Columbia Logo 
















Spectrum Auctions and Sunk Costs*







Michel R. Abood, Neil Abromavage, Kenneth R. Carter, Paul Davis, Bob Dentice, Stelian Dragan


Columbia University

Executive MBA 2002



May 2, 2001



Table of Contents


I.    Introduction. 3

II.      Spectrum Auctions. 3

III.     Spectrum Costs. 3

A.   Accounting Costs vs. Economic Costs. 3

B.   Sunk Costs. 3

IV.     Spectrum Licenses as Sunk Costs. 3

A.   Economic Perspective. 3

B.   Empirical Perspective. 3

V.   Is Spectrum License a Sunk Cost?. 3

A.   Ex Ante Behavior - Irrational Exuberance. 3

B.   Ex Post Behavior – Oligopolies. 3

VI.     Conclusion. 3


I.    Introduction


The use of auctions as means to allocate radio spectrum bands for telecommunications is emerging as a hotly debated policy issue.  Current policy in most OECD countries is for the governmental regulator to auction off exclusive licenses to use certain frequencies.  Central to this debate is the auction’s effects on consumer welfare.  Proponents of auctions assert that spectrums institute market mechanisms into government policy, raising funds for the government and assuring that the spectrum is allocated to its highest value users.  Opponents argue that auctions require communications firms to make large unrecoverable investments prior to being able to offer service and that those costs are passed on to consumers in terms of higher retail prices of wireless services.  The FCC’s current position is that the wireless firms treat spectrum license fees as sunk costs because competition in those services drives prices to marginal cost.  License fees are therefore not passed on to consumers.  The argument assumes that FCC’s policy requires wireless carriers to make survival bids in order to provide services which will not yield a return on the bid.


If the FCC’s position is correct, then current wireless providers will not participate in subsequent auctions, or will only do so at significantly lower bids.  This should have been observed in the PCS auctions recently concluded by the FCC.  Since these auctions set new records for revenues, adjusting for inflation and set-aside preferences for small firms, some other explanation must be produced.  Other explanations include irrational exuberance in the valuation of such licenses or so-called “survival bidding” by current network providers.


This paper seeks to measure the correlation between sunk costs and participation in future auctions in order to test the FCC’s hypothesis.  It does so in two ways.  First, by looking at the behavior of firms in light of the cost of spectrum where carriers pay for license renewal or in cases such as digital broadcasting where the carrier has no spectrum costs.  Second, it proposes an experimental auction which tests whether license fees are sunk costs by looking at ex post behavior towards spectrum costs.  Such an experiment is a two-tiered auction.  In the first tier, qualified firms submit bids for the licenses, while in the second tier, one of the auction winners is selected at random to receive its license for free. The other auction winners will be required to pay for their spectrum licenses.


II.  Spectrum Auctions


Since the late 1920’s, the need to regulate the broadcast of radio signals into the ether has become apparent.  This need is due to a fundamental property of radio waves: as multiple radio operators attempted to use the same, limited number of frequencies, their signals caused interference, the waves amplifying or canceling one another out.  The result was that no one could clearly broadcast or receive signals.  The allocation of the airwaves to avoid interference has been accomplished through a variety of allocation policies, including give-aways, beauty contests, and eventually auctions.


Today in most countries, a central regulator assigns bands of adjacent frequencies to particular applications, and then allocates the exclusive right to those frequencies to minimize the problem of interference.  In most cases, the regulator holds an auction in the hope of ensuring an economic allocation of this scarce resource.  Potential wireless providers do not actually bid for spectrum.  Rather, they receive a license to the exclusive right to emit electromagnetic waves at a given frequency and power level in a specified geographic location.  In most auctions conducted by the Federal Communications Commission (“FCC”), license fees must be paid prior to the beginning of the license which typically lasts for a period of 6 to 15 years.  In February 2001, the FCC in the US completed the re-auctioning of the PCS licenses awarded in the 1996 C-Block auctions.  At the end of the license period, the government may reassign the right to use the frequency.  Once a wireless provider has won a license at auction, it may be bound to provide service and to pay the license fees whether it operates or not.  It may further be restricted from reselling the license.  The recipients of licenses must then make capital investments in network infrastructure to provide these services.


The wireless provider runs the risk of investing in network infrastructure it may not be legally permitted to operate in the future.  This is more than an impairment; the result may be a total loss of the utility of the spectrum license as an asset.  Moreover, it develops a future off-balance-sheet liability, the need to purchase subsequent spectrum licenses.  In such a case, it may not be setting aside sufficient reserves and overstating earnings.


III.       Spectrum Costs

A.   Accounting Costs vs. Economic Costs


Two different types of costs influence a company’s decision when deciding whether to purchase a spectrum license: accounting and economic costs.  The first is a company’s accounting cost.  We define this as the provider’s actual expenses, plus the depreciation for its equipment.  The cellular gear, switching, and trunking equipment necessary to provide service may become more expensive to operate as it gets older.  Such assets may require upgrades and maintenance and are subject to rapid obsolescence, but this does not apply to the spectrum itself.  The spectrum is no less productive later in its useful life.  In fact, the reverse seems to be true.  New equipment allows for increased capacity over the same amount of spectrum.  Since there is a preference for straight-line amortization, it is the most appropriate means for spectrum licenses.  Accounting costs are typically retrospective.


Most managers do not look at accounting costs in decision-making.  They prefer to worry about the future allocation of the firm’s resources.  Management’s primary goal is to be assured that the bids for the spectrum licenses represent the most efficient use of the company’s economic resources.  Thus, they should take a forward-looking view of how to use the firm’s assets.  Accounting costs should play only a small part in the future decisions of managers.  They can provide examples of past expenditures and firm numbers with which to compare upcoming spectrum auctions, yet they should minimally influence decision-makers.


The second cost, economic cost, can be defined as the cost of using the firm’s resources in production.  This also includes the lost opportunities of resources that could have been used elsewhere in a more efficient manner.  Economic costs are forward-looking and thus should be considered by spectrum providers. 


By looking at these costs, managers can allocate their assets to the benefit of the firm.  This entails figuring out what are the best courses of action for the company by using tools such as cost/benefit calculations.  Examining the economic costs also involves looking at all the possible lost opportunities that bidding on the spectrum license has made the company overlook.  The manager must therefore ask himself whether bidding in a spectrum auction is the firm’s most profitable course of action.  This reduces its economic cost.


B.   Sunk Costs


A sunk cost is an expenditure made which is later unrecoverable.  Though a sunk cost is visible only ex post, it can be either ex post or ex ante.  Because sunk costs are unrecoverable, they should not influence current or future behavior of rational economic firms . . . maybe.  When pricing wireless communications services, the network provider will price to market.  To the extent the market allows, the service provider will price above average costs, which include the sunk cost of the investment in the spectrum license.  As competition enters, it will drive service prices to marginal costs.  The service provider will continue to operate as long as it can continue to price above average costs, despite the fact it may loose money in aggregate.  This assumes that the service provider cannot recover its license fee by either providing service or by reselling its license.


IV.        Spectrum Licenses as Sunk Costs


The FCC’s current position is that the wireless firms treat spectrum license fees as sunk costs and competition in those services pushes to marginal cost, and are therefore not passed on to consumers.  The argument says, in essence, our policy has no negative effects for consumer welfare, although the wireless carriers may not be able to earn a return on their investment in spectrum licenses.


Evan Kwerel, of the FCC’s Office of Plans and Policy (“OPP”), concluded in a recent study that the conventional wisdom that auctions for spectrum licenses raise the consumer price of wireless communication services is a widely held misconception.  The argument proceeds if licensees pay for their spectrum instead of getting it for free, they would have higher costs and that these costs would be passed on to their customers in the form of higher prices.  However, Kwerel’s study counter argues this point on both economic theory and empirical evidence.

A.   Economic Perspective


According to the FCC’s study, the amount paid for a spectrum license in an auction is viewed as a sunk cost since the money cannot be recovered from the government.  The historical cost of winning bids at auctions should have no effect on the price or availability of spectrum-based communication services for customers.  Kwerel indicates that while prices do not depend on sunk costs, they do depend on opportunity or avoidable cost, and the opportunity cost of spectrum is independent of the assignment technique. 


The opportunity cost of a license is the amount that a firm forgoes by using the spectrum.  The market-clearing price of the spectrum will be a function of current supply and demand conditions and not the historical cost at which firms acquired their spectrum.  Kwerel concludes that to the extent that auctions assign licenses to the most efficient producers, facilitate efficient aggregation of spectrum, and move spectrum into the production of services consumers value the most, they will tend to expand the supply and reduce the prices of the wireless services most valued by consumers.


Although charging for spectrum licenses will not increase the price which consumers pay for wireless services, it will affect the profits of licensees.  According to Kwerel, theory predicts that when bidding for a spectrum license in an auction, a firm would never knowingly bid more than the discounted value of the expected profits from acquiring the license.  Nevertheless, bidders may sometimes overestimate the value of a license, leading to lower profits.

B.   Empirical Perspective


Kwerel’s conclusions following studies of empirical data were consistent with the theory that sunk costs do not affect prices, regardless of the market structure.  A preliminary analysis of the US cellular telephone market supported the conclusion that paying for a spectrum license does not increase the prices of wireless services.  For this analysis, Kwerel used data on cellular prices and ownership from 1985 to 1998.  The pricing data showed the lowest prices for 160 minutes of use for each of the two cellular carriers in the largest thirty markets.  The conclusion of the study of empirical data was threefold and supported the hypothesis that paying for a license does not lead to higher prices for consumers:


·       prices charged by cellular operators who purchased their licenses in the after-market are not generally higher than those of firms that obtained their licenses for free;


·       a comparison of prices before the sale date and after the sale date indicates no increase in prices following the sale; and


·       paying for a license (the first sale of a license that was initially obtained free of charge) had no effect on prices, or even may have reduced prices as the new licensee gained full control of the business (this particular study looked at average price changes in markets without a license sale three years before and three years after the year of sale allowing a comparison of the change in price before and after a sale relative to the change in price in markets without any sales in the same time period).


Based on economic theory and empirical evidence, Kwerel’s study contradicted the conventional wisdom that if licensees pay for their licenses, they would have higher costs and that these costs would be passed on to their customers in the form of higher prices.


V.    Is Spectrum License a Sunk Cost?


Sunk costs can impact future rationally economic behavior in one way.  Sunk costs become a consideration in the decision whether to reinvest in the asset.  Such reinvestment is typically to repair or replace the asset.  If the asset represents a sunk cost, then the rational economic firm would not reinvest in the asset.  It is not rational to spend additional money when even the original investment is not recoverable.  This factors into spectrum licenses in two ways.  First, spectrum licenses are typically issued for a finite period of time.  Second, as technology progresses more usable spectrum becomes available.  Such “new spectrum” is likely to be auctioned in so-called 3G auctions.  Since these auctions set new records for revenues, adjusting for inflation and set-asides for small and minority-owned firms, some other explanation must be produced.  Other explanations include irrational exuberance in valuation of such licenses or so-called “survival bidding” by current network providers.


One can also look to the behavior of the carrier in light of incentives to recover the cost of the license.  A truly sunk cost offers no incentive to earn such a return.  One such example is the comparison of the PCS licenses to the grant of an additional 6-megahertz of spectrum broadcasters received for HDTV.  The PCS providers paid for their license, while the broadcasters did not pay for their digital TV license.  The broadcasters had no economic incentives to rapidly deploy digital TV in an effort to earn a return on those costs.  Consequently, even several years later, broadcast HDTV is a disappointment.  There has been no effective roll out of new services and little new programming.  By comparison, most metropolitan markets are served by between three to five competing cellular carriers.  Competition obviously lowers price, but all carriers have invested in licenses.  This competition drives price towards marginal cost, which is near zero.  Since cellular prices are still above this price, it can be inferred that wireless carriers are able to charge a price that is above marginal cost, and may be above average cost (including the license fee).  Moreover, the cost of the license is not sunk to the extent recoverable by law in bankruptcy or by economics in that the carrier can resell his license and exit from the market.


A.   Ex Ante Behavior - Irrational Exuberance


An analysis of spectrum auctions would be incomplete without mention of the irrational exuberance present in both the public debt and equity market, affording freely available capital.  No other industry has had as much access to capital as the telecommunications industry.  The availability of capital is fundamental to the telecommunications industry because large capital expenditures are needed prior to offering services and generating cash.  Wall Street was more than happy, generating billions in fees along the way, to assist the telecommunications companies in their need to make massive capital expenditures.


To put this in perspective, telecommunications companies issued $240 billion in high-yield debt from 1996 to 2000.  That is 150% more than all junk bonds issued from 1983 to 2000 during the heyday of leverage.  In addition to high-yield bonds, telecommunications companies borrowed heavily and issued private and public equity.  The availability of capital fit with the government’s plan to make spectrum available to new entrants into the telecommunications market.  The government wanted to make spectrum available to new telecommunications companies and actually allowed them to pay for spectrum with only 10% down and on a payment basis over a six-year period.  Many new companies bid on spectrum licenses without having the cash up-front, but fully believed that they could raise capital to pay for them, after obtaining a license.  In fact, some of the telecommunications companies could not raise capital and fund their business plans without spectrum.  The end result was that the new entrants required the spectrum licenses to survive.  Therefore, they would bid almost anything to acquire the spectrum licenses they needed to complete their business plans.  Freely available capital and a romance with the sector caused these companies to exist and fueled the price of spectrum licenses.


It is very difficult to value spectrum licenses because it requires predicting future demand.  Therefore, purchasing spectrum licenses is very risky.  The impact of this risk on behavior was mitigated for many of competitive carriers bidding for licenses because they knew that someone else would ultimately assume the risk.  The capital providers ended up assuming the lion’s share of this risk. This alone caused the insurgent companies to bid recklessly, forcing the incumbent telecommunications companies to follow suit.  Business Week alluded to this behavior in an article titled “How the FCC Can Clean Up Its Spectrum Auction Mess”:


The so-called C-Block auction for entrepreneurs, which ended last May with $10.2 billion in total bids, went awry when participants ratcheted up their bids to almost three times those of big companies, such as AT&T Wireless and Sprint Spectrum, in an earlier round.  The financial breaks from the FCC “allowed the bidders to gamble” and bid up prices sky-high, says Crispin Vicars, a senior wireless analyst at Yankee Group.  San Diego-based NextWave Telecom Inc., for example pledged $4.2 billion.  Washington based Pocket Communications bid $1.4 billion.  Now, they and four other companies can’t find investors to back them.  Pocket declared bankruptcy this spring.[1]


The spectrum auctions were intended to open the market for spectrum.  The intention was to allow new entrants into the marketplace, and ultimately the benefit consumer.  However, many companies simply acted irresponsibly due to abundant, readily available capital. 


NextWave Telecommunications is a prime example of an upstart telecommunications company that recklessly purchased over 63 spectrum licenses for $4.7 billion based on the belief that they could obtain financing to pay for the licenses as well as to fund their business plan.  NextWave never obtained the necessary financing and ended up paying a mere $500 million to the US Government and is now bankrupt.  Gambles such as NextWave’s forced other telecommunications carries to grossly overpay for spectrum licenses.


Interestingly, capital was not as available in Asia and their auctions have worked out far better.  Asian telecommunications companies experienced tough economic times in the early 1990s.  As a result of their tough times, the Asian telecommunications companies did not have access to capital.  They were unable to bid high prices for spectrum licenses and, hence, the prices were kept down.  The Asian telecommunications companies, unlike their North American and European counterparts who had unlimited access to capital, are in far better financial shape.  The Asian telecommunications companies have manageable debt loads and now spend more money on actually deploying the spectrum.


Over the last five to seven years, capital markets swung too far in the direction of making capital available to telecommunications companies.  Now, as a result of the turmoil in the telecommunications industry, capital markets may be swinging too far in the opposite direction.  At this time, the capital markets are shut for all but a few of the incumbent telecommunications companies.  Lack of capital will reduce the available capital for future spectrum, cause some companies to merge or to be acquired to survive, and will ultimately limit the ability for some companies to renew existing licenses.  The panic that is setting in is evident in a recent article published on The Industry Standard website (March 2, 2001):  Phone Companies Rally to Avoid Credit:

A half-dozen telecom industry executives, lawyers and bankers met Friday with Roger Ferguson, vice chairman of the Federal Reserve's Board of Governors, to discuss the difficulties that upstart carriers are having in raising capital and to talk about the industry's potential to boost productivity in the economy.  Spunky local phone carriers trying to compete against the Baby Bells were the darlings of Wall Street until market sentiment turned against the sector at the end of last year. Now some of the more competitive of those carriers are struggling to find backing in the debt and equity markets.


We have seen the cycle of telecommunications companies who bid on spectrum licenses parallel the availability of capital.  In the future, the lack of capital will be another significant factor in determining the price of spectrum licenses.


B.   Ex Post Behavior – Oligopolies


In the world today, oligopoly is a prevalent form of market structure.  Examples of oligopolistic industries include automobiles, steel, aluminum, petrochemicals, electrical equipment and computers.  In an oligopolistic market, the offerings often are not differentiated.  In our case, spectrum is spectrum.  What would make an oligopoly is if only a few firms had control or possession of most or all of the available spectrum licenses.  As with any oligopoly, a barrier to entry must exist.  With a spectrum license, the barrier to entry is that the government can auction only limited amounts.  This is a natural barrier to entry, because it is basic to the structure of the telecommunications market in its current state.  Furthermore, because the government regulates the availability of spectrum licenses, an incumbent may not even need to initiate strategic actions to deter entry.


Managing an oligopolistic firm is complicated, because all decisions, especially pricing and investment decisions, involve important strategic considerations.  Because only a few firms are competing, each firm must carefully consider how its actions will affect its rivals, and how its rivals are likely to react.  The strategic considerations can be complex.  Furthermore, decisions, reactions, reactions to reactions, and so forth are dynamic, evolving over time.  When managers evaluate the potential consequences of their decisions, they must assume that their competitors are as rational and intelligent as they are.  Then, they must put themselves in their competitor’s place and consider how they would react. 


VI.        Conclusion


Why is this important?  First, if this is the Government’s policy, it should understand its own policy’s efficacy to the maximum extent possible.  A better understanding of the situation would show whether auctions are good or bad policy.  It would also show whether the cost of winning is a curse therefore passed on to the consumer.  Second, it encourages local experimentation in that the FCC can tinker with the policy without affecting the national policy.  Investigational trials could be designed to further refine the policy.  This would increase service providers’ willingness to invest in the licenses, resulting in lower costs to consumers.

* This paper has been submitted to the Twenty-ninth Annual Telecommunications Policy Research Conference.

[1] Yang, Catherine, “How the FCC Can Clean Up Its Spectrum Auction Mess”, Business Week Online Commentary (October 6, 1997).