["The Journal of Industrial Economics, EarlyView. ", "\nABSTRACT\nFavoritism as a mode of corruption is a major concern in public procurement (PP) in many countries. We propose a regression discontinuity design (RDD) test arguing that the winning history of marginal winners and losers can be used to detect favoritism. We apply the empirical test to PP of cleaning services in Finland and find evidence of favoritism even in this relatively low corruption environment. First, winners of close races in terms of win margin tend to have won more also in the past, which is in contrast to competitive environment. Second, to rule out an alternative explanation of collusion based on incumbency, we show that these winners also tend to submit their bids the last, indicating information sharing about rival bids from the buyer to the favored bidder.\n"]