Why Behavioral Economics Needs Causal Evidence
One of the stubborn open questions in behavioral economics is whether the systematic choice errors we document—preference reversals, violated transitivity, inconsistent trade-offs—reflect fixed features of human cognition or something malleable. If rationality is trainable, it changes what we expect from policy interventions. If it is not, we should build systems that work around human limitations rather than trying to fix people.
In October 2018, a team led by Hyuncheol Bryant Kim, then an assistant professor of policy analysis and management at Cornell University, published results in Science that addressed this question with an unusually clean design (Neuroscience News). The study randomized access to educational support among nearly 3,000 female ninth and tenth graders in Malawi and then measured their economic rationality in laboratory experiments. The authors described their design as establishing "causal evidence that an education intervention increases not only educational outcomes but also economic rationality" (Neuroscience News).
That claim deserves careful unpacking—both in what the study actually measured and in what it leaves open.
The Study Design: Setting, Intervention, and Measurement
The intervention was not a classroom curriculum. It was a financial support program arranged by a nongovernmental organization in Malawi, providing cash-based assistance designed to keep girls enrolled in school (Neuroscience News). Participants were female ninth and tenth graders, roughly 3,000 of them, randomly assigned to receive or not receive this educational support.
After the intervention period, researchers administered laboratory experiments to measure what they call "economic rationality." This is a specific technical construct: it refers to the consistency with which a person's choices satisfy standard axioms of rational choice, such as transitivity and the Weak Axiom of Revealed Preference. The experiments were designed to "mimic real-life examples," according to Kim (Neuroscience News).
The key comparison is between students who received the educational support and those who did not. The team found that participants in the education intervention group scored higher on these rationality measures (Neuroscience News).
Co-authors on the paper came from Cornell, Caltech, and the Stockholm School of Economics (Neuroscience News). Funding came from the National Institute of Mental Health, Grand Challenges Canada, and the Bill & Melinda Gates Foundation, among others (Neuroscience News).
What Was Measured, and What Was Not
Here is where precision matters, and where I think overconfident summaries of this study have gone wrong in popular coverage.
The outcome variable is a score. It captures how consistently someone's revealed preferences satisfy consistency axioms in a controlled laboratory task. The source does not report whether the study measured real-world economic decisions, whether participants later saved more, avoided predatory loans, or made better health purchases outside the lab.
The source also does not report whether the authors tested and ruled out specific mechanisms. We do not know from the available coverage whether improvements in numeracy, working memory, or general cognitive ability fully mediated the rationality score differences. The authors say education is "a tool for enhancing an individual's economic decision-making quality" (Neuroscience News), and they frame the finding relative to prior work showing that schooling has "positive effects on a wide range of outcomes, such as income and health" (Neuroscience News). But "economic rationality" as scored in a lab task and "better decisions in daily life" are not interchangeable constructs, even if we hope they correlate.
The randomized design does give the study internal validity for its stated claim: random assignment to educational support predicted higher rationality scores. That is a meaningful result. Whether the same effect would generalize to a different intervention type, a curriculum reform, a financial literacy module, a conditional cash transfer in a different country, remains an open empirical question that this study does not settle.
Cautious Implications for Policy and Research
Kim's framing is that the study "provides evidence of potentially additional benefits coming from improvements in people's decision making" beyond the already-documented income and health returns to schooling (Neuroscience News). That word, potentially, matters. It signals that the authors themselves treat the decision-quality channel as a plausible next step, not a confirmed pathway.
For researchers working in behavioral economics and cognitive neuroscience, the result is interesting because it suggests that at least one component of what we call rationality responds to a structural environmental intervention rather than only to targeted cognitive training. That is a useful data point. It does not mean education is the "best" or "most potent" lever for improving decision-making, nothing in the published result supports a comparative claim of that kind. And it does not tell us whether the effect persists over years, or whether it scales to populations with different baseline educational access.
What the study does demonstrate is a proof of concept: in this specific context, with this specific intervention and measurement, randomizing access to educational support shifted scores on a rationality-consistency task. Replication in other settings, with different measurement approaches and longer follow-up windows, would be the natural next step.
The Broader Picture in Decision Science
The question of what makes choices "rational" and how that quality can be measured connects to a wider literature. Researchers in behavioral economics decision science continue to probe the gap between the normative models we use and the descriptive models that capture what people actually do. The Cornell study adds one piece of evidence on the malleability side of that ledger.
Neuroscience perspectives matter here too. If rationality scores reflect something about executive function, prefrontal engagement, or learned heuristic strategies, then understanding the biological substrate of the intervention effect would require neuroscientific methods the study did not employ. This is not a criticism of the study, it is a boundary condition on what the result can currently tell us about the biology of economic choice. This question also connects to research on neuroscience and political behavior.
The study's contribution sits at an intersection that researchers from Nature Reviews Neuroscience to working behavioral economists keep circling: how much of what looks like stable cognitive ability is actually a function of environmental inputs that can, in principle, be redistributed through policy. The Malawi trial offers a clean, if narrow, data point on one side of that question.
What to Take From This
I want to be explicit about what I think this study supports and what it does not:
- It supports: a randomized financial intervention that increased school participation predicted higher economic rationality scores in a lab task among young women in Malawi.
- It suggests, cautiously: that decision-making consistency may be more responsive to structural interventions than fixed-trait models assume.
- It does not establish: that education "fundamentally upgrades" decision-making in any general sense, that cognitive ability was ruled out as a mediator, or that this represents the most effective tool available for improving economic decisions.
Those are different claims. The first is grounded in the published result. The rest require evidence this study does not provide. Behavioral economics benefits when we keep the distinction sharp.
behavioral economics needs causal evidence
behavioral economics needs causal evidence