ESTIMATING PREFERENCES IN TWO-SIDED MATCHING MARKETS

In two-sided matching markets –such as labor markets – the outcome is a function of agents’ preferences. If we could infer preferences from market outcomes, we could advise market agents on how to improve their odds of matching with a desirable partner. We introduce a model to estimate agents ’ preferences and consideration sets in two-sided matching markets and illustrate its use with data from the marketing assistant professor job market. Our results show that asymmetry of preferences is common. For example, a marketing department may be effective in getting its graduates good placements but may be less attractive as a hiring department. University ranking is a significant factor –but not the most critical – in determining the appeal of the university as a hiring institution as well as the university’s graduates. The most critical factor for candidates is publication in a select few top journals, while for hiring departments it is the establishment of a Ph.D. program. Finally, our estimates explain market outcomes well, in addition to eigenvector centralities, implying that our preference estimates capture valuabl