In classes where there is a majority of boys, relatively low-achieving male school students are more likely to choose to study economics, business or engineering at university. But their increased enrolment in subjects that promise higher earnings in the future does not actually translate into a significant effect on their later life income.

These are among the findings of research by Massimo Anelli and Giovanni Peri, published in the February 2019 edition of The Economic Journal. Their study explores how the gender composition of school classes affects choice of university subject, analysing a unique database on 30,000 graduates of a state secondary school in Milan between 1985 and 2005 and linking this information to their university careers and labour market outcomes.

While boys' choices of university subject are strongly influenced by the gender composition of their school class, there is absolutely no effect on girls' choices. The importance of peer pressure for boys has a clear downside: male students might decide to enrol in economics, business or engineering, although they lack the abilities to graduate from them – and indeed, university dropout rates are high.

But there is also a strong counter to the idea touted by some commentators that single-sex education is a way to reduce the gender gap in science. The new findings suggest that policies to alter the gender balance in school classes are unlikely to be effective in reducing the gender gap in choice of university subjects and later life outcomes.

In a society with persistent gender gaps, understanding the fundamental determinants of pay differentials is of great relevance for both policy-makers and academics. Striking evidence often found in previous research is that despite the fact that girls generally outperform boys during school, their advantage seems to be reversed once they enter the job market.

This phenomenon is partly explained by the fact that choices of ''college majors'' differ substantially across male and female students. Specifically, boys have a higher propensity to enrol in economics, business and engineering, generally associated with higher payoffs after graduation, while women have a natural inclination towards humanities and education, characterised by lower monetary returns.

The new study examines whether the networks that teenagers form during school have an influence on their choice of college major. Specifically, the researchers assess whether the gender composition of school classes affects students' choices of college major and whether it has a long-run impact on their college performance and labour market outcomes.

It has widely been acknowledged that peer environment plays a relevant role in affecting performances and preferences of individuals. Since college major strongly affects an individual's potential earnings, and since women remain segregated in majors with lower expected earnings, it is of great importance to identify the determinants of such segregation and direct policy-makers' intervention.

To carry out their research, Anelli and Peri created a unique database by collecting records for 30,000 individuals who graduated from college preparatory public high school in the municipality of Milan, Italy, between 1985 and 2005, and linked this information to their college career and to labour market outcomes.

What they find is that male students attending a class where more than 80% of students are of their own gender, are six percentage points more likely to enrol in high-paying majors with respect to a baseline probability of 43%. This is equivalent to a 14% increase in the probability of choosing these majors. If the share of males in the class is above 90%, then the same probability increases by 35%.

In contrast with this large effect for boys, the gender share of school peers does not have any effects on the choice of college major of girls, not even in the case of more extreme class composition.

The explanation for these differences across genders lies in the dynamics of friendship networks formation: while female teenagers prefer small networks with peers of the same sex no matter what their gender share is in the class, males tend to form broader and more inclusive ones. In male-dominated classes, boys indeed end up in larger networks, which imply greater information-sharing and larger potential for peer pressure.

What is most striking is that when looking at long-run outcomes, the effect on the choice of major by boys vanishes: university dropout rates are larger for boys from male-dominated classes and there are no measurable long-run effects on labour market outcomes. This suggests that peer imitation leads to sub-optimal choices: because of peer pressure, male students might decide to enrol in economics, business and engineering majors, although they lack the abilities to graduate from them.

These results lead to two important policy considerations and cautionary tales. First, they reveal that when discussing peer effects in school, a myopic focus on short-run outcomes may lead to misleading conclusions. For example, policies to shift people towards STEM majors will not automatically produce positive results; on the contrary, they may cause mismatches and create inefficiencies.

Second, contrary to the widespread idea supporting single-sex education as a way to reduce the gender gap in science, these findings suggest that policies altering the school gender environment are likely to be ineffective in reducing the gender gap.

''The Effects of High School Peers' Gender on College Major, College Performance and Income'' by Massimo Anelli and Giovanni Peri is published in the February 2019 edition of The Economic Journal.