Congratuations to my daughter, Elana Thurston-Milgrom, who has completed her masters degree at the Central European University in Buddapest. Her thesis is “Verny and Wahr from A[ssimilationist]s to Z[ionists] ”
Two big spectrum auctions in India and Germany ended this week.
The 3G auction in India yielded nearly US$15 billion for the Indian treasury for about 40MHz of bandwidth covering the whole nation. This includes the 10MHz that had previously been allocated to the two state-owned telecoms, which now have to pay the prices set in the auction. While revenues in the Indian auction were much higher than the government had forecast, they fell far short of their potential. The bidder that I advised was able to take advantage of the auction’s unique and flawed closing rule to end the auction before the still-remaining competitors had exhausted their demonstrated budgets.
The German auction ended very differently, with total prices of about US$5 billion for 290 MHz of paired spectrum (and some unpaired) covering the profitable German market. The revenues were about half of what some of the leading analysts had forecast. The much lower revenue per unit of bandwidth was realized despite the presence in the German auction of 60MHz of bandwidth in a low frequency (800MHz) block, which accounted for most of the spending in the auction. In this auction, again, the bidder that I advised acted decisively to encourage an early end to the bidding, reducing the prices it paid.
The comparatively poor revenue performance of the German auction compared to the Indian one reflects three big differences. First, the German auction offered a vast 290MHz of supply, leaving plenty for all bidders to acquire valuable packages of spectrum without needing to drive prices very high. At the same time, the auction authorities conveyed extra information to the bidders, enabling them to coordinate expectations about who would get what and further softening the competition. Finally, the participants in the German auction included only incumbent telcos. In contrast, the rapidly growing Indian market (currently adding about 15 million new sim-cards per month!) was a magnet for new entry and active competition. The bidders included a non-incumbent, Etisalat, which brought a large budget to spend, and which exited empty-handed only when the total prices of a nationwide license rose past US$2 billion. In addition, Aircel - an incumbent in some circles in the Indian market, bid aggressively but unsuccessfully to expand its footprint into the Mumbai and Delhi markets, driving up the prices of those circles.
These two now-completed auctions offer many lessons. Bidders should pay particular notice to how much is gained by good strategy and tactics. For my own clients, strategic thinking allowed us to buy at reduced prices or win a preferred combination of spectrum blocks. Auction designers should examine how specific rules affected the revenues in each of these auctions and, more importantly, how participation by an additional bidder can lead to much higher prices. And researchers should notice the sharp differences in the pattern of budget exposure in these auctions compared to the pattern for US spectrum auctions (as reported in my joint research with Professors Jeremy Bulow and Jonathan Levin), which suggests new theories about how real bidders behave.
On January 1, I apparently became the first economist ever to be awarded funds under the Small Business Innovative Research (SBIR) program of the National Science Foundation. I am principal investigator for Auctionomics on a proposal entitled “Incorporating Bidder Budgets in Multi-Item Auctions.” Bidder budgets are important in practice, but normally excluded from the messages communicated by bidders to the auctioneer during the course of an auction. Auctionomics is developing technology to make appropriate use of budget messages.
Over the years, I’ve invented several new multi-item auction mechanisms intended for use in practice. Portugal has now adopted the Day-Milgrom core-selecting auction mechanism for an upcoming radio spectrum auction. Detailed rules can be found here or on the Portuguese regulator’s website.
A paper explaining the logic of this design — “Core-Selecting Auctions” — can be found in the Publications section of this website (link).