Prof Tomoki Fujii

Feature on Tomoki Fujii, Associate Professor of Economics
Associate Dean (Undergraduate Curriculum)
School of Economics, Singapore Management University 

Evaluating policy prescriptions for developing countries  

As a development economist, Associate Professor Tomoki Fujii uses experimental methods to evaluate the effectiveness of development programmes in ‘real-time’.   

When your doctor gives you a prescription you can be sure that it has been through rigorous trials—including randomised control trial. This is not only for safety, but to ensure it is the adequate solution to the health problem. But can the same be said when policymakers prescribe programmes to improve life in developing countries – or anywhere for that matter? How do we know if it will help and not harm the ‘patient,’ in the form of society?  

Associate Professor Tomoki Fujii, of the SMU School of Economics has not only devoted his academic career to this question but has found ways to apply the rigour of drug evaluation to development programmes, for example in health and education.  

In a randomised control trial for drugs, a sample population is randomly divided into two groups–the control group, which receives no intervention (except for a “placebo”), and the treatment group, which does receive the drug. Other factors may be accounted for and isolated as well. Development economists increasingly use randomized control testing to evaluate policies and programs in a similar fashion, and for good reason, says Fujii. 

“Wealthy countries have spent billions of dollars aiding developing countries. But has the success been phenomenal? Success is found here and there, but it's often unclear whether programmes are effective in eradicating poverty or improving education, health and so on.” 

While randomised control trials offer better clarity, applying the method to development programmes is complex, because the ‘patient’ is a society rather than an individual. 

“In medical trials, the drug or treatment typically affects only the individual taking it. But in economics, when you provide cash to impoverished people, they spend the money, and this affects the economic status of others— economists call this the ‘spill-over effect’.” 

Fujii was able to isolate the effects of different versions of policy prescriptions for improving school attendance in Bangladesh, in his paper Conditional Cash Transfer, Loss Framing, and SMS Nudges: Evidence from a Randomized Field Experiment in Bangladesh.   

“Conditional Cash Transfers have become one of the most common policy interventions to increase school attendance, but its cost-effectiveness has not attracted the attention it deserves,” says Fujii. This study compared two policies aimed at improving the effectiveness of school attendance interventions: SMS information nudges; and different methods of conditional cash transfers. 

For the conditional cash transfers, the study drilled even further down to the way the cash transfer was ‘framed’—that is, whether it was pitched to the participants as gaining or losing cash.  

“We split the sample groups into Gain and Loss groups,” Fujii explains. “Participants in the Gain group start with a zero account balance and when they attend school they gain cash. At the end of the intervention period, the money is disbursed to them. Conversely, people in the Loss group start with a full account balance, and when they miss school, they lose cash.” The students and their parents would also receive weekly SMS updates on their school attendance information and account balance. This is where it got interesting.  

“Since we were giving this information by SMS, we also had another treatment group, which only received attendance information, no cash. So the parents were told how many days the child attended, and depending on which group they were in, whether they had gained or lost money.”  

While the study found that loss framing—or loss of cash—had a slightly stronger positive impact on attendance the difference was insignificant, reports Professor Fujii. More important, he says, were the effects of the pure information nudges. In that group, the study detected a positive impact a year later, at the tail end of the intervention. This is particularly the case for girls. This result suggests some important lessons for development policy.  

“The findings suggest that money matters but, once money is gone, that portion of impact is also gone. But if you provide information, in this case via SMS, it might cause a more persistent change in behaviour,” says Fujii. 

Fujii has already been looking at how policymakers can make use of these insights.  

“Bangladesh had one of the longest school closures around the world when the COVID-19 pandemic happened,” he explains. “The learning loss during this period is a serious concern. We are exploring with policymakers the potential to use SMS and other digital technologies to help students’ households stay engaged with education resources even during disrupted times.”   

Finding truth in HIV testing rates  

As well as taking medical research methods into the realm of development, Fujii also maintains an interest in public health issues and the ways in which economics and statistics can help understand them. One example is a recent paper on HIV testing rates in Malawi, Refusal bias in HIV data from the Demographic and Health Surveys: Evaluation, critique and recommendations.  

“In places where HIV and Acquired Immunodeficiency Syndrome (AIDS) are stigmatised, people who are likely to be HIV-positive may refuse to test themselves – because they fear the results,” he says of an issue that persists in many societies.   

He and his colleagues examined this “refusal bias” in HIV data from the Demographic and Health Surveys and the Malawi Diffusion and Ideational Change Project in Malawi, to help health researchers and authorities understand the true HIV prevalence after accounting for the potential presence of selective refusals. “Understanding the consequences of the potential importance of refusals is key to improving the design of policy interventions,” Fujii says.  He and his colleagues are currently working on a related project with updated data.