Alkiviadis Trigas made a comment on a recent post asking about how to address (suspected, apparent) collusion in reverse auctions.
This is a topic in itself and worth a post (or several) in its own right. Although the question is posed with regard to e-auctions in fact it has as much to do with buying in general as it does with e-auctions in particular. So I’ll take a provisional stab at both sides of the question.
Collusion amongst suppliers in general
It’s worth remembering that this is not a new issue. Adam Smith, in The Wealth Of Nations (published 1776) writes that “People of the same trade seldom meet together, even for merriment and diversion, but the conversation ends in a conspiracy against the public, or in some contrivance to raise prices.”
In practice, I have yet to find a buyer who was able (or willing to expend the effort) to actually prove collusion. Most seem to accept that just as buyers want to get as much as possible for as few $$$ as possible, sales people want to deliver as little as possible for as many $$$ as possible. Indeed, most anti-collusion cases come about when one member of the cartel breaks ranks for whatever reason.
In practical terms the best defences buyers have is knowledge of the supply market, and to use that knowledge to broaden the supply market as much as possible. For example – even if there appear to be many distributors, are they only being supplied by a handful of manufacturers? Are there any alternative sources of supply overseas?
Buyers need to have this information – which is only really achievable through deep relationships in the supply market – early on in their sourcing processes. It’s no good waiting until you ask for pricing (whether you are doing an Excel RFQ or a reverse auction) to find out that the supply market is narrower than you thought it was.
Collusive behaviour in e-auctions
Collusive behaviour has been studied in both forward and reverse auctions. I give some examples in my white paper on e-auction design. One of my ‘favourite’ examples is the sale of 3G telecommunications licenses in Switzerland. There were 9 bidders bidding for 4 licenses. Each bidder could win at most 1 license. So the bidders formed joint bidding arrangements such that by the day of the auction there were only 4 entities bidding for 4 licenses. In the end each of these 4 bidders was awarded at the very low reserve price. (A great book to read on auction design that describes the 3G license auctions in detail is Paul Klemperer’s Auctions: Theory and Practice most of which is available online for free at http://www.paulklemperer.org/index.htm)
So what does this mean for reverse auctions?
As with any buying project, start by understanding the supply market and by widening it as much as possible.
From a strictly auction point of view, if you are stuck with what you suspect is a potentially collusive environment then you can still run an auction
Keep the number of bidders confidential
Run a reverse Japanese auction. In a Japanese auction the buyer sets the price level and suppliers accept that price level or withdraw. The buyer then progressively lowers the price level until the last supplier withdraws. I have seen this auction tool achieve quite impressive results, even in tight markets.