Defining Conditional Cash Transfer Programs: An Unconditional Mess
This page in:
Many policymakers are interested in the role of conditions in cash transfer programs. Do they improve outcomes of interest more than money alone? Are there trade-offs? Is there a role for conditions for political rather than technocratic reasons? It’s easy to extend the list of questions for a good while. However, before one can get to these questions, there is a much more basic question that needs to be answered (for any policymaker contemplating running one of these programs at any level): “What do you mean by a conditional (or unconditional) cash transfer program?”
You see; there are a myriad of ways to design and run a cash transfer program. You can identify a behavioral condition related to your outcome of interest (visit health clinics, test negative for drugs, get children vaccinated, keep daughters in school, etc.) and then be vigilant in announcing it (so that everyone understands the rules), monitoring it, and enforcing it. Alternatively, you can treat these dimensions of announcement, monitoring, and enforcement as continuous variables and vary your effort in each of them: this means you can have a ‘conditional’ program that is monitored but not enforced, monitored but enforced only after many warnings and delays, not monitored at all, etc. You can also have programs that are announced to be conditional even when there was never any intention to monitor or enforce the condition. Similarly, you can announce a cash transfer program and have a social marketing campaign that encourages the beneficiaries to invest in their children’s (or their own) human capital. You can have an unconditional cash transfer program for poverty reduction that is run by the Ministry of Finance (MoF) or to reduce dropouts by the Ministry of Education (MoE): they are not the same thing. Again, the list goes on…
The range of options above form an undefined (and hard to define) continuum in which the outcome of interest is being primed in the target population. If, for example, the MoF ran a cash transfer program among the poor (defined, say, by some proxy-means-test) and made it clear that the only rule to receive these transfers was being poor by that proxy-means-test, that would be as unconditional a cash transfer program as you could get. It would also be one that had nothing to do with schooling. Alternatively, you could have the same program being run by the MoE or you could have a MoF program that was called “Child Support Grant” targeted to families with school-age children (where, in both cases, it was also made clear that the transfers were unconditional): now you have an unconditional cash transfer program that is (perhaps lightly) priming schooling. You can continue down this road intensifying the priming of the importance of schooling, imposing explicit conditions, enforcing them lightly, until you have the most extreme CCT: the conditions are announced (and re-emphasized at every opportunity), monitored fairly but rigorously, and penalties applied swiftly in accordancewith clear program rules. The program run by the MoF for poor people without any mention of schooling is perhaps the clearest UCT, while the final example is a pure CCT. Everything else falls in between these two extremes, which happens to be true for most cash transfer programs around the world. And what are these programs called? Well, that’s the messy part…
Let’s take three examples, starting with Ecuador’s Bono de DesarolloHumano (BDH). Three different papers (Araujo and Schady 2008, Osterbeek; Ponce, and Schady 2008; and Edmonds and Schady 2012) describe the program as being designed to mimic PROGRESA/Oportunidades, with clear conditions defined, but never monitored or enforced. In the earliest paper, BDH comes across as a CCT that was not monitored or enforced (still with its effect on schooling mostly due to the beneficiaries’ mistakenly thinking that it was conditional), while in the latest paper, BDH is pretty much interpreted to be a UCT – albeit one that is accompanied by a social marketing campaign by the government that emphasized the importance of children’s health and education. In a separate paper, Ponce and Bedi (2010) describe BDH as a conditional cash transfer. Fiszbein and Schady (2009) also lists BDH as a CCT, with the conditions (for health and education) outlined in a summary table of CCT programs in Latin America and the Carribbean.Further yet, this paper by Fernald and Hidrobo (2011) categorizes BDH an unconditional cash transfer program in its abstract because, as explained later in the paper, it is “... more similar to a UCT due to the lack of verification of compliance with the conditions…” For a program that was conceived as a CCT but ended up with no enforcement of the conditionalities that were only on paper, I find no fault with the description and labeling of the program in each of these publications. But, that is not to say that the end result is not confusing. In fact, quite the contrary: did a CCT lead to those effects in Ecuador? Was it a UCT? Was it a UCT with a social marketing campaign or a CCT with no monitoring? My head hurts…
The second and third examples come from Honduras and Morocco. As I discussed in my last post, PRAF-II in Honduras was designed to improve upon the enforcement of conditions in PRAF-I, but ended up being conditional only on enrollment and not attendance. Nonetheless, PRAF-II has always been referred to as a CCT. In Morocco, the government recently ran a pilot RCT (called Tayssir, in collaboration with the World Bank) with treatment arms providing conditional and unconditional cash transfers. However, the pilot was run by the MoEand because enrollment in the program took place at the local school (conducted by the headmaster), even the UCTs ended up being de facto conditional on enrolling in school at the beginning of the program. So, here we have two very similar programs – both of which were conditional on school enrollment at the outset – but one is called a CCT and the other one a UCT. It is almost as if the initial conception/labeling of programs is sticky and has an undue and potentially misleading influence on the definitions (and, hence, the implied relative merits) of CCT and UCT programs. And, I only discussed examples here where the authors clearly discuss the program with respect to the conditions and their enforcement – there are papers that don’t do much of either.
So, if you’re a donor or a policymaker, it is important not to frame your question to be about the relative effectiveness of “conditional” vs. “unconditional” cash transfer programs: the line between these concepts is too blurry. It turns out that your question needs to be much more precise than that. It is better to define the feasible range of options available to you first (politically, ethically, etc.), and then go after evidence of relative effectiveness of design options along the continuum from a pure UCT to a heavy-handed CCT. Alas, that evidence is the subject of another post…
You see; there are a myriad of ways to design and run a cash transfer program. You can identify a behavioral condition related to your outcome of interest (visit health clinics, test negative for drugs, get children vaccinated, keep daughters in school, etc.) and then be vigilant in announcing it (so that everyone understands the rules), monitoring it, and enforcing it. Alternatively, you can treat these dimensions of announcement, monitoring, and enforcement as continuous variables and vary your effort in each of them: this means you can have a ‘conditional’ program that is monitored but not enforced, monitored but enforced only after many warnings and delays, not monitored at all, etc. You can also have programs that are announced to be conditional even when there was never any intention to monitor or enforce the condition. Similarly, you can announce a cash transfer program and have a social marketing campaign that encourages the beneficiaries to invest in their children’s (or their own) human capital. You can have an unconditional cash transfer program for poverty reduction that is run by the Ministry of Finance (MoF) or to reduce dropouts by the Ministry of Education (MoE): they are not the same thing. Again, the list goes on…
The range of options above form an undefined (and hard to define) continuum in which the outcome of interest is being primed in the target population. If, for example, the MoF ran a cash transfer program among the poor (defined, say, by some proxy-means-test) and made it clear that the only rule to receive these transfers was being poor by that proxy-means-test, that would be as unconditional a cash transfer program as you could get. It would also be one that had nothing to do with schooling. Alternatively, you could have the same program being run by the MoE or you could have a MoF program that was called “Child Support Grant” targeted to families with school-age children (where, in both cases, it was also made clear that the transfers were unconditional): now you have an unconditional cash transfer program that is (perhaps lightly) priming schooling. You can continue down this road intensifying the priming of the importance of schooling, imposing explicit conditions, enforcing them lightly, until you have the most extreme CCT: the conditions are announced (and re-emphasized at every opportunity), monitored fairly but rigorously, and penalties applied swiftly in accordancewith clear program rules. The program run by the MoF for poor people without any mention of schooling is perhaps the clearest UCT, while the final example is a pure CCT. Everything else falls in between these two extremes, which happens to be true for most cash transfer programs around the world. And what are these programs called? Well, that’s the messy part…
Let’s take three examples, starting with Ecuador’s Bono de DesarolloHumano (BDH). Three different papers (Araujo and Schady 2008, Osterbeek; Ponce, and Schady 2008; and Edmonds and Schady 2012) describe the program as being designed to mimic PROGRESA/Oportunidades, with clear conditions defined, but never monitored or enforced. In the earliest paper, BDH comes across as a CCT that was not monitored or enforced (still with its effect on schooling mostly due to the beneficiaries’ mistakenly thinking that it was conditional), while in the latest paper, BDH is pretty much interpreted to be a UCT – albeit one that is accompanied by a social marketing campaign by the government that emphasized the importance of children’s health and education. In a separate paper, Ponce and Bedi (2010) describe BDH as a conditional cash transfer. Fiszbein and Schady (2009) also lists BDH as a CCT, with the conditions (for health and education) outlined in a summary table of CCT programs in Latin America and the Carribbean.Further yet, this paper by Fernald and Hidrobo (2011) categorizes BDH an unconditional cash transfer program in its abstract because, as explained later in the paper, it is “... more similar to a UCT due to the lack of verification of compliance with the conditions…” For a program that was conceived as a CCT but ended up with no enforcement of the conditionalities that were only on paper, I find no fault with the description and labeling of the program in each of these publications. But, that is not to say that the end result is not confusing. In fact, quite the contrary: did a CCT lead to those effects in Ecuador? Was it a UCT? Was it a UCT with a social marketing campaign or a CCT with no monitoring? My head hurts…
The second and third examples come from Honduras and Morocco. As I discussed in my last post, PRAF-II in Honduras was designed to improve upon the enforcement of conditions in PRAF-I, but ended up being conditional only on enrollment and not attendance. Nonetheless, PRAF-II has always been referred to as a CCT. In Morocco, the government recently ran a pilot RCT (called Tayssir, in collaboration with the World Bank) with treatment arms providing conditional and unconditional cash transfers. However, the pilot was run by the MoEand because enrollment in the program took place at the local school (conducted by the headmaster), even the UCTs ended up being de facto conditional on enrolling in school at the beginning of the program. So, here we have two very similar programs – both of which were conditional on school enrollment at the outset – but one is called a CCT and the other one a UCT. It is almost as if the initial conception/labeling of programs is sticky and has an undue and potentially misleading influence on the definitions (and, hence, the implied relative merits) of CCT and UCT programs. And, I only discussed examples here where the authors clearly discuss the program with respect to the conditions and their enforcement – there are papers that don’t do much of either.
So, if you’re a donor or a policymaker, it is important not to frame your question to be about the relative effectiveness of “conditional” vs. “unconditional” cash transfer programs: the line between these concepts is too blurry. It turns out that your question needs to be much more precise than that. It is better to define the feasible range of options available to you first (politically, ethically, etc.), and then go after evidence of relative effectiveness of design options along the continuum from a pure UCT to a heavy-handed CCT. Alas, that evidence is the subject of another post…
Very interesting and useful exposition.
However, even your final conclusion is overly optimistic. CCT and UCTs do do not exist on a smooth continuum, they existing a multi-dimensional design space, some of which is likely to be rugged tather than smooth i.e where small changes may make big performance differences. This point was made by Prichett et al here last year: http://www.wider.unu.edu/publications/working-papers/2012/en_GB/wp2012-…
Given this, the interesting question then is how to effeciently explore this design space?
Hi Rick,
Thanks for your comment. I agree (and am familiar with Lant's paper). The word continuum was used loosely in the final paragraph, but you're really mixing and matching options in a multi-dimensional space. However, I do think that if you wanted to simplify, you could do so by intensifying your efforts. The question is the existence of non-linearities, which are very much plausible given the evidence out there. A tiny 'nudge' somewhere may make a large difference, but I am not aware that people have found it in the context of CT programs...
Berk.
Wonderful article! I would love to discuss this further and often encounter this challenge. What do you think about the definition of conditions put forth in http://mondediplo.com/2013/05/04income? See the line "..There were no conditions regarding wages, employment, caste, gender or age, and the recipients could use the money as they saw fit."
Interesting post. To further complicate things, the “cash” in CCT is often also a misnomer. As more social transfer programs— including heavyweights such as Bolsa Familia, Familias en Acción, and Oportunidades — shift to making payments electronically and through bank accounts, the term “cash transfer” is also misplaced.
Conditional food transfers?
Dear Berk, very interesting post!
Let me suggest that the issue may be elevated even further – the debate could be moved from revolving around UCTs and CCTs to discussing conditional and unconditional transfers more broadly. This would allow factoring into decision-making equations also programs that don’t envision cash as a transfer, but payments in-kind (vouchers or food) – that is, programs like school feeding and take-home rations, which could be considered "conditioanl food transfers", that pursue objectives similar to CCTs’ (e.g., http://siteresources.worldbank.org/INTHDOFFICE/Resources/SchoolFeedE2P…) as well as unconditional food transfers that may aim at income transfer goals a la UCT.
Indeed, the same logic laid out for cash-based transfers may hold for in-kind programs – i.e., in between the extremes of pure conditional and hard-core conditional programs there is a wealth of blends that vary by (intended or unintended) design, monitoring, communication and enforcement features. At the same time, there is evidence that, under the same class of unconditional or conditional programs (whatever the definition), cash and food-based interventions may often lead to different impact pathways (e.g. for Ecuador, see http://www.ifpri.org/publication/cash-food-or-vouchers).
In other words, I think there is a need to integrate these two sets of quandaries – those around conditonalities and those pertaining to transfer selection – under a common decision-making framework to inform about what works best in a given context and for a certain set of objectives.
Hence, building on your final suggestion, with which I fully agree with, it's important not only to explore the comparative performance of different unconditional/conditional shapes in the continuum, but also ensuring that such analysis includes potential costs and impacts of alternative modalities – cash, vouchers, food or combinations thereof. This may be particularly relevant for decision making in many LICs and fragile contexts, although even some CCTs in Latin America, such as Oportunidades in its early days, were de-facto conditional cash and food-based programs.
Very best,
Ugo