Oportunidades Knocks?

Can paying children for attendance and grades boost scores, or will it do more harm than good? In development economics, one of the most popular programs over recent years are conditional cash transfer programs. Modelled on a randomised trial of the Mexican Progresa program (now Oportunidades), a succession of developing nations have put in place programs that pay parents a cash transfer if their children attend school. My read of the development evidence is that there is a strong consensus that conditional cash transfer programs are an efficient way of raising educational attainment (much more effective than, say, simply banning child labour).

But in more affluent countries, conditional cash transfer programs are much more controversial, though according to a piece in the New York Times, they are nonetheless proliferating across US cities:

In New York City and Dallas, high school students are paid for doing well on Advanced Placement tests. …  Another experiment was started last fall in 14 public schools in Washington that are distributing checks for good grades, attendance and behavior. … At 80 tutoring centers in eight states run by Score! Educational Centers, a national for-profit company run by Kaplan Inc., students are encouraged to rack up points for good work and redeem them for prizes like jump-ropes.  

There are plausible theories as to why these programs might work (kids respond to incentives) or cause problems (extrinsic rewards might crowd out intrinsic motivations). Ideally, we’d want some hard evidence from randomised trials that show us the short-run impact (while the cash is on offer) and the long-run impact (after the payments stop). And that’s just what New York City is trying to gather. As researcher Roland Fryer puts it:

We have to get beyond our biases, said Roland Fryer, an economist at Harvard University who is designing and testing several reward programs.  Fortunately, the scientific method allows us to get to most of those biases and let the data do the talking. 

Should we be worried about the ethics of trialling a program that might have negative effects? The New York Times article doesn’t delve into the issue, but it’s certainly an argument that comes up when I chat with people who believe the crowd-out effects are large. While I’d take the argument seriously, my own view is that in places where educational attainment is low and stagnant (eg. US inner cities), it’s worth rigorously testing on a small scale. The downside risk isn’t zero, but if the intervention works, the societal benefit is potentially very large.

In Australia, conditional cash transfer programs could theoretically be implemented in a low-income school anywhere in the country. But the most likely context is in Indigenous education, where we know that Indigenous children are one year behind when they arrive at primary school, and two years behind when they finish primary school. Anyone game for a randomised trial?  

Update: Don Arthur provides a much more detailed backgrounder on the New York experiments, and advocates more theory as well as more empirics. I don’t want to disparage theory (some of my best friends…), but I can’t imagine anyone writing a compelling piece of theory that would satisfy the economists that financial incentives are a failure, or the sociologists that incentives would work.  On the other hand, both sides might – at least grudgingly – accept the results of a well-designed randomised trial.

This entry was posted in Economics of Education, Trade & Development. Bookmark the permalink.

3 Responses to Oportunidades Knocks?

  1. christine says:

    My son’s (grade 1) class has a science class taught by a separate teacher from his regular classes. In this class, for good behaviour, they can earn $ to spend on prizes at certain intervals. My son (perhaps unsurprisingly a relative swot) has already commented on the difference in behaviour of one of his classmates in the science vs regular class, and directly linked it to the extrinsic rewards.

    And it’s not just the prizes, but being able to keep score that matters. Vic’s quite upset about the fact that he’s not collecting money as fast as his more mercenary but less generally well-behaved friends.

    Actually, thinking about this, perhaps there might be people who wouldn’t be happy about paying for explicit performance who might be happy about paying for good behaviour (a la Washington in the example, I guess)? And maybe, given all this discussion of non-cognitive ability, it might have bigger long-term effects?

  2. Don Arthur says:

    Andrew – I think it’s important to distinguish between CCT programs that pay parents from those that reward children.

    Mexico’s PROGRESA/Oportunidades pays mothers when their children are enrolled in school an attend regularly (a minimum attendance rate of 85%, both monthly and annually).

    The intrinsic motivation problem is less of an issue with a design that seeks changes in parental behaviour. In third world contexts there is often an opportunity cost for parents when their children attend school. According to one study:

    Lack of money or need to work is the most common reason given for not continuing school (57 percent), but other important reasons are that the child does not like school or does not learn (23 percent) and that the school is too far away (13 percent).

    While children may enjoy intrinsic rewards from learning, their parents will not.

    In the case of parents, the risk is more that cash rewards will undermine internalised social norms. Parents who send their children to school may experience feelings of pride and those who do not feel guilt or shame. It’s conceivable that paying for ‘virtue’ may weaken the norm.

    In cases where there is no social norm associated with the behaviour, CCT supporters will need to argue with people who think they have some way of establishing new social norms in a community.


    One rationale for CCT is that some behaviours are not intrinsically rewarding for the individual but they lead to outcomes which are. For example, job search is not usually an enjoyable or satisfying behaviour. For many job seekers it involves pestering people who don’t want to talk to them, writing letters and filling out forms that nobody responds to, and being repeatedly rejected.

    However, the outcome of job search is often highly rewarding. For most people, the rewards of work extend beyond the wage they’re paid. A CCT program that rewards people for sustaining their job search efforts might help deal with time preference problems.

    Interesting, the MDRC describes TANF as a CCT program. When an income support payment is conditional on job search and job preparation, this seems a reasonable way to describe it.

    While PROGRESA/Oportunidades has been successful in lifting school attendance rates it’s less clear what impact it’s had on learning outcomes.

  3. Don Arthur says:

    I can’t imagine anyone writing a compelling piece of theory that would satisfy the economists that financial incentives are a failure, or the sociologists that incentives would work.

    Of course not. But the reason many economists are convinced that financial incentives will work is because of their implicit theoretical assumptions.

    What I’m asking, is that policy makers bring these assumptions to the surface so that evaluations can be designed to test them.

    As you know, there are two challenges to validity in evaluations — internal and external.

    Randomised control trials with large sample sizes do a good job of ensuring internal validity. But they don’t necessarily solve the external validity problem.

    For example, imagine that you trial financial incentives in a suburban Australian school and get good results. Does this justify taking the program to scale in remote Indigenous communities?

    I think we stand a better chance of being able to generalise results if we test theories rather than ‘black box’ interventions.

    To understand the differences between populations you don’t necessarily need a different set of theories — you need to know what the relevant variables are and how they operate in different contexts.

    Ronald Fryer’s theoretical work on social interactions and poverty traps hypothesises that students from some minority groups face a different set of incentives in school — a set of incentives that leads them to perform less well than other students.

    This kind of work can be used to inform interventions which then become tests of the underlying theory.

Comments are closed.