch of stuff better than absolutely nothing? – rather than thinking about a simple plausible alternative. Here is how this thinking could play out in the context of a public works program.
How are public works programs different from cash?
1. How do public works programs stand up next to cash in terms of offering direct resources to the poor? In short, they offer less money to the poor than simply transferring cash. In the Sierra Leone project, household income rose only about two-thirds the amount of the income due to the households. In Côte d’Ivoire, the earnings gains were much less than the value of the transfer, in part because of substituting from regular work into the public works program.
So PW programs offer less money to households than cash. But at the same time, public works programs offer a job! (Sort of. A short-term job, within the context of the project.) A job is better than cash, right?
Why might we like jobs better than cash?
2. By encouraging people to work for the money they receive, perhaps PW programs reduce dependency. Last year, Banerjee et al. showed – across 7 government-implemented cash transfer programs – that there is “no systematic evidence that cash transfer programs discourage work.” So even if PW programs do discourage dependency, it’s not an advantage relative to cash transfers.
3. By providing people with work experience, perhaps PW programs increase future employability. People are working. Maybe they’re gaining some basic skills. Of the four programs yesterday, only the Côte d’Ivoire program examined employment after the program. They observe some post-program impacts on self-employment earnings, particularly for participants who did PW combined with self-employment training. Of course, cash transfers have also shown increased earnings several years later in Uganda (Blattman et al.).
4. By providing people with work, perhaps PW programs reduce unrest. We have all these unemployed youth; perhaps PW programs make them less likely to cause social problems. None of the studies examined this directly. The Côte d’Ivoire study showed positive impacts on psychological well-being. We have evidence that cash transfers can also have positive psychological effects in the short-run, in Malawi (Baird et al), although the story is nuanced. They also improved happiness and reduced stress in Kenya (Haushofer & Shapiro). It would be very interesting to see results of PW programs on actual unrest or engagement in disruptive activities.
5. Public works programs may reduce the need for detailed targeting systems. The idea is that by offering a relatively low wage, the only people who will want the public works program are those who really need it. That would remove the need for complicated proxy means tests or community targeted processes, often used in cash transfer programs. However, in Sierra Leone, Malawi, and Côte d’Ivoire, the evidence suggests that the wage isn’t low enough to accomplish that kind of sorting, so the programs don’t end up actually targeting the poorest; and lower wages run into political and potential ethical problems, being lower than national minimum wages. So while this argument may work in theory, it doesn’t seem to be a realized advantage of PW programs in practice.
6. Public works programs actually build something, in addition to giving people resources. Of course, the workers in the PW program are actually building something to improve their community. In Sierra Leone, Malawi, and Côte d’Ivoire, participants did road maintenance. In Sierra Leone and Malawi, they also worked on other community projects. In Ethiopia, participants worked on watershed development (conservation activities that later allow better irrigation). This could be a big plus over a simple cash transfer program: Maybe these communities have great roads at the end of the project, and that has additional benefits on commerce. Only the Ethiopia evaluation showed results on this (although it was simple before-after), but it looked like water access was up and waterborne diseases were down, among other outcomes. Identifying these impacts well is more difficult, especially if randomization of the intervention is within-community, but I’d propose it’s worth trying.
Take away. How do public works programs hold up? On the areas that have actually been measured, it’s not clear that they outperform a simple cash transfer. I hope that these and future evaluations can make sure to report on outcomes that may make PW programs a preferable alternative.
Ultimately, the proposal here is to scrutinize each development program – whether it’s providing training or giving away shoes or a package of services targeted to the ultra-poor or a merit-based scholarship – not simply on whether it has a positive impact. Positive impacts don’t always show up, but they’re still too low a bar. Rather, we can consistently ask whether the program would likely do better than cold, hard cash.
Bonus reading
- India’s workfare program is less cost effective than simply providing cash (by Murgai et al. 2015; ungated). This is a good example of seeking to estimate how a cash transfer program (which didn’t actually happen) would compare to a workfare program that was actually evaluated.
- “More Sweatshops for Africa?” This is an experiment of industrial jobs versus cash transfers in Ethiopia, by Blattman and Dercon. Preliminary results are available here. Updated results exist but don’t seem to be online.
Join the Conversation