By Zoë Hitzig
This post is part of the HISRECO 2018 series. Participants in the 2018 HISRECO conference were asked to write short blogposts to highlight their contribution to the conference for a general audience. The idea of this blogged conference comes from the “Learning by the book” conference in Princeton, published as a series on the History of Knowledge Blog. This post is number 1 of 7.
Veteran auction designer and Stanford economics professor Paul Milgrom claims that the U.S. government’s 2016 “incentive auction” was “by far the most complicated resource reallocation ever attempted, anywhere in the world.” The auction—in which media and telecommunications companies bid for the rights to broadcast over the airwaves (electromagnetic frequencies between 9 kilohertz and 330 gigahertz)—was the result of many years of design, experimentation, and collaboration between a Milgrom-led team of economists and computer scientists and the federal government.
The design aimed to address an age-old problem: How can a regulatory body guide the use of a public good, in this case the airwaves, into its most socially valued uses? While decentralized market-based processes may be highly efficient, treating public goods like private goods raises concerns about equity and access. And yet allocating the good through an entirely centralized process, perhaps through a legislative body in a representative democracy, can be prohibitively inefficient, especially for a good as complex as the airwaves.
Carefully designed auctions theoretically strike a middle ground. Policymakers articulate their goals to economists, who use economic theory to draw up “incentive compatible” auction designs that harness information as in a market mechanism, while ensuring that certain policy goals are met. Policy imperatives act as constraints on the theorist’s designs. In the case of spectrum, in which the uses of the good are varied (wireless, broadband, radio and television broadcasting) and require lengthy consideration of interference issues, the set of constraints becomes stunningly complex.
In the face of such complexity, how did the “incentive auction” designers identify the most socially valued uses? More broadly, in their capacity as consultants to the federal government, how do auction designers ensure that the allocation of the public good aligns with the public will? I argue that the unprecedented complexity of the “incentive auction” gave rise to a new role for the economist in allocating and reallocating public goods.
Milgrom, auction designer extraordinaire, exemplifies this new form of economic expertise, drawing on decades of engagement with the theoretical and practical problems posed by auction design. He was one of the first economic theorists, along with Robert Wilson and Preston McAfee, to draw up designs for the U.S. government’s first ever auction of spectrum licenses in the early 1990s. Since then, he has helped the Federal Communications Commission (FCC) design over 87 spectrum auctions, and he has also advised designers and prospective bidders in power auctions, bankruptcy auctions and spectrum auctions through his consulting firm Auctionomics. Further, he has authored countless theoretical papers in so-called “market design,” a field of economics in which economic theory guides the design of practical solutions to specific, often quite thorny, resource allocation problems.
In this new role, the economist is more than just an adviser or an engineer but also a craftsperson, i.e. a technologist who implements as well as designs, and who creates as well as conceives. This shift in the role of the economist, I argue, corresponds to a shift in the locus of the public’s trust. The public’s trust shifted away from legislative bodies and economic efficiency, and toward trustworthy experts. When experts play this role in the allocation of a public good, the government, perhaps unwittingly, enables a radical redefinition of public interest.
Tragedy of the commons & the anti-commons
The story of spectrum allocation in the U.S. begins in the early 1920s, when the outgrowth of broadcast radio led to a tragedy of the commons on American airwaves. With fledgling media companies, government programs, and techie amateurs all competing for airtime and thus interfering with each other’s signals, the airwaves were cacophonous. Herbert Hoover, Secretary of Commerce at the time, lamented “One of the misfortunes of our present Government structure, and one which needs constructive thought is that we have no bureau or organized authority for dealing with the communications question.”
The initial solution to the tragedy of the commons on the airwaves was tight government regulation. After some confusion and delay, U.S. Congress passed The Radio Act of 1927. The Act declared the spectrum to be “the inalienable possession of the people,” to be regulated by a government organization “as the public interest, convenience, or necessity” requires. Though the regulatory body could deny and grant licenses and assign frequencies and power levels, the commission was not given official power to censor communications.
With the Communications Act of 1934, Congress transferred power over the radio waves to a new commission, the FCC, which also oversaw regulation of telephone services. Congress granted the newly-minted Commission the task of “regulating…to make available so far as possible to all the people…a rapid, efficient, Nation-wide, and world-wide communication service with adequate facilities at reasonable charges.”
Initially, the FCC distributed spectrum licenses to applicants through what came to be known as “beauty contests.” Public interest was thus mediated by lawmakers: Congress developed a system of comparative hearings in which potential licensees demonstrated how their proposed use of the spectrum aligned with public interest. This led to a typical tragedy of the anti-commons in which arbitrary bureaucratic regulation enabled only a few favored companies to receive licenses, thereby destroying any hope of guiding the spectrum into its most socially valued uses.
Economist as adviser
Meanwhile, in a famous 1959 paper, “The Federal Communications Commission,” economist Ronald Coase criticized the government’s approach to spectrum management. Coase proposed a complete system of property rights on the spectrum. If only the government had realized, Coase argued, that the spectrum could be treated like any other resource, they might have seen that the proper approach would have been to allow a full property rights regime. With full property rights for spectrum, he argued, a market for spectrum would most efficiently allocate the public good in the public interest.
Coase’s proposal was met with skepticism for several decades. Shortly after Coase published his seminal article, he was invited to testify before the FCC about the future of broadcasting. So radical was his proposal that FCC commissioner Phillip Cross retorted, “Are you spoofing us? Is this all a big joke?” A few decades later, Senator Daniel Inouye, Chair of the Senate Subcommittee on Communications, argued in 1987 that an auction for spectrum “undercuts the fundamental tenet in communications policy—that the airwaves are a limited public resource [and it] is inappropriate to sell such a resource to the highest bidder.”
Despite the criticism lodged against his proposal following his 1959 publication, Coase nonetheless fulfilled the role of economist-as-adviser during the ensuing decades. He not only published his perspectives in academic journals but also offered these perspectives in government hearings and in a detailed report commissioned by a prominent public policy think-tank.
Economist as engineer
Eventually the Coasian view on spectrum policy gained traction and Congress voted to run the first-ever spectrum auctions in the U.S. To initiate the process, the FCC solicited public comments in an October 1993 “Notice of Proposed Rule-Making.” The Notice received over 200 replies from firms and lobby groups, many of which were written with the help of academic economists. The auction design adopted was a simultaneous ascending auction proposed by Milgrom, Wilson, and McAfee—all academic economists who were employed as consultants for telecommunications companies.
Nobel prize-winning economist Al Roth points to the FCC’s first spectrum auctions as the earliest example in which economists were more like “engineers” than advisers: “Economists were presented with opportunities to take responsibility for the design of detailed rules for complex markets.” Since then, economist-engineers have been involved in the design of the National Residency Matching Program for doctors, a host of public school districts’ student allocation procedures, exchange “markets” for kidneys, and numerous other real-world markets. Roth cemented this nascent identity in his 2002 paper titled “Economist as Engineer”—and a decade later he shared the Nobel Prize with Lloyd Shapley for his theoretical and practical contributions to so-called “market design.”
The social engineering role for economists is a marked departure from that of the economist as adviser. Comparing the role of Coase with the role of the 1994 auction designers illustrates the most salient differences between the two. Coase—with his journal articles, hearings, and think-tank reports—offered an economic perspective on the issue of spectrum allocation. He offered a way of looking at the issue that the policy-maker might want to consider in the process of writing the most effective policy. The auction designers, on the other hand, were involved in every stage of the policy-writing process. They contributed the actual rules that would be implemented in the first ever auction.
Economist as craftsman
The 2016 “incentive auction” was different from the airwave auctions that came before. A rapid change in patterns of demand for spectrum brought on by the rise of wireless broadband technology fundamentally changed the nature of the allocation problem. Two major trends altered the valuation of current licensees’ uses of the spectrum: the decline of television broadcasting and the rise of wireless Internet services over 3G, 4G, and LTE, that ensure “wherever, whenever access to broadband.”
Following Coase’s reasoning, one might think telecommunications companies could simply buy the spectrum licenses currently held by broadcasters who were not using them to their full value. In a market-based allocation procedure, as neoclassical theory goes, prices promise to represent consumer value.
However, given the nature of the spectrum, the problem is more complex. Due to interference problems on the spectrum, it is of critical importance to coordinate various uses proposed by potential users. In addition, it would be near-impossible for telecommunications companies to come up with a strategy for buying individual licenses in a way that ensures nationwide coverage.
The incentive auction, which was initiated in March 2016, marks a departure from previous auctions in that it had 3 phases, instead of just a single round of sales. In the first phase, the “reverse” auction, the FCC bought back spectrum from licensees who were willing “to voluntarily relinquish some or all of their spectrum usage rights”. Second, there was a “repacking” phase, in which the economists repackaged the newly-freed spectrum licenses into blocks suitable for telecommunications companies. Finally, in the third phase, the “forward auction” these newly packaged licenses were sold, mainly to wireless companies.
The repacking phase is where the economists took on an unprecedented role. Solving the repacking issue is computationally hard for even the most sophisticated computers. Its solution does not come from economic theory but rather from complexity theory. The repacking problem is similar to (and yet more complex than) a class of problems that mathematicians call graph coloring problems.
Solutions to graph coloring problems are famously difficult—they are classified as “NP-complete” which means that once the graph coloring problem reaches a certain size, there is no guarantee that the most sophisticated algorithm could solve it in reasonable time. The time it takes to solve a graph coloring problem increases exponentially with the number of constraints. As certain pairs of stations cannot be assigned to the same or adjacent channels (if they are geographically close to each other), the repacking of the 2,000 television stations requires consideration of over 600,000 interference constraints.
The auction designers’ central priority in designing the incentive auction was to simplify the rules of the auction for bidders. “We wanted to put the complexity under the hood,” Milgrom explains, “You step on the gas and you don’t worry how the engine works.” In this capacity, the economists designing the auction were not just “engineers” who proposed an ideal design for the institution, but craftsmen, who both designed and played a key role in implementing the auction. The craftsman role of the economists indicates a shift in the vehicle of public trust: away from the transparency and objectivity of markets and toward trustworthy experts.
|1927 – 1993||1994 – 2015||2016|
|Spectrum policy||Political hearings||Single Phase Auctions||Incentive Auction|
|Vehicle of trust||Lawmakers||Economic efficiency||Trustworthy experts|
|Role of economists||Economist as adviser||Economist as engineer||Economist as craftsperson|
The history of economists’ involvement in spectrum policy thus charts broader shifts in the locus of public trust. A deep trust in the ability of markets to mediate the public will animated Coase’s advice for handling the spectrum. In that setting, public interest was taken to be best represented in a transparent market-based allocation. The incentive mechanism, although it uses all the rhetoric of markets and efficiency, and even though Milgrom in some ways serves as heir to Coase, ironically shows a return to the impenetrable, subjective, and bureaucratic decision-making that the “beauty contest” model (1927-1993) exemplified.
Milgrom et al. were in charge of solving a fiendishly complex problem that even an algorithm cannot solve. His team’s judgment about what to do “where optimization is impossible” was rooted in decades of experience organizing spectrum auctions and in often-inarticulable judgment about the trade-offs between efficiency, complexity, and fairness to participants. What has resulted from this “craftsperson” role for the economist is a changed, narrowed definition of public interest as short-term consumer value.
This high-stakes reallocation of an “inalienable possession of the people” will have effects on the lives of the American public for years to come. And yet, the reliance on craftsperson expertise obscures the relationship between the allocative decisions and the public will. Perhaps this obfuscation is inevitable. After nearly a century of spectrum policy, the words of William Howard Taft, Chief Justice of the Supreme Court during the rise of the radio broadcasting industry in the 1920s, are truer than ever: “interpreting the law on this subject is something like trying to interpret the law of the occult.”
Zoë Hitzig is a doctoral student in economics at Harvard.