Two-sided matching for the US Navy's enlisted detailing process a comparison of deferred acceptance and linear programming via simulation

Abstract

Recent studies of 2-sided matching mechanisms have suggested potential benefits for implementation into the Navy enlisted assignment process. The proposed matching process improves the chance of commands and sailors being assigned either a sailor or billet of their choice. The same studies focused on a particular two-sided Deferred Acceptance (DA) matching algorithm which ensures stable matches, prevents "off-the-site" trades between the matching parties and upholds integrity of the matching system. Although stable matches are important in a voluntary labor market, the DA algorithm may still favor one party depending on whether the command or sailor biased form of the algorithm is used. The Linear Programming (LP) algorithm is an alternative that could optimize system (command and sailor) effectiveness and promote a balanced approach to meeting the preferences of both parties. Although LP does not guarantee stable matches, it is still employed by selective British hospitals for their matching with interns. The extent of the unstable matches has not been examined to measure it against the benefit of higher system effectiveness. This thesis will evaluate if the LP algorithm could serve as a better alternative to DA algorithm through simulation of the Navy enlisted assignment process.