Working Paper: CEPR ID: DP17965
Authors: Yair Antler; Daniel Bird; Daniel Fershtman
Abstract: We study statistical discrimination in a marriage market where agents, characterized by attractiveness (e.g., wealth, education) and background (e.g., race, ethnicity), engage in time-consuming search. Upon meeting, couples date to learn about their match's quality. Following Phelps (1972), different backgrounds impede such learning. We show that even absent any bias, equilibrium features segregation. Moreover, welfare improvements enhance segregation. In particular, radical improvements in search technologies induce complete segregation and a “dating apocalypse'' where agents replace partners frequently. We show that, in line with empirical findings, segregation is decreasing in couples’ attractiveness, and provide conditions for (probabilistic) positive sorting by attractiveness.
Keywords: Statistical discrimination; Segregation; Learning; Search and matching frictions
JEL Codes: J12; J15; D83
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| agents' optimal dating strategies (C78) | segregation arises in equilibrium (D59) |
| noisiness of evaluations (C52) | agents' optimal dating strategies (C78) |
| search technologies (O33) | increased segregation (R23) |
| couples' attractiveness (J12) | segregation (Y40) |
| positive assortative matching by attractiveness (C78) | longer singlehood episodes for less attractive agents (C78) |