IN THE UNITED STATES DISTRICT COURT
FOR THE DISTRICT OF COLUMBIA
United States of America,
Plaintiff,
v.
SBC Communications, Inc. and
AT&T Corp.,
Defendant.
United States of America,
Plaintiff,
v.
Verizon Communications Inc. and
MCI, Inc.,
Defendant.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Civil Action No.: 1:05CV02102 (EGS)
Filed: September 21, 2006
Judge Emmet G. Sullivan
FILED UNDER SEAL PURSUANT
TO PROTECTIVE ORDER
ENTERED AUGUST 4, 2006
REDACTED FOR
PUBLIC INSPECTION
Civil Action No.: 1:05CV02103 (EGS)
|
REPLY DECLARATION OF W. ROBERT MAJURE
- I have been asked to prepare this reply declaration in order to further explain the
basis for the Complaints and relief sought by the Antitrust Division of the United
States Department of Justice (the "Department") and to address the arguments
raised by the various amici. In this declaration, I will address issues raised by
the amici that are focused on the harm alleged by the government and the
remedies contained in the proposed Final Judgments. However, many of the
objections raised by the amici pertain to issues other than whether the proposed
remedies will adequately address the harm alleged in the Complaints. I have
been asked to briefly touch upon the government's analysis of these other issues
for the benefit of the Court and the public.
- It is important to remember that the Department has a specific role in reviewing
mergers. It determines whether in its judgment the merger violates the antitrust
laws, and because it does not have the power to enjoin mergers, whether there is
enough evidence to convince a court that the merger is likely to substantially
lessen competition in a particular market. The Department will be unable to
prevail in court if it has insufficient evidence to prove each element of an
antitrust violation.(1)
- The Department Employed Standard Antitrust Analysis In Evaluating these
Mergers
- Some amici argue that my previous declaration was deficient because
it did not describe the competitive harm using the tools of geographic
market definition or market shares and concentration figures described
in the Department's Horizontal Merger Guidelines,(2)
which outline the analytical framework used in reviewing the competitive
effects of a merger. Others argue that the geographic markets alleged
in the Complaints are too narrow, too broad, or ignore commercial
realities of the telecommunications business. I explain below why
the product and geographic markets alleged in the Complaints are appropriate
based on the characteristics of local private line services, and why
concentration figures are not especially helpful in analyzing these
markets.
- Market Definition
- Market definition is an analytical tool used to assist in determining
the ultimate question of whether a merger will result in harm for
some consumers.(3) It is
useful for describing the scope of potential harm and for identifying
the set of possible competitors. My initial declaration addressed
directly the alleged competitive harm and number of providers that
own a direct connection to the buildings at issue, but it is also
possible to explain the analysis with the tool of market definition.
- The market definition methodology set forth in the Merger Guidelines
starts by looking at the smallest set of goods sold by the merging
parties, and then adds goods that customers consider similar (and
that would serve as a substitute in the face of a price increase)
until a set of goods is found where an owner of all those goods (a
"hypothetical monopolist" over those goods) would find it profitable
to raise price a small but significant amount. The hypothetical monopolist
would not be able to raise price successfully if a number of customers
would switch to other products rather than pay the higher price. Thus,
the methodology is essentially that of identifying the smallest set
of goods over which it is possible for market power to be exercised,
and calling that set the market.(4)
- In defining a product market in these cases, the Department began
its analysis with the local private line products of the merging parties.(5)
This is an appropriate place to start because a local private line(6)
is a fundamental product that is routinely bought and sold. It has
a price and one can consider what would happen if a hypothetical monopolist
raised that price. To confirm that local private
lines is an appropriate product market, I also considered whether
there are other options customers could use instead of purchasing
local private lines. As local private lines are fundamental to any
telecommunications product serving a business customer with any significant
volume of traffic, the only way to avoid directly or indirectly purchasing
local private lines as part of a telecommunications package is to
give up the functionality of a dedicated connection. However, for
the traffic volumes that lead a customer to consider local private
lines in the first place, an undedicated (or switched) circuit would
be insufficient to allow the company to obtain the type and quality
of communication services it requires. I do not believe that undedicated
services are a practical constraint. As far as I can see, there simply
is no other telecommunications service or product that would be an
effective alternative for the customer.
- Probably the closest alternative for a buyer is the same local private
line service provided to a different location. If that other location
is not also monopolized, the customer could get service there from
another carrier and escape the hypothetical monopolist's price increase.
But unless the customer is willing to relocate in order to lower its
telephone bill, this is an impractical alternative. The cost and disruption
of relocating will certainly outweigh the benefits of avoiding a small
price increase.
- The Merger Guidelines methodology therefore supports the
conclusion that local private lines is a relevant product market and
that a relevant geographic market would be the individual buildings
(a definition that fits well with the harm identified in the Complaints).(7)
- Amici NYAG and COMPTEL argue that a single building cannot be a
market because retail customers buy networks, i.e., the ability to
connect to multiple locations.(8) It
is true that many of the customers for local private lines are actually
wholesalers carriers who combine local private lines with other
facilities and services, to sell end-user customers services that
create a local, regional or national network. This fact, however,
does not prevent a connection to a single building from being deemed
a relevant market. Local private lines are priced and sold separately,
and purchasers buy them because they allow dedicated connections to
particular locations. The fact that they are inputs into other services
that are sold in different markets with different characteristics
does not alter the analysis.(9)
- Local private line services to individual buildings is an acceptable
market definition given the nature of the competition between the
merging parties. But, as amici point out, there are some factors that
suggest one could do the analysis by looking at a broader market,
such as a metropolitan area.(10) As
I discuss above, market definition is a tool for assisting with the
analysis of the likely competitive effect. A somewhat different market
definition may serve as an equally useful tool in this regard. To
perform the analysis using the broader geographic market definition
one would need to take into account that different carriers serve
different buildings. Indeed, AT&T's competitive significance in
a particular building depends on whether AT&T is capable of providing
a local private line to that building. It would be inappropriate to
let the choice of market definition cloud the fact that in this case,
ultimately, the predicted harm stems from a loss of competition to
serve the buildings identified in the Appendices to the proposed Final
Judgments.
- Market Shares and Concentration Statistics
- Although concentration measures (e.g., market shares or more complicated
statistics such as HHIs) are often useful tools in antitrust analysis,(11)
sometimes they do not add anything useful to the evaluation of competitive
effects. For example, calculating market shares based on sales can
be misleading in analyzing a local private line market. To illustrate
this, assume that a building has only one tenant who solicited bids
for all of its telecommunications business last year. Assume also
that the tenant obtained competitive bids from five carriers, and
that all of the carriers have lateral connections into the building.
If market shares were calculated based on sales, and if the entire
contract were awarded to one carrier, the market share of the winning
bidder would be 100% and each of the other four firms would have a
zero share. This would incorrectly suggest that one firm has a monopoly
when in fact there was substantial competition for the tenant's business.
- Amicus ACTel suggests that it is appropriate to calculate market
shares and HHIs by giving each carrier with a lateral connection an
equal market share,(12) but that is
effectively the same as just reporting the number of carriers that
have facilities capable of providing local private line service in
a building. For example, in the merger cases at issue here, the Department
has alleged that harm would occur in buildings with two carriers but
only one post-merger. It would not have added anything useful to the
Department's explanation of the harm here to say that HHI figures
would increase from 5,000 to 10,000 in those situations. The Complaints
make clear that customers in those buildings would no longer have
a choice.(13)
- As discussed above,(14) there are
essentially only two ways to generate a building-specific market share
for local private lines (either by using the revenues on sales of
local private lines involving the building or to count capacity to
serve the building). Amicus COMPTEL has cited statistics calculated
by FCC staff for "local voice services" for each state in SBC's territory
based on data from a survey conducted by SBC of medium and large enterprise
retail customers.(15) These figures
are for many reasons inappropriate for evaluating competition in local
private lines. First, there is no way to assess the validity of SBC's
survey results. Second, the data refer to retail sales and may include
purchases of switched as well as dedicated services. Finally, because
the choice of local private line provider is limited to those firms
capable of providing that service at a particular building, any measure
that aggregates buildings where a carrier has facilities with ones
where it does not, would not account for the fact that a carrier may
be competitively significant in the former and not in the latter.
For example, AT&T may be relatively successful in selling service
in buildings where it has facilities, but if one merely averages its
sales across the entire state AT&T appears to be more important
than it really is in some places and less important in the places
where it really is a significant competitor.(16)
- Thus, although it is possible to cite measures reporting significant
concentration in local telecommunications services, it is not clear
that any of these are relevant to the analysis of the markets at issue.
Indeed, all of the ways in which one might calculate such a measure
seem more prone to distortion and misinterpretation than is a simple
description of the facts such as that presented in my initial declaration.(17)
Precisely because mechanical application of concentration measures can
produce flawed conclusions, the Merger Guidelines use such
statistics only to establish initial presumptions, and make clear that
further refinement of the analysis is necessary. In practice, the Department
and the FTC have often found that specific transactions are likely to
have consequences different than these initial presumptions would indicate.(18)
- Entry
- The Merger Guidelines recognize that concentration measures may fail to account
for new firms that are either about to enter or would likely enter. The Merger
Guidelines indicate that if such entry will be timely, likely and sufficient, it should
be included in the analysis.(19)
- In evaluating whether to challenge these mergers in certain buildings
where the parties have the only facilities capable of providing local
private line service, the Department considered the potential for
other CLECs to construct new laterals into these buildings. Amicus
COMPTEL has misconstrued this analysis as a prediction of speculative
entry by these CLECs (i.e., building connections in the hope of using
them at some later date to serve customers).(20)
Although purely speculative entry may be unlikely, there
are clearly instances where a CLEC with fiber close to a building
has bid on business in that building with the intention to build a
connection to serve the customer.(21) Therefore,
the possibility of entrants bidding for new opportunities as they
arise needs to be addressed to correctly depict the likely set of
competitors.(22) The Department attempted
to do that consistently across a wide array of buildings(23) by
using the information provided by CLECs to construct entry screens
that took into account the cost of constructing facilities from each
CLEC's network to the building and the potential revenue opportunities
that might be involved in future competitions.(24)
- As Professor Economides notes in his declaration, it is difficult
to predict precisely where and when entry is likely, and economists
may disagree on the likelihood that a firm will be willing to undertake
the necessary investment.(25) I agree.
But one cannot say that the merger is likely to harm customers in
a building if the possibility of entry cannot be ruled out with some
confidence. If the Department is not sufficiently confident that a
harm can be predicted (and, presumably, proven), it cannot allege
one.
- Amici have argued that the Department's entry analysis is incomplete
because it ignores various factors listed in paragraph 27 of the Complaints.(26)
The entry screens did take into account the most important factors,
which were the potential revenue stream likely to be obtained at some
point in the future and the biggest component of the cost of entering.(27)
But they did not take into account unusual features or circumstances
regarding buildings or customers that could raise costs (e.g., the
existence of railbeds), nor did they consider the possibility of increased
revenues (e.g., if sales could be predicted to increase). Given the
Department's obligation to be able to defend in court its allegations
of where harm is likely to occur, however, I believe that the analysis
used predicts likely entry as closely as is practical.
- The Remedies Are Appropriate to Restore Competition in the Relevant
Buildings.
- My initial declaration explained in detail how the divestitures required by the
proposed Final Judgments will remedy the harm alleged in the Complaints.(28)
I
will not repeat that discussion here. The amici have raised two arguments about
the proposed remedies. First, they allege that the divestitures are not extensive
enough to create another company of the same size and scope as AT&T and
MCI.(29)
The purpose of the remedy is not to recreate the acquired company but to
remedy the alleged competitive harm. AT&T and MCI compete in numerous
markets. For many telecommunications services, such as providing sophisticated
data networks to multinational companies, they faced little or no competition
from their respective merger partners. There is no significant harm stemming
from the lost competition in these markets, so no divestitures are required. The
predicted harm from these mergers stems only from the lost competition to
provide a basic service, local private lines, to more than 700 buildings where entry
appears to be unlikely, and the remedies result in another CLEC stepping into
AT&T's or MCI's shoes in each of these buildings.
- The other argument made by the amici is that based on the small dollar amount
potential buyers are willing to pay for the SBC divestiture assets, the Court
should conclude that the assets will not allow the purchaser to be competitively
significant in the marketplace.(30)
I am not concerned by the actual prices obtained
by SBC for these assets. First of all, the buyer is not acquiring a current revenue
stream, but only the ability to compete for future business opportunities as they
arise. They of course would have been willing to pay more if customers had been
included with the divestiture assets. Amici equate no customers, and
consequently no market share, with no ability to influence prices bid to customers
seeking to buy new services or change providers. This is incorrect. As soon as an
acquirer has the ability to serve these buildings without having to build extensive
new infrastructure, it will be able to compete aggressively for each new business
opportunity.
- Second, it is not uncommon for those acquiring assets to be divested pursuant to
Department consent decrees to pay prices that seem much lower than what those
assets would cost in other contexts. Buyers know that the merged entity must sell
the assets and is anxious to do so under the deadlines imposed by a final
judgment. Economists generally expect that the party negotiating with more
urgency and fewer options will get the worst of an agreement, but how that plays
out in any given sale depends on many other conditions. Thus, specific inferences
of the true value of an asset package cannot be based on the price paid.
- Other Arguments Advanced by Amici Have No Merit
- The amici also contend that the Department should have brought
broader cases than the ones alleged in the Complaints to address:
(1) the effect on local private line services provided to buildings
other than those identified in the proposed Final Judgments (the "3-to-2"
or "4-to-3" situations),(31) (2) AT&T's
and MCI's positions as resellers of local private lines,(32)
(3) potential for collusive behavior between SBC and Verizon post-
merger,(33) (4) loss of competition
in consumer telephony,(34) and (5)
retail telecommunications services.(35)
Although I have been told that these topics are outside the scope
of these proceedings, I offer a brief explanation of the Department's
reasons for not seeking relief in these areas.
- No Harm in "3-to-2" and "4-to-3" Buildings
- Although many mergers will result in the elimination of a competitor in a market,
not all such mergers will result in a substantial lessening of competition. Whether
the elimination of a particular competitor is significant will depend on the
structure of the industry, the nature of competition, the goods or services
involved, and the competitive characteristics of the merging parties. For those
buildings where AT&T or MCI is not the only CLEC with the facilities capable of
providing local private line services, I believe the evidence did not suggest a
significant loss of competition as a result of the mergers.
- Local Private Lines Are Essentially a Commodity
- The Department's investigation produced evidence suggesting that local
private lines are close to a pure commodity. That is, customers perceive no
substantial difference between the various local private line options that might be
available for connecting two points in a city. These are generally sophisticated
buyers who understand the technical nature of the local private line connection
they are buying. They know that the capacity of the connection they are buying is
standardized, and that, for example, the OC-3 grade connection from one carrier,
is interchangeable with the OC-3 grade connection of another carrier. The
competition for these customers generally comes down to price.
- For commodity products, intense competition can occur in a market
with only two firms, absent some type of collusive agreement between
the firms.(36) In addition to describing
how to delineate markets and calculate measures of concentration,
the Merger Guidelines discuss the need to evaluate whether
competitive harm is likely given the specific characteristics of the
market. In these cases, this evaluation is to determine whether there
is an economic story that would explain why the loss of AT&T or
MCI would result in higher prices or other anticompetitive effects.
The Merger Guidelines describe two ways in which a merger
may harm competition.(37) The first
is that the merger would make collusion among the remaining firms
more likely. The second is that the merged firm may be able unilaterally
to raise prices because it has eliminated a competitor who otherwise
would have acted to constrain such behavior. Given the facts of this
case, unilateral effects would only be significant where there were
no other competitors positioned to offer the same commodity service.
- Collusive agreements also are not likely in these markets for several
reasons, including the facts that purchases often involve long-term
contracts, the prices negotiated with customers are not readily observable,
the competitors that each market participant faces vary from building
to building, and each market participant has a strong incentive to
win every customer in order to recover the large fixed costs associated
with building facilities. All of these factors make it difficult for
firms to reach and police collusive agreements.(38)
Consequently, I did not predict such harm in buildings where other
CLECs were present.
- No Evidence that AT&T or MCI Is a Uniquely Important
Competitor
- As I explained in my initial declaration, an additional competitor might
still matter if that competitor brings some unique advantage to the competition.(39)
If, for example, AT&T or MCI was able to provide local private lines at a
particularly low cost, then their presence (even with other competitors) might
push SBC or Verizon to an even lower level of pricing. I did not find evidence to
support a conclusion that AT&T or MCI had unique advantages in selling local
private line services.(40)
-
[REDACTED TEXT]
.(41)
-
[REDACTED TEXT]
.(42)
[REDACTED TEXT]
(43)
[REDACTED TEXT]
.(44)
- Both during the investigation and in its recent filing, ACTel has
argued that the presence of AT&T and MCI in the wholesale market
resulted in lower prices.(45) To support
this argument, ACTel provided the Department with [REDACTED
TEXT] .(46) [REDACTED
TEXT] Attached to this declaration is a summary of the principal
shortcomings identified in [REDACTED TEXT] Professor
Wilkie's analysis, which was prepared by Department economic staff
working under my direction.(47) To
summarize, ACTel's submissions did not provide credible support for
the conclusion that the elimination of AT&T and MCI as competitors
for local private lines would result in any significant loss of competition
where other CLECs remain to offer the service.
- AT&T and MCI Were Not Significant Resellers of
Local Private Lines
- Amici raise concerns about the Department's failure to remedy the
loss of competition due to the elimination of AT&T and MCI as
resellers of local private lines leased from SBC or AT&T.(48)
The amici argue that MCI and AT&T were able to buy local private
lines from the RBOCs at lower rates(49)
than other CLECs and then often resell this access bundled with their
own facilities to other CLECs. Amici allege that whereas this allowed
the CLECs to avoid paying the higher prices charged by the RBOCs for
equivalent circuits, after the merger, CLECs will have to buy access
at higher prices.(50) These arguments
fail because they are factually untrue. [REDACTED TEXT]
AT&T received the same discount from SBC that was available to
any carrier that committed to spending a minimum of $10 million annually.(51)
The $10 million threshold did not preclude other CLECs from qualifying
for discounts under these plans.(52)
Verizon also offered several plans under which discounts were available
to all CLECs including MCI. Not surprisingly, MCI and AT&T did
not resell large dollar volumes of local private lines to other carriers.(53)
- Department Concluded that Collusion Between
RBOCs Was Unlikely
- Several amici repeat arguments previously espoused by COMPTEL criticizing
the Department for failing to consider that after the mergers, Verizon
and SBC might collude by tacitly agreeing not to sell local private
lines to other CLECs in each other's territories.(54)
Specifically, they contend that it is intuitive that Verizon will
not use the MCI facilities in SBC's region to sell services to SBC's
competitors if SBC similarly refrains in Verizon's territory.(55)
As even Sprint and the NYAG recognize, such an agreement would somehow
have to be limited to the sales of these facilities to other CLECs.
Verizon and SBC will want to continue to use these facilities to compete
against each other to serve large enterprise customers. But such a
limited tacit agreement is unlikely to be effective. The facilities
are already deployed, they will presumably be maintained to support
retail sales, which would make cheating on the agreement attractive.
In addition, the agreement may be unsustainable because a wholesale
sale can easily be recharacterized as retail to circumvent the agreement.
Therefore this tacit agreement, that seems intuitive to amici, is
rather unlikely. Indeed, while amici contend that SBC and Verizon
naturally avoid invading each other's territories, in wireless telephony
they have made extensive investments to compete nationwide.
- Consumer Telephony
- Several amici have raised concerns about traditional switched local
and long distance telephony services marketed to consumers.(56)
The Department's decision to not seek relief for these products was
based in large part on the significant changes that had occurred,
and continue to occur, in the markets for these services. First, there
were significant regulatory changes. For example, an FCC order, later
affirmed by the D. C. Circuit, changed existing regulation to alter
the ability of CLECs, including AT&T and MCI, to buy from the
RBOCs low-priced wholesale access services that had allowed them to
serve mass-market customers. AT&T and MCI were relying on these
access services to reach customer locations, and thus, this ruling
diminished AT&T's and MCI's ability to compete. Second, the increased
use of wireless telephones for long distance calling also diminished
AT&T's and MCI's appeal as providers of bundled local and long
distance services. Many customers that had been drawn to AT&T
and MCI because of their reputations and expertise as providers of
long distance services no longer place as high a value on purchasing
from these providers, as most, if not all, of these customers' long
distance calls are now placed on cell phones. Finally, the traditional
cable companies have entered and made significant inroads in offering
both local and long distance telephony. As a result of these continuing
changes, the Department concluded that it would not be able to prove
that the mergers would have a sustained effect on the consumer markets.
- Retail Sales to Enterprise Customers
- The Department concluded that as to retail sales to large enterprise
customers, with the exception of customers in the 2-to-1 buildings,
harm from the mergers was unlikely. This conclusion was based in large
part on the limited competition that exists between the merging parties,
with Verizon and SBC providing primarily local services, and AT&T
and MCI selling primarily long distance and advanced network services.
The lack of extensive direct competition was confirmed by the parties'
documents as well as through extensive interviews of retail customers.
In conducting its interviews, the Department sought to canvass a wide
range of retail customers, and to focus on retail customers that would
be more likely to have concerns about the mergers such as customers
in buildings likely to be affected by the transaction. In total, 211
customer interviews were conducted. These interviews failed to identify
customer concerns supportive of an antitrust theory of harm, but,
more important, many customers told us that in purchasing telecommunications
services they did not consider the merging firms to be competitors.
The Division also looked at data produced by the parties to identify
price differences for retail customers based on the number of competitors
with competing facilities servicing the customers' buildings. No such
relationship was found. In order to make available to the Court some
of the information the Department obtained in customer interviews,
I attached to my initial declaration a large number of customer statements.
The amici attack my use of the 129 customer statements, alleging that
they are not a representative sample of customers' views(57)
and are suspect because they were provided by defendants.(58)
The Department agrees that these statements (submitted by one of the
merging parties) standing alone would be given limited weight. However,
as noted in my initial declaration, these statements are consistent
with the interviews of customers conducted by the Department and the
other information that it reviewed.(59)
- Conclusion
- The issues and evidence presented by the amici are not new. These issues were
considered extensively in the course of the Department's investigation and as part of
the decision to seek the proposed remedies. For the most part, the factual
information that is attached to or cited in their filings was available to the
Department. Notwithstanding the volume and length of these filings, there is
nothing that leads me to question the Department's conclusions that these mergers
pose an identifiable threat to competition for local private line service provided to
more than 700 buildings in Verizon's and SBC's territories, and that the proposed
remedies can reasonably be expected to mitigate this threat with minimal collateral
harm to customers.
I declare under penalty of perjury that the foregoing is true and correct to the best
of my information and belief.
Executed on September __, 2006
_______________________________
W. Robert Majure, Ph.D.
ATTACHMENT: CRITIQUE OF DR. WILKIE'S ANALYSIS
I. Summary of ACTel Data Analysis
[REDACTED TEXT]
The Department worked with
this information to determine if the data would shed light on the markets being reviewed and
possibly support an antitrust theory of harm [REDACTED TEXT] the Division received
Dr. Wilkie's studies as to [REDACTED
TEXT] .(1) The Division analyzed his results. We describe here several significant
issues that led us to discount the data and Dr.Wilkie's studies. REDACTED
- Review of the Pricing Data
[REDACTED TEXT](2) I [REDACTED
TEXT] [REDACTED TEXT]. The data, however, did
not present the clear picture of AT&T as an aggressive competitor painted by Dr. Wilkie.
[REDACTED TEXT]
These facts are inconsistent with
the proposition that the loss of AT&T as a bidder would have a significant impact on competition
but were consistent with other evidence (party documents, sales data, and customer interviews)
that demonstrated that AT&T did not have a large presence in the wholesale market.
[REDACTED TEXT]
Our review of the data also identified issues relating to its use for making predictions
about the entire market for LPL. It can be misleading to use data to predict the future economic
impact of a merger unless it reflects actual commercial transactions. Consequently, information
obtained other than through the normal course of business should be used cautiously, if at all, in
antitrust analysis. Likewise, to form a reliable basis for predicting the economic future, the
number of transactions reflected in the data should be large enough to provide a fair
representation of how competition occurs.(3) The data supplied by ACTel was deficient in both
regards.
[REDACTED TEXT] .
The data Dr. Wilkie addressed include approximately [REDACTED TEXT]
circuits a very small subset of the hundreds of thousands of circuits
that are sold annually. [REDACTED TEXT]
[REDACTED TEXT] The small sample coupled with the unusual circumstances under which
the pricing was generated cast significant doubt on whether reliable predictions could be obtained
from this data.
- Dr. Wilkie's Regressions Are Unreliable
The Division's review of Dr. Wilkie's results demonstrated that they are not reliable. Dr.
Wilkie uses a regression model to analyze the data. Simply put, a regression is a mathematical
model used by economists to understand how different factors (e.g., prices, costs, concentration)
may be interrelated. The models then allow the user to predict the impact of a change to one of
those factors while leaving everything else unchanged. Whether the results are reliable often
depends on whether the analyst selected a model that fits the industry and on the quality of the
data used by the model.
At its most fundamental level, Dr. Wilkie's analysis compares price to a measure of
concentration (number of bidders) to determine if the number of bidders impacts pricing. He
concludes that his studies support the propositions that removing AT&T as a bidder in SBC's
territory and MCI as a bidder in Verizon's territory would result in higher prices (i.e., lower
discounts). The fundamental flaw with Dr. Wilkie's analysis is that he ignores the possibility of
"endogeneity," namely that something other than the variable being tested (in this case the
number of bidders) could be affecting both the price and the number of bidders. For example,
the cost of providing a circuit, which is not included in Dr. Wilkie's analysis, might affect both
the price and the number of bidders. When we tested Dr. Wilkie's results by controlling for
market characteristics at the LATA level, some of which may be causing an endogeneity
problem, we obtained dramatically different results inconsistent with Dr. Wilkie's findings.(4)
As an example, consider
[REDACTED TEXT]
These dramatic differences, which resulted solely from the inclusion of market
characteristics at the LATA level, led us to conclude that Dr. Wilkie's analysis
lacked sufficient robustness to be relied upon. This flaw is present in all
of the regressions performed by Dr. Wilkie.
Finally, ACTel does not submit portions of Dr. Wilkie's analysis that are seemingly
inconsistent with its arguments.
[REDACTED TEXT]
.
FOOTNOTES
1. Some amici fail to understand the analytical
elements of an antitrust investigation or the need for evidence to support
a challenge. They imply, for instance, that the Department could properly
sue based solely on figures that suggest high concentration in a market,
or achieve broad divestitures even though it cannot be proved that entry
is unlikely to alleviate the competitive concerns.
2. U.S. Dep't of Justice & Federal Trade Comm'n,
Horizontal Merger Guidelines § 1.0 (rev. ed. 1997) ("Merger
Guidelines").
3. U.S. Dep't of Justice & Federal Trade Comm'n,
Commentary on the Horizontal Merger Guidelines at 2 (2006) ("Merger
Guidelines Commentary").
4. Merger Guidelines §1.0.
5. There is potentially an even smaller set of goods
one could start with because the parties generally set distinct prices
for local private lines with different capacities (DS1, DS3, OC3). However,
because those prices appear to be more of a menu of "flavors" available
from any carrier capable of providing local private line service, using
the smaller market definition does not substantially alter the analysis
because the same set of providers would participate in each of the smaller
markets.
6. Local private lines are dedicated point-to-point
circuits offered over copper and/or fiber-optic transmission facilities
that originate and terminate within a single metropolitan area. Complaints
¶ 3.
7. The delineation of geographic markets limited
to individual buildings is also consistent with how the Department has
defined markets in other telecommunications merger matters. See,
e.g., Complaint, United States v. Echostar Communications Corp.,
Civ. No. 1:02CV02138 (D.D.C. filed Oct 31, 2002), ¶¶ 30-31
( alleging that each residence constitutes a separate geographic market
but aggregating, for ease of analysis, residences that face the same
choice of providers ). Such aggregating would not be useful here given
the differences in CLEC networks.
8. Memorandum of Eliot Spitzer, Attorney General
of the State of New York, in Response to the August 7, 2006 Submission
of the United States at 8 (Sept. 5, 2006) ("NYAG Resp."); Declaration
of Nicholas Economides (Sept. 5, 2006) ("Economides NYAG Decl."), ¶¶
21, 25, attached to NYAG Resp.; COMPTEL's Response to the Department
of Justice's Supplemental Submission at 20 (Sept. 5, 2006) ("COMPTEL
Resp."); Declaration of Joseph Gillan (Sept. 1, 2006) ("Gillan COMPTEL
Decl."), ¶¶ 10-11, attached to COMPTEL Resp.
9. For example, consider cars and their windshields.
When consumers walk into an automobile showroom they expect to buy an
entire car; however, that car is composed of other products, including
windshields. The fact that an end-user typically buys an entire car
would not preclude the definition of a product market of automotive
windshields, nor would it suggest that car manufacturers could not suffer
harm due to a merger of windshield producers.
Amicus COMPTEL implies that the proper geographic market should be
the entire United States. COMPTEL Resp. at 29-30. COMPTEL's confusion
appears to be linked to its misreading of the Complaints. It interprets
the language "and telecommunications services that rely on local private
lines," Complaints ¶ 19, to mean that the Department filed a case
alleging harm for all telecommunications services provided to business
customers. That is not correct. The Department's case is limited to
local private lines but the language "telecommunications services that
rely on local private lines" refers to the indirect effect of the merger
on customers that purchase services that use local private lines to
the affected buildings as an input. In terms of the example provided
above, this means that if a merger were to cause the price of windshields
to increase, the price of cars might increase.
10. See COMPTEL Resp. at 18-19, 28; Economides
NYAG Decl. ¶¶ 25-27.
11. Merger Guidelines § 1.5.
12. ACTel's Reply Memorandum in Opposition to the
United States' Motion for Entry of Final Judgments at 9 (June 7, 2006).
13. Amici NYAG and COMPTEL have suggested that
HHI figures alone can be used to determine whether competitive harm
is likely from a merger. See NYAG Resp. at 12; Economides NYAG
Decl. ¶ 55; COMPTEL Resp. at 25-26. Although the Merger Guidelines
do establish presumptions for markets being more or less likely
to support a prediction of harm, they go on to describe the other factors
that need to be taken into account before such a conclusion can be reached
(i.e., theory of competitive harm, entry, and efficiencies). Thus, calculating
HHIs is not meaningful unless other information related to the industry
is evaluated. Merger Guidelines Commentary at 15-16. These other\
factors are discussed in my initial declaration as well as below. Declaration
of W. Robert Majure (Aug. 7, 2006) ("Majure Decl."), ¶¶ 12-14,
attached to United States' Submission in Response to the Court's Minute
Order of July 25, 2006 (Aug. 7, 2006) ("United States' Submission").
See also infra ¶¶ 16-18,\ 24-32.
14. See supra ¶¶ 11-12.
15. COMPTEL Resp. at 30-31. The FCC notes that
SBC did not provide a definition for "local voice services." Memorandum
Opinion and Order, In the Matter of SBC Communications Inc. and AT&T
Corp. Applications for Approval of Transfer of Control, 20 F.C.C.R.
18,290 (Nov. 17, 2005), ¶ 70 n.201. In addition, the data is reported
on a statewide basis. Id. ¶ 70 & n.200. COMPTEL fails
to note that the FCC did not find a competitive problem in any retail
enterprise business market despite the figures cited.
16. Three other amici submitted HHIs. New Jersey
Rate Counsel submitted HHIs for retail residential and small and medium
businesses, customers unlikely to be purchasing local private lines.
Reply of the New Jersey Rate Counsel to the United States' Submission
in Response to the Court's Minute Order of July 25, 2006 at 9-10 (Sept.
5, 2006) ("NJRC Reply"). NYAG quotes HHI figures from the report of
the NY Public Service Commission ("NYPSC") staff which were included
in the section that analyzed retail enterprise competition. NYAG Resp.
at 12. The NYPSC staff does not use these figures to evaluate competition
for local private lines. Sprint refers to concentration numbers provided
to the FCC by Dr. Wilkie on behalf of ACTel members XO Communications
and Savvis Communications that were calculated for six metropolitan
areas based on capacity being used to serve customers. Response of Sprint
Nextel Corporation to the United States' Submission in Response to the
Court's Minute Order of July 25, 2006, at 26 (Sept. 5, 2006) ("Sprint
Resp."). Dr. Wilkie's source for the data seems to be building lists
from the parties and less reliable data on other CLECs from a third-party
report. The report appears to include only capacity figures pertaining
to business that was actually won, and therefore may be prone to the
same misinterpretation problems as using revenue figures, as described\
here. See supra ¶ 13.
17. Majure Decl. ¶¶ 6-14.
18. Merger Guidelines Commentary at 15-16.
19. Merger Guidelines §§ 1.32,
3.0.
20.COMPTEL Resp. at 22-24.
21. Id. at 22-23; Sprint Resp. at 24; see
also Majure Decl., Attachs. Tab 9, CLEC Interrogatory Responses.
22. These competitors can drive prices down as
the carrier who already has facilities into the building is forced to
offer prices that are low enough to keep the customer from accepting
the bid from the entrant. This would be significant competition even
if, in the end, the customer decided to use the provider with existing
facilities.
23. The criteria that any particular CLEC uses
in deciding whether to construct a lateral into a building are not necessarily
or absolutely consistent with those used by other CLECs. The Department
estimated possible entry using information gathered from multiple sources
to evaluate whether the potential benefits to a firm could be sufficient
to induce it to build a lateral. The Department found no internal business
documents that analyze the costs and benefits of entering each individual
building in an area.
24. This is, of necessity, an approximation. However,
because the approximation is based on the decisions made under current
conditions, it will tend to understate entry to the extent that there
would be additional incentives to enter following any significant exercise
of market power (e.g., raising prices in the building) by the merged
firm.
25. Economides NYAG Decl. ¶ 11.
26. ACTel's Response to the United States' Submission
Pursuant to the Court's Minute Order of July 25, 2006, at 14-16 (Sept.
5 , 2006) ("ACTel Resp."); Economides NYAG Decl. ¶¶ 68-76;
Declaration of Lee L. Selwyn (Sept. 5, 2006) ("Selwyn NASUCA Decl."),
¶ 46, attached to Reply of Amicus the National Association of State
Utility Consumer Advocates Regarding the Public Interest (Sept. 5, 2006)
("NASUCA Reply").
27. ACTel points out that the Department's screen
did not account for the fact that an acceptable connection point or
node may not be located at the network's closest point to the building.
See ACTel Resp. at 20-21. The Department only accounted for the
cost to construct a connection to the network at its closest point.
It seems likely that the cost of pulling fiber from this point to the
nearest node through an existing conduit is far less than that of actually
constructing new conduit, and so the additional distance to reach a
network node is unlikely to generate costs significant to the entry
decision.
28. Majure Decl. ¶¶ 15-25.
29. Gillan COMPTEL Decl. ¶¶ 21, 23;
Selwyn NASUCA Decl. ¶¶ 35-37; Economides NYAG Decl. ¶¶
67, 88.
30. Gillan COMPTEL Decl. ¶¶ 23-27; Selwyn
NASUCA Decl. ¶¶ 48-49, 52-54.
31. Economides NYAG Decl. ¶ 61.
32. Id. ¶¶ 50-51; Selwyn NASUCA
Decl. ¶¶ 12-16.
33. Sprint Resp. at 29-30; Economides NYAG Decl.
¶ 53.
34. NJRC Reply at 2-3; Selwyn NASUCA Decl. ¶¶
58-61.
35. COMPTEL Resp. at 27-31.
36. Economic theory has long recognized two models
of how industries can be perfectly competitive. The first is where there
is such a large number of suppliers that no one of them can exercise
any influence over the price or total quantity sold. The other is where
there are at least two firms whose products are so nearly identical
that any difference in price drives all of the customers immediately
to the lower-priced firm. These models of competition can be found in
any textbook of industrial organization economics. See, e.g.,
Jean Tirole, The Theory of Industrial Organization 209-11 (1988).
This latter model, known as Bertrand competition, has been studied extensively
by economists. Id. at 211-18. For either merger to have significant
competitive effects, we would have had to establish that the concept
embodied in the Bertrand model did not apply – i.e., establish
that there are significant differences between the services offered
by AT&T or MCI and another CLEC in the same building. This is not
to imply that local private lines markets are an ideal fit to the Bertrand
model. That model does not incorporate important characteristics of
the industry such as regulation and the economics of networks. Nevertheless,
the basic concept in the Bertrand model would apply and limit the degree
to which the presence of additional CLECs in a building results in significantly
more competition.
37. Merger Guidelines § 2.0.
38. Merger Guidelines § 2.1.
39. Amicus Sprint alleges that AT&T and MCI
are unique because it has used competitive offers from these companies
to negotiate lower rates from the RBOCs. Declaration of Keith L. Kassien
(Sept. 1, 2006) ("Kassien Sprint Decl."), ¶ 17, attached to Sprint
Resp. As the fiber maps and lists of on-net buildings attached to my
initial declaration illustrate, there is a significant amount of fiber
available from other CLECs and a great deal of overlap in the facilities
of other CLECs and AT&T and MCI. See Majure Decl., Attachs.
Tab 6, CLEC Network Maps and Building Lists; Tab 7, Overlapping CLEC
Fiber Maps. Sprint has not explained why it will not be able to obtain
equally competitive bids from CLECs to use in negotiations with the
RBOCs after the merger.
40. Amici devote substantial space in their filings
to attempting to establish that AT&T and MCI are unique. However,
it is not enough to point to some characteristic of AT&T and MCI
that is different than other CLECs. The difference must be one that
would support an economic explanation of how the loss of AT&T or
MCI would affect competition for local private lines. Most of the arguments
made by amici focus on the fact that AT&T and MCI are significant
competitors for retail sales to large enterprise customers. Selwyn NASUCA
Decl. ¶¶ 28-32; Sprint Resp. at 14; Economides NYAG Decl.
¶¶ 20-22, 77-80. They fail, however, to draw a connection
between success at the retail level and the ability to be a significant
provider of an input, such as local private lines. Other amici argue
that the scope of the networks operated by AT&T and MCI make them
unique. Selwyn NASUCA Decl. ¶¶ 35-37; Economides NYAG Decl.
¶¶ 20-21, 44-45, 74, 77-78. Network economies might make the
size of the AT&T and MCI networks relevant if these networks were
anywhere near large enough that a significant portion of possible interconnections
a customer might want to make might have both endpoints on the AT&T
or MCI network. Such an advantage may be available to the RBOC given
its ubiquitous coverage in region, but not to a CLEC. As Dr. Selwyn
notes, the vast majority of the buildings served by AT&T are not
on-net, but are served by AT&T primarily through facilities it leases
from the RBOCs. Selwyn NASUCA Decl. ¶ 13. All carriers rely on
interconnecting with other providers in order to offer their customers
the benefits of a larger network than the one they own. Any carrier
who can offer a local private line between a given building and a common
interconnection point offers just as much functionality as AT&T
can offer in that building, except in some very rare cases where the
customer wants to interconnect to another building to which AT&T
has a connection.
41. See Majure Decl., Attachs. Tab 11,
AT&T Documents.
42. See Majure Decl., Attachs. Tab 12,
MCI Documents.
43. [REDACTED TEXT] MCI-DOJ-QQ0001024,
at -1024 (2005).
44. Amicus ACTel finds fault with my previous
declaration because it fails to explain "why prices throughout the relevant
market have reversed a downward trajectory and have consistently increased
since the mergers." ACTel Resp. at 3-4. ACTel's evidence of consistently
increasing prices in the market post-merger consists of selected tariffs
filed in a few states in SBC's territory for one type of local private
line. There are numerous reasons why some prices may increase even without
a merger of firms in a market. ACTel's selection of tariff adjustments
which happen to have involved price increases is far from the kind of
systematic evidence I would want to rely on in predicting the consequences
of a merger. In particular, ACTel does not suggest any way to determine
that other market changes, such as regulatory changes, are not the more
obvious explanations for these particular increases. The fact that other
amici have disputed ACTel's characterization of the pre-merger market
as experiencing decreasing prices, see, e.g., Selwyn NASUCA Decl.
¶¶ 29-32, is yet further reason not to rely on ACTel's interpretation
of these tariff filings.
45. ACTel Resp. at 25-27.
46. [REDACTED TEXT] ACTel had
previously provided a declaration from Dr. Wilkie describing the same
empirical work. Declaration of Dr. Simon J. Wilkie Submitted on Behalf
of the Alliance for Competition in Communications (May 4, 2006), attached
to ACTel's Motion for Amicus Curiae and Intervenor Status Pursuant
to the Tunney Act (May 5, 2006).
47. Critique of Dr. Wilkie's Analysis, Majure
Reply Decl. Attach.
48. Economides NYAG Decl. ¶¶ 50-51,
57, 64; Selwyn NASUCA Decl. ¶¶ 12-16.
49. Dr. Selwyn discusses the ability of a carrier
to purchase a high-capacity circuit and resell it as a collection of
lower capacity circuits (assuming there is enough demand for the lower
capacity circuits between the circuits's end points). Selwyn NASUCA
Decl. ¶ 15. But any carrier can engage in this type of resale,
so AT&T and MCI only have an advantage in dividing and reselling
lines to other CLECs if they received discounts above and beyond those
available to other competitors. As discussed here, they do not.
50. Id. ¶¶ 13-17; Economides
NYAG Decl. ¶¶ 28-30, 39, 50-51.
51. Declaration of Parley C. Casto (May 6, 2005),
¶ 6, attached to Joint Opposition of SBC Communications Inc. and
AT&T Corp. to Petitions to Deny and Reply to Comments, In re:
SBC Communications Inc. and AT&T Corp. Applications for Approval
of Transfer of Control, FCC WC Docket No. 05-65 (May 10, 2005).
52. Id.
53. AT&T reported that [REDACTED TEXT]
of its local private line sales were resales of an SBC facility
( [REDACTED TEXT] ). Declaration of Anthony Fea, Anthony Giovannucci,
Bob Handal, Michael Lesher and C. Michael Pfau (May 9, 2005), ¶
43 (attached to Joint Opposition of SBC Communications Inc. and AT&T
Corp. to Petitions to Deny and Reply to Comments, In re: SBC Communications
Inc. and AT&T Corp. Applications for Transfer of Control, WC
Docket No. 05-65 (filed May 10, 2005)), United States' Reply Submission
Attach. 3. Similarly, MCI stated that [REDACTED TEXT ] of its
wholesale local private lines sales in Verizon East territory were in
part resales ( [REDACTED TEXT] ). [REDACTED TEXT]
54. See, e.g., Sprint Resp. at 30; Economides
NYAG Decl. ¶ 53; cf. COMPTEL's Opposition to the Department
of Justice's Motion for Entry of Final Judgments at 19-20 (Apr. 6, 2006);
Motion and Memorandum in Support of COMPTEL's Motion for Leave to Intervene,
or in the Alternative to Participate as Amicus Curiae at 18-19 (Feb.
8, 2006).
55. I concentrate here on the likelihood of a
tacit agreement as an explicit agreement would be illegal and possibly
criminal.
56. NJRC Reply at 2-3; Selwyn NASUCA Decl. ¶¶
58-61.
57. Amicus NYAG questions the value of these statements
based on the results of what it refers to as a "scientific" survey of
100 "randomly" selected Fortune 1000 companies. NYAG Resp. at 11 &
n.4 (citing Center for Survey Research & Analysis, Views of the
Proposed AT&T/SBC and MCI/Verizon Mergers: From the Perspective
of Fortune 1000 AT&T and MCI Customers (Sept. 2005) (commissioned
by ACTel) ("CSRA Survey")). The survey cited was not scientific and
the sample was not random. The companies were specifically selected
because they had purchased telecommunications services from AT&T
or MCI. The survey respondents were asked 10 questions, only one of
which related to concerns about the merger. In addition, the construction
of the survey tended to bias the respondents' response as to this question.
The survey did not request information about the extent to which the
merging parties competed for the company's business or about the nature
of the concern that the company had about the merger. Although NYAG
reported that two-thirds of the respondents expressed concerns, it did
not reveal that only 11% were very concerned, and that more than 50%
expressed no or little concern. CSRA Survey at 15.
58. COMPTEL Resp. at 27.
59. Majure Decl. ¶ 4 & n.1.
ATTACHMENT FOOTNOTES
1. [REDACTED TEXT] Dr.
Wilkie had previously reported the results of his analysis. Declaration
of Dr. Simon J. Wilkie Submitted on Behalf of the Alliance for Competition
in Communications (May 4, 2006) ("Wilkie ACTel Decl."), attached to
ACTel's Motion for Amicus Curiae and Intervenor Status Pursuant
to the Tunney Act (May 5, 2006). [REDACTED TEXT] .
2. [REDACTED TEXT] [REDACTED
TEXT].
3. [REDACTED TEXT] .
4. These omitted factors can affect both prices
and the number of bidders, and can cause a regression to 'confuse' the
effect of these omitted factors with the effect of the number of bidders.
To try to reduce this problem, we estimated what is commonly called
a "fixed-effects regression" and used LATAs (or regions) as the fixed-effect.
Unlike Dr. Wilkie's analysis, this model controls for factors that are
constant at the LATA level, for example, the fact that the cost of providing
a circuit in New York City may differ from the cost in Kansas City.
In both the [REDACTED TEXT] [REDACTED TEXT]
The dramatically different results we obtained with this reasonable
change to the model led us to suspect that other factors not considered
in Dr. Wilkie's analysis were significantly influencing his results.
|