Transcript Slides (PPT)

Externalities in the Demand for
Privacy: Evidence from the Do-NotCall Registry
Khim-Yong Goh, NUS
Kai-Lung Hui, HKUST
I.P.L. Png, NUS
Privacy and Regulation
• “Privacy is a fundamental right that people do
care about.”
Pamela Jones Harbour, Commissioner of U.S. FTC
• Existing research has been lagging behind:
– Surveys and laboratory experiments
– Little empirical research about real behavior
– Mostly study direct benefits/costs of privacy, not
externalities between consumers
WISSP, October 2010
2
Externalities
• How one person’s decision affect others
– Pollution, public good, etc.
• Two types of externalities among consumers
related to privacy
– Preference externalities: How consumers’ product
preferences affect others
– Strategic marketing avoidance: How consumers’
marketing avoidance efforts affect others
WISSP, October 2010
3
Research Objectives
• Using real registration data from the U.S. Do
Not Call Registry: – Identify the presence of preference externalities
vis-à-vis strategic marketing avoidance in the
demand for privacy
– Draw implications on optimal government
intervention and vendor responses
WISSP, October 2010
4
Theory – Preference Externalities
• Reasons for preference externalities
– Fixed costs in retailing  limited products
– Incomplete information on consumer preferences 
product configuration based on aggregate attributes,
e.g., age and gender
– These apply equally well to direct marketing
• Within-segment: demand  when the size of the
same segment increases
• Cross-segment: demand  when the size of
other segments increases
WISSP, October 2010
5
Theory – Preference Externalities
• The larger the size of a demographic segment
 the more likely a person will find the products
that she wants via direct marketing
 the less likely that she will sign up for DNC
• H1 (Preference externalities): Marketing avoidance in
a particular demographic segment would decrease
with the size of that segment
WISSP, October 2010
6
Theory – Strategic Marketing
Avoidance
• Vendors ignore harms caused by solicitations
• Consumers avoid marketing when the harms
exceed the benefits from direct marketing
• Marketing avoidance facilities serve to “filter”
the population for vendors
• What would vendors do when the population
becomes “richer”?
• How would remaining consumers respond?
WISSP, October 2010
7
Theory – Strategic Marketing
Avoidance
• In general, larger consumer segments
 stronger impact on vendors’ expected profits
due to marketing avoidance
 stronger strategic complementarity
• H2 (Strategic marketing avoidance): Marketing
avoidance in a particular demographic segment
would increase with the size of that segment
WISSP, October 2010
8
The DNC Registry
• Fixed line and mobile, not business numbers
• Telemarketers must check phone numbers
against DNC no less frequently than 31 days
• Registrations were indefinite
• Desirable characteristics
– Real observations of consumer choices
– Free service, low sign up cost, and no competition
– Covers entire USA; lots of variations in consumer
demographics across regions
WISSP, October 2010
9
4
5
Data Sources, Issues, etc.
0
1
2
3
Our equilibrium
state
0
10
20
30
40
50
Week
• County- and MSA-level population data from
U.S. Census Bureau
WISSP, October 2010
10
Preference Externalities: Evidence
WISSP, October 2010
11
Preference Externalities: Evidence
WISSP, October 2010
12
Marketing Avoidance: Evidence
• There was concomitant increase in consumer
purchase of directly marketed items and
marketing avoidance
WISSP, October 2010
13
Empirical Model
• DNC registration rates:
-ve  preference externalities
+ve  strategic marketing avoidance
• With help of population data, specifically,
• Integrating the above equations,
WISSP, October 2010
14
SMA among highincome people
Results
PE among lesseducated people
SMA among
Hispanics
and nonHispanics
SMA among
nonSpanish-only
speakers
WISSP, October 2010
15
Robustness
Different equilibrium
states
Perhaps the effects
were due to direct
communications?
Maybe consumer
heterogeneity mattered?
WISSP, October 2010
16
County-Level Markets
• DNC registration rates:
• Estimation equation,
• Within-segment parameters could be
identified, but not cross-segment parameters
WISSP, October 2010
17
County-Level Markets: Results
Fairly consistent with the MSA-level results…
WISSP, October 2010
18
Summary of findings
• Strategic marketing avoidance: High income,
age, Hispanic ethnicity, non-Spanish-only
speaking consumers
• Preference externalities: Less educated
consumers
– Waldfogel (2003, RJE) and George and Waldfogel
(2003, JPE): Preference externalities among black,
white, and Hispanic ethnicity, but not education in
radio and newspaper markets
– Our (telemarketing) results are opposite
WISSP, October 2010
19
Implications (1)
• Evidence of strategic marketing avoidance
 low-value consumers opted out
 More profitable “market” for the vendors
 Opt out facilities could indeed be helpful
to vendors!
WISSP, October 2010
20
Practical Suggestions
• Direct marketers should support “opt out”
facilities because such facilities help refine
marketing lists
• Once this is understood, there will be less need for
government to offer such services
– Should not “over-respond” to such refinement of
marketing lists
• (otherwise “profitable”) Consumers may strategically
respond by opting out
• Possible exception on competition law so that direct
marketers can agree to limit solicitations?
WISSP, October 2010
21
Implications (2)
• Preference externalities segmented by
education  either:
– Telemarketers have not segmented markets by
education fine enough, or
– Product customization is too low in that market
• Perhaps refine offering by education level?
– Financial/investment services, or educational
services…
WISSP, October 2010
22
Concluding Remarks
• Evidence of externalities  strong justification
for government regulation of privacy
– These are beyond “standard” reasons, such as
consumer myopia or protecting vulnerable
segments, e.g., children
WISSP, October 2010
23
Thank you!
WISSP, October 2010
24