Michael Ostrovsky
Business Economics PhD
Dissertation Chair: Prof. A. Roth
Essays on Matching: Stability in Supply Chain Networks; Trade Patterns under Transportation Cost Heterogeneity; Two-Sided Matching with Common Values
This thesis consists of three essays on matching theory and its applications. The first essay shows that the theory of matching in two-sided markets can be naturally generalized to a much broader setting. The second essay presents a new approach to modeling commodity trade flows, using the techniques of matching theory. The third essay introduces and studies two-sided matching with common values.
Two-sidedness has long been viewed as a critical condition for many key results of matching theory, including the existence of stable matchings. The first essay shows that this restriction is unnecessary, by generalizing the theory of matching in two-sided markets to a much broader setting---supply chain networks. Under natural restrictions, stable networks are guaranteed to exist. The set of stable networks is a lattice, with side-optimal stable networks at the extremes. Several other important results on two-sided matching also extend naturally to the more general setting.
The second essay uses the theory of matching to analyze international commodity trade flows. The standard approaches to modeling such flows assume either perfect substitutability of goods from different countries, which results in unrealistically sharp predictions, or product differentiation by the country of origin, which is plausible for some goods, but is hard to justify for many commodities. The essay introduces an alternative model, which relies on heterogeneity in bilateral transportation costs between the agents who compose countries or regions. This model fits data better than conventional models do, while relying on more natural assumptions and using more realistic parameter values. The theory of matching provides a convenient framework for incorporating the assumptions.
The third essay studies the effect of relaxing another common restriction of matching theory: full information about potential matches. In many situations, there is uncertainty about the qualities of potential matches, and preferences over these matches may therefore depend on the amount of available information. Hence, agents may infer additional information from the actions of others, and change their own actions in response. The essay shows that even in very simple settings with incomplete information and common values, a mechanism producing stable matchings may not exist.
