Average-case analysis of the assignment problem with independent preferences

Abstract

The fundamental assignment problem is in search of welfare maximization mechanisms to allocate items to agents when the private preferences over indivisible items are provided by self-interested agents. The mainstream mechanism Random Priority is asymptotically the best mechanism for this purpose, when comparing its welfare to the optimal social welfare using the canonical worst-case approximation ratio. Despite its popularity, the efficiency loss indicated by the worst-case ratio does not have a constant bound [Filos-Ratsikas et al., 2014]. Recently, [Deng et al., 2017] show that when the agents' preferences are drawn from a uniform distribution, its average-case approximation ratio is upper bounded by 3.718. They left it as an open question of whether a constant ratio holds for general scenarios. In this paper, we offer an affirmative answer to this question by showing that the ratio is bounded by 1/µ when the preference values are independent and identically distributed random variables, where µ is the expectation of the value distribution. This upper bound also improves the upper bound of 3.718 in [Deng et al., 2017] for the Uniform distribution. Moreover, under mild conditions, the ratio has a constant bound for any independent random values. En route to these results, we develop powerful tools to show the insights that in most instances the efficiency loss is small.