Marriage and dating online in Korea

Soohyung Lee has a paper about an online marriage market in Korea, whose findings are similar in some respects to those in the papers described in my two earlier posts today, and different in others: Preferences and Choice Constraints in Marital Sorting: Evidence From Korea.

Like the paper by Hitsch, Hortaçsu, and Ariely, Lee's paper considers detailed transactional data from an online site, this one aimed at marriage rather than more casual dating, and uses the data to estimate preferences for mates, based on a variety of observable traits. Like both Hitsch et al. and the paper on Indian marriages by Banerjee, Duflo, Ghatak and Lafortune, Lee's paper looks at stable matches produced by a deferred acceptance algorithm based on the estimated preferences, and compares them to the observed marriages in population data. Unlike those two papers, the findings in Korea are that the simulated stable matchings look different than the population matchings; they are less assortative on some dimensions (such as employment in the same industry).

The paper suggests that some aspects of Korean marriage data therefore reflect choice constraints on who meets whom, and that as these constraints are relaxed (perhaps by increased use of internet matchmaking), some aspects of Korean life may change.

Market design update (5/12/09): Professor Lee sends me the following email describing how her work has proved useful in the design of the matching site, which uses an algorithm to suggest several possible dates to each participant:
"FYI, the matchmaking company partially adopted my estimates to adjust its matching algorithm. Previously, it assumed all men and all women have the same preference ranking. Under the revised algorithm, it classifies men and women based on their characteristics and allows them to have heterogeneous preference rankings (which I find in my estimation).
With the revised algorithm, the company was able to increase the probability of a proposal turning into an actual first date by a factor of 2. "