# «The Effect of Incumbent Bidding in Set-Aside Auctions: An Analysis of Prices in the Closed and Open Segments of FCC Auction 35 Peter Cramton ...»

In the above regression equation, r is an index for the auction round, and ε is a stochastic shock that is assumed to be of constant variance and zero mean across all auction rounds.

Variable definitions are presented immediately below, and summary statistics for those variables are presented in Table 4.

• LogATTQ = Natural log of AT&T’s smoothed quantity demanded

• Open-ATT = Smoothed, log of quantity demanded in the open, less AT&T’s demand via

• PriceOpen = Price paid for the open licenses

---------------------Place Table 4 here

---------------------Applying OLS to the above regression variables to estimate AT&T’s demand curve for spectrum in Auction 35, the following regression results presented in Equation 15 below are obtained. Standard errors are reported in parentheses below the regression parameters.

The coefficient on price in Equation 15 is interpreted to mean that a $1 increase in price will lead to a 38 percent decrease in the total quantity of spectrum that AT&T demands. The regression parameter on Price is statistically significant at the ninety-nine percent level. Also, the R2 for the regression is 96 percent. The right hand side variable therefore explains most of the variation in the dependent variable—an attractive characteristic because the regression will be used for prediction purposes.

5.3. PREDICTING AT&T’S DEMAND, AND TOTAL DEMAND, IN THE OPEN The parameter estimates from (1) are used to predict AT&T’s quantity demanded in each round of the open segment of the auction. This prediction is outlined by equation 10 above. In particular, the regression parameters from equation 15 are applied to the round-by-round price vector for the open segment of the auction. Performing this prediction yields logATTOpen.25 Using this predicted demand curve for AT&T bidding in the open, had AT&T been precluded from bidding in the closed, total demand in the open is estimated to be [16] Open = (Open − ATT ) + elogATTOpen.

## 5.4. ESTIMATING THE RESULTS IF AT&T ONLY COULD BID IN THE OPEN

The next step in the analysis is to estimate the demand curve in the open, using the calculation of quantity demanded in the open from equation 16 above. To do this, OLS is used to regress the calculation of total demand in the open, Open, on round-by-round prices in the open**segment of the auction. Running this regression yields the following:**

The regression in equation 17 has an R2 of 0.98, meaning that the regression explains 98 percent of the variation in the dependent variable. Further, the standard error of the prediction— that is, the errors between the fitted values on the regression line and the left hand side observations—is small.26 Given equation 17, the clearing price in the open auction is predicted, knowing that approximately 2.32 billion MHzPop was let at auction.27 In log form, the clearing price of $5.39 occurs at a logged quantity of 21.595. This means that prices for the open licenses were suppressed by approximately 53 cents, because AT&T was allowed to bid on the closed licenses through Alaska Native.

To complete this section of the analysis, a price of $5.39 is substituted into AT&T’s demand for spectrum, given by equation 15, to determine the spectrum AT&T would have had it been precluded from bidding in the closed. This quantity is determined to be 403.1 million MHzPop.

5.5. AT&T’S SAVINGS FROM BIDDING IN THE CLOSED By bidding in the closed, there are two effects that allowed AT&T to derive a surplus, beyond what it would have enjoyed, had it only bid on the open licenses. The first effect is that it won a larger quantity of spectrum than it would have otherwise. In particular, AT&T won 649 million MHzPop in Auction 35, whereas it would have won 403 million MHzPop had it only been allowed to bid in the open segment of Auction 35. Because AT&T’s returns model for spectrum is private, the benefits to AT&T of owning the additional quantity cannot be discerned. What is known, however, is that AT&T paid, on average $4.46 per MHzPop for all spectrum (open and closed) that it won at Auction 35. If AT&T had only been allowed to bid in the open, the clearing price in that auction would have been $5.39 per MHzPop. Thus, AT&T would have paid roughly this amount for spectrum, on average, in the open portion of the auction. Thus, AT&T saved, on average, $5.39 - $4.46 = $0.93 per MHzPop on 403 million MHzPop of spectrum, a savings of $374.8 million.28

## 5.6. THE GLOBAL EFFECT ON THE OPEN AND CLOSED AUCTIONS FROM AT&T’S BIDDING FRONT

Because AT&T saved money through its bidding front on the first 403 million MhzPop of spectrum it purchased, one could infer that downward pressure was placed on the combined prices for the open and closed. However, AT&T’s competition in the closed segment allowed it to purchase 649 million MhzPop of spectrum. Consequently, the price of the remaining spectrum that AT&T purchased in the closed, was likely more expensive than it would have been absent AT&T. Further analysis is therefore required to determine the global affect of AT&T’s bidding front on auction prices.Restricting AT&T from bidding in the closed segment of the auction is analogous to subtracting AT&T’s demand for closed spectrum from the total demand for closed spectrum in each round of the auction. According to the estimates, all closed competitors, less AT&T, would have demanded 1.7 billion MhzPop at a price of $2.51 per MhzPop, which occurred by round 32 of the auction. Therefore, AT&T’s participation in the closed increased prices by $3.13 - $2.51 = $0.62 per MhzPop, which fetched an additional $1.042 billion from the closed spectrum.

However, AT&T’s participation in the closed reduced prices in the open. As explained above, prices in the open would have been $5.39 had AT&T not bid in the closed. Prices in the open were $4.96 indicating that $0.43 of revenue was lost on 2.8 billion MhzPop of spectrum.

Therefore, revenue in the open auction was reduced by $981 million. Hence, total auction revenue increased by approximately $61 million ($1.04 billion - $981 million). Consequently, the global revenue effect of AT&T’s bidding in the closed auction was small, given the total financial size of the auction.

6. CONCLUSION If the FCC were attempting to maximize the welfare of consumers of telecommunications services, it would choose an auction design to ensure (1) that the winning bidders placed the greatest value on the spectrum ("efficiency objective") and (2) that there were a sufficient number of winning bidders to yield competitive outcomes ("competition objective"). The diversity mandate imposed by Congress, however, constrained the FCC in its formulation. In particular, the diversity constraint required that the FCC take steps to insure a diversity of spectrum winners, even if the spectrum is most highly valued by the large incumbent carriers.

The FCC decided that, after the bankruptcies in prior set-aside auctions, small firms required additional investment capital to survive in the wireless industry. But that capital did not need to come from incumbent wireless operators. In particular, if additional wireless carriers could survive in the existing market, those carriers would receive backing in a financial market without liquidity constraints. However, the FCC's control standard allowed small firms to act as fronts for large wireless incumbents, thereby defeating the diversity constraint, because the licensees were neither different nor small. Thus, the FCC's standard likely met its efficiency criteria, but it did so at the expense of its competition objective and diversity requirements.

To an economist, the definition of control the FCC used in Auction 35—that is, ownership of a simple majority of voting shares or any other factors listed in its definition of de facto control— is of only limited value. But economic control comes from controlling a “critical” share of the affiliate’s net cash flows. Because AT&T owns 80 percent of Alaska Native’s net cash flows, AT&T likely has sufficient ownership of Alaska Native to possess economic control. The relevant question from a competition perspective is whether an incumbent (such as AT&T) owns a sufficient amount of an entrant’s (such as Alaska Native’s) equity such that it would be induced to raise prices in a unilateral fashion (relative to an outcome in which some third party obtained the new license).

The above test makes clear that neither voting control nor any variables that measure direct management control are sufficient to capture the concept of “control” as that term is normally understood by economists or the Commission. For example, neither “50 percent of the board of directors or management committee” nor the “authority to appoint, promote, demote, and fire senior executives” will affect the incumbent’s primary economic objective—to maximize the sum of its own profits plus its pro-rata share of the profits of its front—when the incumbent sets its price (Auction 35 Public Notice). The Commission’s test for de facto control instead contemplated full consideration of a broad range of control indicators, and under that test, a finding of control would clearly be compelled in this case.

In its second report and order in the matter of the implementation of the Commercial Spectrum Enhancement Act and modernization of the Commission’s Competitive Bidding Rules and Procedures, the FCC revamped its definition of control. The FCC first eliminated the ability of any firm with “impermissable material relationships” from receiving designated entity benefits.

The FCC defined such a relationship as one in which a licensee “has agreements with one or more other entities for the lease (under either spectrum manager or de facto transfer leasing arrangements) or resale (including under a wholesale arrangement) of, on a cumulative basis, more than 50 percent of its spectrum capacity of any individual license.”31 Furthermore, the FCC defined an “attributable material relationship” as one where the designated entity in question leased or resold, on a cumulative basis, more than 25 percent of the capacity for any one license.

Therefore, the FCC’s revised standards base control on ownership share of the essential input of production in question (spectrum), and should therefore be superior to those in force during Auction 35.

Although the FCC’s current implementation of its designated entity program is likely an improvement upon its prior standard of control, the FCC should continue to monitor the relationship between designated entities and incumbent wireless carriers or scrap the program entirely. The prior policy, which resembled an affirmative action program on the surface but actually maintained the status quo, generated at least two types of harm. First, legitimate entrepreneurs who had little chance of winning licenses wasted scarce resources in preparing business plans and raising capital. Second, incumbent carriers wasted resources in creating front organizations to obtain set-aside spectrum. When AT&T exercises its option to acquire the next 40 percent of Alaska Native, the minority shareholders that went along with the scam will be unjustly enriched. The transactions costs of creating fronts are a pure waste of resources.

Consequently, the FCC’s improved control standard should produce a more efficient allocation of resources in the marketplace and should better fulfill the mandate set by Congress by helping true entrants instead of further entrenching current incumbents.

1. In June 2001, the U.S. Court of Appeals for the District of Columbia concluded that “the Commission violated the provision of the bankruptcy code that prohibits governmental entities from revoking debtors’ licenses solely for failure to pay debts dischargeable in bankruptcy,” thereby invalidating the results of Auction 35 (NextWave Personal Communications v. Federal Communications Commission, 254 F.3d 130 (D.C. Cir., June 22, 2001)). One day after the Court of Appeal’s decision in favor of NextWave, the FCC issued a press release stating its intention to pursue NextWave at the Supreme Court (FCC News Release, August 31, 2002). By the summer of 2002, after months of settlement efforts failed, the FCC refused to cancel the reauction results, even though it had no spectrum to give the winners (Wall Street Journal, September 24, 2001: 1).

In September 2002, the FCC reversed course, and determined that the public interest would be best served by allowing the winning bidders to opt out of the auction (Wall Street Journal, September 12, 2002: 1).

2. AT&T WIRELESS SERVICES INC., S.E.C. FORM 10-Q at 4 (filed Aug. 14, 2001).

3. Salmon PCS, LLC, a bidding front set up by Cingular Wireless, also bid in Auction 35.

However, Salmon focused its bidding on the open segment of the auction, whereas Alaska Native bid mostly in the closed auction. In particular, Alaska Native spent $2.7 billion for 29 of the closed licenses, and spent only $201 million for 15 open licenses. Salmon won 35 closed licenses and paid $674 million for those licenses, while winning 44 open licenses worth $1.7 billion.

Because Alaska Native spent 4 times as much for closed spectrum than did Salmon, the paper focuses the analysis on the impact of Alaska Native’s bidding.

4. ALASKA NATIVE WIRELESS L.L.C., F.C.C. FORM 175 APPLICATION, Applicant Identity and Ownership Information at 4 (filed Nov. 6, 2000) (AT&T Wireless PCS Interests “holds 39.9 percent of all member interests in ANW. Under Section 1.2110(c)(2)(ii)(A) of the Commission’s Rules, therefore, AWPI would be considered to hold not more than 80 percent of all member interests on a fully-diluted basis…”).

5. Auction results are available for download at the FCC’s web site http://www.fcc.gov/wtb/auctions/35/charts/35markets.xls.

6. It is quite possible that other incumbent carriers used fronts to gain access to the closed auction. Because Alaska Native so thoroughly dominated the closed auction, however, the paper focuses on Alaska Native’s effect on the results.