Second-Price Auctions in Reality vs. in Theory
In the article “Second-Guessing the Second-Price Auction Model,” the author, Esco Strong, argues that second-price auctions are problematic in real-time bidding marketplaces. This is because the conditions that are necessary in order for the second-price auction model to be effective are not met in real-time bidding auctions. The author argues that this model is only effective if the auction is for a unique item (one without substitutes) and thus that the buyers are willing to bid their [maximum] true private values. However, in real-time bidding auctions, buyers do not submit their true private values because the items are not unique and the buyers have finite budgets that restrict them. Since the auctions are held in a repetitive manner, buyers adjust their bids because they learn over time which bid is optimal in the context of the market and because they try to get the collection of goods that they need at the lowest rates. Therefore, it is no longer a dominant strategy for the buyers to bid their true private values. The author then goes on to argue that this leads to inefficiency and lack of trust in the marketplace because buyers do not operate according to the second-price auction model and sellers assume that they are doing so. In order to combat this, sellers may implement dynamically generated price floors, increasing the prices so that it nears the buyers’ first-prices, yet they still advertise the auctions as second-price. The author then suggests that the real-time bidding marketplace should switch to use a first-price auction model as it would be more straightforward and efficient.
This article could not be understood without the knowledge of what second-price and first-price auctions are. Second-price auctions are those in which the highest bidder wins the auction but pays the second highest price. First-price auctions are those in which the highest bidder wins the auction and pays the value of his/her bid. In a second-price auction, it is a dominant strategy for the bidder to bid the amount at which he/she truly values the item. This is because if the bidder bids higher than his/her true value, he/she risks having a negative payoff because he/she could pay more than his/her true value. If the bidder bids lower than his/her true value, he/she risks not winning the item because he/she unnecessarily bid too low. In a first-price auction, it is a dominant strategy for the bidder to bid below his/her true value because he/she would have a payoff of zero if he/she won the item and bid truthfully. By bidding below his/her true value, the bidder would have a positive payoff if he/she won. However, it is important to realize that a knowledge of second-price and first-price auctions alone is not sufficient in order to determine what is optimal in a marketplace in reality. This is because although these auction models may seem effective in theory, their outcomes may be very different in practice. This is because marketplaces are often complex and it is impossible to predict the behavior of each buyer and seller (since they may not make logical decisions). This is important to consider when determining the efficacy of different auction formats in various situations. Whether you agree with his argument or not, Esco Strong’s article draws attention to an important concept: auction models can operate very differently in application than in theory.
References:
–https://adexchanger.com/data-driven-thinking/second-guessing-the-second-price-auction-model/
–https://blogs.cornell.edu/info2040/2016/10/21/reality-check-second-price-auctions/
-http://www.cs.cornell.edu/home/kleinber/networks-book/networks-book-ch09.pdf