This is the accessible text file for GAO report number GAO-11-733
entitled 'Airline Passenger Protections: More Data and Analysis Needed
to Understand Effects of Flight Delays' which was released on
September 14, 2011.
This text file was formatted by the U.S. Government Accountability
Office (GAO) to be accessible to users with visual impairments, as
part of a longer term project to improve GAO products' accessibility.
Every attempt has been made to maintain the structural and data
integrity of the original printed product. Accessibility features,
such as text descriptions of tables, consecutively numbered footnotes
placed at the end of the file, and the text of agency comment letters,
are provided but may not exactly duplicate the presentation or format
of the printed version. The portable document format (PDF) file is an
exact electronic replica of the printed version. We welcome your
feedback. Please E-mail your comments regarding the contents or
accessibility features of this document to Webmaster@gao.gov.
This is a work of the U.S. government and is not subject to copyright
protection in the United States. It may be reproduced and distributed
in its entirety without further permission from GAO. Because this work
may contain copyrighted images or other material, permission from the
copyright holder may be necessary if you wish to reproduce this
material separately.
United States Government Accountability Office:
GAO:
Report to Congressional Requesters:
September 2011:
Airline Passenger Protections:
More Data and Analysis Needed to Understand Effects of Flight Delays:
GAO-11-733:
GAO Highlights:
Highlights of GAO-11-733, a report to congressional requesters.
Why GAO Did This Study:
Flight delays and cancellations are disruptive and costly for
passengers, airlines, and the economy. Long tarmac delays have created
hardships for some passengers. To enhance passenger protections in the
event of flight disruptions, the U.S. Department of Transportation
(DOT) recently introduced passenger protection regulations, including
a rule that took effect in April 2010 designed to prevent tarmac
delays more than 3 hours (the tarmac delay rule), as well as other
efforts to improve passenger welfare. As requested, this report
addresses (1) whether flight delays and cancellations differ by
community size; (2) how DOT’s tarmac delay rule has affected
passengers and airlines; and (3) how passenger protection requirements
in the United States, Canada, and the European Union (EU) affect
passengers and airlines. GAO analyzed DOT data, including through the
use of regression models, as well as data from FlightStats, a private
source of flight performance information. GAO also reviewed documents
and interviewed government, airline, and consumer group officials in
the United States, Canada, and the EU.
What GAO Found:
Airports in rural communities have higher rates of delays and
cancellations than airports in larger communities, but DOT data
provide an incomplete picture of this difference. DOT’s data include
flights operated by the largest airlines, representing about 70
percent of all scheduled flights. GAO analysis of FlightStats data,
representing about 98 percent of all scheduled flights, show more
substantial differences in flight performance trends by community size
than DOT data. DOT has historically not collected data from smaller
airlines because of the burden it could impose on these airlines, but
without this information, DOT cannot fully achieve the purpose of
providing consumers with information on airlines’ quality of service.
DOT’s tarmac delay rule has nearly eliminated tarmac delays of more
than 3 hours (180 minutes), declining from 693 to 20 incidents in the
12 months following the introduction of the rule in April 2010. While
this has reduced the hardship of long on-board delays for some
passengers, GAO analysis suggests the rule is also correlated with a
greater likelihood of flight cancellations. Such cancellations can
lead to long overall passenger travel times. Airlines and other
aviation stakeholders maintain that the tarmac delay rule has changed
airline decision-making in ways that could make cancellations more
likely. To test this claim, GAO developed two regression models, which
controlled for a variety of factors that can cause cancellations and
measured whether the time period following the imposition of the
tarmac delay rule is correlated with an increase in cancellations. The
two models assessed flights canceled before and after leaving the
gate, for the same 5 months (May through September) in 2009 and 2010.
In both cases, GAO found that there was an increased likelihood of
cancellation in 2010 compared to 2009 (see table).
Table: Percent Difference in Likelihood of Flight Cancellation
Time on tarmac: Before taxi out (at gate);
Increased likelihood of cancellation in 2010 compared to 2009: 24
percent more likely.
Time on tarmac: 1–60 minutes;
Increased likelihood of cancellation in 2010 compared to 2009: 31
percent more likely.
Time on tarmac: 61–120 minutes;
Increased likelihood of cancellation in 2010 compared to 2009: More
than twice as likely (214 percent).
Time on tarmac: 121–180 minutes;
Increased likelihood of cancellation in 2010 compared to 2009: More
than 3 times as likely (359 percent).
Source: GAO analysis of DOT data.
[End of table]
EU requirements provide airline passengers with more extensive
protections, such as care and compensation, for flight delays,
cancellations, and denied boardings than do U.S. or Canadian
requirements. But these protections may also increase costs for
airlines and passengers. For example, some airline officials in the
United States and the EU told GAO that increases in the amount of
denied boarding compensation has increased their overall costs.
Additionally, enhanced passenger protections, such as those in the EU,
can create enforcement challenges if regulations are unclear or not
universally enforced.
What GAO Recommends:
GAO recommends that DOT collect and publicize more comprehensive data
on airlines’ on-time performance and assess the full range of the
tarmac delay rule’s costs and benefits and, if warranted, refine the
rule’s requirements and implementation. DOT did not comment directly
on the recommendations, but indicated that it would soon begin a study
of the effect of the tarmac delay rule.
View [hyperlink, http://www.gao.gov/products/GAO-11-733] or key
components. For more information, contact Susan Fleming at (202) 512-
2834 or flemings@gao.gov.
[End of section]
Contents:
Letter:
Background:
Airports in Rural Communities Have Higher Rates of Delays,
Cancellations, and Diversions than Larger Communities, but DOT Data
Provide an Incomplete Picture of These Trends:
New Rule Has Nearly Eliminated Long Tarmac Delays but is Correlated
with an Increase in the Likelihood of Flight Cancellations:
EU Requirements Provide More Comprehensive Passenger Protections, but
May Also Increase Costs for Airlines and Passengers:
Conclusions:
Recommendations for Executive Action:
Agency Comments and Our Evaluation:
Appendix I: Objectives, Scope, and Methodology:
Appendix II: Delay, Cancellation, and Diversion Trends:
Appendix III: Tarmac Delay Trends Since 2004:
Appendix IV: Tarmac Delay Trends since October 2008:
Appendix V: Tarmac Delay Logistic Regression Analysis:
Appendix VI: GAO Contact and Staff Acknowledgments:
Tables:
Table 1: Cancellations by Stage of Flight, May-September 2009 and 2010:
Table 2: Percent Difference in Likelihood of Flight Cancellation after
Implementation of Tarmac Delay Rule:
Table 3: Comparison of Governmental Requirements in the United States,
Canada, and the EU of Benefits Provided to Passengers by Airlines
during Flight Disruptions:
Table 4: Comparison of Contract of Benefits That May be Provided Under
Domestic Carriage Provisions for Flight Delays and Cancellations for
Selected U.S. Airlines, as of June 2011:
Table 5: U.S. Stakeholders Interviewed:
Table 6: Canadian Stakeholders Interviewed:
Table 7: European Stakeholders Interviewed:
Table 8: Total Number of Tarmac Delays of More than 3 Hours by Stage
of Operation, January 2004-September 2010:
Table 9: Total Number of Tarmac Delays of More than 3 Hours at 35
Airports and Systemwide, January 2004-September 2010:
Table 10: Total Number of Tarmac Delays of More than 3 Hours by Day of
the Week, January 2004-September 2010:
Table 11: Total Number of Tarmac Delays of More than 3 Hours by Hour,
January 2004-September 2010:
Table 12: Tarmac Delays of More than 3 Hours, October 2008-December
2010:
Table 13: Tarmac Delays of More than 3 Hours, May 2010-April 2011:
Table 14: Unadjusted Odds Ratio for Tarmac Cancellation, 2010 versus
2009:
Table 15: Unadjusted Odds Ratio for Tarmac Cancellation, 2010 versus
2009, by Time Spent on Tarmac:
Table 16: Unadjusted Odds Ratio for Gate Cancellation, 2010 versus
2009:
Table 17: Variables and Data Sources:
Table 18: Odds Ratios Estimates for Variables of Interest in Tarmac-
Cancellation Model, 2010 versus 2009:
Table 19: Logistic Regression Results for Tarmac-Cancellation Model
and Other Independent Variables:
Table 20: Logistic Regression Results for Gate-Cancellation Model and
Other Independent Variables:
Figures:
Figure 1: Percentages of Delayed, Canceled, and Diverted Arrival
Flights Systemwide, Calendar Years 2004-2010:
Figure 2: Percentage of Canceled or Diverted Departures by Community
Size in 2010:
Figure 3: Percentage of Delayed Arrivals by Community Size and
Systemwide in 2010:
Figure 4: Percentage of Delayed, Canceled, and Diverted Arrivals by
Community Size in 2010 for DOT and FlightStats Data:
Figure 5: Examples of Sources of Delay by Community Size:
Figure 6: Tarmac Delays of More than 3 Hours during Taxi-Out and Taxi-
In by Year, January 2004-September 2010:
Figure 7: Total Tarmac Delays of More than 3 Hours (180 minutes) by
Minutes of Delay, January 2004-September 2010:
Figure 8: Total Flights on the Tarmac of More than 1 Hour During Taxi-
Out, May-September 2009 and 2010:
Figure 9: Domestic Airlines' Decision-Making Timeline for Flights
Experiencing a Taxi-Out Delay on the Airport Tarmac:
Figure 10: Voluntary and Involuntary Denied Boardings in the United
States as a Percentage of Total Passengers, Calendar Years 2004-2010:
Figure 11: Airline Passenger Complaint Process in the United States,
Canada, and the EU:
Figure 12: Percentage of Departure Flights Canceled and Diverted by
Community Size, Calendar Years 2005-2010:
Figure 13: Percentage of Arrival Flights Canceled and Diverted by
Community Size, Calendar Years 2005-2010:
Figure 14: Percentage of Delayed Departure Flights by Community Size,
Calendar Years 2006-2010:
Figure 15: Percentage of Delayed Arrival Flights by Community Size,
Calendar Years 2006-2010:
Figure 16: Airline-Reported Sources of Delay in Calendar Year 2010:
Figure 17: Airline-Reported Sources of Cancellations in Calendar Year
2010:
Figure 18: Total Number of Tarmac Delays of More than 3 Hours by
Month, January 2004-September 2010:
Figure 19: Airline-Reported Sources of Delay for all Tarmac Delays of
More than 3 Hours, January 2004-September 2010:
Figure 20: Odds of Cancellation for Flights on the Tarmac, 2010 and
2009:
Abbreviations:
ASQP: Airline Service Quality Performance:
BTS: Bureau of Transportation Statistics:
Commission: European Commission:
CTA: Canadian Transportation Agency:
DOT: U.S. Department of Transportation:
ECJ: European Court of Justice:
EU: European Union:
FAA: Federal Aviation Administration:
IG: Department of Transportation Office of Inspector General:
UK: United Kingdom:
[End of section]
United States Government Accountability Office:
Washington, DC 20548:
September 7, 2011:
The Honorable Nick J. Rahall, II:
Ranking Member:
Committee on Transportation and Infrastructure:
House of Representatives:
The Honorable Peter A. DeFazio:
Ranking Member:
Subcommittee on Highways and Transit:
Committee on Transportation and Infrastructure:
House of Representatives:
Flight disruptions, including delays, cancellations, long tarmac
delays, and denied boarding, are costly for passengers, airlines, and
the economy. In recent years, roughly a quarter of all commercial
flights have been delayed or canceled, resulting in tens of thousands
of years in total delayed travel time and billions of dollars in out-
of-pocket expenses and time lost for passengers each year.[Footnote 1]
With fewer empty seats than ever, opportunities for rebooking are
often limited, exacerbating the disruptions and associated costs.
Furthermore, these disruptions may be particularly challenging for
smaller communities that have infrequent service. In a relatively few
but widely publicized cases, long delays have occurred on the tarmac
where passengers were held on planes for hours.[Footnote 2] Passengers
also can find themselves stranded or delayed when they are denied
boarding because an airline has overbooked a flight.
In the United States, airlines are generally not required to
compensate passengers for their time or expenses when flights are
delayed or canceled. In the European Union (EU), however, a regulation
enacted[Footnote 3] in 2004 guarantees care, such as lodging and
meals, and financial compensation for passengers with disrupted travel
plans.[Footnote 4] Canada has promoted guidelines to encourage
airlines to improve passenger protections, but has not otherwise
required airlines to provide passenger care or compensation for flight
disruptions. The U.S. Department of Transportation (DOT) has
implemented some other passenger protections, including a recent
regulation designed to mitigate hardships for airline passengers
during long tarmac delays (the tarmac delay rule).[Footnote 5]
You asked us to examine the extent of passenger protections in the
United States, Canada, and the European Union (EU), as well as assess
the trends in delays and cancellations in different-sized U.S.
communities. We also assessed the effect of DOT's tarmac delay rule.
Accordingly, this report addresses the following questions: (1) Do the
trends in and reasons for flight delays and cancellations in the
United States differ for smaller and larger communities? (2) How has
DOT's tarmac delay rule affected passengers and airlines? (3) How have
the requirements and practices for protecting passengers from flight
delays, cancellations, and denied boardings in the United States,
Canada, and the EU affected passengers and airlines?
To identify and compare the trends in and reasons for flight delays
and cancellations in different-sized U.S. communities, we analyzed
data from DOT and FlightStats, a private data source from Conducive
Technology that records flight performance information. Specifically,
we compared the delay, cancellation, and diversion rates for all
scheduled flights at different-sized airports and communities from
2005 to 2010. To better understand the reason for different trends, we
reviewed a DOT Office of Inspector General report and interviewed
aviation industry experts; representatives of industry associations
and consumer groups; and representatives of the three biggest U.S.
legacy airlines and three biggest low-cost airlines, based on
passenger enplanements. To assess the extent to which the
implementation of DOT's tarmac delay rule was associated with an
increase in cancellations, we examined DOT data on tarmac delay trends
and also constructed two multivariate logistic regression models.
Incorporating data from DOT, the Federal Aviation Administration
(FAA), and the National Oceanic and Atmospheric Administration, the
regression models controlled for factors that can lead to flight
cancellations, including weather, to assess how the rule's
implementation affected the likelihood of flight cancellations. Two
aviation industry experts assessed our models' structure and provided
feedback that we incorporated into our approach as appropriate. We
also spoke with representatives of U.S. airlines, industry
associations, consumer groups, and DOT about the rule's impact on
passengers and airlines. To assess the reliability of DOT and
FlightStats data, we reviewed existing documentation related to the
data sources and interviewed knowledgeable officials at DOT and
FlightStats about the data. We determined that the data were
sufficiently reliable for the purposes of this report. To determine
how the requirements and practices for protecting passengers from
flight delays, cancellations, and denied boardings in the United
States, Canada, and the EU have affected passengers and airlines, we
examined the laws, regulations, international agreements, and
voluntary commitments governing passenger protections in the three
regions. We also reviewed the contracts of carriage for selected
airlines in the three regions and examined DOT data on denied
boardings in the United States. Furthermore, we interviewed airline,
industry association, consumer group, and government officials in all
three regions.
We conducted this performance audit from August 2010 through September
2011 in accordance with generally accepted government auditing
standards. Those standards require that we plan and perform the audit
to obtain sufficient, appropriate evidence to provide a reasonable
basis for our findings and conclusions based on our audit objectives.
We believe that the evidence obtained provides a reasonable basis for
our findings and conclusions based on our audit objectives. See
appendix I for more information on our scope and methodology.
Background:
The U.S. passenger airline industry is principally composed of legacy,
regional, and low-cost airlines. Legacy (sometimes called network)
airlines support large, complex hub-and-spoke operations with
thousands of employees and hundreds of aircraft (of various types),
with flights to domestic communities of all sizes as well as to
international destinations.[Footnote 6] Generally, regional airlines
operate smaller aircraft than legacy airlines--turboprops or regional
jets with up to 100 seats--and often operate flights marketed by a
legacy airline.[Footnote 7] Low-cost airlines generally entered the
marketplace after the U.S. airline industry was deregulated in 1978
and typically have a less extensive network and lower operating costs.
Passengers access flights offered by these various airlines in the
United States through hundreds of commercial-service airports.
[Footnote 8] Primary airports are classified on the basis of passenger
traffic as large, medium, small, and nonhub.[Footnote 9] Passenger
traffic at these airports is highly concentrated: about 70 percent of
passengers enplaned at the 29 largest airports and another 19 percent
enplaned at the 36 next largest airports in 2009, the most recent year
for which these data are available. Some of these largest airports
also face significant congestion and delay issues. As we recently
reported, seven large airports were the source of about 80 percent of
departure delays captured in FAA's Operations Network in 2009.
[Footnote 10]
The national airspace system in which these airlines and airports
operate is a complex, interconnected, and interdependent network of
systems, procedures, facilities, aircraft, airports, and people that
must work together to ensure safe and efficient operations. FAA, DOT,
airlines, and airports all affect the efficiency of national airspace
system operations. In particular, DOT and FAA set policy and operating
standards for aircraft and airports.
As we previously reported, the capacity of the aviation system to meet
the demand of aviation system users is both variable and subject to a
number of interrelated factors.[Footnote 11] The capacity of the
aviation system is affected not only by airports' infrastructure,
including runways and terminal gates, but also by weather conditions
and air traffic control that can, at any given time, result in
disruptions and variation in available airport and system capacity.
For example, some airports have parallel runways that can be used
simultaneously in good weather but are too close together for
simultaneous operations in bad weather. In severe weather, airports
can close, resulting in aircraft being grounded both at the closed
airport and at other airports where aircraft cannot depart for the
closed airport. The number of aircraft that can be safely accommodated
in a given portion of airspace further affects capacity. If too many
aircraft are trying to use the same airspace, some may be delayed on
the ground or en route. For example, delays often occur in the New
York City area because air traffic is so heavy, with three major
airports located within 100 miles of each other. Airlines' scheduling
and business practices can also exacerbate airport congestion and
delays. For instance, some airline business models rely on tight
turnaround times between flights, which can increase the likelihood of
delays for flights scheduled later in the day. Additionally, airlines
sometimes schedule flights during certain periods to accommodate
passenger demand without considering an airport's available capacity.
When flights are disrupted--whether caused by reductions in system
capacity (such as during bad weather) or by internal factors (such as
aircraft mechanical problems or crew shortages)--airlines make trade-
offs between long delays and cancellations, though they generally try
to avoid canceling flights.[Footnote 12] In doing so, they attempt to
minimize disruptions to their network and passengers. For example,
when bad weather reduces airport capacity and fewer flights can take
off or land, airlines must decide how to ration their traffic. They
can hold to their schedule, recognizing that some flights may
experience long delays, or they can cancel some flights to avoid long
delays for the remaining flights. How airlines manage such trade-offs
depends on their business models and the circumstances of each
situation.
As we recently reported, flight delays and cancellations have declined
since 2007, largely because airlines have scheduled fewer flights
during the economic downturn.[Footnote 13] From 2007 through 2010, the
portion of flights that were delayed--that is, arrived at least 15
minutes later than scheduled--or were canceled or diverted decreased
by 6 percentage points, according to DOT data (see figure 1). Indeed,
cancellations rates also peaked in 2007 at 2.16 percent of all
flights, before declining to 1.39 percent in 2009 and 1.76 percent in
2010. Nevertheless, as we previously reported, airports still
experience and contribute substantial delays to the system.[Footnote
14]
Figure 1: Percentages of Delayed, Canceled, and Diverted Arrival
Flights Systemwide, Calendar Years 2004-2010:
[Refer to PDF for image: stacked vertical bar graph]
Year: 2004;
Percentage of arrival flights that are delayed more than 15 minutes:
19.95%;
Percentage of arrival flights that are canceled: 1.79%;
Percentage of arrival flights that are diverted: 0.19%.
Year: 2005;
Percentage of arrival flights that are delayed more than 15 minutes:
20.52%;
Percentage of arrival flights that are canceled: 1.88%;
Percentage of arrival flights that are diverted: 0.2%.
Year: 2006;
Percentage of arrival flights that are delayed more than 15 minutes:
22.61%;
Percentage of arrival flights that are canceled: 1.71%;
Percentage of arrival flights that are diverted: 0.23%.
Year: 2007;
Percentage of arrival flights that are delayed more than 15 minutes:
24.18%;
Percentage of arrival flights that are canceled: 2.16%;
Percentage of arrival flights that are diverted: 0.23%.
Year: 2008;
Percentage of arrival flights that are delayed more than 15 minutes:
21.74%;
Percentage of arrival flights that are canceled: 1.96%;
Percentage of arrival flights that are diverted: 0.25%.
Year: 2009;
Percentage of arrival flights that are delayed more than 15 minutes:
18.88%;
Percentage of arrival flights that are canceled: 1.39%;
Percentage of arrival flights that are diverted: 0.24%.
Year: 2010;
Percentage of arrival flights that are delayed more than 15 minutes:
18.21%;
Percentage of arrival flights that are canceled: 1.76%;
Percentage of arrival flights that are diverted: 0.24%.
Source: GAO analysis of DOT data.
[End of figure]
In recent decades the airline industry's earnings have been extremely
volatile. Despite some periods of strong growth and increased
earnings, airlines have at times suffered such substantial financial
distress that some have filed for bankruptcy. According to a recent
FAA-sponsored research study, U.S. passenger airlines lost more than
$60 billion from 2000 through 2008 on revenues of just more than $1
trillion.[Footnote 15] An inefficient air transportation system that
contributes to flight delays and cancellations increases airline costs
and reduces demand for air travel, compounding these financial
challenges.
Airline industry financial pressures have led airlines to change
certain business practices in order to cut costs and enhance revenue.
For example, airlines have adjusted their capacity to increase
passenger load-factors (i.e., the proportion of available seats filled
with passengers). As a result, a large number of cancellations by an
airline cannot be absorbed easily into later flights and,
increasingly, airlines will not rebook passengers on other airlines'
flights because of the costs involved. Passengers on canceled flights
can then face long overall trip delays. In addition, for decades
airlines have sought to reduce the revenue losses associated with
passengers who do not show up for flights by accepting reservations
for more passengers than they have seats. Because the number of no-
shows is not entirely predictable, there is an element of risk in
overbooking flights. If too many reservations are accepted and more
passengers show up at departure time than the aircraft can carry, the
airline must deal with the costs and customer service issues that
arise when some customers are denied boarding. On the other hand, if
the airline does not accept enough reservations for the flight and the
number of no-shows is greater than expected, the airline loses revenue
from empty seats that could otherwise have been occupied and some
passengers are denied the opportunity to book their first-choice
flight even though that flight could have accommodated them. DOT has
long required airlines to solicit and compensate volunteers on
oversold flights before anyone is bumped involuntarily and has also
mandated financial compensation for passengers who are involuntarily
denied boarding because their flights were oversold[Footnote 16].:
Passenger complaints about delays, cancellations, and denied
boardings, including complaints about being held in an aircraft for
many hours while awaiting takeoff, have led Congress to consider
stronger passenger protections. For instance, after hundreds of
passengers were stuck in planes on snowbound Detroit runways for more
than 8 hours in January 1999, both the House of Representatives and
Senate conducted hearings on airlines' treatment of air travelers and
considered whether to enact a "passenger bill of rights." The Air
Transport Association and its member airlines maintained that they
should have an opportunity to improve their customer service without
legislation and executed an Airline Customer Service Commitment on
June 17, 1999, in which each of the member airlines agreed to prepare
a customer service plan.[Footnote 17] In 2000, AIR-21 mandated a
review by the DOT Office of Inspector General (IG) of the extent to
which each airline met all provisions of its customer service plan.
[Footnote 18] In its 2001 report, the IG found that, overall, airlines
were making progress toward meeting their plan provisions and that
their efforts had been a plus for air travelers. However, the IG also
reported "significant shortfalls in reliable and timely communication
with passengers by the airlines about flight delays and
cancellations." Furthermore, the IG found that the airlines had not
directly addressed the root cause of customer dissatisfaction--flight
delays and cancellations--and had not indicated how they planned to
remedy these problems in areas under their control. Other passenger
rights bills were introduced in Congress in 2001, 2007, 2009, and
2011.[Footnote 19] These bills were also designed to establish and
enhance airline passenger protections, and the 2007, 2009, and 2011
bills explicitly limited tarmac delays to 3 hours. However, the 2001,
2007, and 2009 bills were not enacted, and the 2011 bill has not yet
been enacted during this Congress.
In recent years, DOT has adopted rules to enhance passenger
protections. First, in 2008, it amended its overbooking rule to
increase the required compensation for involuntarily denied boarding,
among other things.[Footnote 20] Second, in late 2009, after a lengthy
rulemaking and a task force report on long tarmac delays, DOT issued
its first "Enhancing Airline Passenger Protections" rule.[Footnote 21]
The final rule, in effect since April 29, 2010, requires certain U.S.
airlines to develop and implement a contingency plan for lengthy
tarmac delays, including an assurance that, for domestic flights,
[Footnote 22] the airline will not allow a tarmac delay to exceed 3
hours unless the pilot-in-command determines that there is a safety-
related or security-related impediment to deplaning passengers, or
that air traffic control has advised the pilot-in-command that
deplaning would significantly disrupt airport operations.[Footnote 23]
The airlines' contingency plans must also include an assurance that
adequate food and potable water will be provided no later than 2 hours
after the aircraft leaves the gate (or touches down, in the case of an
arrival), unless the pilot-in-command determines that safety or
security considerations preclude such service. Failure to comply with
these rules could be considered an unfair or deceptive practice
[Footnote 24] and may subject the airline to enforcement action and a
fine of up to $27,500 per violation.[Footnote 25] Furthermore, under
the rule, the holding out--advertising or operating--of any
chronically delayed flight is considered an unfair and deceptive
practice and an unfair method of competition. The rule also requires a
variety of other actions on the part of airlines to protect and better
inform passengers.
In April 2011, DOT issued its second "Enhancing Airline Passenger
Protections" rule.[Footnote 26] This rule--which partially went into
effect in August 2011 and will be fully implemented in January 2012--
requires airlines, among other things, to reimburse passengers for
baggage fees if their bags are lost, provide consumers with greater
compensation for involuntarily denied boarding, and disclose all fees
for optional services.[Footnote 27] The new rule also expands the
existing tarmac delay rule to cover all U.S. large-, medium-, small-,
and nonhub airports as well as foreign airlines' operations at those
U.S. airports, and establishes a 4-hour time limit on tarmac delays
for international flights of U.S. and foreign airlines, subject to
safety, security, and air traffic control exceptions.
Like the United States, Canada, and EU also have laws, regulations,
and guidance governing consumer protection for air travelers,
including airline responsibilities to passengers when flight plans are
disrupted. U.S., Canadian, and EU airlines generally must adhere to
the passenger protection requirements of the region from which they
are departing.[Footnote 28] Airlines in all three regions also have
contracts of carriage in which they may provide for passenger care,
compensation, or both in the event of a flight disruption.[Footnote
29] Thus, when provided for in law or in a contract of carriage,
[Footnote 30] passengers may be entitled to assistance, compensation,
or both from their airline when a flight delay, cancellation, or
denied boarding occurs.[Footnote 31] For example, under certain
circumstances, some airlines offer food and beverage vouchers during
flight disruptions. Finally, international standards and agreements
also govern the rights of airline passengers, but only on
international flights. Notably, the Montreal Convention, adopted in
1999 and ratified by the United States in 2003, provides that
passengers can bring legal action against an airline for damages
associated with flight delays.[Footnote 32]
Airports in Rural Communities Have Higher Rates of Delays,
Cancellations, and Diversions than Larger Communities, but DOT Data
Provide an Incomplete Picture of These Trends:
Airports in Rural Communities Have a Higher Rate of Delays,
Cancellations, and Diversions than Larger Communities:
The percentages of flights that are canceled or diverted have in
recent years been higher to and from airports in rural communities
than large metropolitan communities, according to FlightStats data.
[Footnote 33] We categorized airports based on the population size of
their surrounding communities to assess the extent to which flight
delays, cancellations, and diversions differ by community size.
[Footnote 34] Our analysis of departure cancellation and diversion
trends, using FlightStats data for all reported flights, shows that,
since 2005, flights from airports in rural communities (communities
with less than 50,000 people) are on average about 3.5 times as likely
to be canceled or diverted as flights from airports in large
metropolitan communities.[Footnote 35] For example, in 2010,
cancellations and diversions accounted for roughly 2 percent of
flights from large, midsized, and small metropolitan communities,
compared with nearly 8 percent of flights from airports in rural
communities (see figure 2).[Footnote 36] Greater cancellation rates
for flights departing rural airports were matched by higher rates of
cancellation for flights arriving at rural airports. (See appendix II
for more information). Such cancellations and diversions can lead to
long overall delay times for passengers. According to one academic
study, the overall average delay time for passengers on canceled
flights is about 5 hours.[Footnote 37]
Figure 2: Percentage of Canceled or Diverted Departures by Community
Size in 2010:
[Refer to PDF for image: stacked vertical bar graph]
Community size: Large metropolitan;
Percentage of flights canceled: 1.79%;
Percentage of flights diverted: 0.16%.
Community size: Midsized metropolitan;
Percentage of flights canceled: 2.05%;
Percentage of flights diverted: 0.21%.
Community size: Small metropolitan;
Percentage of flights canceled: 2.37%;
Percentage of flights diverted: 0.17%.
Community size: Rural;
Percentage of flights canceled: 7.01%;
Percentage of flights diverted: 0.95%.
Source: GAO analysis of FlightStats data.
Note: Community sizes are as follows: large metropolitan (greater than
or equal to 1 million), midsized metropolitan (250,000 to 999,999),
small metropolitan (50,000 to 249,999), and rural (fewer than 50,000).
These data reflect the vast majority of scheduled passenger flights.
In 2010, for example, the data included performance information for
about 98 percent of U.S. scheduled passenger flights from primary
airports excluding airports in U.S. territories.
[End of figure]
The percentage of delayed arrivals--that is, flights that arrived at
least 15 minutes late to their destination--has in recent years been
higher at airports in rural communities than airports in small,
medium, and large metropolitan sized communities, according to
FlightStats data, though the difference is not as substantial for
delays as it is with canceled or diverted flights. As shown in figure
3, while delays occurred in 19.6 percent of all reported flights
systemwide in 2010, delays occurred in 21.7 percent of flights to
airports in rural communities, a 2.1 percentage points or about 11
percent difference in the occurrence of delay. See appendix II for
arrival and departure delay trends since 2005. Such delays can lead to
longer overall trip times for passengers. According to academic
research, the overall average delay time for passengers on a delayed
flight is 37 minutes.[Footnote 38]
Figure 3: Percentage of Delayed Arrivals by Community Size and
Systemwide in 2010:
[Refer to PDF for image: combined vertical bar and line graph]
Community size: Large metropolitan;
Percentage of delayed arrivals: 19.59%;
Percentage of delayed arrivals systemwide: 19.64%.
Community size: Midsized metropolitan;
Percentage of delayed arrivals: 20.05%;
Percentage of delayed arrivals systemwide: 19.64%.
Community size: Small metropolitan;
Percentage of delayed arrivals: 18.72%;
Percentage of delayed arrivals systemwide: 19.64%.
Community size: Rural;
Percentage of delayed arrivals: 21.65%;
Percentage of delayed arrivals systemwide: 19.64%.
Source: GAO analysis of FlightStats data.
Note: Community sizes are as follows: large metropolitan (greater than
or equal to 1 million), midsized metropolitan (250,000 to 999,999),
small metropolitan (50,000 to 249,999), and rural (fewer than 50,000).
These data reflect the vast majority of scheduled passenger flights.
In 2010, for example, the data included performance information for
about 98 percent of U.S. scheduled passenger flights from primary
airports, excluding airports in U.S. territories.
[End of figure]
DOT's Data Provide an Incomplete Picture of Flight Delay,
Cancellation, and Diversion Trends:
DOT's data on flight performance do not show similar disparities
between rural and other airports as do the FlightStats data because
many flights captured by FlightStats are not required to be reported
to DOT. As a result, DOT data provides an incomplete picture of delay,
cancellation, and diversion trends. DOT requires airlines with at
least 1 percent of total domestic scheduled passenger service revenue
to report flight performance data for flights they operate to and from
reportable airports.[Footnote 39] In 2010, 18 airlines reported data,
which accounted for 22 percent of all commercial airlines, 69 percent
of all scheduled flights, and 85 percent of all passengers. The
approximately 31 percent of flights not in DOT's data are scheduled
flights operated by airlines that are not required to report. Some of
these flights are those operated by regional airlines for legacy
airlines, and in general the airlines not required to report to DOT
are small and tend to provide much of the service to airports in small
metropolitan and rural communities. Therefore, DOT's data do not
provide a complete picture of flight performance, especially at
airports in smaller communities. For example, according to DOT's data
for 2010, delays, cancellations, and diversions occurred in 19.6
percent of flights to airports in rural communities and 20.2 percent
of flights to airports in large metropolitan communities. However,
FlightStats' more extensive data show a bigger difference by community
size, with 27.3 percent of flights to airports in rural communities
delayed, canceled, or diverted, compared with 21.6 percent of flights
to airports in large metropolitan communities in 2010 (see figure 4).
[Footnote 40]
Figure 4: Percentage of Delayed, Canceled, and Diverted Arrivals by
Community Size in 2010 for DOT and FlightStats Data:
[Refer to PDF for image: vertical bar graph]
Community size: Large metropolitan;
DOT percentage delayed, canceled, and diverted: 20.19%;
FlightStats percentage delayed, canceled, and diverted: 21.56%.
Community size: Midsized metropolitan;
DOT percentage delayed, canceled, and diverted: 20.85%;
FlightStats percentage delayed, canceled, and diverted: 21.92%.
Community size: Small metropolitan;
DOT percentage delayed, canceled, and diverted: 18.71%;
FlightStats percentage delayed, canceled, and diverted: 20.84%.
Community size: Rural;
DOT percentage delayed, canceled, and diverted: 19.63%;
FlightStats percentage delayed, canceled, and diverted: 27.31%.
Sources: GAO analysis of DOT and FlightStats data.
Note: Community sizes are as follows: large metropolitan (greater than
or equal to 1 million), midsized metropolitan (250,000 to 999,999),
small metropolitan (50,000 to 249,999), and rural (fewer than 50,000).
These data reflect the vast majority of scheduled passenger flights.
In 2010, for example, the data included performance information for
about 98 percent of U.S. scheduled passenger flights from primary
airports, excluding airports in U.S. territories.
[End of figure]
Our analysis of FlightStats' and DOT's delay and cancellation data
suggests that airlines not required to report flight performance
information to DOT have higher delay, cancellation, and diversion
rates than airlines that are required to report. As figure 4 shows,
delay, cancellation, and diversion rates are higher, regardless of
community size, when using FlightStats data, as opposed to DOT data.
FlightStats data includes a greater percentage of all flights than
DOT's data, 98 versus 77 percent, and data trends are similar for
similar flights within each data set. Therefore, airlines not required
to report to DOT likely account for greater rates of delays,
cancellations, and diversions. According to FlightStats data, in 2010,
airlines that were required to report to DOT had lower delay,
cancellation, and diversion rates on average than the 20 largest
airlines not required to report to DOT.[Footnote 41]
This information corroborates what we were told by various
stakeholders, including airline officials and aviation researchers.
According to stakeholders that we spoke with, these differences may
exist for multiple reasons (see figure 5). For example, airlines
operating from smaller airports may have limitations that affect their
on-time performance, such as their use of smaller aircraft, which can
face greater restrictions during certain weather events. As the DOT
Office of Inspector General has reported, airports in rural
communities may have higher delay and cancellation rates because the
airlines serving them may have more limited resources, such as spare
aircraft and crew, at those airports than at metropolitan airports.
[Footnote 42] Furthermore, when FAA institutes traffic management
initiatives to meter air traffic to and from airports, airlines must
choose which of their flights to delay or cancel. According to
previous academic research and aviation stakeholders that we spoke
with, airlines usually prioritize flights by revenue, number of
passengers, aircraft size, route distance, and competition, or flight
frequency.[Footnote 43] In cases where marketing airlines control
operational decisions for their regional partners, the marketing
airlines may disproportionately delay or cancel flights operated by
their smaller, regional partners because those flights tend to be
operated with smaller aircraft with fewer passengers and shorter
routes with less competition from other airlines. Our analysis of two
legacy airlines shows that their regional partners generally have
worse on-time performance. According to FlightStats data, in 2010, two
large legacy airlines canceled 1.96 percent and 1.51 percent of their
own flights, compared with 2.46 percent and 2.43 percent of the
flights regional airlines operated for them. While cancellations to
smaller communities may inconvenience a relatively small number of
passengers, they may result in long trip delays if those smaller
communities have infrequent service. See appendix II for more
information on sources of delay and cancellations.
Figure 5: Examples of Sources of Delay by Community Size:
[Refer to PDF for image: illustrated table]
Factors that may affect delay: Aircraft type;
Details: Smaller airplanes (which generally service small airports in
rural communities) are more likely to be restricted during certain
weather events. For example, according to the DOT Office of the
Inspector General, these planes are not allowed to fly over some weather
patterns or land in some wind conditions, which may delay these planes
or force them to cancel at some airports and not others.
Factors that may affect delay: Airport size;
Details: Airports in rural communities are generally smaller and
therefore may have more air carrier delays because airlines have fewer
resources available to respond to mechanical and staff issues.
Factors that may affect delay: Airline economic resources;
Details: Airline economic decisions may also affect cancellation and
delay trends. When extenuating circumstances, such as weather or
congestion, force airlines to reduce the number of flights they can
operate, airlines must choose which flights to delay or cancel.
Airline decisions that take into consideration the number of
passengers affected, access to alternative routings, or other factors
may affect smaller communities either positively or negatively,
according to the DOT Office of the Inspector General.
Factors that may affect delay: FAA traffic management initiatives;
Details: FAA traffic management initiatives that restrict traffic to
certain airports or geographic areas at certain times in response to
events such as weather or congestion are more prevalent at larger
airports.
Sources: GAO analysis of aviation research and interviews with airline
officials
[End of figure]
DOT has historically not collected flight performance information from
smaller airlines because of the burden it has perceived would be
placed on these airlines.[Footnote 44] Without this information,
though, DOT cannot provide consumers with a complete picture of flight
performance, particularly at airports in smaller communities or for
smaller airlines.[Footnote 45] More comprehensive data would provide
consumers with better information on airlines' performance. Requiring
airlines with a smaller percentage of the total domestic scheduled
passenger service revenue, or airlines that operate flights for other
airlines, to report flight performance information are two ways that
would enhance DOT's data. According to DOT officials, they have
considered reducing the reporting threshold from 1 percent of domestic
scheduled revenue to 0.5 percent to increase the percentage of flights
captured. This change, they estimate, would require an additional 12
airlines to report to DOT and increase coverage from about 85 percent
to more than 96 percent of all passengers. In its December 2009
passenger protections rule, DOT required airlines that are required to
report on-time performance data to DOT to include on-time performance
information on their Web sites for all flights for which their sites
have schedule information. In doing so, it rejected the concern that
airline publication of data from smaller code-sharing airlines on
their Web sites would be overly burdensome, and also noted that flight
performance information was necessary for consumers to make informed
decisions when selecting flights.
New Rule Has Nearly Eliminated Long Tarmac Delays but is Correlated
with an Increase in the Likelihood of Flight Cancellations:
The Number of Long Tarmac Delays Has Declined in Recent Years:
Since 2004, tarmac delays of more than 3 hours peaked in 2007, three
years before the tarmac delay rule was implemented. The decline prior
to the imposition of the rule is likely the result of a combination of
factors, including fewer flights since 2007, runway and other
improvements at some airports, as well as voluntary limits adopted by
some airlines on how long their flights can wait on the tarmac. Tarmac
delays of more than 3 hours, which occur as a plane is taxiing out of
or in to an airport gate ("taxi-out" or "taxi-in"), have historically
been relatively uncommon, accounting for less than 0.1 percent of all
reported flights, according to our analysis of DOT data (see figure
6).[Footnote 46] The vast majority, about 97 percent, of tarmac delays
of more than 3 hours occur during taxi-out (departure), rather than
during taxi-in (arrival).
Figure 6: Tarmac Delays of More than 3 Hours during Taxi-Out and Taxi-
In by Year, January 2004-September 2010:
[Refer to PDF for image: 2 vertical bar graphs]
Number of tarmac delays:
Year: 2004;
Taxi-out: 1,169;
Taxi-in: 70.
Year: 2005;
Taxi-out: 983;
Taxi-in: 32.
Year: 2006;
Taxi-out: 1,220;
Taxi-in: 58.
Year: 2007;
Taxi-out: 1,507;
Taxi-in: 30.
Year: 2008;
Taxi-out: 1,056;
Taxi-in: 17.
Year: 2009;
Taxi-out: 520;
Taxi-in: 2.
Year: 2010;
Taxi-out: 72;
Taxi-in: 4.
Source: GAO analysis of DOT data.
Notes: Beginning in October 2008, DOT required airlines to submit
tarmac delay statistics for three additional categories: flights that
are subsequently canceled or diverted or have multiple gate
departures. For consistency, we have omitted these flights from this
figure. Our analysis also excluded flights reported to DOT by Atlantic
Southeast Airlines, Hawaiian Airlines, and Comair because, according
to DOT, these airlines inconsistently reported gate returns prior to
October 2008. Specifically, some airlines misreported the gate-
departure or takeoff time resulting in an incorrect tarmac delay
calculation. Further, our analysis excluded data for any flights with
tarmac delays of more than 10 hours since there were errors in the
reporting of many such flights. Finally, our analysis did not include
data for international flights. As a result of these exclusions for
data reliability purposes, this figure likely under-reports tarmac
delays of more than 3 hours for these years. See appendix IV for more
information on recent tarmac delays.
[End of figure]
The majority of all tarmac delays of more than 3 hours (180 minutes)
since 2004 are 4 hours (240 minutes) or less (see figure 7).
Specifically, of the 6,740 tarmac delays of more than 3 hours reported
from January 2004 through September 2010, almost 83 percent (or 5,579)
were for 4 hours or less. However, given the length of some of these
delays and the inconvenience or even hardship they sometimes create
for passengers, tarmac delays have received widespread media attention
(see appendix IV for examples of tarmac delays of more than 3 hours
since October 2008, when DOT began collecting more data on such
delays).[Footnote 47]
Figure 7: Total Tarmac Delays of More than 3 Hours (180 minutes) by
Minutes of Delay, January 2004-September 2010:
[Refer to PDF for image: pie-chart]
181 to 240 minutes: 5,579;
241 to 300 minutes: 967;
301 to 360 minutes: 153;
361 minutes or longer: 41.
Source: GAO analysis of DOT data.
Notes: Tarmac delay trends, by delay time, are proportionally similar
across years. As a result, we have combined the data for years 2004
through September 2010. In addition, the average length of taxi-out
and taxi-in tarmac delays is similar, as is the proportion of those
delays that extend beyond 3 hours. Beginning in October 2008, DOT
required airlines to submit tarmac delay statistics for three
additional categories: flights that are subsequently canceled or
diverted or have multiple gate departures. For consistency, we have
omitted these flights from this figure. Our analysis also excluded
flights reported to DOT by Atlantic Southeast Airlines, Hawaiian
Airlines, and Comair because, according to DOT, these airlines
inconsistently reported gate returns prior to October 2008.
Specifically, some airlines misreported the gate-departure or takeoff
time resulting in an incorrect tarmac delay calculation. Further, our
analysis excluded data for any flights with tarmac delays of more than
10 hours since there were errors in the reporting of many such
flights. Finally, our analysis did not include data for international
flights. As a result of these exclusions for data reliability
purposes, this figure likely under-reports tarmac delays of more than
3 hours for these years. See appendix IV for more information on
recent tarmac delays.
[End of figure]
Tarmac delays of more than 3 hours are generally clustered around
certain weather events, during specific times of the year or day, and
at specific airports. For example, tarmac delays of more than 3 hours
most often occur during summer thunderstorms or winter storms, when
airport departures are halted.[Footnote 48] According to our analysis
of DOT data from January 2004 through September 2010, almost two-
thirds of all tarmac delays of more than 3 hours occurred from May
through September. Also, these tarmac delays tend to be clustered on a
select number of days. According to our analysis of DOT data, almost
74 percent of tarmac delays of more than 3 hours, from January 2004
through September 2010, occurred on about 7 percent of the days during
this time period. For example, on July 23, 2008, 113 flights were
delayed more than 3 hours on the tarmac across the national airspace
system during taxi-out.[Footnote 49] Tarmac delays also tend to occur
in the late afternoon, when summer thunderstorms are most likely, and
after delays from the morning and early afternoon are compounded. For
example, since 2004, about half of all tarmac delays of more than 3
hours occurred between 3:00 p.m. and 6:00 p.m. local time. Tarmac
delays are also most prevalent at airports that have high rates of
delays. For example, about 55 percent of tarmac delays of more than 3
hours since 2004 occurred at just seven particularly congested
airports.[Footnote 50] See appendix III for more details on these
trends.
Tarmac Delays More than 3 Hours Have Been Nearly Eliminated Since DOT
Implemented Rule:
DOT instituted new rules in 2010 in response to instances of
passengers subject to lengthy tarmac delays, among other consumer-
related problems.[Footnote 51] Since these rules took effect in April
2010, tarmac delays greater than 4 hours have been eliminated, and
tarmac delays of more than 3 hours nearly eliminated, reducing the
hardship of long on-board delays for some passengers. As mentioned
earlier, these new rules require, among other things, that covered
airlines' contingency plans provide for adequate food and water on all
flights once a flight has been on the tarmac for 2 hours, except when
safety or security preclude such services. Additionally, for domestic
flights, the rule requires that covered airlines should not remain on
the tarmac for more than 3 hours, with exceptions for safety,
security, and disruption of airport operations.[Footnote 52] Violation
of these rules can result in a $27,500 per-violation fine.[Footnote
53] Since the rule went into effect in late April 2010, tarmac delays
of more than 3 hours (180 minutes) have been nearly eliminated (see
figure 8). In the first 12 months since the rule went into effect,
airlines reported tarmac delays of more than 3 hours for 20 flights,
compared with 693 over the same period prior to the rule.[Footnote 54]
Airline consumer groups we spoke with strongly support the tarmac
delay rule instituted by DOT.
Figure 8: Total Flights on the Tarmac of More than 1 Hour During Taxi-
Out, May-September 2009 and 2010:
[Refer to PDF for image: vertical bar graph]
Number of flights on the tarmac:
Time on tarmac: 61-120 minutes;
May-September 2009: 22,364;
May-September 2010: 17,763.
Time on tarmac: 121-180 minutes;
May-September 2009: 2,391;
May-September 2010: 1,257.
Time on tarmac: 181-240 minutes;
May-September 2009: 351;
May-September 2010: 9.
Time on tarmac: 241 minutes or more;
May-September 2009: 62;
May-September 2010: 0.
Source: GAO analysis of DOT data.
Note: This information is based on DOT data for all reporting airlines
systemwide.
[End of figure]
A small number of flights have sat on the tarmac for more than 3 hours
since the rule went into effect, including four that resulted in
violations where airlines were warned. In the first 12 months after
the implementation of the rule, DOT identified 20 incidents where
flights were delayed on the tarmac more than 3 hours and determined
that 11 of these did not violate the tarmac rule, 4 were violations
which resulted in a warning to the airline, and 5 which are still
under investigation (see appendix IV for a list of these flights).
[Footnote 55] Twelve of these 20 flights were canceled and none sat on
the tarmac for more than 4 hours, according to DOT data.[Footnote 56]
DOT has not defined, in the regulation or elsewhere, what constitutes
a violation of the rule that warrants a fine, though DOT enforcement
officials told us that when determining whether to assess a fine, as
well as how much to assess, they consider, among other things, the
nature of the violation, the harm caused to passengers, whether the
delay was preventable, and the size and financial condition of the
airline. According to these officials, airlines are operating under
the assumption that a fine could be assessed at $27,500 per passenger
because DOT's current authority allows for penalties at up to $27,500
"per violation," a phrase which is not defined in statute or
regulation.[Footnote 57]
The Number of Flight Cancellations Have Increased Since DOT
Implemented Rule:
Overall, the number of flight cancellations have increased since the
tarmac delay rule was implemented, according to DOT data, though these
cancellations cannot be directly attributed to the rule.[Footnote 58]
Our analysis of cancellation trends examined flights during the last
two summers, May through September, 2009 and 2010 because they
represent equivalent periods of time before (2009) and after (2010)
the implementation of the rule. Furthermore, as noted previously, the
summer historically accounts for the majority of tarmac delays. While
the number of scheduled flights was similar in these time periods,
total cancellations increased by 5,068 (see table 1).[Footnote 59]
Total cancellations as a percentage of all flights increased from 1
percent in 2009 to 1.2 percent in 2010, a 20 percent increase in the
rate of cancellations. Cancellation rates also increased for the
subset of flights that left the gate and then sat on the tarmac.
[Footnote 60] For example, the percentage of flights that were
canceled after sitting on the tarmac for between 2 and 3 hours (121 to
180 minutes) increased from 6.19 percent in 2009 to 17.34 percent in
2010. As a result of challenges rebooking passengers, such
cancellations can lead to long overall delay times.
Table 1: Cancellations by Stage of Flight, May-September 2009 and 2010:
Canceled before taxi-out;
2009, Total cancellations: 27,161;
2009, Percent canceled of all flights: 0.98%;
2010, Total cancellations: 31,985;
2010, Percent canceled of all flights: 1.16%.
Time on tarmac: 1-60 minutes;
2009, Total flights: 2,711,384;
2009, Total cancellations: 824;
2009, Percent canceled: 0.03%;
2010, Total flights: 2,709,532;
2010, Total cancellations: 926;
2010, Percent canceled: 0.03%.
Time on tarmac: 61-120 minutes;
2009, Total flights: 22,364;
2009, Total cancellations: 259;
2009, Percent canceled: 1.16%;
2010, Total flights: 17,763;
2010, Total cancellations: 396;
2010, Percent canceled: 2.23%.
Time on tarmac: 121-180 minutes;
2009, Total flights: 2,391;
2009, Total cancellations: 148;
2009, Percent canceled: 6.19%;
2010, Total flights: 1,257;
2010, Total cancellations: 218;
2010, Percent canceled: 17.34%.
Time on tarmac: 181-240 minutes;
2009, Total flights: 351;
2009, Total cancellations: 57;
2009, Percent canceled: 16.24%;
2010, Total flights: 9;
2010, Total cancellations: 7;
2010, Percent canceled: 77.78%.
Time on tarmac: 241 or more minutes;
2009, Total flights: 62;
2009, Total cancellations: 15;
2009, Percent canceled: 24.19%;
2010, Total flights: 0;
2010, Total cancellations: 0;
2010, Percent canceled: n/a.
All flights:
2009, Total flights: 2,763,713;
2009, Total cancellations: 28,464;
2009, Percent canceled: 1.0%;
2010, Total flights: 2,760,546;
2010, Total cancellations: 33,532;
2010, Percent canceled: 1.2%.
Source: GAO analysis of DOT data.
Note: This information is based on DOT data for all reporting airlines
systemwide.
[End of table]
According to DOT, although cancellations have increased since the
tarmac delay rule was implemented, few, if any, additional
cancellations can be attributed to the introduction of the tarmac
delay rule. DOT's analysis is limited, though, because it includes
only a portion of all flights, considers the total number of
cancellations instead of the rate of cancellation, and does not
control for other factors that can affect cancellations. In a March
2011 analysis of flight cancellations from 2009 and 2010, DOT found
that, for the period from May through October, the number of flights
canceled after sitting on the tarmac for 2 hours or more increased by
six flights from 2009 to 2010.[Footnote 61] However, as indicated in
table 1, the number of flights that remained on the tarmac for more
than 2 hours (121 or more minutes) declined by more than half--2,804
to 1,266--from 2009 to 2010. As a result, the rate of cancellation
increased from 2009 to 2010. DOT also did not control for the other
factors, such as weather, that can affect an airline's decision to
cancel a flight. When such factors are not controlled for, the
observed changes in cancellations, and any associated costs, cannot be
estimated. A complete consideration of the costs and benefits of the
tarmac delay rule cannot be conducted without, at a minimum,
controlling for these factors. Such a consideration is important
because, according to the Office of Management and Budget, a
fundamental indicator of a publicly acceptable rule is one in which
public benefits exceed public costs.
Airlines Maintain They Have Changed Decision Making in Response to
Tarmac Delay Rule:
Airline and other aviation industry stakeholders that we spoke with
maintained that the tarmac delay rule has changed how airlines balance
the trade-off between the extent to which flights are delayed and
canceled, and that this change has made flight cancellations more
likely. In particular, these officials told us airlines are more often
taking actions to avoid potential DOT fines, including returning
flights to the gate after taxi-out and, because of crew hour limits,
limited gate availability, or the severity in the underlying cause of
delay, some of these flights may then be canceled. Furthermore, when
flights are delayed on the tarmac, airline officials told us they are
now deciding sooner than they did in the past whether to taxi back in
to the gate. A majority of the U.S. airline officials we spoke with
said that, once a flight is delayed on the tarmac, communications
between airline officials and air traffic control officials on how to
handle the delay, such as whether to wait or return to the gate now
starts after about an hour (see figure 9). According to airline
officials we spoke with, uncertain taxi times for take off and the
potential for million-dollar fines have made early decision making
necessary because it may take a significant amount of time for a
flight to return to the gate, if necessary. Additionally, within 2
hours, airlines must provide food and water. Airline officials also
told us that when flights have been on the tarmac for 2 hours, the
pilots begin executing a plan for either takeoff or a return to the
gate within the hour. According to one airline official, this plan
must then be carried out unless they are told by air traffic control
that takeoff is imminent. Officials from one airline told us that
their decision to return to the gate is sometimes put into action
before the flight has been on the tarmac for 2 hours. As a result,
airlines are returning more flights to the gate prior to takeoff. Our
analysis of DOT data found that the number of flights returning to the
gate after waiting on the tarmac for at least an hour has increased by
almost 9 percent from May through September 2009 to May through
September 2010, although it is not possible to definitively attribute
these changes solely to the tarmac delay rule.
Figure 9: Domestic Airlines' Decision-Making Timeline for Flights
Experiencing a Taxi-Out Delay on the Airport Tarmac:
[Refer to PDF for image: illustrated table]
Prior to departure:
If tarmac delay or other disruption, such as weather delay, is likely,
flight may be canceled.
0–1 hour after leaving gate:
* Plane leaves gate and proceeds to ramp area then queues for take-off;
* Delay begins if weather, congestion, or other factors prevents take-
off or increases length of queue.
1–2 hours:
* After an hour on the tarmac, communication may begin between airline
officials, pilots, and FAA’s air traffic controllers at the airport;
* Pilot and airline officials will develop a strategy for either
returning to the gate or taking off by the 3 hour mark.
2–3 hours:
* By the 2 hour mark, the airline is required to provide passengers with
food and water;
* Pilots will execute the airline’s strategy to avoid a continued
tarmac delay and potential fine.
After 3 hours: Flight will have:
* taken off;
* canceled;
* returned to the gate and permitted passengers to deplane prior to
continuing the flight to its original destination; or;
* remained on the tarmac, making it possibly subject to DOT fine.
Sources: GAO analysis of tarmac delay rule and aviation stakeholder
interviews.
Note: The decision-making timeline described in this figure is a
general representation of airline decision-making processes based on
academic studies and interviews with airline officials. The processes
and decisions may vary between airlines based on their business
practices and the specific circumstances.
[End of figure]
In addition to stating that the tarmac delay rule is altering their
decision making during a tarmac delay, airlines maintain that the rule
has increased the likelihood that they will cancel a flight before it
ever leaves the gate. For instance, airline officials told us that
they are precanceling more flights prior to the scheduled departure
time when long tarmac delays are possible, such as during severe
weather, than they did in the past. According to an official from one
airline, its precancellations have increased by 10 percent since late
April 2010, when the rule went into effect. When canceling a flight
before passengers have boarded the plane, airlines have more control
over where they position crew and aircraft to resume normal operations
the following day. According to one major airline, precanceling also
benefits flight crews and airport employees because it gives airlines,
airports, and passengers greater flexibility in rescheduling flights,
work, and personal activities.[Footnote 62]
Our Analysis Finds That the Tarmac Delay Rule is Correlated with an
Increased Likelihood of Flight Cancellations:
Since a variety of factors in addition to the tarmac delay rule may be
correlated with airline cancellation decisions, we developed logistic
regression models that control for several factors that are likely to
be associated with these decisions in order to measure the likely
effect of the tarmac delay rule.[Footnote 63] We used two models to
analyze cancellations. In the first (the tarmac-cancellation model),
we assessed the likelihood of cancellation for all flights that taxi-
out from the gate. In the second (the gate-cancellation model), we
assessed the likelihood of cancellation for flights before they leave
the gate.[Footnote 64] Our analysis examined flights during the last
two summers (May through September 2009 and 2010) because DOT began
collecting more extensive data on tarmac delays that is necessary for
this analysis in October 2008 and, historically, the majority of
tarmac delays occur in the summer.[Footnote 65] Both models control
for several factors that are likely to influence airlines' decisions
about whether to cancel flights, including weather at the origin and
destination airport, airline characteristics, and specific details of
individual flights. Nevertheless, other factors related to
cancellations may not have been fully controlled for. Additionally,
since we used a variable indicating the year as a proxy for the
implementation of the rule in late April 2010, other general changes
in the environment across these two years which affect how airlines
decide whether to cancel a flight may not be fully reflected in our
model. See appendix V for a detailed discussion of the model
structure, a full list of independent variables, and our full results.
Model Results:
Results from the tarmac-cancellation model suggest that the
implementation of the tarmac delay rule is associated with a greater
likelihood of cancellation for flights that taxi-out onto the tarmac.
Furthermore, our results suggest that the greater likelihood of
cancellation increases with the time a plane stays on the tarmac. As
shown in table 2, we grouped flights into hour-long intervals, and for
each group the likelihood of cancellation has increased since the rule
went into effect. This correlation of the rule's implementation with
increased cancellations appears consistent with what airlines have
told us has happened.
Table 2: Percent Difference in Likelihood of Flight Cancellation after
Implementation of Tarmac Delay Rule:
Time on tarmac: Before taxi-out (at gate);
Increased likelihood of cancellation in 2010 compared with 2009: 24
percent more likely.
Time on tarmac: 1-60 minutes;
Increased likelihood of cancellation in 2010 compared with 2009: 31
percent more likely.
Time on tarmac: 61-120 minutes;
Increased likelihood of cancellation in 2010 compared with 2009: More
than twice as likely (214 percent).
Time on tarmac: 121-180 minutes;
Increased likelihood of cancellation in 2010 compared with 2009: More
than 3 times as likely (359 percent).
Source: GAO analysis of DOT data.
Notes:
1. These results are based on analysis of all flights, operated by DOT
reporting airlines, to and from 70 of the FAA's Aviation System
Performance Metrics 77 airports in the United States and Puerto Rico.
2. The percentage differences in the likelihood of cancellation
presented here are based on odds ratios calculated from our models. In
particular, the figure for cancellations before taxi-out is from our
gate-cancellation model, which estimates the effect of different
factors (including the implementation of the tarmac delay rule) on the
odds of flights being canceled at the gate. The figures for
cancellations after taxi-out are from our tarmac-cancellation model,
which estimates the effect of different factors (including the
implementation of the tarmac delay rule) on the odds of flights being
canceled after taxi-out on to the tarmac. All figures reported in the
table are statistically significant at the 1 percent level. Our tarmac-
cancellation model also showed a large, statistically significant
increase in the likelihood of cancellation in 2010 compared with 2009
for flights on the tarmac for more than 3 hours, but we excluded that
information from this table because of the very small number of
flights that experienced such long tarmac delays in 2010.
[End of table]
Results from the gate-cancellation model also indicate that the tarmac
delay rule is associated with a higher rate of flight cancellation. In
particular, when the model controlled for other factors that may be
associated with an airline's decision to cancel a flight, the
likelihood of a gate cancellation was 24 percent higher during May
through September 2010 than it was for the same months in 2009 (see
table 2). The gate-cancellation model also controlled for the same
factors as the tarmac model except for minutes on the tarmac. For both
models, the tarmac delay rule as well as the other factors we included
generally had the expected, and statistically significant, association
with cancellations.[Footnote 66]
EU Requirements Provide More Comprehensive Passenger Protections, but
May Also Increase Costs for Airlines and Passengers:
Passenger Protections Requirements Are More Extensive in the EU than
in the United States or Canada:
Passenger protections requirements for flight delays, cancellations,
and denied boarding are overall more extensive in the EU than they are
in the United States or Canada. While all three regions have enhanced
passenger protections in recent years, EU care and compensation
guarantees are generally more extensive than those in the United
States or Canada. (Table 3 summarizes what airlines are required to
provide passengers for flight delays, cancellations, and denied
boardings in the three regions.) In April 2011, U.S. DOT further
enhanced its airline passenger protections by, among other things,
increasing financial compensation in the event of an involuntary
denied boarding. The Canadian Minister of Transport, Infrastructure
and Communities launched "Flight Rights Canada" in September 2008 to
increase air passenger's awareness of their rights, which includes a
voluntary "Code of Conduct of Canada's Airlines" which, among other
things, recommends Canadian airlines adopt specific provisions related
to flight disruptions in their contracts of carriage.[Footnote 67] In
the EU, a regulation enacted in 2004 entitles passengers to care and
compensation, under specific circumstances, for all three types of
disruptions.[Footnote 68] Officials from the European Commission
(Commission) told us that these rules harmonized levels of customer
service across all EU member states and airlines, ensuring that
passengers can expect to be cared for and compensated if their flight
is canceled or seriously delayed or if the passenger is denied
boarding. Before the current regulation was put into place, according
to these European officials, some airlines were increasingly
overbooking flights, while providing little care or compensation to
those inconvenienced passengers who were denied boarding. The
officials said that the goal of the regulation was not to punish
airlines for delays or cancellations, or even necessarily reduce the
number of disruptions, but rather to make passengers "whole" when
flights are disrupted.
Table 3: Comparison of Governmental Requirements in the United States,
Canada, and the EU of Benefits Provided to Passengers by Airlines
during Flight Disruptions:
Flight disruption: Delay;
United States: No federal requirements for care or compensation[A];
Canada: No requirements for care or compensation;
EU:
* Passengers on flights delayed 2 or more hours from departure
(dependent on flight distance) entitled to care, including, meals,
refreshments, communication services, and, in certain cases, hotel
accommodations[B];
* Passengers on flights delayed more than 5 hours, entitled to
reimbursement in part or in full, dependent upon circumstances,[C] and
a return flight to the first point of departure, at the earliest
opportunity if they decide to pursue their journey.[D]
Flight disruption: Cancellation;
United States: Refund for unused portion of ticket[E];
Canada: No requirements for care or compensation;
EU: Passenger care:
* Passengers may be entitled to care, such as meals, refreshments,
communication services, and, in certain cases, hotel accommodations[F];
* Passengers entitled to the choice between rerouting or
reimbursement[G];
Passenger compensation:[H]
* €250 for flights 1,500 kilometers (km)[I] or less;
* €400 for all intra-community flights more than 1,500 km and all
other flights between 1,500 km and 3,500 km;
* €600 for all other flights;
* May be reduced by 50% if rerouted to destination within certain time
frames[J].
Flight disruption: Denied boarding;
United States: Passenger compensation:[K]
* 200% of the fare (up to $650) if transported to the final
destination (or first stopover, if one) more than 1 hour but less than
2 hours (or more than 1 hour but less than 4 for international
flights) of scheduled arrival;
* 400% of the fare (up to $1,300) if transported to the final
destination (or first stopover, if one) 2 hours or more (or 4 hours or
more for international flights) after the planned arrival time of the
original flight[L];
Canada: No requirements for care or compensation;
EU: Passenger care:
* Passenger entitled to care, including, meals, refreshments,
communication services, and, in certain cases, hotel accommodations[M];
Passenger compensation:[N]
* €250 for flights 1,500 km or less;
* €400 for all intra-community flights more than 1,500 km and all
other flights between 1,500 km and 3,500 km;
* €600 for all other flights;
* May be reduced by 50 percent if rerouted to destination within
certain time frames[O].
Source: GAO analysis of laws and regulations.
[A] According to DOT, passengers who experience an extensive delay are
entitled to the same benefits as those who experience a cancellation,
however, DOT has not defined what constitutes an "extensive" delay. In
instances of tarmac delays, adequate food and potable water must be
provided no later than 2 hours after the aircraft leaves the gate (or
touches down, in the case of an arrival).
[B] Passengers will be entitled to hotel accommodations if the
reasonable expected time of departure is at least the day after the
time of departure previously announced. Regulation (EC) 261/2004, art.
9, 2004 O.J. (L 46/1). See also Id., art. 6(1). But see, infra, C-402/
07 and C-432/07, Sturgeon v. Condor Flugdienst GmbH Böck, Lepuschitz
v. Air France SA, 2009 E.C.R. I-10923.
[C] Regulation (EC) 261/2004, art. 6(1)(iii), 2004 O.J. (L 46/1). See
also Id., art. 8(1)(a).
[D] Id., art. 6(1)(iii). See also Id., art. 8(1)(a).
[E] According to DOT, airlines do not owe refunds to passengers who
accept and travel alternate routes to their destinations.
[F] Regulation (EC) 261/2004, art. 5(1)(b), 2004 O.J. (L 46/1). See
also Id., art. 9.
[G] Id., articles 5(1)(a), 8. See also Id., art. 8.
[H] Id., article 5(1)(c). See also Id., art. 7(1).
[I] In determining distance, the basis shall be the last destination
at which the denial of boarding or cancellation will delay the
passenger's arrival after the scheduled time. Regulation (EC)
261/2004, art. 7(1), 2004 O.J. (L 46/1).
[J] Id., art. 7(2).
[K] 14 C.F.R. § 250.5. On April 25, 2011, DOT issued its final rule on
Enhancing Air Passenger Protections, which, as of August, 23, 2011,
increased the denied boarding compensation limits to what is listed
above, implemented an automatic inflation adjuster for minimum denied
boarding compensation limits every two years, and clarified the
application of the rule and requirements imposed on airlines in this
regard, among other things. 76 Fed. Reg. 23110 (Apr. 25, 2011).
[L] Pasengers are not entitled to compensation if the airline offers
alternate transportation that is planned to arrive at the passenger's
destination or first stopover not later than 1 hour after the planned
arrival time of the passenger's original flight. 14 C.F.R. § 250.5.
[M] Regulation (EC) 261/2004, art. 4(3), 2004 O.J. (L 46/1). See also
Id., art. 9.
[N] Id., art. 4(3). See also Id., art. 7(1).
[O] Id., art. 7(2).
[End of table]
In the event of a flight delay, the EU regulation requires that
airlines offer passengers care and, under certain circumstances, the
option of reimbursement or a return flight to the first point of
departure, while there are no U.S. and Canadian requirements with
similar levels of care or compensation.[Footnote 69] Under EU
regulation, when a flight is delayed at least 2, but less than 5 hours
(depending on the distance of the flight), airlines are required to
provide passengers with certain types of care, including meals and
communication services, and if the delay requires an overnight stay
passengers must be offered hotel accommodations and transportation
between the airport and hotel.[Footnote 70] Furthermore, if the delay
is at least 5 hours, passengers must also be offered reimbursement for
the unused portion of their ticket (and for the part of the journey
already made if the flight no longer serves its original purpose) and,
if necessary, a return flight to the point of departure.[Footnote 71]
They must also be given written notice of the rules for care and
compensation.[Footnote 72] By comparison, passengers on delayed
flights in the United States and Canada are not entitled to care or
compensation by law.[Footnote 73]
In the EU, when a flight is canceled, the EU regulation requires that
passengers receive care in certain circumstances, compensation, and
the option of being rerouted or reimbursed (with a return flight to
the point of departure), while passengers in the United States and
Canada do not have such extensive rights. Passengers on canceled
flights covered by the EU regulation are entitled to the same rights
as those passengers on delayed flights (as described previously) and,
additionally must be offered the choice between being
rerouted[Footnote 74] or reimbursed for part or all of their ticket,
depending on the circumstances along with a return flight to the first
point of departure at the earliest opportunity.[Footnote 75] In
addition, passengers on such flights are entitled to financial
compensation, the amount of which depends on the length of the
canceled flight and may be reduced by 50 percent if the passenger is
rerouted, under certain circumstances.[Footnote 76] An airline may be
exempt from the obligation to pay compensation if it can prove the
cancellation was caused by an extraordinary circumstance that could
not have been avoided even if all reasonable measures had been taken.
[Footnote 77] At the time a flight is canceled, the airline must
provide passengers written notice of the rules for compensation and
assistance.[Footnote 78] By contrast, U.S. rules do not require care
and compensation in the event of a cancellation, but do, require
airlines to offer passengers a refund if they do not wish to accept
and travel alternative routes to their destinations.[Footnote 79] In
Canada, passengers are not entitled to care or compensation in the
event of a cancellation, nor is there a specific requirement that an
airline refund a passenger their ticket price, in whole or in part.
[Footnote 80]
If a passenger is involuntarily denied boarding--generally because an
airline has oversold seats on a flight and cannot find enough
volunteers willing to take another flight--the passenger may be
entitled to benefits, depending on the region. Passengers on flights
covered by the EU regulation have a right to care and financial
compensation. Comparatively, passengers covered under U.S. regulations
are entitled to financial compensation and passengers in Canada are
entitled to neither care nor compensation except as provided under
their contracts of carriage.[Footnote 81] To limit the number of
passengers who are involuntarily denied boarding when a flight is
oversold, airlines in the United States and the EU are required to
first request volunteers to relinquish their confirmed space in
exchange for benefits, such as credit for future travel, before
selecting passengers for denied boarding.[Footnote 82] When selecting
passengers for denied boarding, U.S. airlines are required to use
boarding priority rules that are in compliance with DOT regulations.
[Footnote 83] Under both U.S. and EU requirements, those passengers
selected for denied boarding must be offered financial compensation.
[Footnote 84] Passengers in the EU also are guaranteed the same care
offered to passengers whose flights are delayed or canceled and must
also be offered the option of reimbursement, in whole or in part,
dependent upon the circumstances, and a return flight to the first
point of departure or rerouting the final destination.[Footnote 85] In
both regions, airlines must notify passengers in writing of their
rights.[Footnote 86]
Some airlines in the United States and Canada, as described earlier,
voluntarily include provisions in their contracts of carriage for care
and compensation beyond what is legally required for delays,
cancellations, and denied boarding. These provisions are enforceable
as a legal contract between the airline and the ticket holder. The
airlines we spoke with in the EU do not include any additional care or
compensation beyond the EU requirements discussed previously and the
EU regulation does not require airlines to include those requirements
in their contracts of carriage. We examined the contracts of the seven
largest airlines in the United States and found that five of these
airlines may, in certain circumstances, provide certain types of care,
such as meal vouchers and free phone calls, for delays and
cancellations that extend beyond a certain time (see table 4). Certain
airlines also state in their contracts of carriage that they must also
provide hotel accommodations and ground transportation, under certain
circumstances, when an overnight stay is required. The circumstances
under which airlines provide these amenities vary and may depend on a
number of factors, such as the cause, length, and timing of the flight
disruption. All four of Canada's major airlines have added passenger
protections for delays, cancellations, and denied boarding in response
to the 2008 federal government initiative mentioned above, according
to airline officials in Canada. Although airlines are not required to
adhere to passenger protection provisions outlined in the initiative's
Code of Conduct, because all four of Canada's major airlines have
added its provisions to their contracts of carriage, these become part
of the binding contract between the airline and the passenger.
[Footnote 87] As a result, these airlines now guarantee in their
contracts of carriage that they will provide passengers with a meal
voucher if a flight delay exceeds 4 hours, as well as hotel
accommodations and ground transportation if a flight delay exceeds 8
hours and requires an overnight stay. If a flight is canceled or a
passenger is denied boarding, the airlines will rebook passengers or
refund the unused portion of the ticket.[Footnote 88] Flight Rights
Canada's Code of Conduct does not make the airline responsible for
acts of nature or the acts of third parties.
Table 4: Comparison of Contract of Benefits That May be Provided Under
Domestic Carriage Provisions for Flight Delays and Cancellations for
Selected U.S. Airlines, as of June 2011:
Airline: Alaska Airlines;
Options for transporting passengers:
Rebook on next available flight offered by the airline: [Check];
Rebook on a flight offered by another airline: [Check];
Amenities provided to passengers:
Meals and/or snacks: [Empty];
Hotel accommodation and/or ground transportation: [Check];
Phone calls: [Check];
Credit for future travel: [Check][A].
Airline: American Airlines;
Options for transporting passengers:
Rebook on next available flight offered by the airline: [Check];
Rebook on a flight offered by another airline: [Check];
Amenities provided to passengers:
Meals and/or snacks: [Empty];
Hotel accommodation and/or ground transportation: [Check][B];
Phone calls: [Empty];
Credit for future travel: [Empty].
Airline: Delta Air Lines;
Options for transporting passengers:
Rebook on next available flight offered by the airline: [Check];
Rebook on a flight offered by another airline: [Check];
Amenities provided to passengers:
Meals and/or snacks: [Check];
Hotel accommodation and/or ground transportation: [Check];
Phone calls: [Empty];
Credit for future travel: [Empty].
Airline: JetBlue Airways;
Options for transporting passengers:
Rebook on next available flight offered by the airline: [Check];
Rebook on a flight offered by another airline: [Empty];
Amenities provided to passengers:
Meals and/or snacks: [Check];
Hotel accommodation and/or ground transportation: [Check];
Phone calls: [Empty];
Credit for future travel: [Check][A].
Airline: Southwest Airlines;
Options for transporting passengers:
Rebook on next available flight offered by the airline: [Check];
Rebook on a flight offered by another airline: [Empty];
Amenities provided to passengers:
Meals and/or snacks: [Empty];
Hotel accommodation and/or ground transportation: [Empty];
Phone calls: [Empty];
Credit for future travel: [Empty].
Airline: United Airlines;
Options for transporting passengers:
Rebook on next available flight offered by the airline: [Check];
Rebook on a flight offered by another airline: [Check];
Amenities provided to passengers:
Meals and/or snacks: [Empty];
Hotel accommodation and/or ground transportation: [Check][B];
Phone calls: [Empty];
Credit for future travel: [Empty].
Airline: US Airways;
Options for transporting passengers:
Rebook on next available flight offered by the airline: [Check];
Rebook on a flight offered by another airline: [Check];
Amenities provided to passengers:
Meals and/or snacks: [Empty];
Hotel accommodation and/or ground transportation: [Empty];
Phone calls: [Empty];
Credit for future travel: [Empty].
Source: GAO analysis of airline contracts of carriage as of June 2011.
Note: We selected these airlines from the 10 largest U.S. airlines
using FAA's data on airline available seat miles from the last 12
months for which data were available. Available seat miles are
calculated by multiplying aircraft miles flown in inter-airport
flights by the number of seats available for revenue passenger use on
those flights. Although AirTran Airways and Continental Airlines were
among the 10 largest U.S. airlines, we did not include them in this
analysis because they have recently merged with Southwest Airlines and
United Airways, respectively, and therefore we expect that these
airlines will refer to the contract of carriage of their purchasing
airline. We also did not include SkyWest because it follows the
policies of the airline partner that tickets its flights. SkyWest
partners with AirTran Airways, Delta Air Lines, and United Airlines.
This table does not include provisions that are required by law to be
provided in the event of a lengthy tarmac delay, and also excludes
provisions for refunds. The options for transportation and amenities
offered to passengers vary by airline and by circumstances. For
example, some airlines offer these services to passengers only when
the airline is the cause of the flight disruption and airlines differ
by the amount of time a delay and cancellation must exceed before
providing passengers with amenities.
[A] For flight delays that exceed 2 hours, Alaska Airlines offers
passengers 2,000 Mileage Plan Bonus Miles. JetBlue offers passengers a
$50 credit for future travel when its flights are canceled due to a
controllable irregularity within 4 hours of the scheduled departure
time. JetBlue passengers are also entitled to credit when their flight
is delayed due to a controllable irregularity: $25 if the delay is
greater than 1 hour but less than 2 hours, $50 if the delay is greater
than 2 hours but less than 4 hours, credit equal to the one-way trip
fare if the delay is greater than 5 hours but less than 6 hours, and
credit equal to the roundtrip fare if the delay is greater than 6
hours.
[B] American Airlines provides hotel accommodation if the delay or
cancellation was caused by events within the airline's control and
United Airlines provides hotel accommodations in the event of a flight
diversion.
[End of table]
Passenger Protections May Increase Costs for Airlines and Passengers:
Care and compensation requirements provide protections and benefits
for passengers whose flights are disrupted, but they also increase
costs to airlines and could increase passengers' fares. Airline
officials we spoke with in the EU and the United States maintained
that passenger protections increase their costs, though they did not
provide documentation of specific cost increases because they consider
the information to be confidential.[Footnote 89] While data from
airlines on these costs are unavailable, a February 2010 study of the
EU passenger protection requirements noted that airlines the study
authors spoke with reported the cost of compliance for EU airlines
ranges from 0.1 to 0.5 percent of airlines' annual revenue.[Footnote
90] However, officials from one European airline, as well as officials
from an airline association, maintained that airlines' cost of
compliance exceeds this estimate. Increases in required compensation
for passengers denied boarding have also increased costs for both U.S.
and EU airlines, according to airline officials we spoke with. In the
United States, officials with some U.S. airlines told us that
complying with the requirements to better inform passengers about
routinely delayed and canceled flights and to post information such as
flight on-time performance data on airline Web sites costs hundreds of
thousands of dollars. Officials with one of these airlines estimated
that airline personnel spent about 3 months adding the information at
the airline's Web site. Some airlines in the United States and the EU
told us that compliance costs such as these can lead to higher fares.
However, it is very difficult to isolate the impact of compliance on
fares because they are set based on demand in competitive markets as
well as other factors.[Footnote 91]
Passenger protections can create financial burdens on airlines for
major events outside their control. For example, as noted above,
airlines subject to EU regulations are required to provide certain
care in the event of a delay or cancellation, regardless of whether
the disruption was within the airline's control. These regulations
require an airline to provide passengers with food, lodging, and other
care, dependent on the circumstances, during short-term disruptions in
travel plans. However, when major disruptions to the airspace system
occur, this requirement can obligate airlines to provide passengers
with lodging and other care for extended periods of time at great
cost.[Footnote 92] Such a situation occurred in 2010 when the European
air transport industry was significantly affected by the consequences
of the Eyjafjallajökull volcanic eruption in Iceland. The volcano,
which erupted on April 14, 2010, created a cloud of volcanic ash that
drifted through large sections of European airspace. Volcanic ash
contains substances that may harm aircraft, so national authorities
decided to close affected airspace. As a result, more than 100,000
flights were canceled and millions of passengers were unable to fly.
In many cases, the passengers were stranded in a foreign country
without any immediate possibility to go back home. Representatives of
one EU airline told us that when the eruption occurred, they booked
more than 100,000 hotel rooms for their scheduled passengers and
eventually chartered aircraft to get passengers to their destinations.
The airline's representatives estimated that the incident cost the
airline about $4.5 million. Major disruptions generally result from
unsafe flying conditions. According to airline officials in both the
United States and EU, the possibility of large monetary claims as a
result of such incidents could pressure airlines to operate in
conditions they would otherwise deem unsafe for flight in order to
avoid high costs, but according to Commission officials there are no
available data on the existence or extent of this issue.
While increasing the compensation for denied boarding will increase
airline's costs if airlines don't change their booking policies,
reducing overbooking reduces revenues because fewer seats can be sold,
according to airlines officials we interviewed. Overbooking is a
revenue-producing strategy for many airlines, without which some would
raise fares to offset their losses.[Footnote 93] Additionally, airline
officials said that reductions in overbooking could also limit the
flexibility of passengers when choosing flights, as seat availability
would be reduced and airline policies governing how and when
passengers change their flights could become more restrictive.
However, we found little evidence that increases in denied boarding
compensation in the United States resulted in reduced overbooking.
According to airline officials we spoke with, the 2008 compensation
increase in the United States was not large enough to cause airlines
to reduce their overbooking of flights.[Footnote 94] Additionally,
from 2004 through 2010, the number of voluntary denied boardings in
the United States was less than 0.1 percent of all U.S. passengers
boarded annually, while the number of involuntary denied boardings
rose slightly but remained rare in relation to the total number of
U.S. passengers, at 0.01 percent of all U.S. passengers (see figure
9). In contrast, EU denied boarding compensation, though in some cases
less than U.S. levels, has been significant enough to cause at least
two EU airlines to reduce overbooking of flights, according to
officials from these airlines.[Footnote 95] According to these
officials, this reduction in overbooking has adversely affected
consumers through higher average ticket costs designed to offset the
increased number of unused seats on each flight. However, data showing
whether any such reductions in overbooking have caused EU airlines to
increase their fares are not available.
Figure 10: Voluntary and Involuntary Denied Boardings in the United
States as a Percentage of Total Passengers, Calendar Years 2004-2010:
[Refer to PDF for image: multiple line graph]
Year: 2004;
Percent of passengers voluntarily denied boarding: 0.092%;
Percent of passengers involuntarily denied boarding: 0.006%.
Year: 2005;
Percent of passengers voluntarily denied boarding: 0.073%;
Percent of passengers involuntarily denied boarding: 0.006%.
Year: 2006;
Percent of passengers voluntarily denied boarding: 0.077%;
Percent of passengers involuntarily denied boarding: 0.007%.
Year: 2007;
Percent of passengers voluntarily denied boarding: 0.074%;
Percent of passengers involuntarily denied boarding: 0.008%.
Year: 2008;
Percent of passengers voluntarily denied boarding: 0.077%;
Percent of passengers involuntarily denied boarding: 0.008%.
Year: 2009;
Percent of passengers voluntarily denied boarding: 0.091%;
Percent of passengers involuntarily denied boarding: 0.009%.
Year: 2010;
Percent of passengers voluntarily denied boarding: 0.087%;
Percent of passengers involuntarily denied boarding: 0.008%.
Source: GAO analysis of DOT data.
Note: This figure reflects data from airlines with at least 1 percent
of total domestic scheduled passenger service revenue.
[End of figure]
Enforcing Passenger Protections Can Create Challenges:
Extensive passenger protections, while providing benefits and
guarantees to passengers, can create challenges for the government
entities responsible for enforcing the requirements and for passengers
in obtaining benefits due to them. These challenges include
difficulties enforcing unclear requirements and ineffective passenger
complaint processes. Such challenges can limit the potential for the
requirements to mitigate hardships for airline passengers.
Government enforcement bodies in each region are responsible for
ensuring that airlines comply with their region's requirements. DOT
and the Canadian Transportation Agency (CTA) serve as the enforcement
bodies for the United States and Canada, respectively. In the EU, each
of its 27 member states, and other countries that joined the EU
aviation market (such as Iceland, Norway, and Switzerland),
establishes its own body responsible for enforcing the EU regulation,
which is typically the agency responsible for aviation oversight.
These enforcement bodies use similar activities to monitor airline
compliance, including investigating passenger complaints and issuing
penalties against airlines for noncompliance. Enforcement bodies in
each region receive passenger complaints or information (for example,
through a media report) about a possible violation of passenger
protections and decide whether to investigate.[Footnote 96] DOT
officials told us they will investigate any case alleging a violation
of a DOT rule, but will generally only pursue an enforcement action
against airlines if they discover a pattern or practice of violations
or the incident is particularly egregious. CTA officials and an
enforcement body official from one EU member state told us they
investigate and may pursue enforcement actions against an airline
based on an individual's complaint.[Footnote 97] If officials
determine that an airline has violated passenger protections, they may
fine the airline, depending on the region or the member state.
[Footnote 98] In addition to conducting investigations based on
passenger complaints, enforcement bodies in each region initiate
investigations. For example, DOT officials told us they routinely
investigate each major airline and their investigations have resulted
in the collection of fines.[Footnote 99] In two EU member states,
officials from the enforcement bodies told us they visit airports to
see if airlines are displaying required information about passenger
protections, but have not issued fines.
Lack of Clarity in Regulations and Confusion in Application:
The first challenge to the effective application of passenger
protections arises when there is a lack of clarity in regulations. In
the EU, where passenger protection regulations are more extensive than
in the United States or Canada, officials from the Commission told us
that different interpretations of these regulations by enforcement
bodies in different member states have made it challenging to ensure
successful implementation of the regulation. A 2010 study for the
Commission about the impact of the EU passenger protection regulation
found that more needs to be done to ensure that passengers' rights are
properly protected.[Footnote 100] In particular, the study noted that
in some areas the rights granted by the regulation can lead to
different understandings. The Commission also recently reported that
"the novelty of some provisions of the Regulation has led to different
interpretations, and thus varied application, among airlines and
national enforcement authorities, rendering it difficult for
passengers and stakeholders to understand the scope and limits of the
rights set out."[Footnote 101] Stakeholders told us, for example, that
the following two provisions were unclear and confusing to implement,
respectively:
* Unclear definition of extraordinary circumstances. According to some
airlines, airline associations, and consumer groups we spoke with in
the EU, the definition of this term--which refers to situations in
which airlines are exempt from the passenger compensation requirement
when a flight is canceled--has left room for confusion.[Footnote 102]
A recent ruling by the European Court of Justice (ECJ) provided some
clarification for enforcement bodies when it ruled that technical
issues, such as an airplane malfunction, may constitute an
extraordinary circumstance only when these issues stem from events
outside the normal activities of the airline and are beyond its
control.[Footnote 103] Even so, some enforcement bodies are
interpreting this ruling differently. For example, officials from one
enforcement body told us that even if a technical issue is routine,
they may still consider it an extraordinary circumstance if they
believe the safety risks were too great, whereas other enforcement
bodies in the EU have interpreted the ECJ's ruling more strictly.
Additionally, some stakeholders that we spoke with told us that the
extraordinary circumstance provision in the regulation should be
revised to restrict the amount of assistance an airline must provide
to passengers or to identify an extensive list of scenarios under
which the airline would be exempt from the passenger compensation
requirement. For example, officials from an airline and an airline
association told us that they believe the regulation should be amended
to exempt airlines from paying for weeks of hotel accommodations and
food (not just compensation) in response to major disruptions, such as
the Eyjafjallajökull volcanic eruption. In a recent report, the
Commision stated this incident illustrated the structural limits of
the regulation and that the "proportionality of some current measures,
like the unlimited liability regarding the right to care under major
natural disaster, may merit assessment."
* Confusion over definition of delay. Uncertainty about when
compensation is required for delays and cancellations has also created
enforcement challenges. In the EU, a November 2009 ruling of the ECJ
specified that passengers whose flights are delayed more than 3 hours
experience the same inconvenience as those whose flights are canceled
and therefore both should be entitled to the same financial
compensation payments from airlines.[Footnote 104] This ruling created
confusion in member states and within the industry as to when to
compensate passengers who have been delayed. Airline and some airline
association officials told us that this ruling contradicts the text of
the regulation, which, requires reimbursement (in part or in full),
not compensation, in the event of a delay of more than 5 hours. In the
United Kingdom (UK), according to Commission officials, the
International Air Transport Association, among others, filed a suit in
the UK Court of Justice against the UK's enforcement body's policy to
compensate passengers in line with the ECJ ruling.[Footnote 105] An
official from the UK's enforcement body told us that the UK Court of
Justice submitted questions of law stemming from this case to the ECJ
and until the ECJ responds with further clarification, the enforcement
body has suspended all investigations into complaints on the topic.
[Footnote 106] Uncertainties over these provisions may make it
difficult for airlines and passengers to know when an airline must
compensate its passengers.
The challenges arising from the lack of clarity pertaining to
passenger protection regulations, such as the confusion about the
definition of delay, can be exacerbated when the EU requirement is
applied unevenly across jurisdictions. For instance, the enforcement
of EU regulations has been complicated because member states have
flexibility in structuring their enforcement to account for
differences in their national laws and policies.[Footnote 107] As a
result, enforcement bodies in the Netherlands and Germany, for
example, use different sanction strategies for ensuring that the
airlines comply with the regulation, resulting in varying types and
amounts of penalties for airlines. In particular, the types of
sanctions and amounts of sanctions these enforcement bodies can impose
differ because of laws and policies specific to their member state.
Officials from the Netherlands's enforcement body told us they can
impose only reparatory sanctions, which prevents them from collecting
a fine if the airline makes amends with the passenger, while the
enforcement body in Germany can issue repressive sanctions, which can
be imposed regardless of whether the airline makes amends with the
passenger.[Footnote 108] The amount of a sanction also differs between
the two member states. For example, the Dutch enforcement officials
told us that there is no set amount, but it must be reasonable and
proportionate to the severity of the violation, while in Germany,
officials from the enforcement body told us that the amount of a
sanction is based on the seriousness of the complaint. Different
national laws affect the circumstances in which sanctions can be
issued. For example, the German officials told us that German law
prohibits them from considering ECJ decisions, such as the ruling that
passengers who are delayed more than 3 hours should receive the same
compensation as those whose flights are canceled, and therefore the
German enforcement body is not using the same standards as other
enforcement bodies to sanction airlines.
Ineffective Passenger Complaint Process:
The second challenge to the application of passenger protections
arises when there is a lack of an effective passenger complaint
process. The enforcement processes of the EU, as well as those of the
United States and Canada, demonstrate challenges passengers can face
obtaining benefits due to them. When passengers in the United States,
Canada, or the EU do not receive benefits to which they believe they
are entitled, they can submit a complaint to any or all of three
entities: the airline, the national enforcement body, or the court
system (see figure 11). However, according to government officials,
passengers in the United States and EU can receive financial
compensation only through the airline or the court.[Footnote 109] The
enforcement bodies in these regions cannot award passenger
compensation because their authority does not extend to enforcing
payment by the airlines.[Footnote 110] The 2010 study for the
Commission of the regulation reported inconsistent implementation and
enforcement of the regulation across enforcement bodies and airlines.
[Footnote 111] According to the study, airlines and consumer groups
reported that there are a number of difficulties associated with
passengers in the EU seeking compensation in a court, including the
costs, time burden, availability of small claims courts, and limits on
amounts awarded.[Footnote 112] In the EU, according to Commission
officials, passengers may pay for legal assistance when pursuing
compensation in the courts from a variety of sources, such as from a
consumer protection organization. In some member states, passengers
can also use the commercial claim service EU Claim, but passengers
must pay for these services with a percentage of what they are
awarded.[Footnote 113] Officials from one consumer group told us that
when barriers are imposed on passengers claiming their benefits for
violations of their rights, airlines may not comply with applicable
requirements. Furthermore, despite a number of EU government-sponsored
campaigns to inform passengers of their rights, several EU
stakeholders told us passengers may still not be aware of their rights
and therefore may not submit complaints if they believe their rights
have been violated.[Footnote 114] Additionally, officials from two
consumer protection groups in the EU told us that some passengers may
be confused about their rights under the EU regulation and some
airlines may use that confusion to their advantage.[Footnote 115]
Figure 11: Airline Passenger Complaint Process in the United States,
Canada, and the EU:
[Refer to PDF for image: process map]
Passenger submits complaint:
1) An airline passenger who wants to submit a complaint about a flight
disruption in the United States, Canada, or the EU has three options.
They can submit the complaint to any or all of these venues; however,
passengers in the United States and the EU can only receive
compensation through the airline or court.[A]
Airline:
2) Airline provides passenger with satisfactory response, which may
include compensation.
3) Airline provides passenger with an unsatisfactory response
(passenger can submit a complaint to an enforcement body or in court).
Enforcement Body: (Enforcement bodies include U.S. Department of
Transportation, Canadian Transportation Agency or relevant member
state National Enforcement Body in the EU)
4) Based on approach (either looking for trends of individual
complaints or investigating each complaint) enforcement body may
decide to investigate an airline.
5) Enforcement body finds airline is at fault, which may result in a
penalty for the airline, but not compensation for passengers
(individuals can submit a claim in court or return to the airline for
compensation), except in Canada[A]; or:
6) Enforcement body finds airline is not at fault.
Court:
7) Judge rules in favor of passenger and may award compensation; or:
8) Judge rules in favor of airline, no compensation awarded.
Source: GAO analysis.
[A] In Canada, if the enforcement body finds that an airline has
failed to apply the provisions of its contract of carriage, it can
require the airline to apply the relevant provisions and also award
out-of-pocket expenses that were incurred by the passenger, according
to officials from the CTA. They may also order appropriate corrective
measures pursuant to regulation. Air Transportation Regulations SOR/88-
58, Part V, Div. II, § 113.1(a) (Can). According to DOT, it may obtain
compensation for consumers through settlements of potential violations
with airlines and may also insert such provisions into a consent order.
[End of figure]
Conclusions:
Flight disruptions remain costly for passengers, airlines, and the
economy. DOT has responded by enacting regulations to protect
passengers in the event of tarmac delays and has enhanced involuntary
denied boarding protections. DOT's tarmac delay rule has eliminated
delays greater than 4 hours and nearly eliminated tarmac delays of
more than 3 hours, thereby benefiting tens of thousands of passengers.
Increased compensation for involuntary denied boardings provides for
passengers in the event they are bumped from their reserved flight.
Although DOT's rules have benefited some passengers, DOT's current
flight performance data may not fully inform consumers of airlines'
quality of service as intended. By collecting data only from the
largest airlines, DOT does not obtain and therefore cannot provide
consumers with a complete picture of flight performance, particularly
at airports in rural communities or among smaller airlines. Accurate
flight performance information is necessary for consumers to make
informed decisions when purchasing airline tickets.
Additional information and analysis are also needed to fully
understand the effects of the tarmac delay rule on passengers. Since
the rule went into effect, tarmac delays of more than 3 hours have
been nearly eliminated, with no delays of more than 4 hours, reducing
the hardship for numerous passengers. However, as our analysis has
shown, the rule appears to be associated with an increased number of
cancellations for thousands of additional passengers--far more than
DOT initially predicted--including some who might not have experienced
a tarmac delay. Though it is difficult to know how passengers might
choose between a long tarmac delay and a cancellation, and what costs
and burdens their choices would entail, determining the net benefit to
airline passengers resulting from the rule and assessing whether there
is a causal relationship between the rule and any changes in flight
cancellations will be critical to passengers and airlines.
Additionally, our analysis could only include data from the first
summer of the rule's implementation, so using data through the summer
of 2011 may yield useful information for policymakers. In determining
the impact of the rule, it is important to include both the positive
effects of reducing long on-board delays and the negative effects of
flight cancellations on passengers. Increases in cancellations may be
at least partly due to airlines' assumptions about the significant
enforcement penalties that could result from a violation of the rule.
Although DOT could issue guidance on their penalty structure, it has
chosen not to in order to maintain flexibility under their current
authority.
Recommendations for Executive Action:
To enhance aviation consumers' decision-making, we recommend that the
Secretary of Transportation take the following action:
* Collect and publicize more comprehensive on-time performance data to
ensure that information on most flights, to airports of all sizes, is
included in the Bureau of Transportation Statistics' database. DOT
could accomplish this by, for example, requiring airlines with a
smaller percentage of the total domestic scheduled passenger service
revenue, or airlines that operate flights for other airlines, to
report flight performance information.
To enhance DOT's understanding of the impact of tarmac delays and
flight cancellations, we recommend that the Secretary of
Transportation take the following action:
* Fully assess the impact of the tarmac delay rule, including the
relationship between the rule and any increase in cancellations and
how they effect passengers and, if warranted, refine the rule's
requirements and implementation to maximize passenger welfare and
system efficiency.
Agency Comments and Our Evaluation:
We provided a copy of the draft report to DOT for review and comment.
Senior officials at DOT, including the DOT assistant general counsel
for aviation enforcement proceedings, provided general comments in an
e-mail representing DOT's views on the benefits of the tarmac delay
rule, but did not provide written comments on the recommendations. In
its general comments, DOT stated that, in its view, available data
demonstrate that the tarmac delay rule provided effective consumer
protection for airline passengers. DOT officials believe that the rule
made clear to airlines that, whatever the rationale, it is not
acceptable to leave passengers in aircraft stranded on the ground for
hours on end. Specifically, DOT officials cited data that demonstrate
the rule's effectiveness in preventing extended tarmac delays,
including the elimination of tarmac delays in excess of 4 hours, which
dropped from 105 flights to zero for the year ending April 2011,
completely eliminating these most egregious of delays. Officials also
highlighted the 98 percent drop in 3-hour delays, from 693 flights to
20, during the same period. DOT officials believe that these results
demonstrate the positive impact of the tarmac rule, and that without
it, far more passengers would have been subject to these extended
delays. In response to DOT's general comments, we made changes to the
report to better clarify our findings.
DOT officials said that the information in our report, in their view,
further demonstrates that airlines have gotten the basic message of
the rule and that it has been effective at putting consumers first
when it comes to avoiding lengthy tarmac delays. They cited our
discussion of actions airlines are taking to avoid tarmac rule
violations, including acting more quickly to address delayed flights
and moving more quickly back to gates, affording passengers the
freedom to access the amenities of air terminals. They were also
pleased to see our finding that air carriers are working to comply
with DOT requirements to provide food and water to passengers delayed
on the tarmac for extended periods of time.
Finally, DOT reinforced its commitment to monitor the effects of the
tarmac delay rule to ensure it is achieving intended outcomes and to
address any significant unintended outcomes. DOT initially focused on
comparing the number of flights with 2-hour tarmac delays that are
eventually canceled because in its view this was the best measure of
the effect on cancellations from the rule. According to DOT, they
recently selected a contractor to conduct a comprehensive independent
review and analysis of the impact of the tarmac delay rule now that a
full year of data is available. DOT believes that, at minimum, one
year of data is necessary to assess the rule's effects. DOT's review
will consider on-time performance, cancellations, benefits to
consumers, and other relevant information covering the period back to
2000, to assess the rule's impact on flight delays, cancellations, and
consumers. DOT also provided technical comments, which were
incorporated as appropriate.
As agreed with your offices, unless you publicly announce the contents
of this report earlier, we plan no further distribution until 7 days
from the report date. At that time, we will send copies to interested
congressional committees and the Secretary of Transportation. In
addition, this report will be available at no charge on GAO's Web site
at [hyperlink, http://www.gao.gov].
If you or your staff have any questions about this report, please
contact me at (202) 512-2834 or flemings@gao.gov. Contact points for
our Offices of Congressional Relations and Public Affairs may be found
on the last page of this report. Key contributors to this report are
listed in appendix VI.
Signed by:
Susan Fleming:
Director, Physical Infrastructure:
[End of section]
Appendix I: Objectives, Scope, and Methodology:
In this report, we examined how (1) trends in and reasons for flight
delays and cancellations in the United States differ for smaller and
larger communities; (2) the Department of Transportation's (DOT)
tarmac delay rule has affected passengers and airlines; and (3)
requirements and practices for protecting passengers from flight
delays, cancellations, and denied boardings in the United States,
Canada, and the European Union (EU) have affected passengers and
airlines.
To identify and compare the trends in and reasons for flight delays
and cancellations in different-sized U.S. communities, we examined
trends at airports designated as primary in fiscal year 2009.[Footnote
116] From this group of 367 airports, we excluded the 12 primary
airports in U.S. territories because they operate in different
operational environments than other U.S. airports. We then categorized
the 355 airports by the size of their surrounding community. We used
geographic information system data on the airports' location and
surrounding population. Airports were mapped by the county they are
located in and grouped into one of four categories based on
population: 1,000,000 or greater (large metropolitan), 250,000 to
999,999 (midsized metropolitan), 50,000 to 249,999 (small
metropolitan), and less than 50,000 (rural). This approach controls
for the fact that some small or medium airports--generally secondary
airports such as Hobby Airport in Houston--are actually in large
metropolitan regions. Using these categories, 78 airports were in
large metropolitan communities, 100 were in midsized metropolitan
communities, 122 were in small metropolitan communities, and 55 were
in rural communities.
To analyze flight delay, cancellation, and diversion trends for these
different airport community size categories, we first obtained data
from DOT. These data were drawn from the Airline Service Quality
Performance System (ASQP), which includes information about flight
delays, cancellations, and diversions. ASQP data are based on
information filed by airlines each month with DOT's Bureau of
Transportation Statistics (Office of Airline Information).[Footnote
117] Airlines with 1 percent or more of total domestic scheduled
passenger service revenue are required to report data for their
flights involving any airport in the 48 contiguous states that account
for one percent or more of domestic scheduled service passenger
enplanements.[Footnote 118] We then compared the percentage of flights
that were delayed, canceled, and diverted by community size, by year.
Since DOT does not require all airlines to report on-time performance
information, we also purchased data from FlightStats, a private data
source from Conducive Technology that records flight performance
information for nearly all airlines and airports. We then conducted
the same analysis of delay, cancellation, and diversion trends by
airport community size as we did with ASQP data. We also verified DOT
and FlightStats data as comparable for similar categories of flights.
ASQP and FlightStats did not have data for all 355 primary airports
subject to our examination, because some airports may not have been
considered primary in other years of our analysis (very small airports
may drop below or rise above the 10,000 enplanements threshold year to
year). Furthermore, some airports may have more than 10,000
enplanements annually but generally not be commercial-service
airports. For instance, some military airports may have commercial
flights diverted to them on occasion. As a result, our analysis of the
year 2010, for example, included data for 281 airports using ASQP data
and for 344 airports using FlightStats data. Using our community size
categories, 76 airports were in large metropolitan communities, 99
were in midsized metropolitan communities, 117 were in small
metropolitan communities, and 52 were in rural communities when using
FlightStats data.
We also examined trends in sources of delay and cancellation, based on
DOT's ASQP data as previously described, and compared these trends by
airport community size. FlightStats does not record the sources or
reasons for delays or cancellations.
To assess the reliability of ASQP and FlightStats data, we reviewed
documentation related to both data sources, and interviewed
knowledgeable officials at DOT and Conducive Technology about the
data. We also compared data for the same categories of flights in both
DOT and FlightStats databases, where possible, and found that they
were similar. After excluding certain ASQP flight records for our
analysis of tarmac delay trends, we determined that both ASQP and
FlightStats data were sufficiently reliable for the purposes of this
report.[Footnote 119]
To better understand the reasons for any differing trends in and
sources of flight delays and cancellations, we reviewed a DOT Office
of Inspector General report and interviewed aviation industry experts,
consumer groups, industry associations, and representatives of three
U.S. legacy airlines and three low-cost airlines. For the U.S.
airlines, we selected three legacy airlines that served more than two-
thirds of all legacy airline passengers from 2004 through June 2010
and three low-cost airlines that served more than 80 percent of all
low-cost airline passengers from 2004 through June 2010. These six
airlines served about half of all passengers enplaned on U.S. airlines
from 2004 through June 2010. See table 5 for a list of aviation
industry stakeholders, including airlines, interviewed for this report.
Table 5: U.S. Stakeholders Interviewed:
Government agencies:
Department of Transportation:
* Federal Aviation Administration;
* Bureau of Transportation Statistics;
* Office of the Assistant General Counsel for Aviation Enforcement &
Proceedings;
* Office of the Assistant General Counsel for International Law.
Academics and consultants:
Peter Belobaba, Massachusetts Institute of Technology;
Amy Cohn, University of Michigan;
Mark Hansen, University of California, Berkeley;
Darryl Jenkins, ACG Aviation Consulting Group, LLC;
Joshua Marks, Marks Aviation, LLC;
Lance Sherry, George Mason University.
Consumer groups:
American Society of Travel Agents;
Association for Airline Passenger Rights;
Business Travel Coalition;
Consumer Travel Alliance;
Consumers Union;
FlyersRights.
Industry associations:
Air Carrier Association of America;
Air Transport Association;
International Air Transport Association;
Regional Airline Association.
U.S. legacy airlines:
American Airlines;
Delta Air Lines;
United Airlines.
U.S. low-cost airlines:
JetBlue Airlines;
Southwest Airlines;
Spirit Airlines.
Source: GAO.
[End of table]
To assess how DOT's tarmac delay rule has affected passengers and
airlines, we first examined DOT data on tarmac delay and cancellation
trends since 2004. In order to identify the frequency of tarmac delays
over time, we used DOT's ASQP data to identify all flights with tarmac
delays greater than 3 hours from January 2004 through September 2010.
[Footnote 120] We then analyzed these flights by year; month; time of
day; and type of tarmac delay, such as taxi-in and taxi-out delays
(see appendix III for more information on these trends since 2004).
To better understand the effect of the tarmac delay rule on the
likelihood of flight cancellations, we assessed cancellations in two
contexts. In the first, we assessed the odds of a flight being
canceled after it leaves the gate. In the second, we assessed the odds
of a flight being canceled before it leaves the gate. In order to
isolate the effect of the tarmac delay rule, we analyzed flight data
using models that controlled for a variety of factors that can affect
an airline's decision to cancel a flight. Specifically, we used
logistic regression models to estimate the impact of the tarmac delay
rule on cancellations. Using these models, we were able to control for
other factors that may affect the likelihood of a cancellation,
including weather at the origin and destination airport, airport and
airline characteristics, and specific details of individual flights.
Disruptive weather is a major cause of cancellations, so by including
variables in our model for severe weather events, we were better able
to isolate the rule's correlation with cancellations. Further, the
size of the particular airport, as well as the size and business
practices of airlines, influence cancellation decisions, so we
controlled for certain characteristics of airports and airlines. (See
appendix V for more details on our models.) To verify the strength of
our model, we discussed the models' design and preliminary results
with aviation experts Professor Mark Hansen of the University of
California and Professor Lance Sherry of George Mason University. We
also spoke with representatives of U.S. airlines, industry
associations, consumer groups, and DOT about the impact of the tarmac
delay rule, including changes to airline practices.
Finally, to determine how the requirements and practices for
protecting passengers from flight delays, cancellations, and denied
boardings in the United States, Canada, and the EU have affected
passengers and airlines, we examined the laws, regulations,
international agreements, and voluntary commitments governing
passenger protections in the three regions. In particular, we reviewed
applicable DOT regulations, Regulation (EC) 261/2004, and relevant
provisions of Canada's Air Transportation Regulations and the Montreal
Convention. Additionally, we examined government guidance and
proposals for additional passenger protections, including the Flight
Rights Canada Initiative, European Commission guidance for enforcement
bodies, and Canada's proposed Air Passenger's Bill of Rights. To
describe voluntary passenger protections offered by airlines, we
reviewed the contracts of carriage for nine largest U.S. airlines
based on recent Federal Aviation Administration (FAA) data on the
number of available seat miles. We also spoke with airline officials
from three airlines in Canada and officials from three European
airlines. To further examine the affect that passenger protection
regulations have had on airlines and passengers, we interviewed
airline, industry association, consumer group, and government
officials throughout all three regions. We also assessed DOT data on
denied boardings from 2004 through 2010.[Footnote 121] To document how
regions enforce passenger protection requirements differently, we
visited and spoke with stakeholders in Canada and in the EU, which
were selected based on stakeholder comments and a review of a recent
EC study on the implementation of the EU regulation. In the EU, we
selected The Netherlands and Germany because each has a large aviation
market as well as active and effective enforcement practices, but
which employ different strategies. See tables 6 and 7 for a list of
stakeholders we met with in Canada and the EU.
Table 6: Canadian Stakeholders Interviewed:
Airlines:
Air Canada;
Air Transat;
WestJet.
Consumer groups:
Consumer Association of Canada.
Government agencies:
Canadian Transportation Agency;
NavCanada;
Transport Canada.
Industry association:
Canadian Airports Council.
Source: GAO.
[End of table]
Table 7: European Stakeholders Interviewed:
Airlines:
EasyJet;
KLM;
Lufthansa.
Consumer groups:
Consumentenbond (Netherlands):
Federation of German Consumer Organisations.
Government agencies:
EC, Directorate-General for Mobility and Transport;
Germany, Federal Ministry of Food, Agriculture and Consumer Protection;
Germany, Federal Ministry of Transport, Building and Urban Affairs;
Germany, Luftfahrt-Bundesamt;
The Netherlands, Ministry of Transport, Inspectorate for Transport,
Public Works and Water Management;
United Kingdom, Civil Aviation Authority.
Industry associations:
Association of European Airlines;
European Low Fares Airline Association;
European Regions Airline Association;
German Airline Federation.
Source: GAO.
[End of table]
[End of section]
Appendix II: Delay, Cancellation, and Diversion Trends:
This appendix provides additional information and illustrations of
flight delay, cancellation, and diversion trends from 2005 to 2010,
based on our analysis of FlightStats data. It also provides
information on airline-reported sources of delays and cancellations,
based on our analysis of DOT data.
Figure 12: Percentage of Departure Flights Canceled and Diverted by
Community Size, Calendar Years 2005-2010:
[Refer to PDF for image: multiple line graph]
Percentage of departure flights that are diverted or canceled:
Year: 2005;
Large metropolitan: 1.57%;
Midsized metropolitan: 1.85%;
Small metropolitan: 2.24%;
Rural: 4.16%.
Year: 2006;
Large metropolitan: 1.58%;
Midsized metropolitan: 1.9%;
Small metropolitan: 2.29%;
Rural: 6.25%.
Year: 2007;
Large metropolitan: 2.18%;
Midsized metropolitan: 2.53%;
Small metropolitan: 2.7%;
Rural: 6.43%.
Year: 2008;
Large metropolitan: 2.03%;
Midsized metropolitan: 2.58%;
Small metropolitan: 3.31%;
Rural: 6.94%.
Year: 2009;
Large metropolitan: 1.51%;
Midsized metropolitan: 2.14%;
Small metropolitan: 2.6%;
Rural: 6.24%.
Year: 2010;
Large metropolitan: 1.94%;
Midsized metropolitan: 2.26%;
Small metropolitan: 2.54%;
Rural: 7.96%.
Source: GAO analysis of FlightStats data.
[End of figure]
Figure 13: Percentage of Arrival Flights Canceled and Diverted by
Community Size, Calendar Years 2005-2010:
[Refer to PDF for image: multiple line graph]
Percentage of arrivals that are diverted or canceled:
Year: 2005;
Large metropolitan: 1.58%;
Midsized metropolitan: 1.66%;
Small metropolitan: 2.1%;
Rural: 3.79%.
Year: 2006;
Large metropolitan: 1.61%;
Midsized metropolitan: 1.65%;
Small metropolitan: 2.01%;
Rural: 4.72%.
Year: 2007;
Large metropolitan: 2.17%;
Midsized metropolitan: 2.25%;
Small metropolitan: 2.7%;
Rural: 4.75%.
Year: 2008;
Large metropolitan: 2.18%;
Midsized metropolitan: 2.27%;
Small metropolitan: 2.93%;
Rural: 9.11%.
Year: 2009;
Large metropolitan: 1.58%;
Midsized metropolitan: 1.7%;
Small metropolitan: 2.01%;
Rural: 4.18%.
Year: 2010;
Large metropolitan: 1.97%;
Midsized metropolitan: 1.87%;
Small metropolitan: 2.11%;
Rural: 5.65%.
Source: GAO analysis of FlightStats data.
[End of figure]
Figure 14: Percentage of Delayed Departure Flights by Community Size,
Calendar Years 2006-2010:
[Refer to PDF for image: multiple line graph]
Percentage of departures:
Year: 2006;
Large metropolitan: 13.61%;
Midsized metropolitan: 11.72%;
Small metropolitan: 11.16%;
Rural: 13.16%.
Year: 2007;
Large metropolitan: 16.51%;
Midsized metropolitan: 13.45%;
Small metropolitan: 13.16%;
Rural: 14.09%.
Year: 2008;
Large metropolitan: 13.29%;
Midsized metropolitan: 11.32%;
Small metropolitan: 10.97%;
Rural: 14.23%.
Year: 2009;
Large metropolitan: 15.44%;
Midsized metropolitan: 11.4%;
Small metropolitan: 10.89%;
Rural: 14.01%.
Year: 2010;
Large metropolitan: 14.39%;
Midsized metropolitan: 10.87%;
Small metropolitan: 10.88%;
Rural: 11.84%.
Source: GAO analysis of FlightStats data.
[End of figure]
Figure 15: Percentage of Delayed Arrival Flights by Community Size,
Calendar Years 2006-2010:
[Refer to PDF for image: multiple line graph]
Percentage of arrivals:
Year: 2006;
Large metropolitan: 22.72%;
Mid-sized metropolitan: 22.71%;
Small metropolitan: 19.89%;
Rural: 20.59%.
Year: 2007;
Large metropolitan: 24.5%;
Mid-sized metropolitan: 24.17%;
Small metropolitan: 22.06%;
Rural: 22.81%.
Year: 2008;
Large metropolitan: 22.34%;
Mid-sized metropolitan: 21.69%;
Small metropolitan: 19.99%;
Rural: 22.93%.
Year: 2009;
Large metropolitan: 19.57%;
Mid-sized metropolitan: 19.24%;
Small metropolitan: 18.25%;
Rural: 22.27%.
Year: 2010;
Large metropolitan: 19.59%;
Mid-sized metropolitan: 20.05%;
Small metropolitan: 18.72%;
Rural: 21.65%.
Source: GAO analysis of FlightStats data.
[End of figure]
Figure 16: Airline-Reported Sources of Delay in Calendar Year 2010:
[Refer to PDF for image: pie-chart and horizontal bar graph]
Sources of delay, Systemwide:
Extreme weather: 4%;
National aviation system: 26%;
Airline: 30%;
Late arriving aircraft: 39%.
Sources of delay, by community size:
Large metropolitan:
Extreme weather: 4%;
National aviation system: 28%;
Airline: 29%;
Late arriving aircraft: 39%.
Mid-sized metropolitan:
Extreme weather: 5%;
National aviation system: 15%;
Airline: 39%;
Late arriving aircraft: 41%.
Small metropolitan:
Extreme weather: 5%;
National aviation system: 16%;
Airline: 39%;
Late arriving aircraft: 40%.
Rural:
Extreme weather: 7%;
National aviation system: 16%;
Airline: 35%;
Late arriving aircraft: 42%.
Source: GAO analysis of DOT data.
Notes:
1. Numbers may not add to 100 due to rounding.
2. Security delays do not appear this graphic because they make up
less than 0.5 percent of the delays.
[End of figure]
Figure 17: Airline-Reported Sources of Cancellations in Calendar Year
2010:
[Refer to PDF for image: pie-chart and horizontal bar graph]
Sources of delay, Systemwide:
National aviation system: 13%;
Airline: 37%;
Extreme weather: 50%.
Sources of delay, by community size:
Large metropolitan:
National aviation system: 13%;
Airline: 36%;
Extreme weather: 51%.
Mid-sized metropolitan:
National aviation system: 15%;
Airline: 43%;
Extreme weather: 42%.
Small metropolitan:
National aviation system: 14%;
Airline: 40%;
Extreme weather: 46%.
Rural:
National aviation system: 5%;
Airline: 20%;
Extreme weather: 75%.
Source: GAO analysis of DOT data.
[End of figure]
Notes:
[Refer to PDF for image]
1. Numbers may not add to 100 due to rounding.
2. Security delays do not appear on this graphic because they make up
less than 0.5 percent of cancellations.
[End of figure]
[End of section]
Appendix III: Tarmac Delay Trends Since 2004:
This appendix provides additional information and illustration of
tarmac delays of more than 3 hours, from January 2004 through
September 2010, including how the tarmac delays that occurred during
this period were distributed by year, month, airport, day of the week,
and hour. This appendix also provides information on airline-reported
sources of tarmac delays. This information is based on our analysis of
DOT data.
Table 8: Total Number of Tarmac Delays of More than 3 Hours by Stage
of Operation, January 2004-September 2010:
Total by year: 2004;
Taxi-out tarmac delays: 1,169;
Taxi-in tarmac delays: 70.
Total by year: 2005;
Taxi-out tarmac delays: 983;
Taxi-in tarmac delays: 32.
Total by year: 2006;
Taxi-out tarmac delays: 1,220;
Taxi-in tarmac delays: 58.
Total by year: 2007;
Taxi-out tarmac delays: 1,507;
Taxi-in tarmac delays: 30.
Total by year: 2008;
Taxi-out tarmac delays: 1,056;
Taxi-in tarmac delays: 17.
Total by year: 2009;
Taxi-out tarmac delays: 520;
Taxi-in tarmac delays: 2.
Total by year: 2010;
Taxi-out tarmac delays: 72;
Taxi-in tarmac delays: 4.
Total by year: Total;
Taxi-out tarmac delays: 6,527;
Taxi-in tarmac delays: 213.
Source: GAO analysis of DOT data.
Note: Beginning in October 2008, DOT required airlines to submit
tarmac delay statistics for three additional categories: flights that
are subsequently canceled or diverted or have multiple gate
departures. For consistency, we have omitted these flights from this
figure. Our analysis also excluded flights reported to DOT by Atlantic
Southeast Airlines, Hawaiian Airlines, and Comair because, according
to DOT, these airlines inconsistently reported gate returns, prior to
October 2008. Specifically, some airlines mis-reported the gate-
departure or takeoff time resulting in an incorrect tarmac delay
calculation. Further, our analysis excluded data for any flights with
tarmac delays of more than 10 hours since there were errors in the
reporting of many such flights. Finally, our analysis did not include
data for international flights. As a result of these exclusions for
data reliability purposes, this figure likely under-reports tarmac
delays of more than 3 hours for these years. See appendix IV for more
information on recent tarmac delays.
[End of table]
Figure 18: Total Number of Tarmac Delays of More than 3 Hours by
Month, January 2004-September 2010:
[Refer to PDF for image: line graph]
Month: January;
Tarmac delays of more than 3 hours: 488.
Month: February;
Tarmac delays of more than 3 hours: 377.
Month: March;
Tarmac delays of more than 3 hours: 336.
Month: April;
Tarmac delays of more than 3 hours: 188.
Month: May;
Tarmac delays of more than 3 hours: 574.
Month: June;
Tarmac delays of more than 3 hours: 1,448.
Month: July;
Tarmac delays of more than 3 hours: 1,446.
Month: August;
Tarmac delays of more than 3 hours: 656.
Month: September;
Tarmac delays of more than 3 hours: 313.
Month: October;
Tarmac delays of more than 3 hours: 299.
Month: November;
Tarmac delays of more than 3 hours: 215.
Month: December;
Tarmac delays of more than 3 hours: 400.
Source: GAO analysis of DOT data.
Note: Beginning in October 2008, DOT required airlines to submit
tarmac delay statistics for three additional categories: flights that
are subsequently canceled or diverted or have multiple gate
departures. For consistency, we have omitted these flights from this
figure. Our analysis also excluded flights reported to DOT by Atlantic
Southeast Airlines, Hawaiian Airlines, and Comair because, according
to DOT, these airlines inconsistently reported gate returns, prior to
October 2008. Specifically, some airlines mis-reported the gate-
departure or takeoff time resulting in an incorrect tarmac delay
calculation. Further, our analysis excluded data for any flights with
tarmac delays of more than 10 hours since there were errors in the
reporting of many such flights. Finally, our analysis did not include
data for international flights. As a result of these exclusions for
data reliability purposes, this figure likely under-reports tarmac
delays of more than 3 hours for these years. See appendix IV for more
information on recent tarmac delays.
[End of figure]
Figure 19: Airline-Reported Sources of Delay for all Tarmac Delays of
More than 3 Hours, January 2004-September 2010:
[Refer to PDF for image: pie-chart]
National airspace system: 86.4%;
Late arriving aircraft: 6.2%;
Extreme weather: 4.8%;
Air carrier: 2.6%.
Source: GAO analysis of DOT data.
Notes:
1. Numbers may not add up to 100 due to rounding.
2. Security delays do not appear on this graphic because they make up
less than 1 percent of delays at these airports. DOT collects
cancellation causal data in one of four categories: (1) national
aviation system (a broad set of circumstances affecting airline
flights, such as nonextreme weather that slows down the system, but
does not prevent flying), (2) airline (any delay that was within the
control of the airlines, such as aircraft cleaning, baggage loading,
crew issues, or maintenance), (3) extreme weather (serious weather
conditions that prevent the operation of a flight, such as tornadoes,
snowstorms, or hurricanes), and (4) security (such as, evacuation of
an airport, reboarding because of a security breach, and long lines at
the passenger screening areas). According to DOT, airlines cannot
report "late arriving aircraft" as a reason for a canceled flight
(though they can assign that cause to delayed flights) since
ultimately this reason would fall under the airline category because
airlines can have a spare aircraft available when flights arrive late.
Therefore, "late arriving aircraft" is an airline related cause for a
cancellation that an airline could have prevented. Beginning in
October 2008, DOT required airlines to submit tarmac delay statistics
for three additional categories: flights that are subsequently
canceled or diverted or have multiple gate departures, prior to
October 2008. For consistency, we have omitted these flights from this
figure. Our analysis also excluded flights reported to DOT by Atlantic
Southeast Airlines, Hawaiian Airlines, and Comair because, according
to DOT, these airlines inconsistently reported gate returns.
Specifically, some airlines mis-reported the gate-departure or takeoff
time resulting in an incorrect tarmac delay calculation. Further, our
analysis excluded data for any flights with tarmac delays of more than
10 hours since there were errors in the reporting of many such
flights. Finally, our analysis did not include data for international
flights. As a result of these exclusions for data reliability
purposes, this figure likely under-reports tarmac delays of more than
3 hours for these years. See appendix IV for more information on
recent tarmac delays.
[End of figure]
Table 9: Total Number of Tarmac Delays of More than 3 Hours at 35
Airports and Systemwide, January 2004-September 2010:
Airport: Atlanta Hartsfield International;
Taxi-out tarmac delays: 240;
Taxi-in tarmac delays: 3.
Airport: Baltimore-Washington International;
Taxi-out tarmac delays: 65;
Taxi-in tarmac delays: 6.
Airport: Boston Logan International;
Taxi-out tarmac delays: 140;
Taxi-in tarmac delays: 7.
Airport: Charlotte/Douglas International;
Taxi-out tarmac delays: 123;
Taxi-in tarmac delays: 2.
Airport: Chicago Midway;
Taxi-out tarmac delays: 39;
Taxi-in tarmac delays: 0.
Airport: Chicago O'Hare International;
Taxi-out tarmac delays: 736;
Taxi-in tarmac delays: 3.
Airport: Cincinnati-Northern Kentucky;
Taxi-out tarmac delays: 75;
Taxi-in tarmac delays: 3.
Airport: Cleveland-Hopkins International;
Taxi-out tarmac delays: 29;
Taxi-in tarmac delays: 1.
Airport: Dallas-Fort Worth International;
Taxi-out tarmac delays: 436;
Taxi-in tarmac delays: 11.
Airport: Denver International;
Taxi-out tarmac delays: 75;
Taxi-in tarmac delays: 2.
Airport: Detroit Metro Wayne County;
Taxi-out tarmac delays: 51;
Taxi-in tarmac delays: 2.
Airport: Fort Lauderdale-Hollywood International;
Taxi-out tarmac delays: 25;
Taxi-in tarmac delays: 3.
Airport: George Bush Intercontinental;
Taxi-out tarmac delays: 284;
Taxi-in tarmac delays: 5.
Airport: Greater Pittsburgh International;
Taxi-out tarmac delays: 28;
Taxi-in tarmac delays: 1.
Airport: Lambert St. Louis International;
Taxi-out tarmac delays: 17;
Taxi-in tarmac delays: 0.
Airport: Las Vegas McCarran International;
Taxi-out tarmac delays: 10;
Taxi-in tarmac delays: 8.
Airport: Los Angeles International;
Taxi-out tarmac delays: 15;
Taxi-in tarmac delays: 17.
Airport: Memphis International;
Taxi-out tarmac delays: 26;
Taxi-in tarmac delays: 2.
Airport: Miami International;
Taxi-out tarmac delays: 57;
Taxi-in tarmac delays: 3.
Airport: Minneapolis-St Paul International;
Taxi-out tarmac delays: 55;
Taxi-in tarmac delays: 2.
Airport: New York John F. Kennedy International;
Taxi-out tarmac delays: 888;
Taxi-in tarmac delays: 0.
Airport: New York LaGuardia;
Taxi-out tarmac delays: 495;
Taxi-in tarmac delays: 6.
Airport: Newark International;
Taxi-out tarmac delays: 865;
Taxi-in tarmac delays: 1.
Airport: Orlando International;
Taxi-out tarmac delays: 27;
Taxi-in tarmac delays: 6.
Airport: Philadelphia International;
Taxi-out tarmac delays: 588;
Taxi-in tarmac delays: 2.
Airport: Phoenix Sky Harbor International;
Taxi-out tarmac delays: 16;
Taxi-in tarmac delays: 2.
Airport: Portland International;
Taxi-out tarmac delays: 1;
Taxi-in tarmac delays: 2.
Airport: Ronald Reagan National;
Taxi-out tarmac delays: 143;
Taxi-in tarmac delays: 4.
Airport: Salt Lake City International;
Taxi-out tarmac delays: 34;
Taxi-in tarmac delays: 1.
Airport: San Diego International Lindbergh;
Taxi-out tarmac delays: 0;
Taxi-in tarmac delays: 6.
Airport: San Francisco International;
Taxi-out tarmac delays: 6;
Taxi-in tarmac delays: 5.
Airport: Seattle -Tacoma International;
Taxi-out tarmac delays: 4;
Taxi-in tarmac delays: 4.
Airport: Tampa International;
Taxi-out tarmac delays: 16;
Taxi-in tarmac delays: 5.
Airport: Washington Dulles International;
Taxi-out tarmac delays: 208;
Taxi-in tarmac delays: 4.
Airport: Systemwide;
Taxi-out tarmac delays: 6,527;
Taxi-in tarmac delays: 213.
Source: GAO analysis of DOT data.
Note: According to FAA, the 35 Operational Evolution Partnership
airports listed above are commercial airports with significant
activity and were selected in 2000 on the basis of lists from FAA and
Congress as well as a study that identified the most congested
airports in the United States. Beginning in October 2008, DOT required
airlines to submit tarmac delay statistics for three additional
categories: flights that are subsequently canceled or diverted or have
multiple gate departures. For consistency, we have omitted these
flights from this figure. Our analysis also excluded flights reported
to DOT by Atlantic Southeast Airlines, Hawaiian Airlines, and Comair
because, according to DOT, these airlines inconsistently reported gate
returns. Specifically, some airlines miss-reported the gate-departure
or takeoff time resulting in an incorrect tarmac delay calculation.
Further, our analysis excluded data for any flights with tarmac delays
of more than 10 hours since there were errors in the reporting of many
such flights. Finally, our analysis did not include data for
international flights. As a result of these exclusions for data
reliability purposes, this figure likely under-reports tarmac delays
of more than 3 hours for these years. See appendix IV for more
information on recent tarmac delays.
[End of table]
Table 10: Total Number of Tarmac Delays of More than 3 Hours by Day of
the Week, January 2004-September 2010:
Day of the week: Monday;
Number of tarmac delays: 714.
Day of the week: Tuesday;
Number of tarmac delays: 1,145.
Day of the week: Wednesday;
Number of tarmac delays: 1,395.
Day of the week: Thursday;
Number of tarmac delays: 1,375.
Day of the week: Friday;
Number of tarmac delays: 914.
Day of the week: Saturday;
Number of tarmac delays: 431.
Day of the week: Sunday;
Number of tarmac delays: 766.
Day of the week: Total;
Number of tarmac delays: 6,740.
Source: GAO analysis of DOT data.
Note: Beginning in October 2008, DOT required airlines to submit
tarmac delay statistics for three additional categories: flights that
are subsequently canceled or diverted or have multiple gate
departures. For consistency, we have omitted these flights from this
figure. Our analysis also excluded flights reported to DOT by Atlantic
Southeast Airlines, Hawaiian Airlines, and Comair because, according
to DOT, these airlines inconsistently reported gate returns, prior to
October 2008. Specifically, some airlines mis-reported the gate-
departure or takeoff time resulting in an incorrect tarmac delay
calculation. Further, our analysis excluded data for any flights with
tarmac delays of more than 10 hours since there were errors in the
reporting of many such flights. Finally, our analysis did not include
data for international flights. As a result of these exclusions for
data reliability purposes, this figure likely under-reports tarmac
delays of more than 3 hours for these years. See appendix IV for more
information on recent tarmac delays.
[End of table]
Table 11: Total Number of Tarmac Delays of More than 3 Hours by Hour,
January 2004-September 2010:
Departure hour: 12:00 a.m.;
Taxi-out tarmac delays: 1;
Taxi-in tarmac delays: 0;
Total: 1.
Departure hour: 5:00 a.m.;
Taxi-out tarmac delays: 23;
Taxi-in tarmac delays: 0;
Total: 23.
Departure hour: 6:00 a.m.;
Taxi-out tarmac delays: 135;
Taxi-in tarmac delays: 5;
Total: 140.
Departure hour: 7:00 a.m.;
Taxi-out tarmac delays: 229;
Taxi-in tarmac delays: 13;
Total: 242.
Departure hour: 8:00 a.m.;
Taxi-out tarmac delays: 215;
Taxi-in tarmac delays: 10;
Total: 225.
Departure hour: 9:00 a.m.;
Taxi-out tarmac delays: 192;
Taxi-in tarmac delays: 4;
Total: 196.
Departure hour: 10:00 a.m.;
Taxi-out tarmac delays: 152;
Taxi-in tarmac delays: 5;
Total: 157.
Departure hour: 11:00 a.m.;
Taxi-out tarmac delays: 193;
Taxi-in tarmac delays: 14;
Total: 207.
Departure hour: 12:00 p.m.;
Taxi-out tarmac delays: 269;
Taxi-in tarmac delays: 18;
Total: 287.
Departure hour: 1:00 p.m.;
Taxi-out tarmac delays: 526;
Taxi-in tarmac delays: 13;
Total: 539.
Departure hour: 2:00 p.m.;
Taxi-out tarmac delays: 557;
Taxi-in tarmac delays: 20;
Total: 577.
Departure hour: 3:00 p.m.;
Taxi-out tarmac delays: 747;
Taxi-in tarmac delays: 21;
Total: 768.
Departure hour: 4:00 p.m.;
Taxi-out tarmac delays: 716;
Taxi-in tarmac delays: 21;
Total: 737.
Departure hour: 5:00 p.m.;
Taxi-out tarmac delays: 894;
Taxi-in tarmac delays: 27;
Total: 921.
Departure hour: 6:00 p.m.;
Taxi-out tarmac delays: 717;
Taxi-in tarmac delays: 19;
Total: 736.
Departure hour: 7:00 p.m.;
Taxi-out tarmac delays: 507;
Taxi-in tarmac delays: 9;
Total: 516.
Departure hour: 8:00 p.m.;
Taxi-out tarmac delays: 331;
Taxi-in tarmac delays: 9;
Total: 340.
Departure hour: 9:00 p.m.;
Taxi-out tarmac delays: 94;
Taxi-in tarmac delays: 3;
Total: 97.
Departure hour: 10:00 p.m.;
Taxi-out tarmac delays: 25;
Taxi-in tarmac delays: 1;
Total: 26.
Departure hour: 11:00 p.m.;
Taxi-out tarmac delays: 4;
Taxi-in tarmac delays: 1;
Total: 5.
Departure hour: Total;
Taxi-out tarmac delays: 6,527;
Taxi-in tarmac delays: 213;
Total: 6,740.
Source: GAO analysis of DOT data.
Note: Beginning in October 2008, DOT required airlines to submit
tarmac delay statistics for three additional categories: flights that
are subsequently canceled or diverted or have multiple gate
departures. For consistency, we have omitted these flights from this
figure. Our analysis also excluded flights reported to DOT by Atlantic
Southeast Airlines, Hawaiian Airlines, and Comair because, according
to DOT, these airlines inconsistently reported gate returns, prior to
October 2008. Specifically, some airlines mis-reported the gate-
departure or takeoff time resulting in an incorrect tarmac delay
calculation. Further, our analysis excluded data for any flights with
tarmac delays of more than 10 hours since there were errors in the
reporting of many such flights. Finally, our analysis did not include
data for international flights. As a result of these exclusions for
data reliability purposes, this figure likely under-reports tarmac
delays of more than 3 hours for these years. See appendix IV for more
information on recent tarmac delays.
[End of table]
[End of section]
Appendix IV: Tarmac Delay Trends since October 2008:
Beginning in October 2008, DOT required airlines to submit data on
flights with tarmac delays that were subsequently canceled, diverted,
or had multiple gate departures (see table 12). Previously, DOT had
only captured tarmac delays that occurred during taxi-out or during
taxi-in. While the majority of tarmac delays happen at taxi-out or
taxi-in, the change in reporting captured data for some additional
tarmac delays of more than 3 hours. As a result of these new reporting
requirements, tarmac delays are now captured:
* during taxi-out: the time between when a flight departs the gate at
the origin airport and when it lifts off from that airport (wheels-
off);
* during taxi-in: the time between a flight touching down at its
destination airport (wheels-on) and arriving at the gate;
* prior to cancellation: flight left the gate but was canceled at the
origin airport;
* during a diversion: the tarmac time experienced at an airport other
than the destination airport; or:
* as a result of a multiple gate departure: the flight left the gate,
then returned, and then left again; the tarmac time is the time before
the return to the gate.
Table 12: Tarmac Delays of More than 3 Hours, October 2008-December
2010:
Tarmac times: October 2008: Total.
Month: October 2008;
Tarmac times: Total: 46;
Tarmac times: Percentage of regularly scheduled flights: 0.01%;
Stage of operation:
Prior To cancellation: 2;
Multiple gate departure: 5;
Taxi-out: 33;
Taxi-in: 0;
At diversion airport: 6;
Number of regularly scheduled flights: 556,205.
Month: November 2008;
Tarmac times: Total: 7;
Tarmac times: Percentage of regularly scheduled flights: 0;
Stage of operation:
Prior To cancellation: 0;
Multiple gate departure: 1;
Taxi-out: 4;
Taxi-in: 0;
At diversion airport: 2;
Number of regularly scheduled flights: 523,272.
Month: December 2008;
Tarmac times: Total: 183;
Tarmac times: Percentage of regularly scheduled flights: 0.03;
Stage of operation:
Prior To cancellation: 40;
Multiple gate departure: 13;
Taxi-out: 113;
Taxi-in: 7;
At diversion airport: 10;
Number of regularly scheduled flights: 544,956.
Month: January[A] 2009;
Tarmac times: Total: 85;
Tarmac times: Percentage of regularly scheduled flights: 0.02;
Stage of operation:
Prior To cancellation: 7;
Multiple gate departure: 10;
Taxi-out: 68;
Taxi-in: 0;
At diversion airport: 0;
Number of regularly scheduled flights: 532,339.
Month: February 2009;
Tarmac times: Total: 40;
Tarmac times: Percentage of regularly scheduled flights: 0.01;
Stage of operation:
Prior To cancellation: 5;
Multiple gate departure: 3;
Taxi-out: 32;
Taxi-in: 0;
At diversion airport: 0;
Number of regularly scheduled flights: 488,410.
Month: March 2009;
Tarmac times: Total: 85;
Tarmac times: Percentage of regularly scheduled flights: 0.02;
Stage of operation:
Prior To cancellation: 6;
Multiple gate departure: 9;
Taxi-out: 63;
Taxi-in: 0;
At diversion airport: 7;
Number of regularly scheduled flights: 557,422.
Month: April 2009;
Tarmac times: Total: 74;
Tarmac times: Percentage of regularly scheduled flights: 0.01;
Stage of operation:
Prior To cancellation: 10;
Multiple gate departure: 8;
Taxi-out: 45;
Taxi-in: 0;
At diversion airport: 11;
Number of regularly scheduled flights: 537,793.
Month: May 2009;
Tarmac times: Total: 34;
Tarmac times: Percentage of regularly scheduled flights: 0.01;
Stage of operation:
Prior To cancellation: 7;
Multiple gate departure: 2;
Taxi-out: 24;
Taxi-in: 1;
At diversion airport: 0;
Number of regularly scheduled flights: 546,832.
Month: June 2009;
Tarmac times: Total: 268;
Tarmac times: Percentage of regularly scheduled flights: 0.05;
Stage of operation:
Prior To cancellation: 38;
Multiple gate departure: 40;
Taxi-out: 167;
Taxi-in: 1;
At diversion airport: 22;
Number of regularly scheduled flights: 557,594.
Month: July 2009;
Tarmac times: Total: 161;
Tarmac times: Percentage of regularly scheduled flights: 0.03;
Stage of operation:
Prior To cancellation: 21;
Multiple gate departure: 20;
Taxi-out: 102;
Taxi-in: 0;
At diversion airport: 18;
Number of regularly scheduled flights: 580,134.
Month: August 2009;
Tarmac times: Total: 66;
Tarmac times: Percentage of regularly scheduled flights: 0.01;
Stage of operation:
Prior To cancellation: 6;
Multiple gate departure: 10;
Taxi-out: 43;
Taxi-in: 0;
At diversion airport: 7;
Number of regularly scheduled flights: 568,301.
Month: September 2009;
Tarmac times: Total: 6;
Tarmac times: Percentage of regularly scheduled flights: 0;
Stage of operation:
Prior To cancellation: 0;
Multiple gate departure: 0;
Taxi-out: 4;
Taxi-in: 0;
At diversion airport: 2;
Number of regularly scheduled flights: 510,852.
Month: October 2009;
Tarmac times: Total: 11;
Tarmac times: Percentage of regularly scheduled flights: 0;
Stage of operation:
Prior To cancellation: 0;
Multiple gate departure: 0;
Taxi-out: 11;
Taxi-in: 0;
At diversion airport: 0;
Number of regularly scheduled flights: 531,799.
Month: November 2009;
Tarmac times: Total: 4;
Tarmac times: Percentage of regularly scheduled flights: 0;
Stage of operation:
Prior To cancellation: 0;
Multiple gate departure: 1;
Taxi-out: 2;
Taxi-in: 0;
At diversion airport: 1;
Number of regularly scheduled flights: 509,540.
Month: December 2009;
Tarmac times: Total: 34;
Tarmac times: Percentage of regularly scheduled flights: 0.01;
Stage of operation:
Prior To cancellation: 4;
Multiple gate departure: 3;
Taxi-out: 22;
Taxi-in: 0;
At diversion airport: 5;
Number of regularly scheduled flights: 529,269.
Month: January 2010;
Tarmac times: Total: 20;
Tarmac times: Percentage of regularly scheduled flights: 0;
Stage of operation:
Prior To cancellation: 1;
Multiple gate departure: 3;
Taxi-out: 11;
Taxi-in: 2;
At diversion airport: 3;
Number of regularly scheduled flights: 521,809.
Month: February 2010;
Tarmac times: Total: 60;
Tarmac times: Percentage of regularly scheduled flights: 0.01;
Stage of operation:
Prior To cancellation: 5;
Multiple gate departure: 1;
Taxi-out: 52;
Taxi-in: 1;
At diversion airport: 1;
Number of regularly scheduled flights: 483,270.
Month: March 2010;
Tarmac times: Total: 25;
Tarmac times: Percentage of regularly scheduled flights: 0;
Stage of operation:
Prior To cancellation: 9;
Multiple gate departure: 2;
Taxi-out: 11;
Taxi-in: 1;
At diversion airport: 2;
Number of regularly scheduled flights: 549,262.
Month: April 2010;
Tarmac times: Total: 4;
Tarmac times: Percentage of regularly scheduled flights: 0;
Stage of operation:
Prior To cancellation: 0;
Multiple gate departure: 0;
Taxi-out: 1;
Taxi-in: 0;
At diversion airport: 3;
Number of regularly scheduled flights: 529,330.
Month: May 2010;
Tarmac times: Total: 1;
Tarmac times: Percentage of regularly scheduled flights: 0;
Stage of operation:
Prior To cancellation: 0;
Multiple gate departure: 0;
Taxi-out: 1;
Taxi-in: 0;
At diversion airport: 0;
Number of regularly scheduled flights: 542,747.
Month: June 2010;
Tarmac times: Total: 3;
Tarmac times: Percentage of regularly scheduled flights: 0;
Stage of operation:
Prior To cancellation: 2;
Multiple gate departure: 0;
Taxi-out: 1;
Taxi-in: 0;
At diversion airport: 0;
Number of regularly scheduled flights: 551,687.
Month: July 2010;
Tarmac times: Total: 3;
Tarmac times: Percentage of regularly scheduled flights: 0;
Stage of operation:
Prior To cancellation: 3;
Multiple gate departure: 0;
Taxi-out: 0;
Taxi-in: 0;
At diversion airport: 0;
Number of regularly scheduled flights: 570,788.
Month: August 2010;
Tarmac times: Total: 1;
Tarmac times: Percentage of regularly scheduled flights: 0;
Stage of operation:
Prior To cancellation: 0;
Multiple gate departure: 0;
Taxi-out: 0;
Taxi-in: 0;
At diversion airport: 1;
Number of regularly scheduled flights: 569,217.
Month: September 2010;
Tarmac times: Total: 4;
Tarmac times: Percentage of regularly scheduled flights: 0;
Stage of operation:
Prior To cancellation: 2;
Multiple gate departure: 2;
Taxi-out: 0;
Taxi-in: 0;
At diversion airport: 0;
Number of regularly scheduled flights: 526,107.
Month: October 2010;
Tarmac times: Total: 0;
Tarmac times: Percentage of regularly scheduled flights: 0;
Stage of operation:
Prior To cancellation: 0;
Multiple gate departure: 0;
Taxi-out: 0;
Taxi-in: 0;
At diversion airport: 0;
Number of regularly scheduled flights: 545,519.
Month: November 2010;
Tarmac times: Total: 0;
Tarmac times: Percentage of regularly scheduled flights: 0;
Stage of operation:
Prior To cancellation: 0;
Multiple gate departure: 0;
Taxi-out: 0;
Taxi-in: 0;
At diversion airport: 0;
Number of regularly scheduled flights: 520,999.
Month: December 2010;
Tarmac times: Total: 3;
Tarmac times: Percentage of regularly scheduled flights: 0;
Stage of operation:
Prior To cancellation: 1;
Multiple gate departure: 1;
Taxi-out: 0;
Taxi-in: 0;
At diversion airport: 1;
Number of regularly scheduled flights: 539,382.
Source: DOT.
[A] According to DOT, January 2009 includes one flight with two
separate 3-hour tarmac times. Northwest Flight 1491, on January 28,
2009, was on the tarmac for 188 minutes before returning to the gate.
The flight departed the gate a second time and was on the tarmac for
199 minutes before wheels-off. Details of the flight are listed as a 3-
hour multiple gate departure delay and a 3-hour taxi-out delay.
[End of table]
Table 13: Tarmac Delays of More than 3 Hours, May 2010-April 2011:
Date: 5/28/10;
Stage of flight: During taxi-out;
Airline: Delta Air Lines;
Origin: ATL;
Destination: DFW;
Delay or cancellation cause: National aviation system;
Minutes over 3 hours: 2;
DOT outcome: No violation.
Date: 6/18/10;
Stage of flight: Prior to cancellation;
Airline: United Airlines;
Origin: ORD;
Destination: LGA;
Delay or cancellation cause: Extreme weather;
Minutes over 3 hours: 5;
DOT outcome: Violation; airline warned.
Date: 6/18/10;
Stage of flight: Prior to cancellation;
Airline: United Airlines;
Origin: ORD;
Destination: ATL;
Delay or cancellation cause: Extreme weather;
Minutes over 3 hours: 3;
DOT outcome: Violation; airline warned.
Date: 6/18/10;
Stage of flight: During taxi-out;
Airline: United Airlines;
Origin: ORD;
Destination: IAD;
Delay or cancellation cause: National aviation system;
Minutes over 3 hours: 2;
DOT outcome: Violation; airline warned.
Date: 7/23/10;
Stage of flight: Prior to cancellation;
Airline: American Eagle;
Origin: ORD;
Destination: BNA;
Delay or cancellation cause: Extreme weather;
Minutes over 3 hours: 34;
DOT outcome: No violation.
Date: 7/23/10;
Stage of flight: Prior to cancellation;
Airline: American Eagle;
Origin: ORD;
Destination: RDU;
Delay or cancellation cause: Extreme weather;
Minutes over 3 hours: 19;
DOT outcome: No violation.
Date: 7/23/10;
Stage of flight: Prior to cancellation;
Airline: American Eagle;
Origin: ORD;
Destination: BWI;
Delay or cancellation cause: Extreme weather;
Minutes over 3 hours: 18;
DOT outcome: No violation.
Date: 8/5/10;
Stage of flight: At diversion airport;
Airline: United Airlines;
Origin: SJU;
Destination: IAD;
Delay or cancellation cause: n/a;
Minutes over 3 hours: 20;
DOT outcome: No violation.
Date: 9/22/10;
Stage of flight: Prior to cancellation;
Airline: US Airways;
Origin: JFK;
Destination: CLT;
Delay or cancellation cause: Air carrier;
Minutes over 3 hours: 17;
DOT outcome: No violation.
Date: 9/22/10;
Stage of flight: Prior to cancellation;
Airline: Delta Air Lines;
Origin: PHL;
Destination: DTW;
Delay or cancellation cause: Extreme weather;
Minutes over 3 hours: 3;
DOT outcome: No violation.
Date: 9/22/10;
Stage of flight: During multiple gate departure;
Airline: Southwest;
Origin: PHL;
Destination: STL;
Delay or cancellation cause: Extreme weather and national aviation
system;
Minutes over 3 hours: 19;
DOT outcome: Violation; airline warned.
Date: 9/22/10;
Stage of flight: During multiple gate departure;
Airline: Pinnacle;
Origin: JFK;
Destination: ORD;
Delay or cancellation cause: Extreme weather and national aviation
system;
Minutes over 3 hours: 5;
DOT outcome: No violation.
Date: 12/12/10;
Stage of flight: During multiple gate departure;
Airline: Delta Air Lines;
Origin: DTW;
Destination: MIA;
Delay or cancellation cause: Extreme weather and national aviation
system;
Minutes over 3 hours: 12;
DOT outcome: No violation.
Date: 12/12/10;
Stage of flight: Prior to cancellation;
Airline: Pinnacle;
Origin: DTW;
Destination: CID;
Delay or cancellation cause: Extreme weather;
Minutes over 3 hours: 6;
DOT outcome: No violation.
Date: 12/27/10;
Stage of flight: At diversion airport;
Airline: United Airlines;
Origin: SAN;
Destination: SFO;
Delay or cancellation cause: n/a;
Minutes over 3 hours: 9;
DOT outcome: No violation.
Date: 1/10/11;
Stage of flight: Prior to cancellation;
Airline: Delta Air Lines;
Origin: ATL;
Destination: HNL;
Delay or cancellation cause: Extreme weather;
Minutes over 3 hours: 31;
DOT outcome: Under investigation.
Date: 4/24/11;
Stage of flight: Prior to cancellation;
Airline: United Airlines;
Origin: JFK;
Destination: SFO;
Delay or cancellation cause: Extreme weather;
Minutes over 3 hours: 22;
DOT outcome: Under investigation.
Date: 4/27/11;
Stage of flight: Prior to cancellation;
Airline: Delta Air Lines;
Origin: ATL;
Destination: SLC;
Delay or cancellation cause: Air carrier;
Minutes over 3 hours: 22;
DOT outcome: Under investigation.
Date: 4/27/11;
Stage of flight: Prior to cancellation;
Airline: Delta Air Lines;
Origin: ATL;
Destination: ONT;
Delay or cancellation cause: Air carrier;
Minutes over 3 hours: 20;
DOT outcome: Under investigation.
Date: 4/27/11;
Stage of flight: During taxi-out;
Airline: Delta Air LInes;
Origin: ATL;
Destination: FLL;
Delay or cancellation cause: National aviation system and air carrier;
Minutes over 3 hours: 5;
DOT outcome: Under investigation.
Source: GAO analysis of DOT data.
Note: ATL (Atlanta Hartsfield International), BNA (Nashville
International Airport), BWI (Baltimore-Washington International), CID
(Cedar Rapids Iowa), CLT (Charlotte/Douglas International), DFW
(Dallas-Fort Worth International), DTW (Detroit Metro Wayne County),
FLL (Fort Lauderdale Hollywood International), HNL (Honolulu
International), IAD (Washington Dulles International), JFK (New York
John F. Kennedy International), LGA (LaGuardia International), MIA
(Miami International), ONT (Ontario International), ORD (Chicago
O'Hare International), PHL (Philadelphia International), RDU (Raleigh-
Durham International Airport), SAN (San Diego International
Lindbergh), SFO (San Francisco International), SJU (Luis Munoz Marin
International Airport), SLC (Salt Lake City International), and STL
(Lambert St. Louis International).
[End of table]
[End of section]
Appendix V: Tarmac Delay Logistic Regression Analysis:
This appendix describes two models that we designed to assess whether
DOT's tarmac delay rule is correlated with an increase in airline
cancellations. Both models use data for the same months before and
after the rule went into effect to analyze whether and how a variety
of factors--including the imposition of the rule--are associated with
the likelihood (or odds) that a flight will be canceled. One model
analyzes the likelihood of cancellation after a flight has left the
gate and gone onto the tarmac; the other analyzes the likelihood of
cancellation at the gate. Specifically, this appendix discusses (1)
the incidence of cancellations since the rule's implementation, (2)
the conceptual framework for examining these issues through modeling,
(3) variable calculations and data sources, and (4) the models'
results.
Incidence of Flight Cancellations in 2010 Relative to 2009:
To examine the incidence of flight cancellations before and after the
tarmac rule's implementation, we collected data on flights for May
through September in 2009--before the rule went into effect--and for
the same months in 2010, after the rule's implementation.[Footnote
122] We examined the incidence of cancellation for flights that were
canceled after they left the gate and went onto the tarmac, and for
flights before they left the gate. The data cover flights reported to
the Bureau of Transportation Statistics (BTS) at 70 airports in the
continental United States.[Footnote 123]
Table 14 provides information, for the time frame of this analysis, on
the number of flights in each time period that left the gate and took
off, and the number of flights that left the gate but eventually were
canceled. From that information we calculate the odds of cancellation
in each of the two years. These odds equal the number of flights that
were canceled divided by the number of flights that were not canceled.
For example, in 2009, the odds of cancellation is (808/1,868,189),
which equals 0.000433. Thus, roughly 4 out of every 10,000 flights
that exited the gate were ultimately canceled in that year. Finally,
we calculated the odds ratio of a flight being canceled in 2010
compared with 2009, which is a ratio of the odds of cancellation in
2010 to the odds of cancellation with 2009.
Table 14: Unadjusted Odds Ratio for Tarmac Cancellation, 2010 versus
2009:
Year (May-September): 2009;
Number of flights not canceled: 1,868,189;
Number of flights canceled: 808;
Odds of cancellation: 0.000433.
Year (May-September): 2010;
Number of flights not canceled: 1,845,296;
Number of flights canceled: 992;
Odds of cancellation: 0.000538.
Odds ratio, 2010 compared with 2009: 1.24.
Source: GAO analysis of DOT data.
Notes:
1. "Unadjusted" means that this analysis has not controlled for other
factors that may influence the incidence of cancellation.
2. These results are based on analysis of all flights, operated by DOT
reporting airlines, to and from 70 of the FAA's Aviation System
Performance Metrics 77 airports in the United States and Puerto Rico.
These results differ from the cancellation trends presented in table 1
of the report because table 1 illustrated trends by these same
airlines across the entire system and not just at these 70 airports.
[End of table]
The data show that flights are rarely canceled after leaving the gate.
In both years, a very small fraction of flights that left the gate
were ultimately canceled. As noted, in 2009 roughly 4 flights (that
left the gate) were canceled for every 10,000 flights that took off.
However, the odds of cancellation for a flight that has left the gate
did appear to rise in 2010 compared with 2009. The odds ratio is the
odds of a tarmac cancellation in 2010 divided by the odds of such a
cancellation in 2009. The odds ratio exceeds 1, indicating that
cancellations were more likely to occur in 2010. Specifically there
was about a 24 percent increase in the odds of cancellation in 2010
compared with a year earlier.
Because we hypothesized that the likelihood of cancellation for a
flight that has left the gate may be greater the longer it sits on the
tarmac, we assessed the odds of cancellation based on how long a
flight sits on the tarmac, as shown in table 15. These data reveal
that in both 2009 and 2010, the odds of cancellation rise
substantially for flights that have been on the tarmac for longer
periods of time. For example, in 2009 the odds of cancellation for
flights on the tarmac 60 minutes or less are only a small fraction of
a percent, but for flights on the tarmac for 121 to 180 minutes, the
odds rise substantially to 6 percent in that year. Using these odds,
we calculate odds ratios showing the relative odds of cancellation for
each hour category compared with the base hour (up to 1 hour of
delay), within each year. As shown, the odds ratios rise dramatically
as more time passes on the tarmac--a 42-fold increase in the odds of
cancellation when a plane has been sitting on the tarmac for 61 to 120
minutes compared with a delay of 60 or less minutes in 2009.
Table 15: Unadjusted Odds Ratio for Tarmac Cancellation, 2010 versus
2009, by Time Spent on Tarmac:
Year (May-September): 2009;
Time on tarmac (minutes): 0-60;
Number of flights not canceled: 1,848,276;
Number of flights canceled: 452;
Odds of cancellation: 0.000244.
Year (May-September): 2009;
Time on tarmac (minutes): 61-120;
Number of flights not canceled: 17,707;
Number of flights canceled: 184;
Odds of cancellation: 0.01039;
Odds ratio, time on tarmac compared with 0 to 60 minute time, within a
given year: 42.
Year (May-September): 2009;
Time on tarmac (minutes): 121-180;
Number of flights not canceled: 1,912;
Number of flights canceled: 116;
Odds of cancellation: 0.06067;
Odds ratio, time on tarmac compared with 0 to 60 minute time, within a
given year: 248.
Year (May-September): 2010;
Time on tarmac (minutes): 0-0;
Number of flights not canceled: 1,830,874;
Number of flights canceled: 561;
Odds of cancellation: 0.000306;
Odds ratio, time on tarmac compared with 0 to 60 minute time, within a
given year: [Empty];
Odds ratio, time on tarmac in 2010 compared with same time on tarmac
2009: 1.25.
Year (May-September): 2010;
Time on tarmac (minutes): 61-120;
Number of flights not canceled: 13,591;
Number of flights canceled: 266;
Odds of cancellation: 0.019571;
Odds ratio, time on tarmac compared with 0 to 60 minute time, within a
given year: 63;
Odds ratio, time on tarmac in 2010 compared with same time on tarmac
2009: 1.88.
Year (May-September): 2010;
Time on tarmac (minutes): 121-180;
Number of flights not canceled: 829;
Number of flights canceled: 159;
Odds of cancellation: 0.191797;
Odds ratio, time on tarmac compared with 0 to 60 minute time, within a
given year: 626;
Odds ratio, time on tarmac in 2010 compared with same time on tarmac
2009: 3.16.
Source: GAO analysis of DOT data.
Notes:
1. "Unadjusted" means that this analysis has not controlled for other
factors that may influence the incidence of cancellation.
2. These values may not exactly equal the values based on division
because the odds ratios presented here are based on unrounded numbers.
Data for tarmac delays of more than 3 hours are not included in the
table because for 2010 the number of such delays was only eight and we
decided that odds ratios based on such small numbers may not be
representative.
3. These results are based on analysis of all flights, operated by DOT
reporting airlines, to and from 70 of the FAA's Aviation System
Performance Metrics 77 airports in the United States. and Puerto Rico.
These results differ from the cancellation trends presented in table 1
of the report because table 1 illustrated trends by these same
airlines across the entire system and not just at these 70 airports.
[End of table]
These data provided in table 15 also reveal that for every "time-on-
the-tarmac" category, the odds of cancellation in 2010 exceeded the
odds of cancellation in 2009 because all of the odds ratios (shown in
the far right column) exceed 1. We calculated these odds ratios by
taking the odds of cancellation in one tarmac time category in 2010
and dividing it by the odds of cancellation for the same tarmac time
category in 2009. These data further show that the differential
between the likelihood for cancellation in 2010 over 2009 rose the
longer a flight was on the tarmac. While the odds of cancellation for
flights on the tarmac for 60 or less minutes were 25 percent greater
in 2010 than in 2009, for flights on the tarmac 121 to 180 minutes,
there was a threefold greater odds of cancellation in 2010 compared to
2009. Figure 20 shows how the relative odds of flight cancellation in
2010 compared to 2009 increases the longer a flight sits on the tarmac.
Figure 20: Odds of Cancellation for Flights on the Tarmac, 2010 and
2009:
[Refer to PDF for image: multiple line graph]
Odds of Cancellation:
Time on the tarmac: 0-60 minutes;
2010: 0.000306;
2009: 0.000244.
Time on the tarmac: 61-120 minutes;
2010: 0.019571;
2009: 0.001039.
Time on the tarmac: 121-180 minutes;
2010: 0.191797;
2009: 0.006067.
Source: GAO analysis of DOT data.
[End of figure]
Finally, we calculated odds ratios to examine the relative odds of
flight cancellations at the gate in 2010 and 2009. Table 16 shows the
odds of cancellation each year and the odds ratio for gate
cancellations in 2010 compared with 2009. The odds of a gate
cancellation were 13 percent greater in 2010 compared with 2009.
Table 16: Unadjusted Odds Ratio for Gate Cancellation, 2010 versus
2009:
Year (May-September): 2009;
Number of flights that taxi away from the gate: 1,868,997;
Number of flights canceled at the gate: 16,850;
Odds of Gate cancellation: 0.00902.
Year (May-September): 2010;
Number of flights that taxi away from the gate: 1,846,288;
Number of flights canceled at the gate: 18,807;
Odds of Gate cancellation: 0.01019.
Odds ratio, 2010 compared with 2009: 1.13.
Source: GAO analysis of DOT data.
Notes:
1. "Unadjusted" means that this analysis has not controlled for other
factors that may influence the incidence of cancellation.
2. These results are based on analysis of all flights, operated by DOT
reporting airlines, to and from 70 of the FAA's Aviation System
Performance Metrics 77 airports in the United States and Puerto Rico.
These results differ from the cancellation trends presented in table 1
of the report because table 1 illustrated trends by these same
airlines across the entire system and not just at these 70 airports.
[End of table]
Conceptual Framework of Models:
While the unadjusted odds ratios indicate that the likelihood of both
tarmac and gate cancellations increased in May through September 2010
relative to the same time period in 2009, this increase may or may not
be attributable to the tarmac delay rule in 2010. Many factors may
contribute to flight cancellations, and there could be an observed
difference across two years for a number of reasons. For example,
weather events may disrupt traffic more in one year than in another,
or airline scheduling or traffic patterns could change over time. To
develop a model to examine this issue, it is helpful to first consider
whether there is any reason why the tarmac rule might be correlated
with flight cancellations. In particular, what is it about airline
behavior that could be influenced by the tarmac delay rule?
Model Hypothesis:
When there are flight disruptions, airlines face a trade-off between
the consequences of delays they might incur and cancellations. For
example, when bad weather reduces airport capacity, thus slowing the
rate at which flights can take off or land at an airport, airlines
must decide how to ration their traffic. They can choose to hold to
their schedule and fly all their flights, but risk long delays.
Alternatively, they can choose to cancel some of their flights, thus
mitigating the capacity constraint they face and reducing the amount
of delays for their remaining flights. Although airlines have some
control over these trade-offs, airport capacity--both in gate space
and on the tarmac--sometimes becomes so constrained that cancellations
are unavoidable. In managing these circumstances, airlines attempt to
minimize disruptions to passengers and costs to themselves. How an
airline makes decisions within the context of this trade-off will vary
among airlines depending on their business models and the particular
situation at hand.
The DOT's tarmac delay rule requires airlines to limit the time
flights spend on the tarmac to less than 3 hours or face the
possibility of a substantial fine. Our hypothesis is that if the
tarmac rule is associated with a greater incidence of flight
cancellations, this may occur because the rule may have altered
airlines' calculus in analyzing the trade-off between delay and
cancellation. According to airline representatives we spoke with,
flights that sit on the tarmac for a significant period of time may
have to return to the gate to avoid a fine and, because of crew hour
limits or because of the severity of the underlying cause of the
delay, these flights may be canceled. In addition, airline officials
and aviation stakeholders told us that the rule has increased the
likelihood that they will precancel a flight--that is, cancel a flight
before it ever leaves the gate. First, if a flight is returning to the
terminal to avoid a tarmac fine, a flight that has not yet left the
gate might need to be canceled to free gate space for the returning
flight. Airline officials also told us that they are precanceling more
flights before their scheduled departure time when weather or other
factors indicate that long tarmac delays are possible. We were also
told by one airline official that precancelations may be preferable if
a long tarmac delay seems likely because passengers are likely to have
more rebooking options if they are precanceled than if they wait for
some time on the tarmac and attempt to rebook later in the day.
There are several limitations to this analysis. First, important
factors related to cancellations may not be controlled for. For
example, we do not have information on flights that were canceled for
mechanical problems. This, along with other factors that might be
relevant, could not be controlled for because we do not have adequate
data to assess all factors that could be associated with
cancellations. Also, the analysis provides a suggestion as to the
factors that are correlated with cancellations, but does not
necessarily suggest a causal relationship.
Estimation Method:
To isolate the correlation between the rule and cancellations, as well
as to better understand what other key factors are associated with the
rate of cancellations, we developed two models to examine whether the
rule may be correlated with a change in the incidence of flight
cancellations. Because we are estimating the likelihood of a discreet
event--whether a given flight is canceled--we applied a logistic
regression (or logit method) for the estimation. This method enables
us to assess how each of a set of independent factors correlates with
the odds of a binary event--in this case the cancellation or
noncancellation of an airline flight. We examined two contexts in
which a flight may be canceled, after some time on the tarmac or at
the gate.[Footnote 124]
Tarmac-cancellation model. In the first model, we assessed whether
flights that left the gate were more likely to be canceled during May
through September 2010 than during the same time period in 2009.
Although the tarmac rule considers fines only for flights on the
tarmac more than 3 hours, our discussions with airline officials and
experts suggested that airlines begin to assess the risk of tarmac
violation well before prolonged tarmac delay begins. We grouped
flights into hour-long categories based on the amount of time a flight
sat on the tarmac in order to assess whether the length of time on the
tarmac is associated with the odds of cancellation. For example, if a
flight sat on the tarmac for 72 minutes, we placed it in the 60 to 121
minutes tarmac time category. For the tarmac-cancellation model, we
assessed 3,715,219 flight records for the 10 months included in the
model; 1,799 of these were ultimately canceled.
Gate-cancellation model. In the second model, we examined whether
flights were more likely to be precanceled after the rule went into
effect. A precancellation occurs when a scheduled flight is canceled
before it ever leaves the gate. Thus, even a flight that goes onto the
tarmac but is later canceled is treated as a flight that was not
canceled at the gate in this analysis. This model included 3,750,868
flight records, of which 35,649 were precanceled at the gate.
Many factors affect the possibility of a flight's cancellation and,
therefore, we attempted to account for these other factors in the
model. By controlling for this array of other influences on
cancellations, the model is designed to determine whether the tarmac
rule is independently correlated with the odds of a flight being
canceled. Based on our research and discussions with airline
representatives and academic experts, we identified factors that
contribute to flight cancellations, including factors related to (1)
the origin and destination airports, including circumstances at those
airports at the time a flight is scheduled to depart, such as weather
conditions; (2) characteristics of specific airlines and their
operations; and (3) the scheduled city-pair route and the individual
flight. Our hypothesis is that these same factors contribute to both
types of cancellations, although we do not expect that the
relationship between any of the factors and the odds of cancellation
will necessarily be exactly the same in the two models. Each specific
variable and its data source are discussed below.
Variables and Data Sources:
The primary source of data for the model is flight level data from
BTS's ASQP system. This system includes flight-level data, each record
of which provides an array of information about a single-leg
flight,[Footnote 125] such as the origin and destination airports, the
date and time the flight was scheduled to depart and arrive, the
airline, the taxi-out time, cause of any delay, and whether the flight
was canceled. The BTS data form the level of observation for the
models--which is a given flight--and data from other sources are
merged into these observations. Airlines that account for at least 1
percent of total domestic scheduled passenger service revenue are
required to file this flight information with BTS. Because this
required filing leaves out smaller airlines, not all flights are
included in the model. Moreover, our analysis includes data for 70
airports. The remainder of this section describes the rationale for
including each of the variables in the model, how each is calculated,
and the source of the data.
The dependent variable. The dependent variable for the two models is a
dummy variable--that is a variable that takes a value of one or zero
depending on the presence or absence of some characteristic. For the
tarmac-cancellation model this variable takes a value of one if a
flight that left the gate returned to the gate after going onto the
tarmac and then was canceled, and otherwise takes a value of zero. For
the gate-cancellation model, the variable is set to one if the flight
was canceled before taxiing out from the gate and is otherwise set to
zero.
Variable of interest: implementation of tarmac delay rule. In both
models, we include data on flights from May through September in 2009
and for the same months in 2010. Since the rule went into effect in
late April 2010, a dummy variable indicating whether a flight took
place in 2010 is used. For the tarmac model we also include a set of
dummy variables indicating how long a flight was on the tarmac before
taking off or arriving back at the gate. We classify hour-long
categories of tarmac time: 0 to 60 minutes, 61 to 120 minutes, 121 to
180 minutes, and more than 180 minutes, and include three dummy
variables (using the 0 to 60 minute tarmac time as the reference
category and therefore leaving it out of the regression) to test
whether cancellations become more likely with longer tarmac times.
Additionally, we multiply these three dummy variables by the dummy
variable indicating whether the flight took place in 2009 or 2010.
Creating such interactions allows the measured impact of the tarmac
time on cancellation to be different before and after the
implementation of the tarmac delay rule. Thus, we include six time-on-
tarmac dummy variables in the tarmac delay model.
Variables related to airports and conditions at airports. Several of
the independent factors that might affect the odds of a flight being
canceled are related to the airports at which a flight begins and ends
and certain conditions at those airports:
* Dummy variables for congested airports. It is well known that
certain airports suffer more than others from congestion and delays.
Because some airports have more delay-related issues, we believe that
flights involving these airports may be more likely to be canceled,
holding other factors constant. In a previous report we found that,
according to FAA data, seven airports were the source of 80 percent of
departure delays.[Footnote 126] Because this issue could affect
flights both on the tarmac and at the gate, we use two dummy variables
in both models to denote whether a flight either started or ended at
one of these seven airports. We expect that flights involving these
airports are more likely to be canceled.
* Endpoint airport weather conditions. One of the factors likely to
influence flight cancellations is the weather. Certain weather
conditions can disrupt an airport's realized capacity level and cause
traffic to flow more slowly or even halt for a time. We obtained data
from the FAA for the National Oceanic and Atmospheric Administration's
reporting of weather conditions for each hour at each of the airports
included in our model. The data source provided information on the
incidence of 32 types of weather conditions, such as fog, snow,
thunderstorms, and hail. Additionally, FAA ranks each of the 32
weather conditions from 1 to 3 to indicate the impact of that
particular weather condition on aviation activity. For example,
thunderstorms can be highly disruptive to air traffic and is assigned
a value of 3, while rain is assigned a value of 2, and haze a value of
1, indicating that haze usually presents only minor problems for air
traffic. Using these data, we developed two variables to denote the
occurrence of potentially disruptive weather conditions at the origin
or the destination airport around the time a flight was scheduled for
departure or arrival, respectively. In particular, to characterize the
weather at the origin airport, we designated the hour of scheduled
departure as the anchor timeframe, but we also took into account
weather conditions in the hour before and the hour after the scheduled
takeoff. This variable is set to 1 if a weather condition with a value
of 3 (a significant weather condition) existed at the origin airport
during the hour before, at the hour of, or the hour after the
scheduled departure time. Similarly, the second variable is set to 1
if a weather event of value 3 occurred at the destination airport
within the 3-hour window around the scheduled arrival time of the
flight. Poor weather is expected to be associated with a greater
likelihood of cancellation for flights already on the tarmac as well
as those at the gate.
* Ground delays and ground stops. This variable considers whether FAA
has initiated programs to slow or stop traffic at an airport because
of weather conditions, congestion, or some other reason. We obtained
data on such programs--either ground stops or delays--at all the
airports in our sample, by hour, across the 10 months of our analysis.
Using the scheduled departure hour, we created two dummy variables
that were set equal to 1 if the origin or destination airports,
respectively, had any program in place to slow or stop traffic at the
hour that a flight was scheduled to depart. We expected that flights
affected by a ground stop or ground delay program would be associated
with greater odds of a flight cancellation both on the tarmac and at
the gate.
* Airport on-time performance. A final measure that we included in the
model to capture how well each airport is handling its scheduled
traffic at a given point in time is the rate of the on-time
performance. Data for this analysis come from the Aviation System
Performance Metrics database maintained by FAA. We obtained on-time
arrival and on-time departure performance information for the airports
in the model by hour.[Footnote 127] In the model, a variable for the
on-time departure performance for the origin airport is anchored at
the hour of scheduled departure. Similarly, another variable is
constructed for the on-time arrival performance at the destination
airport. We expected that lower on-time performance measures would
indicate difficulties in flowing the scheduled traffic and would thus
be associated with a greater odds of flight cancellation.
Variables related to airlines and their operations. Some factors that
might be correlated with the odds of a flight cancellation are related
to the airline that is operating the flight and how the flight fits
into that airline's network:
* Size of airline. Certain airlines may be more inclined to cancel
flights than other airlines. We separated airlines into three
categories: the legacy airlines, which are typically the larger
networked airlines; low-cost airlines, which include Southwest and
AirTran; and the smaller airlines, such as regional airlines, that
tend to fly shorter routes with smaller aircraft and often operate
flights for legacy airlines. We did not include the third airline
classification in the model; instead we use it as the reference
category against which the other two categories of airlines are
compared.[Footnote 128]
* Airline hub. Many airlines operate a network through which
particular airports--called hubs--are used for the transfer of traffic
so that a larger number of routes can be served. Even though our model
looks at the odds of cancellation for a single leg flight and we do
not examine itineraries of more than one flight leg, an airline
considers, when deciding whether to cancel a flight, how its flights
are interrelated and how passengers transfer among them. If a flight
takes off from an airport that is a hub for the airline operating that
flight, we deemed this an origin/hub flight. Likewise, if a flight is
destined to an airport that the carrier of record states is one of its
hubs, we designated it as a destination/hub flight. If an airport is a
hub for an airline, we expect this could affect the decision about
whether to cancel a flight.
* Average passengers per flight (on an airline-route basis). This
variable is designed to take into account the likelihood that airlines
will attempt to deliver as many passengers as possible to their
destination and so might be more inclined to cancel flights with fewer
passengers onboard when circumstances disrupt traffic flow. Because
data were not available on the number of passengers onboard each
particular flight, we used the average number of passengers for a
particular airline on a given route over the course of a month,
divided by 10. Thus, the results indicate the change in the odds of
cancellation for each additional 10 passengers on a given airline's
flight for that route.
Variable related to the route and flight. The following variables
provide information about the origin-to-destination route and the
specific flight.
* Route distance. Some past research has shown that airlines are less
likely to cancel longer distance flights. We placed routes in four
categories according to distance: less than 500 miles, and three
categories that were more than 750 miles. We did not include the
flights that fell into the 500 to 750 distance because it is the
reference category that other distance dummy variables are compared to.
* Day of the week. Since traffic patterns vary across the days of the
week, particularly weekdays versus weekend days, we included a dummy
variable for flights that took place on the weekend. We expected that
weekend flights will be canceled less often than weekday flights
because less traffic is scheduled on the weekend making a given set of
circumstances on the weekend less likely to disrupt traffic on these
days.
* Scheduled departure hour. Airlines may be more or less reluctant to
cancel flights at certain times of the day than at other times. For
example, canceling early flights may be less problematic because there
will be more options for rebooking passenger that day than there would
be later in the day. Additionally, airlines may need to consider where
an aircraft ends the day in preparation for the next day's traffic,
and so may prefer not to cancel flights late in the day. We created
four categories for departure hours: overnight, morning, afternoon, or
evening. The afternoon category is not included in the model because
it is the reference group we compare the three other dummy variables
against.
Table 17 provides information on the source of data for each of the
variables.
Table 17: Variables and Data Sources:
Variable: Dependent variable: indicator that a flight was canceled;
Data source: BTS data.
Variable: Tarmac delay rule dummy variable;
Data source: n/a.
Variable: Indicator of length of time on tarmac (used as dummy
variable and as interaction with tarmac delay dummy variable);
Data source: BTS data.
Variable: Dummy variable for congested airports;
Data source: Identified as seven airports with most delay, based on
previous GAO report.
Variable: Endpoint weather variable;
Data source: National Oceanic and Atmospheric Administration weather
data and FAA indicators of severity of weather conditions for aviation.
Variable: Ground stops and ground delays;
Data source: FAA's Operations Network data.
Variable: Airport on-time performance;
Data source: FAA Aviation System Performance Metrics data.
Variable: Airline size categories;
Data source: BTS Form 41 data on airline cost structures and previous
GAO analysis.
Variable: Airline hub airports;
Data source: As identified by airlines.
Variable: Average airline passengers per flight on a route;
Data source: BTS Form 41 filings.
Variable: Route distance;
Data source: BTS data.
Variable: Day of the week;
Data source: BTS data.
Variable: Scheduled departure hour;
Data source: BTS data.
Source: GAO.
[End of table]
Model Results:
This section provides results for both the tarmac-cancellation and
gate-cancellation models.
Results for Tarmac-Cancellation Model:
We used output from the logistic regression model for the rule change
dummy variable and the six dummy variables related to time on the
tarmac to ascertain the relative odds of flight cancellations before
and after the implementation of the tarmac rule. Table 18 shows, based
on the model that controlled for other factors, how the odds of
cancellation in each tarmac time category in 2010 compared with the
odds of cancellation for the same tarmac time in 2009--specifically,
we show the ratio of those odds. In all hour categories of tarmac
time, the odds of cancellation were greater in 2010 than in 2009
because all of the odds ratios exceed 1. Moreover, the differential in
the odds ratio of cancellations across the 2 years increased with the
time a flight was on the tarmac. For flights that were on the tarmac
for less than an hour, the odds of a cancellation were about one-third
higher in 2010 than in 2009. But the longer a flight remained on the
tarmac the more the relative odds of cancellation were greater in 2010
than in 2009. For flights with 61 to 120 minutes of tarmac delay, the
odds ratio rose to 2.14, indicating that the odds of a cancellation
more than doubled in 2010 compared with 2009, and for flights with 121
to 180 minutes of tarmac delay, the odds of cancellation more than
tripled in that same time period. Finally, the odds ratios in table 18
are very similar to those presented in table 15, indicating that the
inclusion of key variables to control for other factors did not have
much effect on our findings related to the tarmac rule.
Table 18: Odds Ratios Estimates for Variables of Interest in Tarmac-
Cancellation Model, 2010 versus 2009:
Variable: 0-60 minutes on tarmac;
Odds ratio; 2010 hour category compared with same hour in 2009:
1.31[A].
Variable: 61-120 minutes on tarmac;
Odds ratio; 2010 hour category compared with same hour in 2009:
2.14[A].
Variable: 121-180 minutes on tarmac;
Odds ratio; 2010 hour category compared with same hour in 2009:
3.59[A].
Source: GAO analysis of DOT data.
Note: Data for tarmac delays in excess of 3 hours are not included in
the table because for 2010 the number of such delays was only eight
and we decided that odds ratios based on such small numbers may not be
representative.
[A] Denotes an odds ratio that is significant at the 1 percent level.
The significance of the coefficients in the models we fit was
evaluated using a simple Wald test statistic, which is asymptotically
equivalent to the likelihood-ratio chi-square statistic. We regarded
as significant all coefficients which yielded a test statistic with an
associated probability of less than 1 percent, which means that there
is less than a 1 percent probability of finding an effect (or
association) as large as the one indicated by the estimated
coefficient just by chance, or as a result of random fluctuations.
[End of table]
Table 19 provides the odds ratios from the logistic regression model
for all other variables included in the tarmac-cancellation model.
Some of the key findings are:
* Flights departing from or destined to an airline's hub airport are
less likely to be canceled.
* Flights in evening hours are less likely to be canceled than flights
departing in the afternoon.
* Flights of greater than 750 miles are less likely to be canceled
than flights of 500 to 750 miles.
* Flights are more likely to be canceled if the departure airport or
arrival airport is experiencing severe weather at or around the time
of scheduled departure or arrival, respectively.
* Flights are more likely to be canceled if a ground stop or ground
delay was in effect at either the departure airport or the arrival
airport at the scheduled time of departure.
Table 44: Logistic Regression Results for Tarmac-Cancellation Model
and Other Independent Variables:
Variable: Departure airport is one of seven most congested, compared
to all other airports;
Odds ratio estimate: 0.99.
Variable: Destination airport is one of seven most congested, compared
to all other airports;
Odds ratio estimate: 1.26[A].
Variable: Legacy airline, compared to smaller airlines;
Odds ratio estimate: 1.84[A].
Variable: Low-cost airline, compared to smaller airlines;
Odds ratio estimate: 0.24[A].
Variable: Weekend day, compared to weekday;
Odds ratio estimate: 0.86[A].
Variable: Scheduled departure in morning hours, compared to afternoon
hours;
Odds ratio estimate: 1.04.
Variable: Scheduled departure in evening hours, compared to afternoon
hours;
Odds ratio estimate: 0.51[A].
Variable: Scheduled departure in overnight hours, compared to
afternoon hours;
Odds ratio estimate: 1.08.
Variable: Departure airport is a hub for airline;
Odds ratio estimate: 0.69[A].
Variable: Destination airport is a hub for airline;
Odds ratio estimate: 0.70[A].
Variable: Severe weather at departure airport around time of scheduled
departure;
Odds ratio estimate: 1.45[A].
Variable: Severe weather at destination airport around time of
scheduled arrival;
Odds ratio estimate: 2.87[A].
Variable: On-time performance, departures at departure airport;
Odds ratio estimate: 0.73[B].
Variable: On-time performance, arrivals at arrival airport;
Odds ratio estimate: 0.43[A].
Variable: Flight is less than 500 miles, compared to flights of 500-
750 miles;
Odds ratio estimate: 1.09.
Variable: Flight is 750-1,000 miles, compared to flights of 500-750
miles; Odds ratio estimate: 0.77[A].
Variable: Flight is 1000-1500 miles, compared to flights of 500-750
miles;
Odds ratio estimate: 0.71[A].
Variable: Flight is 1,500 or more miles, compared to flights of 500-
750 miles;
Odds ratio estimate: 0.60[A].
Variable: Average number of passengers per flight, by airline, route,
and month (effect is per 10 passengers);
Odds ratio estimate: 0.92[A].
Variable: Ground delay or stop in effect at departure airport around
departure time;
Odds ratio estimate: 1.31[A].
Variable: Ground delay or stop in effect at destination airport around
departure time;
Odds ratio estimate: 1.98[A].
Source: GAO analysis of DOT data.
[A] Denotes an odds ratio that is significant at the 1-percent level.
The significance of the coefficients in the models we fit was
evaluated using a simple Wald test statistic, which is asymptotically
equivalent to the likelihood-ratio chi-square statistic. We regarded
as significant all coefficients which yielded a test statistic with an
associated probability of less than 1 percent, which means that there
is less than a 1 percent probability of finding an effect (or
association) as large as the one indicated by the estimated
coefficient just by chance, or as a result of random fluctuations. It
is not possible, strictly speaking, to characterize a logistic
regression model in terms of the amount of variance in the dependent
variable that is explained by the independent variables included in
the model. We do note, however, that the percentage of canceled and
noncanceled flights that are correctly classified using this model and
the independent variables it includes is 86.3 percent.
[B] Denotes an odds ratio that is significant at the 5-percent level
according to a Wald test statistic.
[End of table]
Results for the Gate-Cancellation Model:
Table 20 provides the findings for the gate-cancellation model, which
assesses the likelihood of precancellations, adjusted to account for
factors other than the tarmac delay rule that may influence the
incidence of cancellation. One significant finding is that the odds
ratio for the rule change is substantially greater, when adjusted,
than indicated by the simple unadjusted odds ratio shown in table 16.
The model results indicate that the odds of gate cancellations rose by
24 percent after the rule went into effect, whereas the simple result
indicated only a 13 percent increase in those odds. This suggests that
to understand the independent correlation between the tarmac delay
rule and likelihood of gate cancellation, it is important to control
for the other factors that are likely correlated with such
cancellations.
Findings from the gate-cancellation model suggest:
* Gate cancellations are more common when a flight is departing from
or destined to one of the seven most congested airports in the U.S.
* Gate cancellations are less common for flights scheduled to depart
in the evening, compared to flights departing in the afternoon.
* Gate cancellations are more common when severe weather is affecting
either endpoint airport of a flight at the relevant hour.
* Gate cancellations are more common for very short flights, compared
to flights of 500 to 750 miles in distance.
* Gate cancellations are less common for flights of a more than 750
miles, compared to flights of 500 to 750 miles.
* Gate cancellations are more common if a ground delay or ground stop
was in place at the origin or destination airport at the time of
scheduled departure.
* In this case it appears that flights to an airline's hub airport are
more likely to be canceled.
Table 20: Logistic Regression Results for Gate-Cancellation Model and
Other Independent Variables:
Variable: Rule change dummy;
Odds ratio estimate: 1.24[A].
Variable: Departure airport is one of seven most congested, compared
to all other airports;
Odds ratio estimate: 1.27[A].
Variable: Destination airport is one of seven most congested, compared
to all other airports;
Odds ratio estimate: 1.33[A].
Variable: Legacy airline, compared to smaller airlines;
Odds ratio estimate: 1.57[A].
Variable: Low-cost airline, compared to smaller airlines;
Odds ratio estimate: 0.70[A].
Variable: Weekend day, compared to weekday;
Odds ratio estimate: 0.91[A].
Variable: Scheduled departure in morning hours, compared to afternoon
hours;
Odds ratio estimate: 1.43[A].
Variable: Scheduled departure in evening hours, compared to afternoon
hours;
Odds ratio estimate: 0.86[A].
Variable: Scheduled departure overnight hours; compared to afternoon
hours;
Odds ratio estimate: 1.45[A].
Variable: Departure airport is a hub for airline;
Odds ratio estimate: 1.03.
Variable: Destination airport is a hub for airline;
Odds ratio estimate: 1.34[A].
Variable: Severe weather at departure airport around time of scheduled
departure;
Odds ratio estimate: 1.36[A].
Variable: Severe weather at destination airport around time of
scheduled arrival;
Odds ratio estimate: 1.30[A].
Variable: On-time-performance, departures at departure airport;
Odds ratio estimate: 0.12[A].
Variable: On-time-performance, arrivals at arrival airport;
Odds ratio estimate: 0.15[A].
Variable: Flight is 0-500 miles, compared to flights of 500-750 miles;
Odds ratio estimate: 1.28[A].
Variable: Flight is 750-1000 miles, compared to flights of 500-750
miles;
Odds ratio estimate: 0.86[A].
Variable: Flight is 1000-1500 miles, compared to flights of 500-750
miles;
Odds ratio estimate: 0.88[A].
Variable: Flight is 1500 or more miles, compared to flights of 500-750
miles;
Odds ratio estimate: 0.57[A].
Variable: Average number of passengers per flight, by airline, route,
and month (effect is per 10 passengers);
Odds ratio estimate: 0.88[A].
Variable: Ground delay or stop in effect at departure airport around
departure time;
Odds ratio estimate: 2.14[A].
Variable: Ground delay or stop in effect at destination airport around
departure time;
Odds ratio estimate: 1.80[A].
Source: GAO analysis of DOT data.
[A] Denotes an odds ratio that is significant at the 1 percent level.
The significance of the coefficients in the models we fit was
evaluated using a simple Wald test statistic, which is asymtotically
equivalent to the likelihood-ratio chi-square statistic. We regarded
as significant all coefficients which yielded a test statistic with an
associated probability of less than 1 percent, which means that there
is less than a 1 percent probability of finding an effect (or
association) as large as the one indicated by the estimated
coefficient just by chance, or as a result of random fluctuations. It
is not possible, strictly speaking, to characterize a logistic
regression model in terms of the amount of variance in the dependent
variable that is explained by the independent variables included in
the model. We do note, however, that the percentage of canceled and
noncanceled flights that are correctly classified using this model and
the independent variables it includes is 75.8 percent.
[End of table]
Sensitivity Analysis:
We ran the models using several other specifications, most of which
involved alternative variable specifications. These runs indicated
that our findings for the tarmac rule were robust across these
specifications. Alternatives included the following:
* Variations on how the specific airlines were grouped. In the base
case we classified airlines into three categories: legacy airline, low-
cost airline, and all others. In an alternative specification, we
classified airline as large or small based on the number of
enplanements.
* Variations for characterization of origin and destination airports.
In the base case models, we included dummy variables to indicate that
the airport (origin or destination) was one of the seven most
congested airports in the United States. In an alternative
specification, we used 62 dummy variables to indicate whether the
airport (origin or destination), was one of the 31 largest airports.
* Alternative measure for poor weather conditions. In the base case,
we classified weather at endpoint airports as severe if a weather
event occurring around the time of the flight would be considered
highly disruptive to aviation activity. In an alternative
specification we included both severe and moderately disruptive
weather conditions.
* Alternative distance measure. In the base case, we classified
distance into broad mileage categories. In an alternative
specification, we entered distance divided by 100 as a continuous
variable.
* Elimination of flights that were canceled after a tarmac delay for
the gate model. For the gate model, we included flights that left the
gate, even if they were later canceled. We did so because the airlines
were attempting to get these flights off the ground when they were
making gate-cancellation decisions, and so we treated these flights as
nongate cancellations. In one sensitivity run, we eliminated any
flights that left the gate but were later canceled.
[End of section]
Appendix VI: GAO Contact and Staff Acknowledgments:
GAO Contact:
Susan Fleming, (202) 512-2834 or flemings@gao.gov:
Staff Acknowledgments:
In addition to the contact named above, Paul Aussendorf, Assistant
Director; Amy Abramowitz; Kyle Browning; Lauren Calhoun; Anne Doré;
Grant Mallie; Michael Mgebroff; Sara Ann Moessbauer; Josh Ormond; and
Melissa Swearingen made key contributions to this report.
[End of section]
Footnotes:
[1] See Senate Joint Economic Committee, Your Flight Has Been Delayed
Again: Flight Delays Cost Passengers, Airline and the U.S. Economy
Billions (Washington, D.C.: May 2008), and M. Ball, C. Barnhart, M.
Dresner, et al, Total Delay Impact Study A Comprehensive Assessment of
the Costs and Impacts of Flight Delay in the United States, National
Center of Excellence for Aviation Operations Research, Federal
Aviation Administration (November 2010). This latter study estimated
that the total cost of all U.S. air transportation delays in 2007 was
$31.2 billion, with the $16.7 billion passenger component based on the
passenger time lost due to schedule buffer, delayed flights, flight
cancellations, and missed connections.
[2] According to the U.S. Department of Transportation, a "tarmac
delay means the holding of an aircraft on the ground either before
taking off or after landing with no opportunity for its passengers to
deplane."
[3] In the EU, the European Commission, whose mission is to promote
the general interest of the EU, initiates the legislative process by
drafting specific pieces of legislation and proposing them to the
Council of the European Union and European Parliament who together
serve as the EU's legislative branch. Under the EU's co-decision
procedure, both the Council of the EU and the European Parliament must
approve legislation in order to enact a law. The European Union
currently has 27 member states.
[4] Regulation (EC) 261/2004, art. 3, 2004 O.J. (L 46/1). The
regulation was passed by the European Parliament and the Council of
the European Union in 2004, but came into force in February, 2005.
[5] Throughout this report, we refer to the "tarmac delay rule," which
is one component (contingency plans for lengthy tarmac delays) of the
Enhancing Airline Passenger Protections Final Rule. 74 Fed. Reg. 68983
(Dec. 30, 2009). Unless otherwise stated, reference to the "tarmac
delay rule" refers to this single component of the rule.
[6] The modifier "legacy" derives from the fact that these airlines
were operating when the industry deregulated in 1978. Although the
airline industry is now largely free of economic regulation, it
remains regulated in other areas, most notably safety, security, and
operating standards.
[7] Many regional airline flights are operated under a fee-for-service
arrangement whereby the airline marketing the flight has decision-
making control over most of the inventory of the airline operating the
flight, including the decision to cancel flights.
[8] Commercial service airports are defined by 49 U.S.C. § 47102 as
public airports that the Secretary of Transportation determines have
2,500 passenger boardings (enplanements) each year and receive
scheduled passenger aircraft service.
[9] FAA divides commercial service airports into primary airports
(enplaning more than 10,000 passengers annually) and other commercial
service airports. Among primary airports, "nonhub" airports enplane
fewer than 0.05 percent of systemwide passengers, "small" hub airports
enplane at least 0.05 percent but fewer than 0.25 percent of
systemwide passengers, "medium" hub airports enplane at least 0.25
percent but fewer than 1 percent of systemwide passengers, and "large"
hub airports enplane at least 1 percent of systemwide passengers. 49
U.S.C. § 47102.
[10] FAA's Operations Network provides information on which airports
delays were attributed to--that is, which facility instituted a
traffic management initiative that resulted in flights being delayed.
In 2010, we reported that seven airports--Newark Liberty
International, New York John F. Kennedy International, New York
LaGuardia, Atlanta Hartsfield International, Philadelphia
International, Chicago O'Hare International, and San Francisco
International--were the main drivers of departure delay across the
system and that these airports accounted for about 80 percent of
departure delays at airports across the national airspace system. See
GAO, National Airspace System: Setting On-Time Performance Targets at
Congested Airports Could Help Focus FAA's Actions, [hyperlink,
http://www.gao.gov/products/GAO-10-542] (Washington, D.C.: May 26,
2010).
[11] [hyperlink, http://www.gao.gov/products/GAO-10-542]; National
Airspace System: Long-Term Capacity Planning Needed Despite Recent
Reduction in Flight Delays, [hyperlink,
http://www.gao.gov/products/GAO-02-185] (Washington, D.C.: Dec. 14,
2001); and GAO, Air Traffic Control: Role of FAA's Modernization
Program in Reducing Delays and Congestion, [hyperlink,
http://www.gao.gov/products/GAO-01-725T] (Washington, D.C.: May 10,
2001).
[12] See Bengi Manley and Lance Sherry, The Impact of Ground Delay
Program (GDP) Rationing Rules on Passenger and Airline Equity, Center
for Air Transportation Systems Research, George Mason University
(2008); Nicholas G. Rupp, An Investigation Into the Determinants of
Flight Cancellations, East Carolina University (2005); and Zalman
Shavell, The Effects of Schedule Disruptions on the Economics of
Airline Operations, Mitre Corporation (2000).
[13] [hyperlink, http://www.gao.gov/products/GAO-10-542].
[14] [hyperlink, http://www.gao.gov/products/GAO-10-542].
[15] M. Ball, C. Barnhart, M. Dresner, et al, Total Delay Impact Study
A Comprehensive Assessment of the Costs and Impacts of Flight Delay in
the United States, National Center of Excellence for Aviation
Operations Research, FAA (November 2010).
[16] 14 C.F.R. part 250.
[17] Though an Airline Passenger Bill of Rights Act was introduced in
1999 to provide "enhanced protections for airline passengers," it did
not come to a vote. Airline Passenger Bill of Rights Act of 1999, H.R.
700, 106th Cong. (1999).
[18] Wendell H. Ford Aviation Investment and Reform Act for the 21st
Century (AIR-21), Pub. L. No. 106-181, § 224, 114 Stat. 61, 103 (2000).
[19] Airline Passenger Bill of Rights Act, H.R. 1734, 107th Cong.
(2001); Airline Passenger Bill of Rights Act of 2007, S.678, 110th
Cong. (2007); Airline Passenger Bill of Rights Act of 2009, S.213,
111th Cong. (2009); Airline Passenger Bill of Rights Act of 2011, H.R.
729, 112th Cong. (2011).
[20] 73 Fed. Reg. 21026 (Apr. 18, 2008).
[21] 74 Fed. Reg. 68983 (Dec. 30, 2009).
[22] For international flights that depart from or arrive at a U.S.
airport, a contingency plan must provide that an airline will not
permit an aircraft to remain on the tarmac for more than a set number
of hours, as determined by the airline, before allowing passengers to
deplane. The exceptions for domestic flights apply to international
flights, as well.
[23] 74 Fed. Reg. 68983 (Dec. 30, 2009), codified at 14 C.F.R. §
259.4. This rule applies to all certificated and commuter air carriers
(i.e., U.S. airlines) that operate scheduled passenger service or
public charter service using any aircraft originally designed to have
a passenger capacity of 30 or more seats. These requirements also only
apply to flights at large-and medium-hub airports.
[24] 49 U.S.C. § 41712.
[25] 14 C.F.R. § 383.2(a) prescribes penalties for civil violations,
including those under 49 U.S.C. § 41712.
[26] 79 Fed. Reg. 23110 (Apr. 25, 2011).
[27] On July 20, 2011, DOT extended the effective date of certain
provisions of the rule from August 23, 2011 to January 24, 2012.
Specifically, DOT extended the effective date of requirements
pertaining to full fare advertising, specific baggage fee disclosures,
post-purchase price increases, flight status notifications (e.g.,
notifications to passengers in the event of delays, cancellations, or
diversions), and holding a reservation without payment. DOT did not
extend the effective date for the tarmac delay provisions. 76 Fed.
Reg. 45181 (July 28, 2011). GAO, Commercial Aviation: Consumers Could
Benefit from Better Information about Airline-Imposed Fees and
Refundability of Government-Imposed Taxes and Fees, [hyperlink,
http://www.gao.gov/products/GAO-10-785] (Washington, D.C.: July 14,
2010).
[28] EU airlines departing outside of the EU but bound for an EU
airport are required to adhere to EU regulations, except where the
country of departure has provided benefits or compensation or
assistance. Regulation (EC) 261/2004, art. 3, 2004 O.J. (L 46/1).
[29] Additionally, these three regions also prescribe requirements for
what must addressed in an airline contract of carriage. For example,
Canada's Air Transportation Regulations require airlines to clearly
state in their contracts of carriage their policies related to the
treatment of passengers, including their policies related to flight
delays, cancellations, and denied boarding. Air Transportation
Regulations SOR/88-58, Part V, Div. I, § 107(1)(n) (Can).
[30] According to DOT, many provisions of the contracts of carriage
are also filed in the airlines' tariffs with DOT and with their
governments.
[31] In addition, in the United States, regulations require certain
U.S. and foreign airlines to develop customer service plans to post on-
time performance data for each flight on their Web sites. Airlines
must state in these customer service plans certain commitments to
passengers, such as the commitment to reimburse for any fee charged to
transport a bag that is lost, notify customers of flight disruptions
within 30 minutes of the airline's becoming aware of the disruption,
and handle bumped passengers with fairness and consistency.
[32] Convention for the Unification of Certain Rules for International
Carriage by Air (Montreal Convention), ch.3, art. 19, May 28, 1999
(entered into force on Nov. 4, 2003), reprinted in S. Treaty Doc.
No.106-45. The Montreal Convention also limits airlines' liability for
delays to 4150 Special Drawing Rights (or $6,658.87 as of July 22,
2011) per passenger. Montreal Convention, art. 22.
[33] FlightStats is an operating platform, owned and operated by
Conducive Technology, which records flight performance information.
This database includes more flight performance information than DOT's
Airline Service Quality Performance System (ASQP) database. For
example, the data we obtained included performance information for
about 98 percent of U.S. scheduled passenger flights in 2010 from
airports considered primary for fiscal year 2009, excluding airports
in U.S. territories. ASQP data included about 77 percent of flights
from those airports. In general, the ASQP data were more complete for
airports in larger communities. For example, only about 45 percent of
all flights from airports in rural communities were captured by ASQP.
[34] Specifically, we grouped all airports considered primary for
fiscal year 2009, excluding airports in U.S. territories, into four
categories based on population: greater than or equal to 1 million
(large metropolitan), 250,000 to 999,999 (midsized metropolitan),
50,000 to 249,999 (small metropolitan), and fewer than 50,000 (rural).
This approach controls for the fact that some small or medium airports-
-generally secondary airports such as Hobby Airport in Houston--are
actually in large metropolitan regions. Using these categories, in
2010, we examined data from 76 airports in large metropolitan
communities, 99 in midsized metropolitan communities, 117 in small
metropolitan communities, and 52 in rural communities. We excluded
commercial service, nonprimary airports because they handle fewer than
10,000 enplanements annually and therefore are not considered airports
that provide connectivity to the national airspace system.
[35] A flight is "diverted" if it lands at an airport other than its
scheduled destination because of extreme weather or security concerns,
for example.
[36] This 8 percent figure for 2010 represented 6,844 departure
cancellations and 932 departure diversions from airports in rural
communities.
[37] Sherry, Lance, Passenger Trip Delay Statistics for 2010, Center
for Air Transportation Systems Research at George Mason University. To
be presented at the Transportation Research Board 91ST Annual Meeting
(January 2012).
[38] Lance Sherry, Passenger Trip Delay Statistics for 2010. DOT also
reports average delay times for aircraft. Since 2005, the average
delay per delayed aircraft arrival is about 55 minutes.
[39] 14 C.F.R. part 234.
[40] Departure data also showed differences between FlightStats and
DOT data, with FlightStats showing a slightly higher percentage of
flights from airports in rural communities delayed, canceled, or
diverted. However, FlightStats showed lower percentages of flights
from airports in large, midsized, and small communities delayed,
canceled, or diverted than did DOT data.
[41] Based on data for all reported flights, DOT reporting airlines'
flights were delayed 19.54 percent, canceled 1.59 percent, and
diverted 0.16 percent of the time. Nonreporting airlines were delayed
20.27 percent, canceled 2.48 percent, and diverted 0.25 percent of the
time.
[42] Zinser, Todd, Report on the Audit of Small Community Aviation
Delays and Cancellations, Report No. CR-2006-049 (May 19, 2006).
[43] N. Rupp and G. Holmes, An Investigation into the Determinants of
Flight Cancellations. Also see, Jing Xiong, Revealed Preference of
Airlines' Behavior under Air Traffic Management Initiatives,
dissertation (2010).
[44] FlightStats uses flight schedules to track the actual performance
of flights based on data from various airport and on-line sources.
This generates extensive, real-time data on flights, which is then
archived in its database. After reviewing the data and supporting
documentation, as well as discussions with relevant officials, we
determined that FlightStats data were sufficiently reliable for the
purposes of this report.
[45] The purpose of this information is to provide consumers with
information on airlines' quality of service. See 14 C.F.R. § 234.1.
[46] DOT data include information on tarmac delays of more than 3
hours for all domestic airlines that handle at least 1 percent of all
domestic scheduled passenger service revenue. Since DOT determines
annually which airlines meet the threshold and therefore must report
flight performance data, the number of airlines required to report
varies. In 2010, 16 airlines were required to report and 2 reported
voluntarily. Since October 2008, DOT has required airlines to submit
additional tarmac delay statistics. As of August 23, 2011, covered
airlines, both United States and foreign, are required under 14 C.F.R.
§ 244.2 to report all passenger operations that experience a tarmac
time of more than 3 hours at a U.S. airport, in a form as set forth in
regulation. A covered airline includes all U.S. certificated air
carriers, U.S. commuter airlines, and foreign airlines that operate
passenger service to or from a U.S. airport with at least one aircraft
that has an original manufacturer's design capacity of 30 or more
seats. See 76 Fed. Reg. 23110, 23161 (Apr. 25, 2011).
[47] Starting in 2008, DOT began monitoring tarmac delays more
closely, requiring reporting airlines to submit data on all tarmac
delays. Prior to October 2008, airlines reported data on tarmac delays
that occurred during taxi-out or taxi-in. Since October 2008, airlines
have been required to submit long tarmac delay statistics for three
additional categories: flights that are canceled after the tarmac
delay, flights that return to the gate after the tarmac delay before
taking off, and flights that experience extended tarmac delays during
taxi-in at a diversion airport. In 2009, of the 868 long tarmac
delays, one-third (283 flights) were captured in these new categories.
[48] For the purposes of this report we consider the summer season to
run from May through September.
[49] Since 2003, airlines have reported the cause of delay to DOT in
one of five broad categories: late-arriving aircraft, airline,
national airspace system, extreme weather, and security. Tarmac delays
of more than 3 hours during both taxi-out and taxi-in are usually
officially attributed by airlines to issues with the national airspace
system. For example, since 2004, 86 percent of such taxi-out delays
were attributed to the national airspace system, according to DOT
data. The extent to which delays attributed to the national airspace
system are related to weather events is unknown, since this reporting
category includes delays caused by air traffic control issues, weather
that slows down operations at an airport, or other ground congestion.
Delays during taxi-in may also occur at diversion airports when
Transportation Security Administration or Customs and Border
Protection officials are not available at airports where passengers
are arriving, according to one airline association. For example, a
regional airline flight from Houston to Minneapolis that was diverted
to Rochester, Minnesota, in the summer of 2009, sat on the tarmac in
Rochester for more than 5 hours because there was severe weather and
confusion over the Transportation Security Administration's after
hours operating procedures. DOT fined Continental Airlines, which had
contracted the ExpressJet flight, and ExpressJet each $50,000 for this
incident. In addition, DOT assessed a civil penalty of $75,000 against
Mesaba Airlines, which provided voluntary ground handling to
ExpressJet, for providing inaccurate information to ExpressJet about
deplaning passengers, which was found to be an unfair and deceptive
practice in violation of 49 U.S.C. § 41712. DOT Order 2009-11-16 (Nov.
24, 2009).
[50] In 2010 we reported that, according to FAA's Operations Network
data, these seven airports--Newark Liberty International, New York
John F. Kennedy International, New York LaGuardia, Atlanta Hartsfield
International, Philadelphia International, Chicago O'Hare
International, and San Francisco International--were the main drivers
of departure delay across the system, and that these airports
accounted for about 80 percent of departure delays at airports across
the national airspace system. See [hyperlink,
http://www.gao.gov/products/GAO-10-542].
[51] Enhancing Airline Passenger Protections, 74 Fed. Reg. 68983 (Dec.
30, 2009).
[52] 14 C.F.R. § 259.4(b). As noted previously, as of August 23, 2011,
international flights of U.S. and foreign carriers are subject to a 4-
hour time limit on tarmac delays, subject to exceptions for safety,
security, and disruption of airport operations.
[53] 14 C.F.R. § 383.2(a) prescribes penalties for civil violations,
including those under 49 U.S.C. § 41712. The term "violation" is not
defined in statute or regulation.
[54] Over these same time periods, tarmac delays of more than 4 hours
went from 105 to 0.
[55] To date, two airlines have been fined in connection with the
tarmac delay rule, but not for exceeding the 3-hour limit. United
Airlines and Pinnacle Airlines were fined $12,000 and $10,000,
respectively, in September 2010 for misreporting tarmac delays. Both
airlines reported incurring delays of more than 3 hours to DOT on
Bureau of Transportation Statistics Form 234 (On-Time Flight
Performance Report). However, after an investigation was initiated by
DOT, the department, or United Airlines in their respective situation,
determined that they had made an erroneous report and the delay was
under 3 hours. The airlines were fined for reporting inaccurate data.
DOT Orders 2010-9-22, 2010-9-11. As of the spring 2011, DOT was also
investigating potential violations of other aspects of the rule
including the requirement to provide food and water service by 2
hours, cabin temperature, and the availability of lavatory services.
[56] Of the remaining flights, three were able to take off shortly
after 3 hours, three returned to the gate prior to takeoff, and two
were at a diverted airport and able to reach the gate.
[57] DOT officials told us that, although they are not required to,
they could issue guidance on their penalty structure as it pertains to
the tarmac delay rule, but have chosen not to in order to maintain
flexibility under their current authority.
[58] We used DOT's ASQP flight performance data because FlightStats
data are not sufficiently detailed for the analysis we present in this
section. Further, though DOT data are more limited for smaller
airlines and airports than FlightStats, the tarmac rule implemented on
April 30, 2010, applied only to flights at large-and medium-hub
airports. However, the April 2011 rule, effective August 23, 2011,
expands the rule to small and nonhub airports.
[59] According to DOT data, scheduled flights only declined by 0.1
percent (or 3,167 flights) from 2009 to 2010.
[60] Overall, fewer flights were delayed on the tarmac for more than 1
hour in 2010 than were delayed in 2009, although the cancellation rate
was higher for the 2010 flights. Specifically, from May through
September 2010, 17,763 flights sat on the tarmac for more than 1 hour,
compared with 22,364 during the same time period in 2009.
[61] DOT limited its analysis to cancellations of flights that were
delayed on the tarmac at least 2 hours because "they are virtually
certain to be directly related to a lengthy tarmac delay." Using that
criterion, in its December 2009 regulatory analysis of the final rule,
DOT estimated that the rule would result in 41 additional canceled
flights per year, affecting 3,176 passengers. According to DOT, the
impact of additional cancellations on passengers varies. For
passengers who boarded a flight before it is canceled, the majority
will wait longer for another flight the same day, while others will
return to the airport the following day.
[62] We also heard from some airlines that they have diverted more
flights to avoid congested airports that may lead to long tarmac
delays during taxi-in for those flights. According to officials we
spoke with, this has included re-routing flights away from hub
airports and "landing short" of their destination at an airport where
they have resources to accommodate the diverted plane. Furthermore, to
avoid challenges at the diversion airport, at least some airlines have
altered where they divert to. Prior to the rule, airlines might have
just landed at the closest airport. Now, however, they are being more
proactive, choosing to land at airports where they have supplies
(i.e., gate space, employees, and equipment) and are more equipped to
handle the traffic. Since the implementation of the rule, arrival
diversions have increased almost 5 percent from 7,272 to 7,617,
according to DOT data for the summers of 2009 and 2010, when flight
traffic levels were similar. However, this increase cannot necessarily
be attributed to the implementation of the tarmac delay rule.
[63] A logistic regression (or logit method) provides an indication of
which independent variables are correlated with the dependent
variable--which in this case, is the incidence of a flight
cancellation. Moreover, it provides this measure of correlation
independent of the effects of the other independent variables included
in the model. It is important to note, however, that this type of
statistical method only suggests correlations between variables and
not causation. That is, our findings do not provide an indication that
any of the independent variables we included in the model actually
caused (or did not cause) cancellations, but only how those variables
are correlated with the incidence of cancellations.
[64] Both models analyze flights between 70 airports in the United
States and Puerto Rico. The FAA Aviation System Performance Metrics
database includes performance data on operations at 77 airports. Our
analysis included domestic flights operated by DOT reporting airlines
to and from these airports, except for three in Alaska and Hawaii (Ted
Stevens Anchorage International Airport, Honolulu International
Airport, and Kahului Airport) which were not included and four (Gary
Chicago International Airport, Greater Rockford Airport, Teterboro
Airport, and Van Nuys Airport) that had no reported flights to or from
the other 70 airports during the months of our analysis. The 70
airports used in our analysis mostly overlap with the large-and medium-
hub airports at which flights were required to comply with the tarmac
delay rule in 2010.
[65] Since our models used data only from 2009 and 2010, we did not
need to exclude any particular flight records due to data reporting
issues because, as noted earlier, DOT enhanced its tarmac delay data
collection procedures in October 2008, improving its data reliability.
[66] For example, weather events and ground delay programs at either
the origin or destination airport were associated with a greater
incidence of cancellation. Much like previous academic work on
cancellations, our results also showed that the likelihood of
cancellation decreases as the length of the flight increases.
[67] As discussed later in this report, all four of Canada's major
airlines have added these provisions to their contracts of carriage in
response to this initiative.
[68] Regulation (EC) 261/2004, 2004 O.J. (L 46/1).
[69] Regulation (EC) 261/2004 applies to the following: (1) all
flights departing an EU signatory member states' airport (regardless
of the destination) and (2) all EU airline flights departing outside
the EU (a third country) into the EU unless the country of departure
has provided benefits or compensation or assistance. Regulation (EC)
261/2004, art. 3, 2004 O.J. (L 46/1). This regulation applies to
operating airlines, but the liability falls upon the airline which is
contractually obligated to the passenger. Under article (3)(5), when
an operating airline which has no contract with the passenger performs
obligations under this regulation, it shall be regarded as doing so on
behalf of the person having a contract with that passenger.
[70] Regulation (EC) 261/2004, articles 6 and 9, 2004 O.J. (L 46/1).
According to a recent European Commission report of delay data
voluntarily provided by airlines, on average, less than 1.2 percent of
all flights departing from and arriving in the EU between 2006 and
2009 potentially fall under the scope of their passenger protections
regarding long delays (i.e., where flights are delayed by at least 2
hours).
[71] Regulation (EC) 261/2004, articles 6(1)(iii) and 8(1)(a), 2004
O.J. (L 46/1). But see, infra, C-402/07 and C-432/07, Sturgeon v.
Condor Flugdienst GmbH Böck, Lepuschitz v. Air France SA, 2009 E.C.R.
I-10923.
[72] Regulation (EC) 261/2004, art. 14, 2004 O.J. (L 46/1).
[73] As mentioned earlier, passengers who experience a long delay on
the airport tarmac in the United States are entitled to certain care,
such as food and water no later than 2 hours, and the assurance a
plane will not remain on the tarmac greater than 3 hours. 14 C.F.R.
part 259. Additionally, the Montreal Convention may also be applicable
when international passengers are delayed, allowing passengers to take
legal action against an airline for damages. Convention for the
Unification of Certain Rules for International Carriage by Air
(Montreal Convention), ch.3, art. 19, May 28, 1999 (entered into force
on Nov. 4, 2003), reprinted in S. Treaty Doc. No.106-45. Furthermore,
Canadian airlines are required to address the "failure to operate on
schedule" in their contract of carriage. Air Transportation
Regulations SOR/88-58, Part V, Div. I, § 107(1)(n)(vi) (Can). Finally,
DOT officials told us that they can, in certain circumstances, take
enforcement actions and issue fines against airlines unless the
airline provides passengers with refunds after cancellations and
extensive delays.
[74] Rerouting shall be provided under comparable transport
conditions, to the passenger's final destination at the earliest
opportunity; or, under comparable transport conditions, to the
passenger's final destination at a later date at the passenger's
convenience, subject to the availability of seats. Regulation (EC)
261/2004, art. 8, 2004 O.J. (L 46/1). The rerouting does not
necessarily need to be operated by the airline the passenger booked
with. See, Information Document of the Directorate-General for Energy
and Transport: Answers to Questions on the Application of Regulation
(EC) 261/2004, February 17, 2008.
[75] Regulation (EC) 261/2004, articles 5 and 8, 2004 O.J. (L 46/1).
Reimbursement must be within seven days and must provide the full cost
of the ticket at the price at which it was bought, for the part(s) of
the journey not made, and for the part(s) of the journey made if the
flight is no longer serving any purpose in relation to the passenger's
original travel plans. Regulation (EC) 261/2004, art. 8(1)(a), 2004
O.J. (L 46/1).
[76] The operator may reduce the compensation by 50 percent if the
arrival time on the reroute does not exceed the originally scheduled
arrival time by (1) 2 hours for all flights 1,500 kilometers (km) or
less, (2) by 3 hours for all intra-community flights more than 1,500
km and all other flights between 1,500 km and 3,500 km, and (3) by 4
hours for all other flights. Regulation (EC) 261/2004, art. 7(2), 2004
O.J. (L 46/1).
[77] Regulation (EC) 261/2004, art. 5(3), 2004 O.J. (L 46/1). See also
C-549/07, Wallentin-Herman v. Alitalia--Linee Aeree Italiane SpA, 2007
E.C.R. I-nyr.
[78] Regulation (EC) 261/2004, art. 14, 2004 O.J. (L 46/1). According
to a recent Commission report of delay data voluntarily provided by
airlines, on average, about 0.5 percent of scheduled flights in Europe
between 2006 and 2009 were canceled and potentially fall under the
scope of their passenger protections regarding cancellations.
[79] Consumers have a private right of action for refunds under the
Consumer Credit Protection Act, 15 U.S.C. § 1601 et seq.; DOT has
codified these requirements at 14 C.F.R. part 374. Additionally,
according to officials at DOT, it can take enforcement action against
airlines, under 49 U.S.C. § 41712, for the refusal to issue refunds
for the purchase price of a ticket, within a certain number of days,
to passengers who wish to cancel their trip as a result of a flight
cancellation or significant schedule change (for example, flight
delay).
[80] Cancellations and refund policies, however, are required to be
addressed in an airline's contract of carriage. Air Transportation
Regulations SOR/88-58, Part V, Div. I, § 107(1)(n)(vi), (vii), 122(c)
(Can).
[81] Canadian regulations require airlines to set forth policies
related to compensation for denial of boarding in their contracts of
carriage. Air Transportation Regulations SOR/88-58, Part V, Div. I, §
107(1)(n)(iii), (vii), 122(c) (Can).
[82] 14 C.F.R. § 250.2b; Regulation (EC) 261/2004, art. 4, 2004 O.J.
(L 46/1). Under the EU regulation, passengers who volunteer to
surrender their reservations must also be offered the choice between
reimbursement and a return flight to the first point of departure or
rerouting to their final destination at the earliest opportunity.
[83] Boarding priority factors may include, but are not limited to,
the following: a passenger's time of check-in, whether a passenger has
a seat assignment before reaching the gate for airlines that assign
seats, the fare paid by a passenger, a passenger's frequent flyer
status, and a passenger's disability or status as an unaccompanied
minor. 14 C.F.R. § 250.3.
[84] The factors that determine the amount of financial compensation
are different in the EU and the U.S. In the EU, compensation depends
on the time passengers reach their final destination after being re-
routed and the distance of the flight, and in the United States, it
depends on the time passengers are re-routed and whether it is a
domestic or international flight. Regulation (EC) 261/2004, articles 4
and 7, 2004 O.J. (L 46/1); 14 C.F.R. § 250.5. Passengers who are
involuntarily denied boarding will not be entitled to compensation if
they are offered alternative transportation that, at the time the
arrangement is made, is planned to arrive at the first stop-over, or
if none, the final destination, not later than 1 hour after the
original flight's arrival time. U.S. airlines may offer free or
reduced rate air transportation in lieu of cash if the value is equal
to or greater than the cash payment otherwise required and the airline
informs the passenger of the amount of cash compensation that would
otherwise be due and that the passenger may decline the transportation
benefit and receive the cash payment. 14 C.F.R. § 250.5(b).
Additionally, in accordance with DOT's Final Rule on Enhancing Airline
Passenger Protections, as of August 23, 2011, the airline must also
disclose all material restrictions on airline travel vouchers that may
apply. 76 Fed. Reg. 23110 (Apr. 25, 2011), at 49 C.F.R. § 250.5(c)(3).
[85] Regulation (EC) 261/2004, articles 8 and 9, 2004 O.J. (L 46/1).
[86] Regulation (EC) 261/2004, art. 14, 2004 O.J. (L 46/1); 14 C.F.R.
§ 250.9. Additionally, in accordance with DOT's Final Rule on
Enhancing Airline Passenger Protections, as of August 23, 2011, the
airline must also provide verbal notification of such rights. 76 Fed.
Reg. 23110 (Apr. 25, 2011). Furthermore, although not required, the
four major Canadian airlines include in their contracts of carriage
similar denied boarding procedures as in the United States and the EU.
[87] Officials from the Canadian Transportation Agency identified the
four major Canadian airlines as Air Canada, WestJet, Air Transat, and
Air Canada Jazz.
[88] The Code of Conduct also specifies that airlines should provide
refreshments to passengers who experience a delay while on the tarmac,
if it is safe, practical, and timely to do so, and the option to
deplane after 90 minutes if circumstances permit.
[89] Additionally, several airline officials told us that they do not
document their overall costs for complying with passenger service
requirements.
[90] Steer Davies Gleave, Evaluation of Regulation 261/2004, European
Commission, Directorate-General Energy and Transport (London, UK,
February 2010).
[91] See GAO, Airline Deregulation: Reregulating the Airline Industry
Would Likely Reverse Consumer Benefits and Not Save Airline Pensions,
[hyperlink, http://www.gao.gov/products/GAO-06-630] (Washington, D.C.:
June 9, 2006).
[92] During the last 5 years, flight cancellations have generally been
less frequent in the EU than in the United States. According to
Commission estimates, flight cancellations comprised about 0.5 percent
of all flights in Europe from 2006 through 2009. According to DOT
data, canceled flights in the United States ranged from about 1.4
percent to 2.1 percent of U.S. flights during the same period. The EC
reported that flight cancellations in Europe increased to about 2.3
percent of all flights during 2010 primarily because of unfavorable
flight conditions caused by the eruption of the Eyjafjallajökull
volcano in Iceland.
[93] Increases in required compensation for passengers involuntarily
denied boarding could motivate airlines to be more aggressive when
seeking volunteers for denied boarding, such as increasing the value
of vouchers offered to passengers that willingly deplane, since they
may be able to offer these volunteers compensation that is less than
the amount required for passengers involuntarily denied boarding.
[94] Officials of one low-cost U.S. airline, though, did tell us that
the cost associated with the 2008 increase in denied boarding
compensation forced the airline to reduce overbooking levels by
approximately 25 percent.
[95] Prior to implementation of the current EU requirements
(Regulation (EC) 261/2004), EU law required airlines to provide
passengers denied boarding on flights of 3,500 kilometers or less with
150 euros and passengers denied boarding on flights of more than 3,500
kilometers with 300 euros. Regulation (EC) 261/2004 increased the
required denied boarding compensation to 250 euros for flights of
1,500 kilometers or less, 400 euros for flights of more than 1,500
kilometers but less than 3,500 kilometers, and 600 euros for flights
of more than 3,500 kilometers.
[96] Officials from enforcement bodies we spoke with in all three
regions told us they attempt to resolve passenger complaints with the
airline informally before conducting a formal investigation.
[97] DOT officials told us that they closed 717 passenger complaint
cases in 2009 and 828 cases in 2010. As of May 26, 2011, they have
closed 328 cases and 454 cases remain open. In 2009-2010, CTA received
a total of 642 air travel complaints for processing through its
dispute resolution processes, 599 for informal facilitation and 43 for
formal adjudication.
[98] Officials at the CTA told us that enforcement staff can impose
financial penalties against an airline or take administrative actions
against the airline, including issuing formal reprimands, cease and
desist orders, license suspensions, and license cancellations. CTA can
also order an airline to modify their contract of carriage and if the
airline has failed to apply its contract of carriage correctly, they
can award a passenger their out-of-pocket expenses as a result of the
airline's failure to apply the contract of carriage. They may also
order appropriate corrective measures pursuant to regulation. Air
Transportation Regulations SOR/88-58, Part V, Div. II, § 113.1(a)
(Can).
[99] According to DOT officials, these proactive investigations began
in 2008 and involve on-site examination of airlines' compliance with
DOT's air travel consumer protection regulations covering advertising,
refunds, carriage of passengers with disabilities, baggage, and
oversales, as well as to review the airline's customer service
commitments and manuals and training materials relevant to DOT's
consumer protection requirements. DOT officials said they conduct
about six such investigations each year.
[100] Steer Davies Gleave, Evaluation of Regulation 261/2004.
[101] European Commission, Communication From the Commission to the
European Parliament and the Council (April 2011).
[102] "Extraordinary circumstance" is not defined in Regulation (EC)
261/2004. The regulation exempts an airline from the obligation to pay
compensation in the event of a cancellation if the airline can prove
that the cancellation is caused by extraordinary circumstances that
could not have been avoided even if all reasonable measures had been
taken. Regulation (EC) 261/2004, art. 5(3), 2004 O.J. (L 46/1). In
Wallentin-Herman, the ECJ defined extraordinary circumstances as
events which by their nature are not inherent in the normal exercise
of the activity of the airline and are beyond its actual control.
Wallentin-Herman v. Alitalia--Linee Aeree Italiane SpA, 2007 E.C.R. I-
nyr.
[103] Wallentin-Herman v. Alitalia-Linee Aeree Italiane SpA, 2007
E.C.R. I-nyr.
[104] Sturgeon v. Condor Flugdienst GmbH Böck, Lepuschitz v. Air
France SA, 2009 E.C.R. I-10923.
[105] TUI Travel Plc, British Airways Plc, Easyjet Airline Company Ltd
and International Air Transport Association v. The Civil Aviation
Authority, Case CO/6569/2010.
[106] According to Commission officials, passengers in the UK may
still submit a complaint related to the ECJ ruling through the court
system.
[107] The Commission does strive to ensure that the application of EU
regulations is harmonized so that passengers are provided with equal
treatment across the EU. In order to create harmonization, the EU
encourages enforcement bodies to discuss issues and standardize their
practices through meetings and the development of voluntary agreements
and issuance of Commission guidance. Additionally, the ECJ provides
uniform and authoritative interpretations of EU legislation to ensure
that they are applied in the same way in all EU member states.
Ultimately, though, member states are given latitude in their
enforcement practices.
[108] Officials from the Netherlands's enforcement body told us that
legislative changes made to national law (Aviation Act) were made in
December 2009 to permit this body to impose repressive sanctions.
[109] In Canada, CTA can order an airline to pay out-of-pocket
expenses to passengers if the airline has failed to apply the
provisions in the contract of carriage. Passengers can also submit
claims through an alternative dispute resolution (ADR) process in the
EU, but this option is available in only four EU member states, the
rulings are not always binding on the airline, and ADR staff may lack
of expertise to handle cases, especially those related to
extraordinary circumstances.
[110] Although DOT cannot adjudicate claims and make monetary awards,
DOT officials noted that its enforcement staff periodically intervenes
with airlines to obtain compensation for a passenger to which he or
she is entitled. Additionally, according to DOT, it may obtain
compensation for consumers through settlements of potential violations
with airlines and may also insert such provisions into a consent order.
[111] An even more recent report announced that the Commission will
work with the network of national enforcement bodies to agree on
harmonized interpretation and enforcement of the regulation. In 2011
it will launch an Impact Assessment to assess the proportionality of
the current measures in the light of experience and the costs of the
regulation for stakeholders, with a view to propose further measures
on Air Passenger Rights, including of a legislative nature, in 2012.
[112] Passengers can use the European small claims procedure
(Regulation (EC) 861/2007, 2007 O.J. (L 199/1)), which Commission
officials told us aims to accelerate the settlement of small cross-
border disputes and is available to every resident in a member state.
[113] EU Claim is a privately funded organization headquartered in the
Netherlands. Based on flight data it has collected, EU Claim advises
clients about the validity of their claims against airlines pursuant
to Regulation (EC) 261/2004 and may choose to represent these clients
before the airline and, if necessary, court. The fee for the
organization's services is a percentage of any compensation awarded to
the client.
[114] In 2009, the Netherlands collected about 750 passenger
complaints, whereas Germany collected about 3,000. In Europe in 2009,
a total of 35,198 complaints were received. The majority of these
complaints (41 percent) were due to cancellations, and 25 percent were
due to delays.
[115] The Commission reported in 2011 that it will work with national
enforcement bodies to promote a more uniform and quick handling of
complaints as well as encourage airlines to establish reasonable and
precise time frames for handling passenger claims.
[116] We excluded commercial service, nonprimary airports because they
handle fewer than 10,000 enplanements annually and therefore are not
considered "hub" airports that provide connectivity to the national
airspace system.
[117] 14 C.F.R. part 234.
[118] The regulation also provides for the voluntary reporting of an
airline's entire domestic system and voluntary reporting by other
airlines. For calendar year 2010, the reporting airlines were AirTran
Airways, Alaska Airlines, American Airlines, American Eagle Airlines,
Atlantic Southeast Airlines, Comair, Continental Airlines, Delta Air
Lines, ExpressJet Airlines, Frontier Airlines, Hawaiian Airlines,
JetBlue Airways, Mesa Airlines, Northwest Airlines, Pinnacle Airlines,
SkyWest Airlines, Southwest Airlines, United Airlines, and US Airways.
Other airlines reported data in other years of our analysis when they
accounted for at least one percent of total domestic scheduled
passenger service revenue.
[119] Beginning in October 2008, DOT required airlines to submit
tarmac delay statistics to ASQP for three additional categories:
flights that are subsequently canceled or diverted or have multiple
gate departures. For consistency, we omitted these flights from our
analysis of tarmac delay trends from 2004 to 2010. Our analysis of
tarmac delay trends also excluded flights reported to DOT by Atlantic
Southeast Airlines, Hawaiian Airlines, and Comair because, according
to DOT, these airlines inconsistently reported gate returns.
Specifically, some airlines misreported the gate-departure or takeoff
time resulting in an incorrect tarmac delay calculation. Further, our
analysis excluded data for any flights with tarmac delays of more than
10 hours since there were errors in the reporting of many such
flights. Finally, our analysis of tarmac delay trends did not include
data for international flights.
[120] Additionally, our analysis excluded flights reported to DOT by
Atlantic Southeast Airlines, Comair, and Hawaiian Airlines due to
errors in how they reported the data to DOT. Our analysis also
eliminated flights delayed more than 10 hours because there were
errors in the reporting as confirmed by DOT.
[121] Similar data was not publicly available in Canada or the EU.
[122] While data were available for later months in 2010, which we
could have compared with the same months in 2009, we believe that
factors affecting cancellations during winter months differ
considerably from factors that influence cancellations during summer
months, when weather is less predictable ahead of time. Thus, we
believe that if the rule has increased cancellations, it would do so
primarily during summertime weather events, and, that is therefore,
the focus of this model.
[123] Both models analyze flights between 70 airports in the United
States and Puerto Rico. The FAA Aviation System Performance Metrics
database includes performance data on operations at 77 airports. Our
analysis included domestic flights operated by DOT reporting airlines
to and from these airports, with the exception of three in Alaska and
Hawaii (Ted Stevens Anchorage International Airport, Honolulu
International Airport, and Kahului Airport) which we chose not to
include, and four airports (Gary Chicago International Airport,
Greater Rockford Airport, Teterboro Airport, and Van Nuys Airport)
that had no reported flights to or from any of the other ASPM airports
during the months of our analysis. In addition, for the tarmac model
(but not for the gate model) we did not include flights that were
diverted and thus did not land at the scheduled destination airport
and flights that had longer than a 60 minute taxi-in delay at the
destination airport.
[124] The number of flights included in both the tarmac and gate-
cancellation models are somewhat less than the number of flights used
to develop the unadjusted odds ratio because some observations that
had missing data were deleted from the logistic model.
[125] Each observation is one flight segment. Many passengers change
planes and have two or more segments on a one-way trip.
[126] Specifically, according to data on delays occurring when flights
are under FAA control, we found that 80 percent of departure delays
across the national airspace system were due to conditions at just 7
of the 35 Operational Evolution Partnership airports--Newark Liberty
International, New York LaGuardia, New York John F. Kennedy
International, Chicago O'Hare International, Atlanta Hartsfield
International, San Francisco International, and Philadelphia
International. See [hyperlink, http://www.gao.gov/products/GAO-10-542].
[127] Data by airport was available for the 35 Operational Evolution
Partnership airports and for the remaining airports the on-time
performance for the entire national airspace at the relevant hours was
used.
[128] The airlines classified as legacy included Delta Air Lines,
United Airlines, American Airlines, US Airways, and Continental
Airlines. Low-cost airlines included AirTran Airways, Alaska Airlines,
JetBlue, and Southwest. The remaining airlines were grouped as "other."
[129] See, for example, Jing Xiong, Revealed Preference of Airlines'
Behavior Under Air Traffic Management Initiative.
[End of section]
GAO's Mission:
The Government Accountability Office, the audit, evaluation and
investigative arm of Congress, exists to support Congress in meeting
its constitutional responsibilities and to help improve the performance
and accountability of the federal government for the American people.
GAO examines the use of public funds; evaluates federal programs and
policies; and provides analyses, recommendations, and other assistance
to help Congress make informed oversight, policy, and funding
decisions. GAO's commitment to good government is reflected in its core
values of accountability, integrity, and reliability.
Obtaining Copies of GAO Reports and Testimony:
The fastest and easiest way to obtain copies of GAO documents at no
cost is through GAO's Web site [hyperlink, http://www.gao.gov]. Each
weekday, GAO posts newly released reports, testimony, and
correspondence on its Web site. To have GAO e-mail you a list of newly
posted products every afternoon, go to [hyperlink, http://www.gao.gov]
and select "E-mail Updates."
Order by Phone:
The price of each GAO publication reflects GAO’s actual cost of
production and distribution and depends on the number of pages in the
publication and whether the publication is printed in color or black and
white. Pricing and ordering information is posted on GAO’s Web site,
[hyperlink, http://www.gao.gov/ordering.htm].
Place orders by calling (202) 512-6000, toll free (866) 801-7077, or
TDD (202) 512-2537.
Orders may be paid for using American Express, Discover Card,
MasterCard, Visa, check, or money order. Call for additional
information.
To Report Fraud, Waste, and Abuse in Federal Programs:
Contact:
Web site: [hyperlink, http://www.gao.gov/fraudnet/fraudnet.htm]:
E-mail: fraudnet@gao.gov:
Automated answering system: (800) 424-5454 or (202) 512-7470:
Congressional Relations:
Ralph Dawn, Managing Director, dawnr@gao.gov:
(202) 512-4400:
U.S. Government Accountability Office:
441 G Street NW, Room 7125:
Washington, D.C. 20548:
Public Affairs:
Chuck Young, Managing Director, youngc1@gao.gov:
(202) 512-4800:
U.S. Government Accountability Office:
441 G Street NW, Room 7149:
Washington, D.C. 20548: