In medicine, researchers use randomised controlled trials (RCTs) to study the effectiveness of potential treatments, such as a pill or a diet. By randomly selecting a “treatment group” and a “control group”, some patients get the medication and some get a placebo (sugar pill), and the difference in outcomes is measured.

This approach is also being used in education research. A study might involve putting students in a smaller class size and the control group in a standard one, and measuring the difference in outcomes to assess the impact of teacher-student ratio on learning.

Our research looked at the impact of financial incentives, which could be for children, teachers, or parents. And they could be for outputs like test scores, or for inputs like reading books, doing mathematics homework, school attendance, or behaviour.

The very idea of paying kids to learn raises a number of questions and concerns. Does it work?
What are the side effects on other subjects? Do incentives turn education into a “transaction”
and damage intrinsic motivation?

These are all good questions, but they are empirical ones. In principle the answers could cut for
or against incentives. To understand the different effects requires a large-scale RCT. That is precisely what my co-authors Roland Fryer, Tanaya Devi and I did in schools in Houston,
Texas and Washington D.C.

In Houston we provided incentives for kids to do mathematics homework problems. Fifth graders in the treatment group in 25 schools got $2 per homework problem that they mastered on curriculum-integrated, specifically provided educational software. Parents of the children got $2 per problem mastered. Teachers were eligible for bonuses of up to $10,000. The 25 control schools got the identical educational software and training, but no financial incentives.

These “vertical incentives” (for students, parents and teachers) did, indeed, cause children to do dramatically more mathematics homework problems. It also led to a fairly large increase in performance on state-level standardised maths tests.

What economic theory tells us will happen is that extra incentives in one area will crowd out effort in other areas. So, we weren’t surprised (though definitely concerned) to find that scores on English Language Arts (ELA) tests went down by almost the exact same amount as the maths scores went up.

In fact, the message from Houston was even more depressing. The most able 20 per cent of students, based on their prior-year test scores, did way better on maths and exhibited no decline in ELA scores. The least able 20 per cent, however, did lots more maths problems, but performed no better on maths test scores, and far worse on ELA scores. Vertical incentives for the least able students were a disaster.

Worse still, these effects persisted for two years after we stopped the experiment and incentives (but kept measuring student performance). It was as if the incentives had taught the higher-ability kids to work hard and the lower-ability kids not to bother.

By contrast, in Washington D.C. we provided balanced student-only (horizontal) incentives for measures such as attendance, behaviour, short-cycle assessments, and two inputs chosen by each school. Treatment students were 3 per cent more likely to attend school, committed 28 per cent fewer behavioural offences, and were 13.5 per cent more likely to report completing most or all of their homework relative to control students.

These financial incentives led to a very large increase in reading and maths test scores. This led to a 17 per cent increase in students scoring at or above proficiency for their grade in maths and a 15 per cent increase in reading per year.

Indeed, we estimated that the return on investment to providing these balanced incentives was a staggering 32 per cent per annum. Not only did these incentives work, they provide great bang for the buck.

Most people want Australia to have great schools and provide educational opportunities for all our children. Many of us see it as perhaps the most important way for those from less advantaged backgrounds to have opportunity and success later in life.

The real issue is how we do it and what it costs. Careful research in the US has identified the positive causal effect of a range of interventions: tutoring, out-of-school programs, smaller class size, better teachers, a culture of high expectations and, yes, financial incentives.

We need to be open to all these possibilities, and to carefully assess both the causal effect of them in Australia and the return on investment of additional public resources.

Richard Holden (@profholden) is professor of economics at UNSW Business School.

Continue the conversation at our SMH Student Facebook group.