Will New Media Destroy The Local Media?

Will New Media Destroy the Local Media?
Susan Athey, Emilio Calvano, Joshua Gans

Monash University Seminar
October, 2009
Tuesday, 10 November 2009

Introduction Issues
Baseline Model Research Questions
Extensions Literature
Conclusion Outline
Media is under pressure
Will New Media Destroy the Local Media? 2 /36

Introduction Issues
Conclusion Outline
Fall in advertising revenue

Introduction Issues
Conclusion Outline
Fall in news share of ad revenue

Introduction Issues
Conclusion Outline
Web-ads less effective than print

Introduction Issues
Conclusion Outline
Fall in advertising revenue Loss of ‘good’ journalism

Introduction Issues
Conclusion Outline
Fall in news share of ad revenue Local newspaper bankruptcies

Introduction Issues
Conclusion Outline
Fall in news share of ad revenue Local newspaper bankruptcies
Web-ads less effective than print Rise of aggregators/blogs

Introduction Issues
Conclusion Outline
New Media has ...

Introduction Issues
Conclusion Outline
New Media has ...
increased content variety

Introduction Issues
Conclusion Outline
New Media has ...
allowed consumers to choose their own

content bundles

Introduction Issues
Conclusion Outline
New Media has ...

content bundles
reduced advertising effectiveness

Introduction Issues
Conclusion Outline
New Media has ...

content bundles
allowed new consumer targeting

technologies

Introduction Issues
Conclusion Outline
New Media has ...

content bundles

technologies

Introduction Issues
Conclusion Outline
New Media has ...
increased content variety Competition

content bundles

technologies

Introduction Issues
Conclusion Outline
New Media has ...

content bundles

technologies

Introduction Issues
Conclusion Outline
New Media has ...

content bundles

Efficiency technologies

Introduction Issues
Conclusion Outline
... but the fundamentals are unchanged

Introduction Issues
Conclusion Outline

“… in an information-rich world, the wealth of information means a dearth of
something else: a scarcity of whatever it is that information consumes. What
information consumes is rather obvious: it consumes the attention of its recipients.
Hence, a wealth of information creates a poverty of attention and a need to allocate
that attention efficiently among the overabundance of information sources that might
consume it.”

Introduction Issues
Conclusion Outline

consume it.”
(Herbert Simon, 1971, pp.40-41)

Introduction Issues
Conclusion Outline

consume it.”

Introduction Issues
Conclusion Outline

consume it.”
“I know that half of my advertising dollars are wasted … I just don’t know which half.”

Introduction Issues
Conclusion Outline

consume it.”
(John Wanamaker)

Introduction Issues
Conclusion Outline

consume it.”
(John Wanamaker)

Introduction Issues
Conclusion Outline

consume it.”
(John Wanamaker)
“Newspaper readers are ‘better’ than Web visitors. Online readers are a notoriously fickle
bunch, and apparently are getting more so by the day. Web visitors barely stick around, yet
they are counted in broad traffic statistics as if they were the same as the reader who
lingers over his Sunday paper.”

Introduction Issues
Conclusion Outline

consume it.”
(John Wanamaker)
(Paul Farhi, Washington Post)

Introduction Issues
Conclusion Outline

consume it.”
(John Wanamaker)
(Paul Farhi, Washington Post)

Introduction Issues
Conclusion Outline
The news for local media is mixed

Introduction Issues
Conclusion Outline
Local media’s tailored content allowed consumer self-selection

Introduction Issues
Conclusion Outline

... threatened by geographic targeting

Introduction Issues
Conclusion Outline

The Internet is allowing consumers to switch between outlets and unbundle
content

Introduction Issues
Conclusion Outline

content
... remove local media bundling rents vs hyper-local opportunities

Introduction Issues
Conclusion Outline

content
The Internet is competing for consumer attention

Introduction Issues
Conclusion Outline

content
The Internet is competing for consumer attention
... so there is less share to traditional media

Introduction Policy Issue
Conclusion Outline
Research Questions

Conclusion Outline
Research Questions
What is the impact of consumer switching and geographic

targeting on local media?

Conclusion Outline
Research Questions
What is the impact of consumer switching and geographic

targeting on local media?
Provide a full equilibrium analysis

Conclusion Outline
Existing Approaches

Conclusion Outline
Existing Approaches
Two-Sided Markets
Rochet & Tirole, Armstrong: use prices to different sides of the market to
maximise platform profits
Armstrong & Wright: competitive bottlenecks can arise when one side
single-homes while the other multi-homes. Rents to single-homers.

Conclusion Outline
Existing Approaches
Two-Sided Markets
Rochet & Tirole, Armstrong: use prices to different sides of the market to
maximise platform profits
Armstrong & Wright: competitive bottlenecks can arise when one side
single-homes while the other multi-homes. Rents to single-homers.
Traditional Media Economics
Anderson & Coate (2005): broadcasters compete for viewers and then sell
advertising according to a revenue function
Consumers are assumed to single-home while advertisers (implicitly) are
assumed to multi-home
Charge monopoly price for access to viewers

Conclusion Outline
Outline

Conclusion Outline
Outline
Start with traditional media economics environment (i.e., single-homing

consumers) and examine local versus general outlet competition

Conclusion Outline
Outline

The impact of geographic targeting

Conclusion Outline
Outline

The impact of blogs

Conclusion Outline
Outline

The impact of blogs
Extensions (generate competition for advertisers)

Conclusion Outline
Outline

The impact of blogs

Capacity constrained advertisers

Conclusion Outline
Outline

The impact of blogs

Consumers who switch between outlets

Conclusion Outline
Outline

The impact of blogs

Consumers who switch between outlets
Perfect tracking

Introduction Elements
Baseline Model Equilibrium
Extensions Impact of Targeting
Conclusion Impact of Blogs
Consumers
Consumers (readers)
Endowed with T periods of attention
Live (and make purchases) in local market, m (there are M local markets)
Outlet choices
Case C-SH: single-homing
Case C-VS: variety seeking
Case C-SB: stochastic browsing

Advertisers
Advertisers
Are located in a specific local market, m
Symmetric
Only one impression per consumer from a given advertiser is valuable over T
periods of attention
Value of impressing a consumer, v, distributed F(v) (special case, U[0,1])
Advertiser valuations
Case A-CV: constant valuations (unlimited demand)
Case A-CC: capacity constrained (desired limited ‘conversions)
Will New Media Destroy the Local Media? 10/36

Outlets
An outlet, i, has advertising capacity of ai per unit of attention.
The outlet can ensure that each consumer is reached just once (e.g. ads on the
sports page)
Thus, if they capture a consumer’s attention for t periods, the total number of
advertisers they can supply to is tai.
Advertising effectiveness:
θi,m: the probability that an impression is on the ‘right’ consumer for an advertiser from
market m.
Depends on the mix of consumers served by the outlet
Local outlets (lm) only attracts consumers from market m
General outlet’s (g) readers are from all markets.
Let xm be the share of consumers from market m served by the local outlet, and 1-xm
be the share served by the general outlet
Start with this exogenous, then endogenize through quality investments qi
Tailoring
Advertisements in local outlets are more effective than those in general

ones
An advertiser in location m’s expected return for a general outlet
impression is:
1 − xm
θ g, m =
M − ∑ m ′ ≠ m xm ′
More localities implies higher wasted impressions on general outlet
For symmetric local markets:
1
θg =
M
Advertising demand (C-SH, A-CV)
pi,m: impression price for outlet i in location m

Demand: advertisers with valuations, v > pi,m
Because consumers single-home, advertisers multi-home (if there is
capacity)
Di, m ( pi, m ) = 1 − F( pi, m / θ i, m )
Pi, m (ai ) = θ i, m F −1 (1 − ai )

Advertising supply (C-SH, A-CV)
Consumers choose one outlet for all attention periods

Each advertiser will be supplied one impression per consumer per outlet
How many impressions can a local outlet supply?
For each consumer, it can supply Tai impressions
How many impressions can a general outlet supply to advertisers in m?
Tag − ∑ m ′ ≠ m Dg, m ′


Tag − ∑ m ′ ≠ m Dg, m ′


Tag − ∑ m ′ ≠ m Dg, m ′
Impressions
to consumers in all localities


Tag − ∑ m ′ ≠ m Dg, m ′
Impressions
to consumers in all localities


Tag − ∑ m ′ ≠ m Dg, m ′
Impressions Impressions supplied

to consumers in all localities to advertisers
outside of m

Example
Suppose that ag = alm = 1, T = 2 and M = 2
Suppose that in each locality there are two advertisers with values V and v (V > v)
What will the general outlet earn?
As the general outlet cannot tell the location of the consumer, its capacity will be sold only to
the higher value advertisers with value V
Note, however, that these advertisers will pay for impressions in each locality and so the value
of those impressions will be V/2
Thus, the general outlet can charge pg = V/2 and will earn per consumer profits of:
π g = pgTag = V
What will local outlets earn?
Can supply both advertisers in the locality.
Can earn a price of v
So its profits will be:
π lm = plmTal = 2v
Market Clearing Prices
Local outlet:
D( pl, m ) = Tal, m ⇒ pl, m = P(Tal, m )
General outlet:
∑ D( p g θ g, m ) = Tag
m
with symmetry:
D( pg θ g ) = Tag M ⇒ pg = (1 M )P(Tag M )

What happens with Geo Targeting?


UK website


Australian ad
UK website

What happens to prices/profits?

How do prices compare if al,m = ag = a?
pl, m > pg ⇒ P(Ta) > 1
M P( M1 Ta)
Intuition:
General outlets sell less to each advertising market, which pushes up
prices, while inefficiency pushes them down.
Shape of the demand function matters, i.e., is P(a) > αP(αa)?
Trade-off between inefficiency and market thickness hinges on whether
reducing demand to an Mth of its current level increase prices by more
than a factor of M
If advertiser demand is uniform, the comparison becomes:
pl, m > pg ⇒ M (M + 1) > Ta

Endogenous Advertising Capacity
Local outlet’s problem:

max al ,m Tal, m P(Tal, m )
General outlet’s problem:

1
max ag M Tag P( M1 Tag )
Note that the general outlet’s problem is the same as the local outlet’s problem
except for the scaling factor (1/M). So we will get:
a *
l, m = 1
M a *
g p *
l, m = Mp *
g
Profit comparison per consumer: equal profits!

Each type of outlet can sell its own consumers’ attention for the same amount.

Summing up Impact of Geo Targeting

Case (i): exogenous ad capacity (e.g., page constraints)
• Targeting may make general outlet better or worse off, depending on
slope of demand curve; if worse off, general outlet won’t adopt
technology
• If targeting makes general outlet better off, local outlets are not directly
impacted, but are indirectly impacted through quality investment
• Then ‘local news media’ may stop investing in quality
• Social value of Geo Targeting greater than private value (more advertisers
supplied)
Case (ii): endogenous ad capacity
• Targeting does not affect profits: local and general make same profits per
consumer with and without geo targeting
• General has higher investment incentives with and without targeting
Entry of Blogs
• Blogs may or may not be able to sell advertising

• Since outlets do not compete for advertising, blogs do not affect
advertising prices
• Blogs capture attention from traditional outlets and reduce incentives to
invest
• The outlets that lose most consumers to blogs will lose further
consumer share from the investment effect

Introduction
Baseline Model Capacity Constrained Advertisers
Variety Seeking Consumers
Extensions
Perfect Ad Tracking
Conclusion
Diminishing value for consumers

• Capacity constrained advertisers -- looking to hold expected
“conversions” fixed
• General impressions yield 1/M of local conversions
• M will not be the optimal number of impressions for a general outlet but
we start by assuming advertisers buy M general for every local
impression
• With C-SH and A-CC, the outlets do directly compete for advertisers
• Generates single quality-adjusted price in the market
• One unit of supply from the general outlet counts as 1/M
• Market clearing quality-adjusted price (in units for local):
pl = P(T (al + 1
M ag )) pg = 1
M pl
Introduction
Extensions
Perfect Ad Tracking
Conclusion
Impact of Geo Targeting
• Local outlet profits per consumer

Tal P(T (al + 1
M ag ))
• General outlet profits per consumer
Tag 1
M P(T (al + 1
M ag ))
• With fixed advertising capacity
• General gets an Mth the local profit per consumer
• With endogenous advertising capacity
• General selects the same effective capacity as local and gets the same
profit per consumer
• So introduction of targeting does not matter

Introduction
Extensions
Perfect Ad Tracking
Conclusion
Optimal Choice of General Impressions
Advertisers on general outlet choose their impression budget (n) to solve:
n (v; pg ) = arg max n 1 − (

*
( M )
) v − npg
M −1 n
n* will differ from M and is decreasing in v.

Expected surplus from advertising on local outlet always higher (for equal
impression prices) than advertising on general outlet. This implies that pl will
exceed pg.
High value advertisers sort onto local outlet with price constraints:
vl − pl = 1 − ( ( *
M −1 n (vl ; pg )
M ) ) vl − n* (vl ; pg )pg

Introduction
Extensions
Perfect Ad Tracking
Conclusion
Preliminary Results (C-SH, A-CC)
Supply-side: marginal advertisers on each outlet
1 − F(vl ) = xlTal
vl
M ∫ n* (v; pg )dF(v) = xgTag
vg
Implications:
• Advertisers are segmented by quality
• Outlets compete for advertisers at the margin
• Difference in profits depends on M
• Increased share of non-advertising blogs increases pl and pg prices

Introduction
Extensions
Perfect Ad Tracking
Conclusion
Targeting Impact (C-SH, A-CC)
• Case (i): exogenous ad capacity

• Targeting makes general outlets better in proportion to the efficiency
gain (unlike A-CV, this is unambiguous and substantial)
• Case (ii): endogenous ad capacity (same as A-CV)
• General outlets do worse than local outlets without targeting, due to
declining market value of non-targeted consumers
• Both cases
• General outlets have greatly enhanced incentives to invest in quality as a
result of targeting

Introduction
Extensions
Perfect Ad Tracking
Conclusion
Blog Impact (C-SH, A-CC)

• Blogs may or may not be able to sell advertising
• Unlike A-CV, outlets do compete for advertising
• If Blogs capture consumer attention and do not provide advertising (due to
choice or friction in advertising market), they increase equilibrium prices but
decrease scale for existing outlets.
• If blogs do have advertising, they increase supply and lower equilibrium prices
• Blogs are likely to be price takers, so will put up as much advertising as
consumers will watch
• Blog advertising may be inefficient since advertisers don’t know the audience,
so their contribution to total supply may not be as great as it appears
• Innovation in ad platforms may make it more efficient in the future
• Like A-CV, blogs take attention away from traditional outlets and reduce
incentives to invest
• YouTube: lots of attention, few ads
Introduction
Capacity Constrained Advertisers
Baseline Model
Extensions
Perfect Ad Tracking
Conclusion
Multi-homing consumers (C-VS, A-CV)
• Suppose now that consumers switch between outlets

• This is a proxy for what might arise if they used web sites rather than
traditional media
• Assume in C-VS that consumers visit each outlet half the time.
• The demand from advertisers does not change overall but they now
choose to advertise on just one outlet (single-home)
• Outlets service twice the consumers but have half the attention from each.
• Expositional ease: suppose that T = 2

Introduction
Baseline Model
Extensions
Perfect Ad Tracking
Conclusion
Market-clearing impression prices
•vi: marginal advertiser on i

• Assortative matching between advertiser value and effectiveness
means that the highest value advertisers will place ads on the local
outlet and the next tranche on the general outlet
• Two critical values; one that is indifference between local and general;
and one that is indifferent between general and nothing
• Local vs. general: vl –pl = vl /M - pg
• General vs. none: vg/M= pg
• Market clearing:
1 − F(vl ) = al
(
M F(vl ) − F(vg ) = ag )
⇒ 1 − F(vg ) = al + 1
M ag
Introduction
Baseline Model
Extensions
Perfect Ad Tracking
Conclusion
Outlet profits
•Comparing prices
pg = 1
M P(al + 1
M ag )
pl = M −1
M P(al ) + pg = 1
M ((M − 1)P(a ) + P(a + l l
1
M ag ) )
• Local price is determined by indifference between local and general,
where marginal advertiser gets greater value (M-1)/M from local
• General price set by aggregate supply of advertising and discounted for
low effectiveness
• Local outlet serves high-value advertisers and gets a greater price

Introduction
Baseline Model
Extensions
Perfect Ad Tracking
Conclusion
Impact of Geo Targeting
•Profits per consumers:

πg = 1
M P(al + 1
M ag )ag
πl = 1
M ((M − 1)P(a ) + P(a + l l
1
M ag ) al)
• Local always earns higher prices, gets higher profits
• With uniform distribution, local outlet produces more effective
advertising space
• Targeting removes asymmetry in effectiveness
• General outlet benefits and local outlet is harmed

Introduction
Baseline Model
Extensions
Perfect Ad Tracking
Conclusion
Endogenous Ad Capacity (Example)
•Uniform values, M = 2
π g = 4 (2 − ag − al1 − al 2 )ag
1
π l1 = ( 1
2 (1 − al1 ) + (2 − ag − al1 − al 2 ) al1
1
4 )
• Competitive externalities between localities due to impact on general
outlet impressions
• Equilibrium: each outlet supplies 1/2; pg = 1/8 and pl1 = 3/8. Local outlets
earn 3 times profit per consumer
• With targeting each supplies 1/3 and impression price is 1/3
• Profits now 1/9 per consumer

Introduction
Baseline Model
Extensions
Perfect Ad Tracking
Conclusion
Summary
• If the Internet allows consumers to bundle more diverse news outlets

together (through switching) ...
• Increases the local advantage in competition with general outlets for
advertisers
• Merit to the hyper-local push
• If the Internet allows for geographic (or specialised) consumer tracking

• Increases the effectiveness of general outlet advertising
• Decreases the local advantage in competition for advertisers with
general outlets
• Suggests hyper-local moves may be misplaced.

Introduction
Baseline Model
Extensions
Perfect Ad Tracking
Conclusion
Perfect Ad-Tracking
• Suppose that there exists a price-taking ad platform that can perfectly track
consumers: they can say to advertisers, “I will place an impression in front of
a consumer and charge you p for it.”
• Advertisers need only contract with the platform for p per impression.
• Suppose that ad capacity is symmetric and fixed at a
• The total supply of advertising space to a consumer is: 2a
• why? the platform tracks the consumer and so wherever they are places
ads in front of them. The most they can place is 2a.
• The total demand for advertising space is: v = p
• Therefore: 2a = 1 - F(v) implying that p = P(2a)
• Outlet profits:
π i = xi P(2a)2a
• This is the same as under traditional media with single-homing consumers;
thus, even a local outlet earns the same as before.
Introduction
Baseline Model Conclusions
Extensions Future Directions
Conclusion
Conclusions
• Closed model of supply and demand for advertising
• Models with fixed conversions can be used to account for the fact that
proliferation of ad space and ad impressions does not necessarily dry
up the residual demand for advertising
• New technologies for targeting only hurt local news media in cases where
advertisers compete, and then only under certain conditions
• Competition is induced by the internet, when consumers begin browsing
multiple sites
• Competition also induced by capacity constrained advertisers
• Blogs can be good for an outlet if they steal attention from someone else
• Forces for consolidation: big outlets invest more in quality to grab
attention
• High quality general content, low-quality/blog-based local content
Introduction
Baseline Model Conclusions
Extensions Future Directions
Conclusion
Future Directions
• Model with mixture of single and multi-homing consumers

• Wasted impressions
• Frequency capping technologies
• Examine perfect ad tracking under more general conditions
• Examine content provision
• Bundling
• Average versus marginal content quality


Will New Media Destroy The Local Media?

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Will New Media Destroy The Local Media?

Uploaded by

Copyright:

Available Formats

Will New Media Destroy the Local Media?

Susan Athey, Emilio Calvano, Joshua Gans

Tuesday, 10 November 2009

Media is under pressure