Knowledge Problem

Lynne Kiesling

Courtesy of Slashdot, and following up on Mike’s spectrum auction post from earlier this month, the 700 MHz spectrum auction has come to a close. From the New York Times blog post mentioned in the Slashdot article:

The winning bids totaled $19,592,420,000. That’s nearly double the amount the commission had hoped to raise from the spectrum being abandoned next year as television stations switch to new frequencies. On the scale of billions, the total has hardly changed in a month. But bidding continued on little blocks of frequencies around the country that cellphone companies are using to fill in gaps in their service. The last bid in the auction was $91,000 for frequencies around Vieques, P.R.

Lynne Kiesling

The palpable buzz around the release of the iPhone last year was exciting, and indicated some of the great design and functionality we can expect from mobile phones in the future (even if I won’t get one yet because I won’t give AT&T my business, and I haven’t seen a local vendor selling them unlocked yet for me to take to another provider). But at the same time the iPhone’s functionality was limited relative to its capabilities; it could do so much more than the web-based applications and the Apple applications installed on it.

One way to achieve that potential is for Apple to allow third-party developers to develop applications for the iPhone. Developers are chomping at the bit to do so. So yesterday Apple announced a Software Developer’s Kit (SDK) that developers can use to get the technical information they need to develop applications for the iPhone. Moreover, the venture capital firm Kleiner Perkins Caufield & Byers had established a $100 million venture capital fund for iPhone entrepreneurs, called the iFund.

The title of this PC World article about the SDK gets at the interesting economics of Apple’s decisions, from a property rights perspective.

Apple’s SDK blows open the process of creating native apps for the iPhone by letting most any would-be coder get started. Developers can sign up and download the SDK for free, which in turn allows Apple to reach out to a wider cross-section of would-be coders than they might have otherwise.

According to iPhoneDevCamp co-founder Raven Zachary, “The fear [in the development community] today was that Apple was going to constrain the ability for third-party developers to distribute apps, in the same way they did with the iPod games market.” There, Zachary notes, Apple made it very difficult for small developers to create and release a game: “You have to get Apple’s approval, have them approve the source code, and then they take a large percentage of the profits for the distribution of that app.

“What we’ve seen instead is Apple opening up the marketplace in the same way they’ve opened up the podcast directory in iTunes Music Store. They will be far more open about letting developers list their apps,” says Zachary.

Apple’s software and design are the reasons that this device is so popular, is the phone to beat in the mobile market, and will become increasingly useful in business environments (which is one of the objectives of the various iPhone moves that Apple is currently making). Apple has a long history of keeping its software, its operating system, etc. proprietary; in fact, in the 1980s one of the biggest criticisms of Apple was that IBM and Microsoft’s operating system proliferated and gained larger installed base and market share because of the extent to which they shared information about how to write software interfaces to their operating system.

So here is Apple, apparently having learned from its 25-year history of deciding how far down the property right chain they should retain use rights. And that’s exactly what an SDK does — it grants use rights to some of Apple’s property, to enable them to develop applications that Apple hopes will make that property, and the iPhone, more valuable to more possible customers. This is a far cry from the 1980s Apple approach to software development, and as the quote above suggests, is not necessarily universal across the entire Apple product line.

I think this CNET article hits the nail on the head:

To me, the most interesting thing about the development of the smartphone industry is the wide-open nature of the race. This time around, a winner is not going to be picked in the early stages of the competition. Several huge important companies–Apple, Microsoft, Google, Nokia, RIM, and don’t forget about Palm just yet–have already had an impact on the development of the product, and will continue to do so well into the future.

Despite sitting out the first few years, Apple has arguably vaulted ahead of its competition in just 12 months. The other players in this industry came into smartphones building them for businesspeople and their IT masters. Then they tried to woo the consumer.

Apple has done the complete opposite, hooking those who had never used a smartphone before with the iPhone’s interface, and now giving them the opportunity to use it for both work and play.

The first era of the mobile-computing industry was about hardware. The second part will be about software. And right now, no one is developing mobile software like Apple.

Apple is allowing developers more generous use rights in its software to strengthen the value of the iPhone, and to open the door to further innovation in mobile computing and communications. I hope they continue to turn the mobile communications market on its head: this type of creative destruction is the foundation of our economic well-being.

Michael Giberson

I was wrong a few weeks ago when I wrote, “While the FCC has sponsored research and considered deployment [of package bidding and combinatorial auctions], so far they have been dissuaded by the complexity of the auction design and opposition from commenters.” The FCC has tried package bidding once before, in 2003, when it auctioned six regional narrowband PCS licenses in the 900 MHz band. It wasn’t much of a test of combinatorial auction designs, however: only two bidders participated in the auction, and the FCC sold only five of the six licenses offered.

Much more interesting is the limited package bidding to be allowed in the auction of 700 MHz spectrum scheduled to begin January 24. This is the auction that has attracted the interest of Google along with just about every wireless company out there. (Wired reports on why the auction is such a big deal.)

One piece of the 700 MHz spectrum being offered, the “C Block,” is attracting the most interest, and it is the block that the FCC is offering under new “Hierarchical Package Bidding” (HPB) rules. HPB is a complicated name, but it is a pretty simple idea. In effect, the FCC is offering two ways to bid on the 12 regional C Block licenses: companies can submit individual bids on one or more of the 12 licenses, and companies can submit bids on three super-regional packages of licenses. The most interest will likely be in the package containing regional licences 1-8, covering the fifty U.S. states. (The other four licenses are divided into an Atlantic package, covering Puerto Rico, the U.S. Virgin Islands and the Gulf of Mexico, and a Pacific package, which covers U.S. Pacific territories.)

The HPB rules emerged from the experimental auction design work of Charles Holt and Jacob Goeree while they were testing a variant of the FCC’s simultaneous multi-round auction rules that permitted package bidding. (See links below.) Relatively poor performance of the FCC’s package bidding auction design led Holt and Georee to examine alternative approaches, and they eventually developed and tested the HPB design.

Various experimental tests examined the FCC’s standard simultaneous mult-round ascending auction, the FCC’s variant SMR with package bidding, an approach called RAD, the combinatorial clock design, and the HPB design that Holt and Goeree proposed as an alternative. In accepting the HPB, the FCC said:

[W]e will use HPB in part because the mechanism for calculating [prices] is significantly simpler than other package bidding pricing mechanisms. … In addition, we find that . . . HPB procedures in general strike a careful balance between permitting bidders adequate bidding flexibility and discouraging insincere and anticompetitive bidding behavior.

The design has been developed and tested in the economics lab. We are about to see how well it performs in practice.

NOTES: The FCC provides much more information on the 700 MHz auction online. I extracted the FCC quoted material from a Caltech press release that also has comments from Goeree and additional commentary. The results of the tests have been presented in numerous reports authored by Goeree, Holt, and others:

Hierarchical Package Bidding: A Paper & Pencil Combinatorial Auction (October 2007)

An Experimental Test of Flexible Combinatorial Spectrum Auction Formats (September 2007)

An Experimental Comparison of Flexible and Tiered Package Bidding (May 2007)

An Experimental Comparison of the FCC’s Combinatorial and Non-Combinatorial Simultaneous Multiple Round Auctions (July 2006)

Comparing the FCC’s Combinatorial and Non-Combinatorial Simultaneous Multiple Round Auctions: Experimental Design Report (April 2005)

Michael Giberson

David Porter and Vernon Smith have put together an assessment of the FCC’s 12 years of experience with spectrum auction design, forthcoming in the Journal of Law Economics and Policy. From the introduction:

Recent policy discussions regarding broadband Internet access have revived debates about various methods for allocating electromagnetic spectrum rights and the appropriateness of spectrum auctions. Debates over the assignment of spectrum rights via auction are hardly new. Economists have long argued that auctions would promote efficiency… Still, auctions have attracted vigorous opposition from defenders of traditional “public interest” licensing… When this market-oriented approach was actually implemented in the United States in the early 1990s, these debates ceased being purely theoretical. Twelve years of actual experience with spectrum auctions now allow us to look back and assess what lessons can be learned from the adoption of this approach.

The paper provides a focused, non-technical look at market design issues that have arisen as the FCC implemented a 1993 federal law permitting spectrum auctions. Many of the “best and brightest” among auction economists have participated in the research and analysis of the FCC’s auction design proposal. Perhaps not surprisingly, auction theory didn’t contain an answer for every broad design issue or narrow auction implementation task that came along. As a result, there has been a lot of trial-and-error and learning along the way.

Porter and Smith are experimental economists, and their writing is heavily informed by their background. After assessing some “unexpected” strategic behaviors in early auctions – jump bidding, parking, retaliatory bidding, and various other signaling games – Porter and Smith note that the “behaviors observed in these auctions may have been surprising to the FCC and some economists, but they had been observed many times in laboratory experiments.”

The article then turns to a discussion of “new, untested rules” imposed by the FCC in the attempt to curtail strategic bidding and speed up auctions, and two areas of current interest: information policy and package bidding. Information policy concerns how much information about other bid and bidders during the course of the auction. Generally, the more information disclosed, the more effort bidders will put into strategic bids – often to the detriment of auction efficiency and seller revenue. The FCC proposed a strict “no information” policy for the Advanced Wireless Services auction, a position endorsed by the FTC and Department of Justice, but then decided to reveal all information if certain (easily met) conditions were in fact met. (They were.) Porter and Smith remark that information policy can be tested in the lab.

Package bidding, which requires a combinatorial auction design, gives bidders tools to bid for just the combinations of licenses desired, something complicated or impossible to do in the FCC’s standard simultaneous ascending auction design. Package bidding can be particular valuable in spectrum auctions because depending upon a bidder’s business strategy and existing resources, any one license may be either a complement or substitute for other licenses being offered. Both Smith and Porter (among others) have long advocated use of combinatorial auctions for the sale of spectrum. While the FCC has sponsored research and considered deployment, so far they have been dissuaded by the complexity of the auction design and opposition from commenters. [See UPDATE, below.] (Combinatorial auctions have been used by others, see the citations in the articles for examples.)

A sort of motto or theme that permeates the work of economists involved in experimental testing of market designs is, in effect, “You can pay a little, and learn from your mistakes in the lab, or you can pay a lot when you learn from your mistakes in the field.” While Porter and Smith don’t quite state the idea outright, it is fundamental to their outlook. In fact, the idea is embedded in their title, “FCC Spectrum Auction Design: A 12-year Experiment.” The FCC hasn’t avoided testing their auction design ideas in experiments, it is just that they have chosen to experiment in high-stakes, hard-to-control, live action auctions with billions of dollars of resources at stake. How many resources have been wasted in the FCC’s 12 years of poorly controlled experiments?

Of course, the auctions, even if not as efficient as they could be, are vast improvements over allocation by government fiat or allocation by lottery, the approaches auctions have replaced.

Porter and Smith say, “a fresh reexamination of the FCC auction design protocols is overdue,” and they suggest a few criteria: reduce participation costs to bidders, make it easier to assemble desired packages, and minimize incentives for strategic behavior. And, of course, “Whatever the proposed redesign, it should be thoroughly tested first in the laboratory.”

UPDATE: I was wrong, the FCC hasn’t been dissuaded completely from trying package bidding, as I explain in a January 2008 post, the FCC tried package bidding in a small way in 2003 and is about to try a new package bidding design in the big 700 MHz auction. See that post for additional detail and links.