Revenue and efficiency ranking in large multi-unit and bundle auctions

Should a seller use a multi-unit auction for identical and indivisible units of a good? We show, under specific assumptions on the value distributions of the bidders, that in large markets the multi-unit format generates higher (lower) expected revenue compared to the bundled format when the supply is relatively scarce (abundant). In contrast, a large market is shown to be always more efficient under the multi-unit format than its bundled counterpart. Thus under these assumptions a profit maximizing seller is expected to choose the relatively efficient multi-unit format when supply is scarce.