5.2. EFFECTS OF DIGITIZATION
Scholars have studied how the shift from analog to digital distribution of mass media is
affecting consumers, advertisers, and media platforms. For consumers, the most studied
phenomena are the increased control digitization offers by, for example, allowing con-
sumers to selectively avoid advertisements. For advertisers, the two-way nature of digital
media networks enhances their ability to target their messages to consumers. For media
content platforms, digital signal delivery offers new business models. Taken as a whole,
these effects change the nature of competition in mass media markets and the range of
products provided.
5.2.1 Consumer Control over Advertising Exposure
A burgeoning empirical literature documents how increased viewer control has affected
viewers’ exposure to advertising.
3
Esteves-Sorenson and Perretti (2012) found that tele-
vision remains largely a passive activity, with strong state persistence despite the small cost
of switching channels. However, viewers do switch when they are annoyed by commer-
cials.
Interian et al. (2009) analyzed a large dataset of tuneaway behavior during commer-
cials, showing that the typical advertisement loses 1–3% of the audience at the beginning
of the ad slot.
Schweidel and Kent (2010) presented moment-by-moment viewing data
culled from digital STBs showing that program ratings fall by an average of 10% during ad
breaks and that program genre can reliably predict audience retention during commercial
breaks.
Wilbur (2015) investigated household-level zapping behavior and found that
advertising content characteristics reliably predict commercial avoidance, after control-
ling for program content, audience composition, and audience heterogeneity. Movie ads
were zapped less often than average while auto insurance and website commercials were
avoided more frequently.
Tuchman et al. (2015) report that the median household in
their data skips about 10% of the ads that it is exposed to.
Have elevated levels of advertising avoidance diminished the profitability of advertis-
ing? The limited evidence available suggests that the answer may be no.
Zufryden et al.
(1993)
surprisingly found that ad exposures interrupted by zapping were more strongly
associated with sales than uninterrupted exposures; they hypothesized that this difference
was due to viewers’ heightened attention while changing channels. More recently,
Bronnenberg et al. (2010) examined data produced by a field study in which digital video
recorders were given to households in a scanner panel data sample. Households’ adver-
tising exposures and brand purchases in the following year were compared to baseline
3
It is important to remember that advertising avoidance is as old as advertising; Wilbur (2008) reviewed the
literature on traditional modes of avoiding ads, which included leaving the room, muting the television,
changing channels, or diverting attention to other activities, or engaging in conversation.
213
Recent Developments in Mass Media
brand purchases in the previous year. The confidence interval for the effect of DVR
acquisition on advertising effectiveness was tightly centered around zero, suggesting
no diminution in advertising effectiveness, a result they attributed to fairly low rates
of advertising avoidance among panelists.
Is this a puzzle? As advertising avoidance has gotten easier, commercial avoidance
seems to become more frequent. Yet there is no evidence of diminished advertising effec-
tiveness, and television advertising revenues have not fallen. There are three possible
explanations. One is that advertisements may be designed to retain their effectiveness,
even when played at high speed. For example, prominent displays of the brand logo
reduce advertising avoidance (
Teixeira et al., 2010), and can increase brand attitude,
intention to purchase, and choice behavior (
Brasel and Gips, 2008). A second possibility,
which to the best of my knowledge remains unexplored, is that perhaps households do
not avoid ads when they are in the market for the advertised product. In other words, an
advertisement exposure that a consumer actively chooses to avoid might have offered
little value to the advertiser in the event of an exposure. Further, many commercials
are repeated with such frequency that even frequent ad-avoiders may receive some min-
imum number of exposures to the most common commercials. The third possibility is
that (at least some) ads may be complements to product consumption, as originally pro-
posed by
Becker and Murphy (1993). Tuchman et al. (2015) investigated this possibility
using a single-source panel database of purchases and advertising exposures. They found
that greater recent brand consumption led to a lower probability of zapping the brand’s
television advertisement in the future.
A number of theoretical analyses have produced competing predictions about how
increased levels of advertising avoidance will affect media markets.
Anderson and
Gans (2011)
modeled a monopoly platform and examined how consumer adoption of
advertising avoidance technology would alter its strategy. In equilibrium, consumers
who value programming the most adopt advertising avoidance technology first, leading
to rising advertising time, falling welfare and content quality, and more mass-market con-
tent.
Athey et al. (2013) extended this framework to allow for competing outlets, imper-
fect measurement of advertising exposure, and endogenous multihoming by advertisers.
As consumer switching increases, advertising levels rise and the premium paid for large
audience increases, but total advertising expenditure falls.
4
In contrast, Ghosh and Stock
(2010)
did not model media platforms but they did consider the effect of informative
advertisements on price competition in the product market. In their model, advertising
avoidance leads to some consumer ignorance, which can sometimes increase advertisers’
product prices and profits in equilibrium, raising the demand for advertising.
4
See also Chapter 10.
214 Handbook of Media Economics
5.2.2 Increased Targeting of Advertising
Digital distribution networks enable media platforms to identify individual recipients of
programs and to choose the level of targeting offered to advertisers. Media vary in the
targeting capabilities they offer. For example, television signal distributors could theoret-
ically sell commercial breaks personalized to individual households, but they have so far
only engaged in limited tests of more-aggregated targeting based on geodemographic
segments (
Deng, 2015). Most mass media platforms online allow varying degrees of audi-
ence targeting.
Bergemann and Bonatti (2011, 2014), Dukes and Gal-Or (2003), and Ghosh et al.
(2013)
highlight the essential trade-offs in platforms’ choice of ad targeting capabilities.
5
Untargeted advertising produces many low-value, wasted exposures because most con-
sumers are uninterested in most products. Highly targeted advertising eliminates wasted
exposures, but it increases advertiser competition in the product market, driving down
rents and willingness to pay for advertising. Therefore, the media platform will optimally
choose between offering some intermediate degree of ad targeting, or combining ad tar-
geting with exclusive advertising rights.
Empirical evidence on the increased targeting of advertising is more scarce. Perhaps
the best known papers are
Goldfarb and Tucker (2011a,b). They first investigated several
thousand randomized advertising experiments, showing that matching advertising con-
tent to site content or making advertisements more obtrusive each increases advertising
effectiveness individually, but doing both in conjunction backfires. This effect is likely
because consumers realize they are being targeted and react negatively, as the negative
interaction was largest in sensitive product categories such as health and financial services.
Second, they applied a difference-in-differences framework to estimate the effect of a
European restriction in advertising targeting on advertising effectiveness. They compared
post-policy European advertising response to pre-policy European advertising response,
as well as to US advertising response in both periods. The results indicated that display
advertising became far less effective, especially so on niche-oriented websites, for smaller
advertisements, and for advertisements without obtrusive (e.g., interactive, video, audio)
features.
5.2.3 New Media Business Models
Although it has long been known that advertising causes audience loss, television net-
works have never charged per unit of audience delivered. Nielsen audience estimates
are too noisy to precisely estimate audience sizes by commercial. Instead, television net-
works have always charged per unit of time (i.e., per 30-s spot). Advertising sales
5
See also Chapter 10.
215Recent Developments in Mass Media
contracts do include minimum audience thresholds, but these thresholds are often set
approximately two standard deviations below the expected audience size. When
Nielsen’s point estimate of the audience size falls short of the promised level, then the
network provides “make-goods”—free advertising inventory to compensate the adver-
tiser for the difference between the minimum guaranteed audience size and the point
estimate of the audience size.
The possibility of substantial inefficiencies in advertising sales arises because of a diver-
gence in television advertisers and networks’ interests. Advertisers seek to “break through
the clutter” by getting their advertisements noticed, and therefore include obtrusive stim-
uli which help to gain attention and reinforce their selling message. These stimuli may be
aversive to some viewers who are not in the market for the advertised good, leading them
to change the channel or start fast-forwarding. Often, these viewers will miss the rest of
the commercial break and possibly subsequent breaks as well. Therefore, the most effec-
tive ads (i.e., those that sell very effectively) may actually reduce the network’s stock of
viewer attention available for sale to other advertisers. This problem is exacerbated by
television networks’ traditional practice of ordering advertisements within the break
on a quasi-random basis.
Goldstein et al. (2013) devised an ingenious test of this hypothesis. They first hired
workers online to complete a standard email categorization task while simultaneously
being exposed to banner advertisements. The workers were randomly assigned to one
of three pay rates and one of three advertising conditions: no ads, annoying ads, or innoc-
uous ads. The researchers then observed each worker’s output, allowing them to estimate
the compensating variation of the various types of advertising content. They found that
innocuous ads and no ads led to approximately equal levels of worker output, but annoy-
ing ads led to faster task abandonment. To achieve the same rate of task abandonment as
caused by annoying ads, the researchers would have had to reduce the workers’ pay rate
by approximately $1 per 1000 impressions. This effect exceeds the market price that
many websites charge for display advertising. One might speculate the effect is even larger
in the context of television, as video advertisements are substantially more obtrusive than
banner advertisements.
Wilbur et al. (2013) note that the rapid proliferation of digital television signal
delivery networks allows the possibility of redesigning television networks’ business
models to better align networks’ and advertisers’ interests. They designed an algorithm
to select, order, and price advertisements in a commercial break of endogenous
length. The main idea of the algorithm is to first find the selection and ordering of
advertisements that maximizes total welfare of all advertisers in the commercial break.
Doing so requires the use of dynamic programming to optimize the networks’ oppos-
ing goals of retaining audience and allocating each slot in the break to the advertiser
with the highest willingness to pay. Payments are assigned using a second-price auction
to incentivize truthful bidding and to reward advertisers with low zapping rates.
216 Handbook of Media Economics
The authors estimated the primitives of their model using market data and used a
series of simulations to show that it consistently finds the globally optimal solution,
increases advertiser welfare and network revenues, and runs fast enough to implement
at scale.
5.2.4 Media Market Outcomes
Digitization of media distribution networks is changing media markets by lowering dis-
tribution and entry costs, by altering consumer discovery of new and niche products, and
by changing the resulting range of product quality offered.
Digitization has brought dramatic change in print newspaper markets.
George and
Waldfogel (2006)
studied the effects of entry and expansion by a high-quality national
newspaper (the New York Times) into local newspaper markets in the United States. They
found that readers targeted by the national newspaper tended to adopt it in place of local
newspapers, and that local papers responded by increasing content appealing to the
national newspaper’s non-targeted segments.
The effects of digital distribution have been studied extensively in the context of
consumer products.
Brynjolfsson et al. (2011) found that online distribution produces
a longer tail of product sales than physical distribution even holding the selection of
available products constant. They attributed the increase in the size of the tail to the
improved consumer search and recommendation engines enabled by digital delivery.
On the other hand,
Ellison and Wolitsky (2012) investigated how firms might respond
to increased competition online by raising consumers’ search costs. They showed that
firms have incentives to obfuscate (e.g., complicate consumer evaluation of available
options) when consumer search costs are convex in effort or when consumers have
imperfect information about the exogenous components of search costs.
Waldfogel (2013) examined the joint impact of digitization on sales, production, and
distribution costs, and product quality in the recorded music industry. Digitization trig-
gered three substantial changes: music piracy increased, barriers to entry fell, and new
modes of music discovery (e.g., Internet radio) were invented. The result was a sharp
decrease in album production by traditional music labels, a huge influx of new producers,
and a steadily increasing share of commercially successful albums coming from indepen-
dent producers.
Anecdotally, the print newspaper industry has experienced some changes similar to
those documented by
Waldfogel (2013). A common question is whether similar changes
will eventually occur in the television/video industry? In the author’s view, one possible
differentiator is that in online video, current bandwidth constraints bind tightly. Video
consumption is too high, video program files are too large, and today’s cable and telco
networks are too slow for all consumers to simultaneously switch from consuming broad-
cast video to consuming individual video streams over home Internet connections
217Recent Developments in Mass Media
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.