A breakthrough study of 2,204 Chinese fifth graders suggests that girls are not inherently more risk-averse than boys; when cognitive errors and guessing are accounted for, the gender gap in risk-taking actually depends on the size of the potential reward. The researchers’ findings are published in the NBER Working Paper Series.
Traditional economic models presume decision‑making is perfectly rational, with choices directly revealing true preferences. By contrast, behavioural economists recognise that people often commit cognitive mistakes or rely on mental shortcuts, complicating how researchers interpret risk attitudes.
This measurement challenge is particularly acute with children, whose decision‑making skills and attention spans are still developing. The study’s authors sought to separate a child’s genuine risk tolerance from random guessing and lapses in attention.
Economists Cesar Chavez Padilla and James J. Heckman of the University of Chicago, with Shuaizhang Feng of Jinan University and Zhe Yang of Liaoning University, led the research. Their aim was to isolate true preferences from cognitive noise and guessing.
The team analysed data from a large dataset of 2,204 fifth graders in China, aged ten to eleven. The children completed interactive games on tablet computers designed to measure their willingness to take risks.
The games used a format known as multiple price lists. Participants faced a sequence of choices between a guaranteed token payoff and a gamble—such as a coin flip with a high or low payoff. As the list progressed, the gamble became more attractive, and the point at which a child shifts from the safe option to the gamble signals their risk tolerance.
A perfectly rational individual would switch from the safe option to the gamble just once. However, many children switched back and forth within a single list. Rather than discarding these inconsistent responses, the researchers treated them as informative about the quality of the child’s decision‑making.
To investigate the psychology behind choices, intelligence was measured through standard visual puzzles, and personality was assessed using the Big Five traits. Teacher evaluations—rather than self‑reports by the pupils—were used to gauge traits such as conscientiousness and openness, with observations drawn from extended classroom contexts.
The researchers developed a mathematical model to separate true risk preferences from two types of error: deliberation noise, where reasoning goes awry, and guessing, where a child randomises without solving the problem.
The model distinguished between temporary lapses and persistent individual traits. Some mistakes arise from momentary distraction, while others stem from enduring cognitive habits that make some children consistently noisier across tasks.
Instead of the rigid assumption of constant risk aversion across all prize sizes, the study employed a flexible Expo‑Power utility framework. This approach allows a person’s willingness to take risks to vary with the scale of potential rewards.
When the analysis used only a basic model that ignored cognitive noise and guessing, the results echoed a common finding in behavioural economics: girls appeared more hesitant to take risks than boys.
However, once the full model accounted for guessing, decision noise, and changing stake sizes, the gender gap disappeared. The data revealed that risk aversion is highly dependent on the size of the stakes: for small rewards, boys were more risk-averse than girls, while for the largest gambles, girls were more risk-averse than boys.
Beyond these findings, the study showed that intelligence and personality significantly influence decision‑making consistency. Higher intelligence scores correlated with less random guessing across both genders, and with lower deliberation noise particularly among boys. Overall, roughly 30 percent of the choices were categorised as mistakes rather than reflections of true preferences.
Breaking down the sources of error, transitory random noise accounted for about 12 percent, persistent individual noise for around nine percent, and outright guessing for roughly nine percent of all choices.
Personality appeared to shape behaviour differently by gender. For boys, traits such as conscientiousness directly affected baseline risk preferences, with conscientious boys being more risk-averse and exhibiting less deliberation noise. For girls, personality mainly influenced the consistency of their decision‑making rather than their actual risk tolerance; traits affected how often they guessed and how noisy their thinking was, but did not measurably shift core risk preferences.
Socioeconomic background, including parental education and household stability, had little direct impact on risk preferences or the propensity to make decision errors. Instead, socioeconomic factors seemed to influence choices indirectly by shaping intelligence and personality traits.
A limitation acknowledged by the researchers is that the study captures only a single moment in time, limiting observations about long‑term development. It remains possible that economic preferences and environments shape personality traits over time, suggesting avenues for future longitudinal work.
Another challenge concerns defining a baseline notion of wealth for children, a prerequisite for assessing perceived financial risk. Because young students do not manage household finances, wealth is an abstract concept in this context. The authors tested alternative wealth assumptions and found that the principal conclusions held across these specifications.
The study is titled Relating Cognitive Skills and Personality Traits to Economic Preferences: Decision-Making under Cognitive Shocks and Guessing and was authored by Cesar Chavez Padilla, Shuaizhang Feng, James J. Heckman, and Zhe Yang.
