Essays on matching markets with correlated preferences

Date of Completion

January 2009


Economics, General|Economics, Theory




Since it was first introduced in 1962, matching theory has been very well studied in all aspects but one, namely correlation in the preference lists. This missing factor in the preferences might cause underestimation in the results of the studies on the matching markets. Moreover, it might be misleading in the analysis and the evaluation of the market outcome. The objective of this dissertation is to fill this gap in the matching theory literature, that is, to explore, via simulations, the effect of correlation in the preference lists on the aggregate satisfaction of the participants. ^ In the first chapter, a general methodology is presented to introduce correlation in the preference lists that can be used in any kind of matching market. The second chapter focuses on the simplest two-sided and one-to-one matching market, that is, a marriage matching model, using the men-propose Gale and Shapley algorithm. The third chapter focuses on a one-sided matching market, namely the roommates problem, using the extended version of the Gale and Shapley algorithm. For each of the matching markets in question, a measure to quantify the level of the correlation is also provided which enables us to sort the preference profiles according to their correlation levels and makes it possible to do statistical analysis. ^ Results show that the correlation is an important factor that affects the aggregate satisfaction levels of the participants. Correlated preferences mean that everybody is competing for the few popular candidates and competition results in significantly less satisfied participants. The second and third chapters also provide regression equations to predict the aggregate satisfaction level before running the algorithm once the correlation levels are known. ^