1. How did the settlement affect the Airlines’ behaviors?2. Does this case suggest any restrictions on the pricing behaviors used by merchants in online marketplace?

Unformatted Attachment Preview

Rapid Price Communication
and Coordination:
The Airline Tariff Publishing Case (1994)
Severin Borenstein
A firm announces a price increase and shortly thereafter its competitor announces its own increase to the same price level. Is that price fixing? Most
antitrust economists and lawyers would say no. What if the announcements
are made and changed rapidly? What if each firm makes many announcements before they settle down at identical prices? Finally, what if the prices
being announced are to take effect at some future date so that no sales actually take place at these prices while the announcements are being made?
This is a gray area of the antitrust laws. While an agreement among competitors to fix prices is per se illegal, computer technology that permits
rapid announcements and responses has blurred the meaning of “agreement” and has made it difficult for antitrust authorities to distinguish public
announcements from conversations among competitors.
These were some of the issues that were raised in the U.S. Department
of Justice’s investigation of the major U.S. airlines and the Airline Tariff
Publishing Company (ATPCO), which is owned by the airlines and disseminates price change information to airline and travel agent computer
systems. The investigation began in 1991, and the resulting case was settled
with a consent decree in March 1994. The case never went to trial, and
therefore it set no formal precedent. Still, the legal pleadings, negotiations,
and the final consent decree indicate some of the difficult antitrust issues
that are raised by rapid price announcements as well as the impact of new
communication technologies on those issues.
Severin Borenstein consulted for the U.S. Department of Justice in this matter. For helpful comments and discussions, I am grateful to Dennis Carlton, Rob Gertner, Rich Gilbert, John Kwoka,
Carl Shapiro, and Larry White.
From the time of airline deregulation in 1978 through the early 1990s, the
airline industry was in a state of nearly constant upheaval. At the time of
deregulation, the U.S. domestic jet air travel market included about twenty
competitors ranging from small regional carriers to the largest national and
international airlines. Immediately following deregulation, many startup
airlines entered the domestic market. Concentration at both the national and
the route level fell in the first five years following deregulation. Most of the
startups were small regional airlines. Under the financial pressure of competition, few were ever able to turn a profit. By the mid-1980s, airlines were
folding faster than new ones were entering, and many were disappearing
through merger. A merger wave in the middle and late 1980s, along with
more exits during that time, raised the Herfindahl-Hirschman Index for domestic air travel from 854 at the beginning of 1985 to 1074 at the beginning
of 1990.
Two of the most important and unforeseen developments in the industry following deregulation were the airlines’ moves to hub-and-spoke networks and their increased sophistication in pricing and marketing their products. The hub system created a natural area of dominance for a carrier. Since
most cities do not have sufficient traffic to support hubs for two different carriers, a typical big-city airport is likely to be dominated by a single airline.
From its hub, an airline would offer nonstop service to many or most of the
country’s other large cities and many small cities in the region of the hub.
Such a network also lends itself naturally to offering change-of-plane service between airports for which the hub is a convenient intermediate stop.
These routings could compete with nonstop service offered by a carrier that
has a hub at one of the end-point airports. For instance, while Northwest was
the only carrier to offer nonstop service between its Detroit hub and Los Angeles, it had less than 60 percent of the traffic in that market. The remaining
traffic flowed over competing carriers’ hubs: with Continental, changing
planes in Denver; with American, changing planes in Chicago; with United,
changing planes in Denver or Chicago; with Delta, changing planes in Salt
Lake City.
The major carriers also became very sophisticated in marketing and
pricing their products following deregulation. They developed customer
loyalty plans that reinforced their dominance at their hubs. These included
frequent flyer programs and travel agent commission override programs
(TACOs). TACOs are effectively “frequent booker” programs for travel
agents: Agents are rewarded for directing a high proportion of their bookings to the airline. Airlines also offered corporate discount programs that
rewarded a corporation for concentrating its air travel with just one airline.
The debate about the efficiency versus market power enhancement proper-
Case 9:
The Airline Tariff Publishing Case (1994)
ties of these programs is still active, but there is general agreement that
these programs led to greater airport concentration.1
Believing that hubs delivered significant competitive advantages, and
forecasting continued growth in the industry, the remaining carriers in the
late 1980s invested heavily in new equipment. The world’s largest producer
of jet aircraft, Boeing, reported record sales, and the delivery lag on some
aircraft grew to many years. Carriers continued to establish new hubs in
ever smaller cities, including Dayton, Raleigh-Durham, and Kansas City.
By early 1990, however, it was apparent that the industry had overinvested
in aircraft capacity. Demand was not expanding as rapidly as expected, and
hubs at smaller airports were frequently turning out to result in more costs
than benefits to the hub carrier. As the 1990–1991 recession hit and the Gulf
War reduced even domestic air travel demand, newspapers reported that
fleets of commercial aircraft were being grounded and stored in the Mohave
desert. Many carriers went into financial distress and a number entered
Chapter 11 bankruptcy proceedings: Eastern (1989), Continental (1989),
Braniff (1989), Pan Am (1991), America West (1991), Midway (1991), and
TWA (1992). As a whole, the industry reported record losses in 1990, 1991,
and 1992.
By the time the ATPCO investigation began in 1991, airlines had developed
very sophisticated systems for setting fares, determining the number of
seats available at each price, and disseminating this information to customers, travel agents, and other airlines. On a single route, such as Minneapolis-Atlanta, a carrier was likely to have more than a dozen different
fares available at a time, and to have still more listed fares that were unavailable because no seats had been allocated to that fare category.
ATPCO is a central clearinghouse for distribution of fare change information. Each day airlines send to ATPCO information on new fares to be
added, old fares to be removed, and existing fares to be changed. At least
once a day, ATPCO produces a compilation of all industry fare change information and sends that computer file, which includes thousands of fare
changes, to a list of recipients. The list includes, among others, all of the
major airlines and all four of the computer reservation systems (CRSs) in
the United States—Sabre, Apollo, Worldspan, and System One. Each CRS
company operates a networks of computers that travel agents and airlines
use to access flight, fare, and seat availability information on virtually any
a description of market power that might result from these loyalty plans and increasing hub
concentration, see Borenstein (1992). For an alternative view, see Carlton and Bamberger (1996).
airline in the world. Thus, information sent to the CRS becomes available
to consumers, travel agents, and all other airlines by the following morning.
To follow the ATPCO case, it is important to understand the information that is transmitted by ATPCO. Fare information that airlines submit to
ATPCO includes a fare basis code (a “name” of the fare), the origin and
destination airports, the price, first and last ticket dates, first and last travel
dates, and any restrictions on the use of the fare (e.g., advance purchase;
minimum stay; blackout dates; type of consumer who can buy it, such as
clergy; or a specific routing or set of flights to which the fare applies). First
and last ticket dates indicate when the carrier or a travel agent can sell tickets on the fare, while first and last travel dates indicate the range of dates on
which travel can occur under the fare. By setting a future first ticket date on
a new fare or setting a future last ticket date on an existing fare, a carrier
could announce a fare increase, but delay its implementation until a specific
future date. This ability to pre-announce price increases became a central
focus of the investigation.
Ticket and travel dates, and all restrictions, are submitted as a “footnote” to the fare. Each footnote has a “footnote designator,” which is a
name for the footnote. Footnotes do not follow a numerical order, but are
simply given footnote designators by the submitting airline. The use of fare
basis codes and footnote designators as possible modes of communication
also became a focus of the investigation.
While the ATPCO case involved a complex set of institutions and markets,
one of the issues at its core was quite basic: When a small number of firms
selling a homogeneous good can monitor one another’s prices and respond
to changes almost immediately, what is the likely outcome? In such a case,
collusive pricing can result even without any sort of explicit communication among the firms. Acting unilaterally, each firm recognizes that price
cuts will be matched immediately, so cutting price makes sense only if the
firm would prefer an equilibrium in which all firms charged the new lower
price. This greatly reduces the incentive to compete on price.
Gertner (1994) explored the outcome in such a market when firms
have different costs and capacity constraints. His work, which is motivated
by and refers frequently to the airline industry, concludes that if firms are
not too different, the outcome in immediate-response markets will still be
close to the collusive outcome and the price will be dictated by the firm that
prefers the lowest price. This occurs because higher-cost firms have nothing
to offer a low-cost firm in return for its agreeing to a price above its own
profit-maximizing levels. Of course, if the firm that prefers the lowest price
Case 9:
The Airline Tariff Publishing Case (1994)
differs across markets, there may well be room for trades in which each firm
agrees to a higher price than it would like in one market in return for increasing price closer to its preferred level in another market. Gertner also
finds that if firms differ sufficiently in costs or other attributes, one firm may
be able to sustain a lower price than others with none wanting to change its
price given the prices charged by others. This result relies on the lower-cost
firms’ having a capacity constraint. In such a case, the higher-cost firms are
better off allowing the low-cost firm to fill its capacity and then selling to
the remaining demand than matching the price of the lower-cost firm and
gaining a higher market share. Thus, even though the airlines differed in
costs and other attributes, the ability to monitor one another’s prices closely
and respond very quickly could still result in prices well above the competitive level.
This line of economic research, however, has a mixed message for antitrust. On the one hand, low-cost monitoring and quick response raise concern that prices will end up at supracompetitive levels and will harm consumers. On the other hand, this may happen without any further facilitating
circumstances—that is, without any actions that are clearly in violation of
antitrust laws. It is not an antitrust violation for a firm unilaterally to charge
high prices. Not only does such a circumstance present a dilemma for the
prosecution of an antitrust case, it also makes it difficult to devise a remedy
to the situation. Neither “charge lower prices” nor “stop responding to the
actions of other firms” are realistic remedies under the antitrust laws (though
the former is the basis for much of economic regulation).
On December 21, 1992, the U.S. Department of Justice filed antitrust
charges against ATPCO and eight major airlines.2 The complaint charged
that the airlines, through ATPCO, had colluded to raise price and restrict
competition in the airline industry. The Justice Department argued that the
airlines had carried on detailed conversations and negotiations over prices
through ATPCO. It pointed to numerous instances in which one carrier on a
route had announced a fare increase to take effect a number of weeks in the
future. Other carriers had then announced increases on the same route,
though possibly to a different fare level. In many cases cited, the airlines
had iterated back and forth until they reached a point where they were announcing the same fare increase to take effect on the same date. In cases
where one airline did not announce that it would post the same fare increase
as the others, the increase generally did not take place. In such situations it
was common for carriers to “roll” their fare increases—that is, to move the
airlines were Alaska, American, Continental, Delta, Northwest, Trans World, United, and
effective date further into the future, in order to give the carrier that had not
announced a matching fare increase more time to do so.
The DOJ garnered this information simply from the records of the
ATPCO. It also had documents from each airline’s daily internal fare change
reports, which included phrases of the nature “we are waiting to see if [carrier name] is going to go along with our proposed increase,” “we are abandoning the increase on [city1]-[city2] because [carrier name] has not
matched,” and “[carrier name] is now on board for the [date] increase to [fare
level] on [city1]-[city2].” The DOJ argued that the announcement of fares
that are to take effect at a later date allowed the airlines to negotiate over
prices without ever offering those prices to the public. While none of the
announcements was binding, such “cheap talk” can still aid collusive
The DOJ’s case also was based on patterns of multimarket coordination that it claimed to have identified. The complaint argued that the carriers were using fare basis codes and footnote designators to communicate to
other airlines linkages between fares on different routes. A typical example
of the allegation went something like this: Say that airline A1 has a hub at
city C1 from which it serves a route to city C3 with nonstop flights, as illustrated in Figure 9-1. Airline A2 has a hub at C2, which is between C1 and
C3. Airline A2 is offering a relatively low fare in the C1–C3 market with
service that requires a plane change at C2. This low fare is siphoning customers from the nonstop service that A1 offers on the route. A1 would like
A2 to raise its fare on the C1–C3 route.
If that were the whole story, however, A1 would not have much ability
to bribe or coerce A2. However, A2 serves C2–C4 with nonstop service, and
A1 offers change-of-plane service on that route over its hub at C1—exactly
the reverse of the previous situation. A1 could strike a deal with A2 in
which each carrier agrees not to undercut the other’s nonstop service with
its own fares that require a plane change at its own hub.
The DOJ argued that in such situations the ATPCO system of fare basis
codes and footnote designators offered the sort of sophisticated communication necessary to spell out and agree upon such a deal. Here’s one way the
DOJ said it would work: A1 would institute a new fare on C2–C4 that undercut A2’s fare on that route, and A1 would give this new fare the same or
a similar fare basis code as A2 was using for the fare A1 was unhappy with
on C1–C3, thus signaling to A2 the connection between the two fares. A1
would then put a short last-ticket date on this new fare, indicating that it
would be available for only, say, two weeks. It would also put in a fare on
the C2–C4 route that matched A2’s current fare and would give that fare a
first-ticket date that was the same as its last-ticket date for the cheaper fare.
A1 would then wait to see if A2 got the message. If it did, A2 would put a
last-ticket date on its fare on C1–C3 that was the same as the last-ticket date
Farrell and Rabin (1996) for a thorough discussion of the effects of cheap talk on collusion.
Case 9:
The Airline Tariff Publishing Case (1994)
FIGURE 9-1 Carriers with Overlapping Networks.
A1 had put on its cheap C2–C4 fare and would add a new fare that matched
A1’s fare on C1–C3 and had the same date for its first-ticket date. If that
happened, then two weeks hence each carrier, without further action, would
raise its fare on the other’s nonstop route so that it was no longer undercutting the nonstop route with change-of-plane service.
If A2 did not get the message or respond in the way that A1 wished, A1
could roll forward its last-ticket date on its cheap C2–C4 route. By refiling
the fare with a different last-ticket date, A1 could also make sure that this
fare again showed up on A2’s daily list of new fares, just in case A2 had
overlooked it the previous time.
The DOJ argued that the combination of future first-ticket dates and
fare basis codes or footnote designators that allowed an airline to highlight
a link between two fares on different routes made it much easier than it
would otherwise be for two airlines to “negotiate” over fares on different
routes. With these facilitating devices, the Department asserted, the airlines
could make clear the “trades” they were offering: raising price on one route
in return for a rival raising price on another route.
While under Section 1 of the Sherman Act blatant price fixing has been
found to be per se illegal, in reality there are many cases that do not fit that
mold. Often, as in the ATPCO case, the action under scrutiny is not secret
meetings of executives in smoke-filled rooms. In this case, no face-to-face
meetings were alleged. Rather, the airlines were accused of making very
frequent statements that amounted to a negotiation over price. The statements were also made in public insofar as travel agents and others with ac239
cess to a CRS could follow the rounds of announced future price changes.
The basis of the Supreme Court’s view that blatant price fixing is per se illegal is that there is no credible argument that such behavior could benefit
consumers or competition. The airlines asserted that there was a clear argument that the actions at issue could benefit consumers, so that any examination should be under a rule-of-reason standard.4
The airlines responded to the specific DOJ charges by pointing out that
all firms price in response to the actions of their competitors. They argued
that each carrier was acting in its own independent best interest when it
raised price. It would be unrealistic to think that a carrier would set its fares
without considering the response that they could anticipate from other airlines. Once it was recognized that it is legitimate for an airline to consider
the likely response from it competitors, the airlines argued that the DOJ allegations were indistinguishable from competitive behavior. A carrier probably would not want to cut price if all its competitors would matc …
Purchase answer to see full