Too Close Collaboration Between Students Reduces Overall Performance: Classroom Evidence On The Effects Of Social Interactions

Barcelona GSE Focus

The Royal Economic Society and the Economic Journal have selected Giacomo De Giorgi and Michele Pellizzari to receive the 2015 Royal Economic Society Prize for their paper, “Understanding Social Interactions: Evidence from the Classroom.” The prize organizers released the following briefing and video of the winning paper.


University students who go to many of the same classes will help each other when one of them is off sick or otherwise distracted from their studies. But knowing that this ‘mutual insurance’ is in place leads to average performance declining as everybody in the group makes less effort. These are among the findings of a study of the behaviour of economics and management students at Bocconi University in Milan by Professors Giacomo De Giorgi and Michele Pellizzari.

Their research, which is published in the September 2014 issue of the Economic Journal, suggests that in situations where social interactions between people are too frequent, they may lead not to productive cooperation or competition but to an overall decline in performance. The possibility of such an outcome might apply not just to university students but also to corporate colleagues or co-authors of research papers.

There is little doubt that social interactions are a pervasive feature of life, the researchers note. In economics, such importance has been underlined by a large number of empirical and theoretical studies. But still very little is known about the exact economic mechanism that generates such interactions.

For example, in the school environment, a setting in which social interactions have been extensively studied, students may interact by either competing or collaborating with each other – or perhaps by mutually insuring against shocks to their learning ability (for example, health problems or other distractions). Without knowledge of the mechanism of interaction, it is extremely difficult to think about policy intervention.

This study combines theory and empirical analysis to develop a methodology that makes it possible to understand the behaviour underlying social interactions. The researchers analyse students at Bocconi University (a selective higher education institution specialising in economics and management), exploiting the random process the university uses to allocate students to classes for each of their compulsory courses.

The researchers find that pairs of students who were randomly assigned to the same classes more often and hence spend more time together perform more similarly compared with pairs who interact less frequently. This result is important because it confirms the presence of ‘peer effects’ in this setting, but it is compatible with many different interaction mechanisms.

To disentangle the exact mode of interaction, the researchers further investigate the data in the light of a theoretical model. They document that the pairs that are more frequently assigned to the same classes display more similar but also lower grades, a result that is incompatible with their collaborating or competing and consistent with a mechanism of mutual insurance.

In other words, students interact to shield themselves from unexpected events that affect their learning capacity. For example, by exchanging notes, students may catch up with course material they may miss when they are sick. Knowing that they can get notes from their peers, they would be more likely to skip lectures.

Students may also be distracted by various occurrences during the lectures (for example, a text message, a passing car, chit chat, etc.) and if they know they can ask their fellow students for the content of the lecture while otherwise engaged, they will give in to those distractions more easily. As a result, everybody in the group makes less effort and average performance declines. This is a standard ‘moral hazard’ argument that applies to the insurance mechanism.

To provide further support for this interpretation of the data, the researchers also find that when pairs of students interact more frequently, it is more likely that one peer consistently outperforms the other, a result that is consistent with the smoothing away of health or distraction shocks.

The authors conclude:

‘Our analysis sheds new light on the role of social interactions in education. But the conclusions can easily be extended to a number of other settings, such as teamwork in firms or co-authorship in academic research.

‘From a policy perspective, our findings suggest that if people interact too frequently, then they may engage in (formal or informal) mutual insurance rather than cooperation or competition and, as a consequence, average performance may decline.’

More information on the RES Prize is available on the Royal Economic Society’s website