Matching mechanisms, such as matching students to schools, is one of the most successful exports from microeconomic theory to public policy. These mechanisms can entice truthful revelation of preferences and result in efficient and stable outcomes. Efficient in the sense that there are no students who would agree to the move without making another student worse-off and stable in the sense that a student does not prefer to move to another school where there are open places, or this student has priority over another student already accepted to that school.

‘Matching and Chatting’ by Tingting Ding and Andrew Schotter (2016) asks how these mechanisms are affected in the face of communication between students (or parents of students). In a lab experiment, the participants are offered to submit preferences over three objects and told how they value the objects (4, 16, or 24 dollars). Boston and Gale-Shapley matching mechanisms assign objects to participants.

Let’s move back to the school choice example and call the objects schools and the participants students. The main treatment allows students to communicate, with chat boxes on the computer, before playing the game a second time. Crucially, the chatting network is part of the experimental design. Each of the 20 students in the experiment is only allowed to chat with between one and four other students.

The paper answers a range of different questions, but I will focus in on the chatting treatment. Chatting does have an effect. Firstly, chatting is more likely to lead to stable outcomes. Students who are allowed to chat are far more likely to change their strategy than students who are isolated. If students chat in homogenous groups (where they have the same preferences over schools), they are more likely to change their strategy. However, chatting in heterogenous groups is more likely to increase payoffs. Chatting to someone like you causes you to change your strategy, but this decision may also cause a decreased payoff.

Certain students have priority over certain schools. Interestingly, the students who do not have priority benefit the most from chatting, which suggests their may be a welfare improvement from extending the social networks of low income families (who are more likely to have low quality schools in their neighborhood). My feeling is that it is a stretch to link these results to the welfare of low income families. Students (or parents) choose their friends and the information they share with each friend. It would be very difficult to force communication that does not arise naturally – even if it is beneficial for one side of the chat.