Building political collusion : evidence from procurement auctions

We investigate the relationship between the time politicians stay in office and the functioning of public procurement. To this purpose, we collect a data set on the Italian municipal governments and all the procurement auctions they administered between 2000 and 2005. Identification is achieved through the introduction of a two-term limit for the mayor in March 1993: since elections were not coordinated across cities, and previous terms were not counted in the limit, mayors appointed right before the reform could be reelected for two additional terms, while the others for one only. Our primary finding is that one extra term in office deteriorates public spending. In fact, it decreases the number of bidders and, most importantly, the winning rebate. Interestingly, we also find that the probability that the same firm is awarded more auctions, or that the winning firm is local, increases with time in office. These results are compatible with the predictions of a model of favoritism in repeated procurement auctions, where time reveals collusive types, thus increasing the value of illegal connections at the expense of higher procurement costs. -- Procurement auction ; collusion ; public works ; time in office