You are on page 1of 26

1

Total Delay Impact Study


Mark Hansen, Thea Graham, Bo Zou,
Chieh-Yu Hsao et. al
TRB
January 11, 2009

2
Outline
Motivation
Goals and Scope
Results to Date
3
Impact of Aviation on the U.S. Economy
Aviation accounts for
11 million aviation-related jobs
$1.2 trillion in economic activity
5.6 Percent Contribution to Gross
Domestic Product
Adds $61 Billion to the U.S. Trade
Balance
The Aviation Industry contributes to the U.S. Economy and International Trade
4
Questions to be Addressed
What are the ripple effects?
Quality of life
Economic productivity
Global economic position
Transportation choice
Societal costs?
Environmental consequences
from uncontained delay?
5
Project Deliverables
Estimate the total cost of airline delay on the
economy and society
Direct costs including passenger and air carriers
Indirect costs, i.e. economic productivity, lost wages, etc.
Tertiary costs including societal and quality of life costs
Estimate the impact of aviation infrastructure
investment on US economic productivity
Final Report January 2010

6
Approach
2007 as case study year
Quantify flight delays and related
metrics
Quantify passenger delays
Assess economic impact
Costs to airlines and passengers
Costs to other sectors
Firm-level case studies
Adaptation cast studies


7
Results to Date
Impact of flight delays on
passenger route demand
Impact of flight delays on airline
costs
8
Impact of Delays on Passenger
Route Demand
9
Introduction
The model
Predicts aggregate link flows from flows in particular city-pair
markets
Deals with demand generation and demand assignment in a
single model (consistent with random utility theory), and quantifies
induced air travel
Models both multiple routes and multiple airports within regions
Capture patterns of correlations among alternatives
Answer empirical questions
How is air service demand affected by causal factors, such as fare
and delay?
What are the welfare impacts of different scenarios, including
different levels of congestion and scheduled flight time?



10
The Demand Model
Model city-pair demand generation and its assignment
among airports and routes



City A City B
H1
H2
O1
O2
O3
D1
D2
Origin airport
Connecting airport
Destination airport
Travel by other
modes, or non-
travel
11
Estimation Results
NL3B model is the preferred model
Routes between two cities (within a
nest) are correlated
A route is
more likely to be substituted by another
route of the same O-D airport pair than the
routes of the other O-D airport pairs
least likely to be substituted by the non-air
alternative
Most coefficients are significant and
have expected signs

Non-
Air
O1
D1
NL3B
O1
D2
O1
H1
D1
O1
H2
D1
O1
H2
D2
O1
H3
D2
Air
OD Airport
Variable NL3B
-OLS
NL3B
-IV
Fare (hundreds of 2004 dollars) -0.160
***
-1.546
***

[0.005] [0.206]
ln(Frequency)Direct
(flights per quarter)
1.337
***
1.240
***

[0.016] [0.029]
ln(Max frequency of two segments)
Connecting (flights per quarter)
0.440
***
0.627
***

[0.009] [0.030]
ln(Min frequency of two segments)
Connecting (flights per quarter)
0.822
***
0.957
***

[0.007] [0.023]
Scheduled flight timeDirect
(minutes)
-0.019
***
-0.004
[0.000] [0.002]
Scheduled flight time Connecting
(minutes)
-0.019
***
-0.006
**

[0.000] [0.002]
Dummy for direct routes
(=1, if direct route)
3.874
***
6.066
***

[0.141] [0.397]
Positive hub arrival delay
t-1
(minutes per flight)
-0.002
**
-0.006
***

[0.001] [0.002]
Positive hub arrival delay
t-4

(minutes per flight)
-0.002
**
-0.007
***

[0.001] [0.002]
Inclusive value of level 3
(parameter=
a p
/ )
0.937
***
0.664
***

[0.011] [0.014]
Inclusive value of level 2
(parameter=
m a
/ )
0.711
***
0.795
***

[0.009] [0.010]
Inclusive value of level 2 *market
distance
-0.008
***
-0.012
***

[0.001] [0.001]
Market distance (hundreds of miles)

0.018
***
-0.024
***

[0.005] [0.005]
ln(market distance) 1.888
***
1.575
***

[0.048] [0.046]
Per capita personal income of market
(thousands of 2004 dollars)
0.015
***
0.038
***

[0.003] [0.003]
Constant (level 1) -17.316
***
-16.229
***

[0.116] [0.102]

12
System Delay Experiment:
Hub Delay
The delay of the 30 hub airports reduce by 100%
Changes in traffic (from NL3B-IV):
Connecting routes: +6.6 million trips
Direct routes: -1.9 million trips
All air routes: +4.7million trips

13
System Delay Experiment
Impacts on LADC Market
The delay of the 30 hub airports reduce by 100%
LA DC
ORD
ATL
+23,093
+6,437 (10.4%)
+6,941(15.4%)
+3,501(4.5%)
Other Hubs: +11,673(6.4%)
-12,985 (3.4%)
LAX
ONT
SNA
BUR
LGB
PSP
DEN
DFW
PHX
+3,884 (5.1%)
+3,643 (5.7%)
DCA
IAD
BWI
+9,472(1.6%)
+8,309(6.1%)
+3,770(5.1%)
+1,367(5.2%)
+786(5.1%)
-611(1.2%)
+1,605(0.6%)
+12,501(6.0%)
+8,987(2.0%)
14
System Delay Experiment:
Schedule Padding
Calculated from individual flight data: ASQP (2004)
For each route-quarter,
Padding time=
Average scheduled flight time -
20
th
percentile of actual gate to gate time
Scenarios for schedule padding,
New scheduled flight time
=Average scheduled flight time - padding time
= 20
th
percentile of actual gate to gate time




15
Welfare Analysis
User benefits approximated by the rule of





is calculated from and fare elasticity of
market demand




ly. respective ts, improvemen with the and ut witho
market a of traffic the are Q and
ly; respective ts, improvemen with the and ut witho
market a of fare the are and where,
2 / ) )( (
1 0
1 0
1 0 1 0
m m
m m
m
m m m m
Q
P P
Q Q P P

+
1
m
P
1
Q
m
16
Welfare Analysis
Scenarios
Net Changes
in Traffic
(Million
Trips)
Total User
Benefits
(Billion
Dollars)
Benefits per
User
($)
(1) Hub Delay
-100%
4.71 0.74 2.48
(2) No Schedule
Padding
7.62 1.17 3.87
(3) Both(1)
and(2)
12.60 1.98 6.44
17
Impact of Delays on Airline
Costs
18
Approaches to Estimating
Cost of Delays to Airlines
Cost Factor
Block-hr DOC (Schumer report)
Block-hr Total Cost (Schumer)
Airline/Airport/ANSP Surveys
(Cook)
Econometric
Economic Cost Function
Unit Cost Change Model
(Discussed here)


19
Unit Cost Change Model
Panel data
9 major airlines
17 years (90-07)
Use changes of cost and influencing
factors
Dependent Variable: % Change in
Operating Cost per ASM
Independent Variables (% or absolute)
Change in delay (using different metrics)
Change in other factors (e.g. costs)


it it it it
c | | | + A + A + = A ) factor other ( y_factor) punctualit ( * ) ost unit_opr_c (
2 1 0
20
Punctuality Factors
Arrival delay
Departure delay
In this study use difference in minutes
between scheduled
and actual arrival/departure time
On-time percentage
a flight is considered on-time when it
arrives less than 15
minutes after its published arrival time
Cancellation rate



21
Metric 1&2: dep./arr. delay
Dependent variable Independent variables Parameter estimates
(t-statistics)
R
2

Percent_cost_change Dep_delay_change
CCI_change
0.007296(3.15)
0.001771(2.98)
0.1281
Dep_delay_change
Percent_CCI_change
0.007759(3.35)
0.2046(2.68)
0.1186
Percent_dep_delay_change
CCI_change
0.0552(2.87)
0.001261(2.67)
0.1187
Percent_dep_delay_change
Percent_CCI_change
0.0586(3.05)
0.1782(2.29)
0.1078
arr_delay_change
CCI_change
0.007429(3.03)
0.001930(3.27)
0.1239
arr_delay_change
Percent_CCI_change
0.007675(3.11)
0.2182(2.86)
0.1097
Percent_arr_delay_change
CCI_change
0.0831(2.94)
0.001928(3.26)
0.1211
Percent_arr_delay_change
Percent_CCI_change
0.0851(2.99)
0.2155(2.81)
0.1057
22
Estimating total delay cost
Our
Estimation
ATA (2007) Schumer (2008)
Cost Factor
Approach
Cost Factor
Approach
Total Cost Approach
Measure of
Delay
A
Arrival delay
> 15 min
C D A B C D
Total
Airline-side
Delay Cost
(billion $)
8.82/9.10
(10.9/11.3)
8.09 3.6 6.1 12.2 21.7 19.1 23.4
delay / cost/cost) (
1
A A = |
( ) ) ASM Total ( * ]
1
1 [ * ) cost unit current ( cost delay total
) delay arr. avg. ( *
1
|
e
=
A: Difference in minutes between scheduled and actual arrival time. Early arrivals set to 0.
B: Actual elapse time-5
th
percentile of observed elapse time for given segment in a given month.
C: Delay leaving the gate +delay during taxi (relative to FAA nominal taxi times) + (airborne delay relative to smaller
of flight plan and block time minus nominal taxi times).
D: Delay leaving the gate +delay during taxi (relative to FAA nominal taxi times) + (airborne delay relative to 5
th

percentile of observed airborne time for given segment in a given month).

23
Estimating marginal delay cost
at a system level
Diseconomies of scale: airlines would suffer more for
marginal increase of delay when delay is at a higher level

24
Estimating avg. delay cost
($/min)
delay / cost/cost) (
1
A A = |
( ) ) capacity seat avg ( * ) length stage avg ( * )
$
( * cost/min delay
1
ASM
| =
Our approach
ATA,
2007
Barnett et
al. 2001
Cook et
al. 2004
ITA, 2000
9 airlines
System-
wide
Delay cost
for airlines
($/min)
87.6/90.5 81.7/84.4 60.46 51.70 60.4 78.5
market-dominant airlines tend to suffer more from
delays than do carriers of relatively small size
25
Summary of Cost Estimates
Approach Metrics
System-wide
cost
Our approach
Model 5/6 Arrival delay type A 8.8/9.1
Model 13~16 On-time percentage 8.0/8.2/9.2/9.5
Model 19/23 Cancellation 2.4/2.7*
ATA (2007) Cost factor based
Arrival delay > 15
min
6.9**
Schumer (2008)
Cost factor based
Arrival delay type C 3.6
Arrival delay type D 6.1
Total cost based
Arrival delay type A 12.2
Arrival delay type B 21.7
Arrival delay type C 19.1
Arrival delay type D 23.4
Cancellation cost
Our estimate of system-wide delay cost is in the range of $ 8-9.5 billion.
If cancellation is taken into account, total cost reaches $10.5-12 billion.

26
Future work
Airline/Aircraft type cost data
Additional explanatory variables
Other punctuality metrics
Variance of delay
Long delays
Diversion
Incorporating buffer delay

You might also like