Testimony of

J. Howard Beales III

Professor of Strategic Management and Public Policy

Before the

Subcommittee on Digital Commerce and Consumer Protection

On

Understanding the Digital Advertising Ecosystem

you for the opportunity to testify today on the Digital Advertising Ecosystem. I am Howard

Beales, a Professor of Strategic Management and Public Policy at the George Washington

School of Business. I have published a number of academic articles on privacy regulation. From

2001 to 2004, I was the Director of the Bureau of Consumer Protection. During that time, the

Commission re-thought its approach to privacy regulation and promulgated the National Do Not

Call Registry.

I want to make three essential points this morning. First, internet content is a public

good: it is not used up in consumption. Private market provision of such public goods has

generally depended on revenue from advertising, as does internet content today. Second, the

value of advertising depends critically on the availability of information about the likely viewer.

When information is available, advertising prices are roughly 3 times higher than when there is

no information about the viewer. Impairing the flow of information would significantly reduce

the revenues available to support internet content, an impact that would be particularly

problematic for smaller publishers. Third, advertising is actually beneficial to consumers. It

leads to more competitive markets, with lower prices and more product improvements. It also

narrows the differences between different demographic groups.

Internet Content is a Public Good

Among a nearly infinite variety of possibilities, consumers can now listen to radio broadcasts,

watch television programs, read the daily paper, or just hang out with their friends online.

Although these activities have considerable value to consumers, they are frequently supplied to

1
consumers free of charge. Instead, Internet content is largely funded by advertisers who pay to

have their ads included along with the online content.

From an economic perspective, Internet content is a “public good.” Unlike private goods,

public goods are not “used up” in consumption, and instead remain available for other consumers

to enjoy. A classic example of a public good is free broadcast radio or television. Any number

of consumers can enjoy the content, without any additional costs of providing it.

Long before the Internet, publishers developed effective mechanisms to finance content

that consumers wanted despite the public good nature of their product. Conventional media

markets face the same underlying economic issues, and offer valuable insights into successful

models for the provision of content.

The most common market mechanism for providing public goods is advertising. In

effect, advertising converts the public good of media content into a private good of exposures to

advertising. Content becomes a way for the publisher, to attract an audience that in turn can be

sold to advertisers. Because advertisers ultimately want to reach individual consumers, a larger

audience is more valuable than a smaller one – it produces more advertising exposures available

for sale.

The business of producing content and selling advertising is a “two-sided” or “platform”

market. Content must attract an audience, but the platform must also attract advertisers. The

financial support for the content comes from advertising revenue. In some circumstances, such

as directories or fashion magazines, advertising may increase the overall value of the product to

consumers. In other circumstances, however, advertising is a nuisance: Too much advertising,

or advertising that is too intrusive or offensive to consumers, may drive away some of the

2
audience, thereby reducing the number of advertising exposures that can be sold. The publisher

must consider both sides of the market in deciding what content to provide and how much

advertising to offer.

Throughout history, advertiser support has been a vital revenue source for media

companies. Many, such as free broadcast radio or television, depend almost entirely on

advertising revenue for survival. Also common are mixed models, such as the typical magazine

or newspaper, or cable television programming, where subscription payments from consumers

provide some revenue, but typically advertising revenue remains vital and is frequently the

largest source of revenue.

There are, of course, some models that are purely supported by subscription revenues,

such as satellite radio or premium cable TV channels. Market behavior makes clear, however,

that most consumers most of the time are not willing to pay a premium price to avoid advertising

content.

There is nothing fundamentally different in the provision of online content from

providing similar content in conventional media markets. Publishers, ranging from major media

companies to specialty sites that specialize in particular niches, must cover the costs of

producing the content they provide. Although there are other models, by far the most common

business model supporting the provision of Internet content is advertising. Given the long history

of advertiser supported media markets, that fact should not be surprising.

3
The Value of Advertising Depends on Information

In any media market, the price of advertising depends on the characteristics of the

audience. In conventional media, where large numbers of consumers of necessity see the same

advertisement, advertisers choose where to advertise based on the demographic characteristics of

the audience as a whole. Not surprisingly, some audiences are more valuable than others,

because more advertisers are interested in reaching them or they are harder to attract to

programming. Advertising prices therefore depend on audience demographics as well.

Online, each consumer who visits a website can be served a different advertisement.

What advertisers are willing to pay for that slot, however, depends critically on what they know

about the viewer. And in turn, what advertisers are willing to pay determines the resources

available to support the content of that particular website. Anonymity may appear attractive to

an individual viewer, but because it reduces the price of the advertisement, it reduces the revenue

available to support the content of the website that the viewer is enjoying. It is, in short, a subtle

form of free riding on the contributions of others.

There are two predominant forms of online advertising: search advertising and, broadly

speaking, display advertising. Search advertising is purchased based on the keywords that a

consumer has just entered in a search engine and is usually sold on a cost per click basis. That is,

the web page is paid based on the number of clicks on the advertisement, rather than the number

of consumers who see it. Advertisers bid for keywords, and the search engine provider will

select which advertisements to include in the results based on the bid price and its own estimate

of the likelihood that this consumer will find the advertisement sufficiently interesting to click on

4
it. Information that enables the search provider to make better estimates of the likelihood that a

consumer will click on the link will increase the provider’s revenue.

The other major category of online advertising is display advertising, which includes

display and banner ads, rich media, and digital video ads. Display advertising is generally sold

on a cost per thousand (CPM) basis. Third party intermediaries, including advertising networks

and ad exchanges, are key participants in this marketplace. Advertising networks pool inventory

from numerous, usually small publishers. Advertising is increasingly sold in real-time auctions,

with advertisers bidding for particular advertising availabilities based on what, if anything, they

know about the viewer. In the auction, the highest bidder wins the advertisement, at the price

offered by the second highest bidder. Information about the viewer is obtained through cookies,

which enable advertising networks and others to determine what other websites that particular

user has visited.

In two separate studies, I have examined the impact of better information on the price of

digital advertising. In a 2010 study, I surveyed 12 of the 15 largest advertising networks to

determine the impact of behavioral targeting, which uses data based on user browsing behavior

across multiple web sites to categorize likely consumer interest in a given advertisement. I

compared the price of advertising on a CPM basis when it was sold based on behavioral targeting

with the price when the advertisement was sold on a “run of network” basis, meaning that it

could appear anywhere on the network with no specification as to the characteristics of the user.

I found that the CPM for behaviorally targeted advertising was roughly 3 times higher than the

5
price of run of network advertising – a substantial price premium. I also found that the majority

of advertising revenue was passed through to the publisher.1

A second study, with Jeffrey Eisenach, analyzed 2013 impression-level data from two

anonymous operators of automated advertising exchanges to determine how better information

influenced the auction price. We found that more information led to a price premium that was

both statistically and economically significant. If there was a cookie available with the

impression, the price was roughly 3 times higher than if there was no cookie. Moreover, the

longer the cookie had been in place, the greater was the increase in price. The price of an

impression with a cookie that had been in place for 90 days was 3.7 times higher than the price

with no cookie on one exchange, and 7.1 times higher on the other. The study also used data

from Adomic, which measured the relative prevalence of different advertising sales models

across the top 4,000 Internet publishers. Even the largest publishers sold about half of their

advertising availabilities through third-party technologies, while smaller, “long-tail” publishers

relied on these approaches for up to two thirds of their advertising sales.2

Other studies support the same conclusion: the value of online advertising, and hence the

revenue available to support the production and development of online content, depends

critically on the availability of information about the likely viewer of the ad. Regulatory

requirements that impair the flow of information will significantly reduce the revenue available

1
Howard Beales, “The Value of Behavioral Targeting,” published online by Network Advertising Initiative,
available at http://www.networkadvertising.org/pdfs/Beales_NAI_Study.pdf, March, 2010.
2
J. Howard Beales and Jeffrey A. Eisenach, “An Empirical Analysis of the Value of Information Sharing
in the Market for Online Content,” published online by Digital Advertising Alliance, available at
http://www.aboutads.info/resource/fullvalueinfostudy.pdf, January, 2014.

6
to online content producers, leading to a less vibrant Internet. The impact will be greatest on the

smaller publishers, who are most dependent on third-party technologies for advertising revenue.

It is also vital to recognize that regulatory rules are likely to have very different impacts

on different companies. Companies that utilize sign-ins are likely to have the most information,

because they can typically observe the consumer’s behavior whenever he or she is signed in to

the service. Thus, Facebook and Google likely have significant informational advantages over

other participants in the online advertising marketplace. Some large publishers with many

different content pages will have information about behavior as the consumer moves around their

various offerings. Other important participants in the online marketplace, however, are not

consumer-facing at all. Instead, they work with publishers or advertisers to observe behavior

across independent websites through the use of cookies. There are numerous such companies,

most of whom consumers have never heard of – for example, 33across, Accuen, Acuity, and

Adara, which happen to be the first four names on the the list of members of the Network

Advertising Initiative. More elaborate consent requirements could seriously disadvantage these

companies, and help protect the market shares of the current leaders in the online advertising

market: Facebook and Google. As in many other areas, large players in online advertising

markets have incentives to agree to regulatory requirements that they can satisfy more easily than

their smaller competitors. And as in any other market, creating regulatory barriers that have the

effect of protecting market leaders from competition is bad for consumers.

7
Advertising Provides Important Benefits for Consumers

Individually, we may think of advertising as a nuisance, and many times it is. The ability

to advertise, however, is critical to maintaining effective competition in markets for goods and

services.

The competitive benefits of advertising are by now well known. In the words of Nobel

Laureate George Stigler, “advertising is an immensely powerful instrument for the elimination of

ignorance.”3 Informed consumers drive the competitive process, benefitting all consumers as

sellers compete for the informed minority.4 Numerous economic studies have shown that

restrictions on advertising increase prices to consumers, even when advertising does not mention

price.5

Advertising also stimulates innovation. If sellers cannot advertise innovative products, or

if they cannot tell consumers why new product characteristics are important, there is less

incentive to make improvements in the first place.6 One of the best studied examples involves

Kellogg’s 1984 claims for All Bran cereal, conveying the then novel recommendation of the

National Cancer Institute (“NCI”) that diets high in fiber may reduce the risk of some cancers.7

The science, which was based largely on epidemiology rather than human clinical trials, was

3
George J. Stigler, The Economics of Information, 64 J. POL. ECON. 213, 220 (1961).
4
See, e.g., Alan Schwartz and Louis L. Wilde, Intervening in Markets on the Basis of Imperfect Information:
A Legal and Economic Analysis, 127 U. PA. L. REV. 630 (1978-1979).
5
The FTC itself has summarized the empirical evidence regarding the impact of advertising on prices. See
In re Polygram, 2003 WL 21770765 (FTC), Docket No. 9298 (July 24, 2003), at note 52.
6
Advertising is an intangible investment, whose value can only be recovered through repeat sales. Sellers
invest in and maintain product quality to generate repeat business. See Phillip Nelson, Advertising as Information,
82 J. POL. ECON. 729 (1974).
7
The Kellogg incident is discussed in J. Howard Beales, Timothy J. Muris, and Robert Pitofsky, “In Defense
of the Pfizer Factors,” in James C. Cooper, Ed., The Regulatory Revolution at the FTC: A Thirty-Year Perspective
on Competition and Consumer Protection (Oxford University Press, 2013), pp. 83-108.

8
uncertain. Citing these uncertainties, the FDA threatened to seize All Bran as an unapproved

new drug. When the FTC and the NCI defended Kellogg, the FDA changed course.

An FTC Staff Report documented the impact of the Kellogg campaign and its aftermath.8

Increased advertising about fiber content and its relationship to cancer risks led to significant

changes in cereals.9 Claims about the relationship between diet and disease increased elsewhere

as well, with similar marketplace impacts. For example, claims about the relationship between

diet and heart disease rose from less than 2 percent of food advertising in 1984 to more than 8

percent in 1989;10 consumption of fat and saturated fat, the primary dietary risk factors for heart

disease, fell far more sharply after 1985.11 Again, advertising led to beneficial changes in diet.

Advertising is particularly important to less advantaged groups. The FTC Staff Report

documented that although fiber consumption increased for all groups, it increased more among

racial minorities and single parent households.12 Another study has shown that the least

educated paid the highest increase in prices when eyeglass advertising was restricted. 13

Online advertising can be expected to have similar effects to any other advertising, and

those effects are generally good for consumers. Restrictions that impair its effectiveness can

only reduce those benefits.

8
Pauline Ippolito & Alan Mathios, Health Claims in Advertising and Labeling: A Study of the Cereal
Market, FTC Staff Report (1989), available at http://www.ftc.gov/be/econrpt/232187.pdf.
9
For example, the fiber content of new cereals increased 52 percent, and the weighted average content of
cereals (reflecting both product changes and changes in consumer choices) increased at a significantly higher rate
than before health claim advertising began. Ippolito and Mathios, supra note 8.
10
Pauline Ippolito & Janice Pappalardo, Advertising Nutrition & Health: Evidence from Food Advertising,
1977–1997, FTC Staff Report (2002), available at http://www.ftc.gov/opa/2002/10/advertisingfinal.pdf.
11
Pauline Ippolito & Alan Mathios, Information and Advertising Policy: A Study of Fat and Cholesterol
Consumption in the United States, 1977–1990, FTC Staff Report (1996), available at
http://www.ftc.gov/be/consumerbehavior/docs/reports/IppolitoMathios96_fat_long.pdf.
12
Ippolito and Mathios, supra note 8.
13
Lee Benham & Alexandra Benham, Regulating through the Professions: A Perspective on Information
Control, 18 J.L. & Econ. 421 (1975).

9
Thank you again for the opportunity to testify today. I look forward to your questions.

10