Recently, I talked about the tension between spectrum efficiency and competition policy in auctions. Briefly, for reasons I will elaborate below, the largest wireless providers (AT&T and Verizon) can extract more value out of a wireless license than their significantly smaller rivals (especially when we include the foreclosure value of keeping the license out of the hands of competitors). As a result, we should expect over time that the biggest wireless companies will eventually have an unbeatable edge in wireless capacity unless the FCC takes some measures to balance out the spectrum holdings.
Not surprisingly, the same problem surfaces when companies buy spectrum licenses from each other. After all, a license transfer is essentially a private auction (only with less transparency and higher transaction cost – factors that work in favor of the largest companies). We should therefore expect to see the same tension around spectrum efficiency and concern for competition policy play out in license transfers.
Two recent transactions put these concerns in stark relief: AT&T’s recent acquisition of spectrum from Qualcomm and Verizon’s proposed acquisition of licenses from Spectrumco. Both cases arguably represent an improvement in spectrum efficiency by moving the licenses from those who were ultimately unable (or unwilling) to use them efficiently to those able to pay the most for them (and therefore, presumably, extract the greatest value). At the same time, however, the transfers aggravate the existing spectrum imbalance between the largest wireless providers and competitors.
I explore the trade offs posed by these transactions, and discuss how they are essentially another version of the same spectrum efficiency v. competition policy I discussed in auctions, below . . .
(Warning, I’ve been reading a lot of Jared Diamond stuff lately, so this blog post ends up sounding a lot like a spectrum version of Guns, Germs and Steel. Honestly, I’ll try to be funnier next time.)
The Coase Doctrine: Better Spectrum Efficiency Through Markets
First, allow me to restate my thesis. Ender existing policy of relying on exclusive geographic-based licenses for spectrum capacity, spectrum access – the one utterly essential input for mobile services – remains a scarce, finite commodity. In 1993, we embraced Coase’s proposal to use market mechanisms to distribute spectrum. Under this theory, the entity that values the spectrum most, and has the capacity to pay for it, will get the license. We justify this as a matter of social utility by observing that for profit maximizing firms (the vast majority of license holders under this system) “values the most” means “best able to extract value,” presumably in the form of providing goods and services.
Despite the oft repeated mantra by policymakers short on actual expertise but long on ideological faith that auctions “place spectrum in the hands of those who use put it to their highest best use,” they don’t — and Coase never said they did. Instead, Coase postulated that standard market similar to that for other scarce goods would likely produce more efficient spectrum use. While the initial distribution by auction will (according to Coase) almost certainly prove inefficient (but no better or worse than any other distribution method, so we might as well use it and collect a chunk of change for the government), subsequent secondary market transaction (‘sales’ for you non-economists,) will correct these inefficiencies over time.
As a consequence, the ability of firms to buy and sell licenses plays as important a role in Coasian theory of spectrum efficiency (and therefore as important a role in our Coasian spectrum policy) as auctions. Accordingly, spectrum policy at the FCC in the last ten years has also focused on increasing the “liquidity” of spectrum, making it easier for parties to transfer spectrum usage rights and easier for parties to use spectrum flexibly. This makes sense even if we believed that auctions automatically and always put licenses in the best possible hands for the best possible uses. After all, technology changes, consumer demand shifts, companies go bankrupt or abandon lines of business, or any of a number of things may happen to render even a previously efficient allocation inefficient. So secondary market transactions keep spectrum use efficient for the same reason auctions do — the company that can and will pay the most for it probably can get the most value from it. And if the buyer guesses wrong, they will sell it to someone who can.
Why Would Spectrum Efficiency Create Competition Problems?
All well and good, but this has the tendency to concentrate the bulk of the wireless licenses in the hands of one or two companies. Why? The bigger a wireless company gets, the greater its economies of scale, making it better able to extract value from the spectrum. Because that enables the company to offer better service, it grows at the expense of its spectrum-starved rivals,. The company with the spectrum advantage can therefore acquire additional licenses more easily as time goes by, while the spectrum deprived firms must pay higher and higher prices for spectrum they can use less efficiently simply to try to maintain their existing customer base.
Worse, spectrum licenses in the right range for cellular service are not merely scarce, they are static. Barring a rare, well publicized event like a government auction, there are a set number of wireless licenses for a particular band or use. Nor does an auction improve matters, as everyone knows how many licenses the government will auction. At any time, all potential providers know exactly how much wireless capacity exists in the world and who holds the rights to that capacity. It is therefore in theory possible for a large enough player or players to literally “corner the market” on spectrum.
Even if swallowing all the existing licenses seems to much of a gulp, the number of suitable licenses in an auction, at least, is small enough — and the two largest players are big enough — to swallow the lion’s share of new wireless capacity and thus keep potential rivals lean and hungry. This ability to foreclose rivals from needed inputs (called, unsurprisingly, “foreclosure value”) adds to the value of the spectrum to the largest players. In other words, a company might spend $2 billion on a license which can only produce $1 billion in value, if the company believes it will make a $1 billion or more by keeping its rivals weak.
What Works For Auctions Works For Secondary Markets
Companies can acquire spectrum licenses in two ways: from the government at auction, or from each other. If a T-Mobile or Sprint or Metro need more spectrum capacity, why don’t they just buy it from someone else?
The same factors that favor the largest players in government auctions also favor the same players in secondary market purchases – in fact even more so. At heart, a purchase of spectrum licenses by one company from another (possibly with built networks and subscribers as well) is a form of private auction. If Sprint or US Cellular goes to a licensee and says: “Nice spectrum you got there, interested in selling it?” The licensee would be a fool not to run to AT&T or Verizon and say: “Someone else is offering me this much for my spectrum, how much will you offer me for it?” Even with the most careful precautions and non-disclosure agreements, a company actively interested in selling its spectrum (or itself) can find ways to signal to other potential bidders “I’m available, make me an offer.” At this point, all the other factors favoring the larger player kick in: it can extract more value, it enjoys foreclosure value, and it can pay more money due to size and better access to financing.
Other factors work to the advantage of big players in the secondary market beyond even the auction factors. A government spectrum auction is a known event, scheduled well in advance with well understood rules designed to promote transparency. By contrast, finding companies willing to sell spectrum (or themselves) is more difficult, unless the company puts itself up for sale. Negotiating sales takes a good deal of time, management resources and money, something a smaller firm has in less supply than a larger firm. This lack of transparency (meaning the difficulty in finding out what is for sale and how much it is really worth) and high transaction cost (meaning the high cost of finding a willing seller and negotiating a deal) all work in favor of the biggest firms.
Finally, larger firms may have other advantages beyond spectrum. For example, AT&T and Verizon control backhaul facilities that their rivals lack, and therefore must lease. This allows AT&T and Verizon to save money while allowing them to increase cost to competitors, further cutting into profit they might use to buy spectrum. In addition, as we will see from the examples below, the ability of larger firms to offer better terms may extend to related businesses. AT&T could offer DT seats on its board and an interest in its entire corporate wireline and wireless concerns in exchange for T-Mobile. Verizon can offer the cable operators in Spectrumco resale agreements and the use of its vast network of wireless stores as outlets for cable services.
As a result of all these advantages, nearly every significant spectrum assignment in the last five years has involved AT&T or Verizon buying. Even when the Department of Justice Antitrust Division (DOJ) or Federal Communications Commission (FCC) forced Verizon or AT&T to divest licenses in certain local markets as part of a larger purchase, they usually ended up selling to each other. The ability of the largest rival to pay more cash, and possibly the hope of reciprocal deals in the future, apparently outweigh any concern about making the closest rival stronger.
A closer look at some recent real world examples helps illustrate the problem of spectrum efficiency v. competition theory when evaluating mergers. This doesn’t tell us which we should favor (I will save that for a later installment, but you can probably guess my sympathies). Rather, it demonstrates that — contrary to the claims of AT&T and Verizon and their supporters — we really do face a choice between spectrum efficiency and robust wireless competition if we continue to allow Verizon and AT&T to acquire more spectrum licenses to meet the “spectrum crunch” faced by all providers.
Real World Example One: AT&T/T-Mobile
Last year, T-Mobile’s German parent, DeutscheTelekom (DT) decided to exit the American wireless market by selling T-Mo to AT&T for a whopping $39 billion. This placed a premium of about $11 billion over what analysts estimated DT could get by spinning T-Mo off as a separate company and selling the stock. AT&T had good reason to bid so high if it wanted to win DT’s private auction of t-Mo. As DT admitted later (despite protestations to the contrary to regulators), it had five different offers for T-Mo in addition to AT&T, but AT&T offered the most attractive terms. In addition to the $39 billion, DT would receive a substantial chunk of AT&T stock (which includes ownership of AT&T generally, not just its wholly-owned subsidiary AT&T Wireless) and two seats on AT&T’s board of directors. To finance the deal, AT&T secured a $20 billion letter of credit.
Why would AT&T pay such a fantastic premium for T-Mobile, including taking on $20 billion of new debt? AT&T claimed it needed the spectrum to address capacity shortage. But as critics pointed out at the time, this made no sense on just the pure value of spectrum as spectrum. T-Mobile’s network was already operating at peak capacity (indeed, it was more constrained than AT&T’s network, since it has significantly less spectrum). Also, with the T-Mo purchase no more, AT&T has discovered (as merger critics predicted) it has other, less expensive ways to use its own spectrum more efficiently.
We can only explain AT&T’s willingness and ability to outbid potential rivals by acknowledging that AT&T would gain significantly more value from T-Mobile and its spectrum than would rival firms for reasons having nothing to do with the intrinsic value of T-Mobile’s spectrum or the use to which T-Mobile put the spectrum. AT&T was not going to repurpose T-Mo’s spectrum for some higher, best use. T-Mo was using the spectrum for exactly the same purpose AT&T intended to use it. Rather, as AT&T argued, it would enjoy “synergies.” Combining with T-Mo would make it bigger, better able to extract the value (by laying off “redundant” T-Mo staff and other cost cutting measures). Critics also noted that AT&T would eliminate a significant (and disruptive) rival, thus allowing AT&T to charge higher prices as a result of weaker competition. And, of course, only a company as big as AT&T could afford to pay such a premium, or secure the financing for it.
Economist can (and did) argue that that these factors constitute a more efficient use of spectrum. After all, if we can get the same value out of the spectrum with 20,000 fewer employees, the number AT&T was apparently prepared to lay off from T-Mobile, that represents a far more efficient use overall. But this more efficient use comes at a cost (beyond the 20,000 laid off workers) in terms of competition, with no clear way to replace the competitor.
Indeed, DT’s interest in selling T-Mo in the first place underscores the competitive problem and how spectrum advantage and size undermine smaller rivals. T-Mo used its existing spectrum exceedingly efficiently, offering services comparable to AT&T and turning a fairly substantial profit as a result. DT feared its inability to acquire enough spectrum for future needs (or, as they insisted at the time, they had “no path to LTE”). As we have discussed above, this fear seems well grounded in reality in light of existing spectrum policy and its impact on competing firms.
Real World Example #2: Naked License Transfers, AT&T/Qualcomm and Verizon/Spectrumco.
The AT&T/T-Mo deal involved existing competitors. We can easily understand how the addition of T-Mobile’s customers to AT&T’s existing marketshare and the elimination of one of only four national providers threatened competition. But what about a naked assignment of licenses? Do the same synergies exist that would favor the two largest firms, AT&T and Verizon, over their weaker rivals?
Two recent deals, AT&T’s acquisition of Qualcomm’s Lower 700 MHz spectrum and Verizon’s purchase of AWS licenses from Spectrumco and Cox, provide us with an excellent test. Despite strong demand for spectrum of this quality, we once again see the prime spectrum gravitating to the two largest wireless providers: AT&T and Verizon. A closer examination again shows why absent any policy response, spectrum concentration will continue and will continue to undermine the ability of other firms to compete with AT&T and Verizon.
In looking at these deals here, I want to set aside the questions that usually arise about whether AT&T or Verizon genuinely “need” the spectrum, whether they are “spectrum warehousing” or whether the party selling the spectrum is a “speculator” which obtained spectrum solely for the purpose of selling it and thus realizing an unjust enrichment. These questions have their place in FCC reviews of license transfers – particularly the question of need. But I want to examine the best case for spectrum efficiency and see if, left unchecked, the natural tendency will drive to consolidation. So let us take the most favorable case to all players in the transaction and see where that leads us.
Back in 2002, despite the utter lack of any certainty as to when broadcasters would return their analog spectrum and free up most of the 700 MHz band, the FCC auctioned certain licenses in what we now call the Lower 700 MHz band (you can find the details on the FCC Auction 44 and Auction 49 webpages). The 2002 Lower 700 MHz band auction contained just about every element ripe for disaster: an aborted FCC plan to implement incentive auctions under questionable statutory authority, Congressional micromanagement, uncertainty as to when the licenses would become available, which in turn created uncertainty over how to optimize the band or what technology would work with this particular band plan. Unsurprisingly to anyone not convinced that spectrum auctions automatically mean billion dollar payouts, the auction fetched a pittance and attracted little interest from commercial mobile service providers such as AT&T.
Qualcomm, however, had a technology for a mobile television service called Mediaflo. Qualcomm therefore went out and acquired a national footprint in the Lower 700 MHz band and began diligently testing and trying to deploy Mediaflo. Unfortunately for Qualcomm, the service went nowhere. It suffered the usual problems of new technologies trying to go from the lab to reality, accompanied by difficulties to acquiring rights to content and other issues.
This left Qualcomm with potentially valuable spectrum for which it had no use. In the old “command and control” days, we would expect Qualcomm to turn in its licenses and have the FCC reauction or repurpose the spectrum in some other way. But in the new Coasian universe, we don’t need to wait for some silly regulator to act. Who says the FCC could do a better job of discovering who could buy the spectrum and repurpose it? Let Qualcomm go and find a buyer instead.
As it turned out, however, finding a buyer raised a bunch of problems. Qualcomm wanted a buyer for the entire national footprint. It would do Qualcomm no good to sell off the juicy bits and remain stuck with a bunch of spectrum it couldn’t use and no one wanted to buy. That automatically eliminated anyone too small to utilize (or afford) a national footprint. Also, for technical reasons, Qualcomm’s little sliver of spectrum needed the right partner. It was too small to boost someone to national coverage on its own, so it needed a buyer with a potentially compatible network. In addition, repurposing Qualcomm’s spectrum would require handset manufacturers to develop a whole new chipset for a band no one had anticipated including in mobile voice/data service – a challenge made more expensive and complicated for various interference reasons unique to the nature of the band.
This combination of factors worked to eliminate just about anyone but AT&T from buying the Qualcomm spectrum. AT&T had significant 700 MHz spectrum and could afford the expense of buying Qualcomm’s national footprint. Other national carriers, such as Sprint or T-Mobile, lacked a compatible network. Other 700 MHz licensees, such as C. Spire, lacked a national network and the resources to bid against AT&T. In addition, AT&T had a sufficiently large customer base that handset manufacturers would be willing to undergo the expense of designing and building new handsets to use the Qualcomm spectrum.
This is not to say that the other carriers such as Sprint or C. Spire didn’t need access to the Qualcomm spectrum – of course they did. The Qualcomm spectrum also undeniably added capacity to AT&T, aggravating an already bad competitive situation. Absent any other regulatory fix, such as roaming and/or interoperability, this deal created a stark choice between putting spectrum to productive use at the expense of competition, or keeping valuable spectrum unproductive to keep the wireless market competitive. Added to this, allowing AT&T to repurpose the spectrum also increased the risk of harmful interference to competitors using the Lower 700 MHz band.
Still pending, we have the Verizon/Spectrumco/Cox transaction. God willing, I shall find the time to write more about this transaction (which implicates so many aspects of telecom policy) at some future date. If you want some background, you can see what my colleagues at Public Knowledge have written here, here, here and this podcast here. For right now, suffice it to say that Spectrumco, a consortium of Comcast, Time Warner Cable, and Bright House, hold a bunch of licenses from the “Advanced Wireless Service” (AWS) licenses. At one point Cox was also part of Spectrumco, but they exited back in ‘08 to try to start their own wireless company. (If you really want the backstory you can see some of my blog posts from back in the day. AWS Auction, Spectrumco/Sprint/Clearwire/Cox, et al.)
Comcast, et al. never did figure out what to do with the AWS spectrum. Meanwhile, Cox tried to start its own wireless business, only to discover that breaking into the already mature wireless business against the entrenched incumbents is not so easy. In fact, it is damn expensive and damn hard. So after a few years of struggling, Cox scaled back its wireless efforts, until finally dropping them altogether.
A bunch of us hoped that T-Mobile would team with Spectrumco (and possibly Cox) to put their unused AWS spectrum to good use. Instead, Verizon Wireless swooped in and gobbled it up. After negotiating a deal with Spectrumco for their spectrum, Verizon captured Cox’s spectrum a few weeks later.
Why did Verizon end up with the Spectrumco licenses and not Sprint/Clearwire (with whom Spectrumco had a longstanding relationship) or T-Mo? For that matter, why couldn’t an upstart like Metro find the money to tempt Spectrumco or Cox? The price Verizon paid, aprox. $2.3 billion for Spectrumco’s licenses and $315 million for Cox’s licenses, is steep but not prohibitively expensive. Unlike the Qualcomm spectrum, a network operator could have integrated the empty AWS spectrum into any system as easily as it could newly cleared and auctioned spectrum. Yes, T-Mo was arguably paralyzed because AT&T and DT took a good, long time to recognize the inevitable and finally give up. But everyone could see it was coming. If Comcast, et al. wanted to get a bid from T-Mo, it could have done so. What made Verizon the irresistible partner of choice?
Turns out that, once again, size matters. Here, Verizon could offer Spectrumco and Cox a set of side deals that – and I am putting the happiest, most neutral face on this – would confer additional advantages to the cash from selling the spectrum. Under the “side agreements,” Verizon Wireless will also market Comcast, TWC, BH and Cox’s video services. So the cable guys not only get cash, they gain a new sales engine for their products. Verizon has about 40% of the wireless market, while its nearest non-AT&T competitor has about half that. If cable operators put a premium on this additional feature of the deal, no other provider (with the possible exception of AT&T) can possibly match it. Indeed, statements by Comcast Senior VP David Cohen make it clear that without the side deals, Spectrumco would never have sold its licenses to Verizon — at least not at this price.
Which brings us back to the efficiency problem. Let us take Spectrumco and Cox at their word that they will not use the spectrum anytime soon, if ever. No one else can offer Spectrumco and Cox what they want in terms of cash and added value from the side deals (what we might think of as a ‘competitors with benefits’ arrangement). Absent action by the FCC or DOJ, a large chunk of incredibly valuable spectrum will end up in the hands of the largest wireless provider, reenforced by new exclusive marketing agreements with the largest cable companies.
Unlike AT&T/T-Mo, it clearly improves spectrum efficiency to move the spectrum from the unproductive Spectrumco to the productive Verizon. Again, if all we cared about were spectrum efficiency, then we would say this proposed deal (absent the side agreements, which I will discuss some other time) serves the public interest because productive spectrum is better than unproductive spectrum. But we care about more than spectrum efficiency. We also care about competition. From a competition standpoint, the transfer of a huge swath of undeveloped spectrum to the largest provider is a potential competitive catastrophe.
Consider this comparison. Once upon a time, we looked at land management as an issue of economic development. We looked at swamps as inefficient uses of land and potential sources of disease. So we drained swamps and converted them into farmland and real estate developments. After a while, we noticed that “swamps” played a very important role in our ecology, and that destroying wetlands for agricultural or industrial uses caused real problems. So we adopted land use policies that try to balance the benefits of preserving wetlands with our desire for economic development.
The analogy to sustainable development works at another level as well. Incrementally, it is damn hard to know when you hit the tipping point between sustainable economic development and ecological disaster. Similarly, it can be extremely hard to tell when the spectrum advantage for the largest (or two largest) wireless companies becomes so great it becomes impossible for competitors to catch up. The only thing we can be sure of is, all things being equal, if we aren’t there yet, and nothing else changes, we will reach the point of competitive collapse eventually.
Or will we? And if we do, would it really be so bad? In Part III, I will address why I think that other answers provided – such as the availability of wifi or more efficient spectrum use – don’t solve the competition problem. I will also look at whether, given the advantages of spectrum efficiency, we ought to care about the possible collapse of wireless competition from spectrum starvation of competitors.
Stay tuned . . . .
Two questions for you:
1) Our incumbent, too big to fail wireless carriers claim that they must acquire most available spectrum to achieve efficiency through scale (size) and to accrue positive networking externalities for subscribers. The FCC (but curiously not Ofcom in the UK) buys this argument, albeit without any empirical research. Is this claim so credible as to justify the elimination of spectrum caps? Had the FCC retained spectrum caps, would consumers suffer by having a fragmented marketplace, less efficiency and lower positive networking externalities?
2) Airport authorities for the most part have resisted sponsored researchers’ assertion that auctioning off takeoff and landing slots would solve all congestion/efficiency problems. Empirical research shows that airports serving as a fortress hub, dominated by a single carrier, have significantly higher airfares than airports with greater competition. Additionally research shows that slot auctions result in warehousing, or substitution of smaller aircraft, which might exacerbate congestion, for “use or lose” auctions. Does the aviation example provide a true rebuttal to the prevailing wisdom that auctions always promote efficiency and enhance consumer welfare?
Thanks. I will address the first question in my next installment. As for the aviation example, it strikes me as consistent, but I’m hesitant to say anything definitive since I am not familiar enough with the industry or the research.