Willingness to Pay and

The Demand for Broadband Access

Paul Rappoport, Temple University

Presentation to
BROADBAND MARKETS:
BRINGING TOGETHER SUPPLY AND DEMAND
New Zealand Institute for the Study of Competition and Regulation
October 17, 2003
Presentation Outline
o Statistics
o Questions
o Broadband Demand
n Descriptive analysis
o Broadband Demand
n Models
o Conclusions
Statistics
 Cautionary Note

“If data analysis is to be well done, much of it must

be a matter of judgment, and ‘theory’, whether
statistical or non-statistical, will have to guide, not
command.”

John W. Tukey, Annals of Mathematical Statistics, Vol 33,

Number 1, March 1962
Also, keep in mind
o “Poor data and good reasoning give poor results.
Good data and poor reasoning give poor results. Poor
data and poor reasoning give rotten results.” Edmund
Berkeley, Computers and Automation, V 10, Num 10, Sept, 1969

o “By no process of sound reasoning can a conclusion

drawn from limited data have more than a limited
application” J.W. Mellor, Higher Mathematics for Students of
Chemistry and Physics
Questions
o Are the “drivers” of broadband the same as the
“drivers” of dial-up Internet access?
o Is there evidence of a broadband “Digital
Divide” ?
o Do “applications” correlate with the choice of
access ?
o How important is price?
Descriptive Assessment
o Broadband Access
n Income
n Education
n Household Size
n Age
Distribution of Internet Penetration by Income
70 18
16
60
14
50
12
40 10
30 8
6
20
4
10 2
0 0
<$

$1

$2

$3

$5

$7

>$
15

5-

5-

5-

0-

5-

1
00
$2

$3

$5

$7

$1
5

00
Dial-up Broadband
Source: MSG Centris Omnibus Survey. Unconditional Rates
Internet and Education
45
40
35
30
25
20
15
10
5
0
< HS HS Some College College Grad

No internet Dial-up Broadband

Source: MSG Centris Omnibus Survey. Unconditional Rates
Internet Penetration and Household Size
40
35
30
25
20
15
10
5
0
1 2 3 4 >4

No internet Dial Broadband

Internet Penetration by Age
25

20

15

10

0
No internet Dial Broadband

<20 20-29 30-39 40-49 50-59 60-74 >74

Source: MSG Centris Omnibus Survey. Unconditional Rates
Analysis of Basic Demographics

Whereas the availability of broadband

service matters in the choice of
broadband, household demographics
are of limited value when attempting
to classify households as broadband
or narrowband!
Descriptive Analysis
Household Internet Activity
Household Activity
o Type of sites visited
o Number of sites visited
o Frequency of sites visited
o Time site was visited
o Duration of a visit
o Other activity derived from the visit
Distribution of Sites by Access
Distribution of type of site visited
Distribution of hours on line
 Distribution by time of day
Usage Distribution By Time of Day

45
Narrowband
40
Broadband
35
30
Minutes of Use

25
20
15
10
5
0
t

am

pm
m

m
on
gh

2a

4a

6a

8a

2p

4p

6p

8p
no
ni

10

10
id
M
Usage Characteristics
Usage Characteristics

600
Narrowband
500 Broadband
Minutes of Use

400

300

200

100

0
mean median standard deviation
Average Share of Visits by Category
30
25
20
15
10
5
0
Dial-Up Broadband

Travel Finance Online Shop Business

Information Svc Entertainment Internet
Downloading Intensity
50
45
40
MB/Month

35
30
25
20
15
10
5
0
Fi In O In En Bu Tr
na te nl fo te sin av
nc rne in rm rt el
ia eS at ai es
lS t ho io nm s
er pp n en
v ice ing Se t
rv
s ice
s

Dial-up Broadband

Source: Click-stream data from Plurimus, Inc.

Models
o Discriminant Models
o Discrete Choice
o Discrete – Continuous Choice
o Contingent Valuation
Discriminant Analysis
o Classification of broadband users
n Higher levels of income
n Higher levels of education
n Age 30- 50
n Visits to business sites, financial sites and
entertainment sites
o Internet activity important in classifying
access
Discrete Choice
o Choices
n Dial-up vs No Internet
n Dial-up vs Cable modem
n Dial-up vs ADSL
n Dial-up vs Cable modem or ADSL
n Cable modem vs ADSL
 Dial-up vs CM Access Elasticities
Dial-up CM

Dial-up -0.230 0.518

CM 0.010 -0.895
Dial-up vs ASDL Elasticities
Dial-up ADSL

Dial-up -0.168 0.423

ADSL 0.040 -1.364

Discrete – Continuous Choice
 Problem: Choice and
Usage Jointly Determined

Activity
Time
online

Choice of Type of sites

Access visited
Discrete – Continuous Choice
o Issues
n Access choice and usage jointly dependent
n All other measures of usage such as sites visited,
frequency of visits, duration of visit are also
endogenous
n Modeling requires (at a minimum) a nested logit
framework estimated within a simultaneous equations
framework
n Estimation requires detailed household specific
variables
Willingness to Pay
Contingent Valuation: Overview
o Method that requires asking people directly,
in a survey, how much they would be willing
to pay for a specific service.
o “Contingent” in the sense that people are
asked their willingness to pay, contingent on
specific hypothetical scenario.
Problem(s)
1. Can willingness to pay (WTP) information
be obtained from surveys and used to
describe “demand” ?
2. How do estimates of elasticities for
broadband services compare to published
estimates?
3. Can the use of WTP be generalized and
applied to other products and services?
 WTP and the Demand for
Broadband Service

o Focus is on the price of the service – thus

economic value associated with a service is
generally bounded
o Application is directed towards the estimation
of price elasticities
Lognormal Demand Curves
be the tolerance price of the ith household be the actual market
o Let
1 if 0 otherwiseioiiqppq=≥=
o Then

oip
o Assuming that is distributed as a
2 and ppms
lognormal with parameters
Lognormal Demand
We have:

222P(1|)P()1(;,)Let Q represent the expected proportion of buyers we h

Data
o 12,000 responses to an omnibus survey
administered during the second quarter, 2003.
o Questions included for broadband service
(ADSL, Cable Modem) – WTP.
Application to Broadband
o Question 1 What is the least price at which
the respondent would consider the item too
expensive
o Question 2 What is the highest price at which
he would dismiss it as a shoddy article of
inferior quality
Computation
o Compute the fraction of respondents quoting a
threshold price that exceeds a price p.
o Plot Q(p) against p
o Estimate lognormal parameters from the data
logQ()log()dpdpp=
o Elasticity given by
Demand for Cable Modem Service

Figure 1: Cable Modem Demand

40
35
30
Proportion

25
20
15
10
5
0
0

0
10

20

30

40

50

60

70

80

90

10
Price (WTP)
 Cable Modem Elasticity
Price Elasticity
$20 -0.45
$30 -0.65
$40 -0.87
$50 -1.23
$60 -1.58
$70 -1.83
Elasticity
o Initial estimates are in line with previously published
values (evaluated at price of $40-$50)
n Rappoport, Taylor, Kridel
o CM (-0.81 -1.05)
o DSL (-1.17 -1.55)
n WTP
o CM (-0.87 -1.23)
Conclusions
o Theory of consumer choice appears to
“work” (easily implemented)
o Illustrates potential value using WTP
approach
o Derived elasticities in line with other
published results
Conclusions
o Demographics play a minor role in predicting
broadband vs narrowband users
o Internet activity (usage) helps discriminate
between narrowband and broadband users
o Price matters
o Keep in mind, in the U.S., the top three usage-
based activities are: (1) Gaming, (2) gambling
and (3) other entertainment
Conclusion
o Prices for broadband will fall
o Entertainment is the focus of content
providers, ISPs and telecom companies
o No evidence of a digital divide, though less
densely populated areas are less like to have
broadband
o No compelling reason for government
subsidies or policies promoting broadband
Contact Information
o Paul Rappoport
n Prapp4@comcast.net or
n Prapp@sbm.temple.edu
o Temple Fox School of Business
n 215-204-5025
n 215-913-1775 (cell)