Commons with increasing marginal costs: random priority versus average cost*
比较了平均成本与随机优先两种机制在边际成本递增的公共资源分配中的效率，发现当用户数量超过可负担产出2.4倍时，随机优先机制显著优于平均成本。
Indivisible units are produced with increasing marginal costs. Under average cost , each user pays average cost. Under random priority , users are randomly ordered (without bias) and successively offered to buy at the true marginal cost. Both average cost (AC) and random priority (RP) inefficiently overproduce. RP tends to overproduce less, but which game collects more surplus depends much on the demand configuration. We show that a key to compare the welfare properties of the two mechanisms is the crowding factor , i.e., the number of potential users over the number of units of output users can afford: The more crowded the commons, the more RP outperforms AC. In the quadratic cost case, beyond the threshold value of 2.4 for the crowding factor, RP strongly outperforms AC; beneath it AC only mildly outperforms RP. Thus the RP mechanism manages crowded commons better than AC.