Information and Brand

1
Information and Brand
2
Last Time

• We learned about search
• We talked about the experiment brought about by
  the Internet
• Commerce appears less frictionless than expected
• One possible major friction reputation, brand

3
Klein Leffler

• Common belief without some third-party enforcer
  to sanction stealing and reneging, market
  exchange would be impossible
• But reputation
• Paper examines nongovernmental repeat-purchase
  contract enforcement mechanism
• Answer market prices above the competitive price
  and the presence of nonsalvageable capital are
  means of enforcing quality promises.

4
Nuts bolts

• Can produce high or low qual, consumers can only
  tell after purchase
• If sell low, consumers all know next period
• How can market reward high-quality?
• Imagine competitive equil at P1.
• Can a firm increase profits by increasing output
  and selling low-quality output?

5
Model

• Costs C C(x,q) F(q) MC FC
• qquality, xquantity of good
• All FC non-sunk
• Consider figure can produce high quality and
  charge P1 no profits
• Or can cheat produce low qual, sell for P1 for
  one period.

6
(No Transcript)
7

• Benefit from cheating
• Hence, at comp price, firms will always cheat
• Maybe theres a higher price that motivates
  honesty

8
A Higher Price, above P1

• That motivates honest production and allows some
  consumer surplus
• Suppose P tilda is the premium over P1
• Gain from not cheating
• Perpetual stream of rents

9
Gain from cheating

• One-time payoff
• Reputational incentives allow market to support
  high quality product if W2W3
• Price premium is protection money paid by
  consumers to induce contract performance

10
But firm charging premium earns profits

• So cant be an equilibrium
• Competition must take non-price form
• Firm-specific investments
• Purchase assets with nonsalvageable costs
  capitalized value of premium rent stream
• Examples sunk investments in firm logo
• Expensive sign promoting firm
• advertising

11
Equilibrium

• High price needed to assure quality would invite
  entry, which would dissipate profit
• Solution non-recoverable firm-specific
  investments
• (lost if renege)
• Similar to collateral
• Tradeoff between high prices and non-recoverable
  investments to assure high quality
• Advertising even non-informative
• Investment in brand, lost if cheat

12
Would you eat at this restaurant?

• (Suppose contracts nonexistent)
• Why eat here?
• enjoy steak? What else?

13
Reputational Incentives for Restaurant Hygiene

• Jin and Leslie (again)
• Quality is difficult to observe, costly to
  produce
• Prior to grade cards, do restaurant owners have
  incentives to maintain quality?
• How does this vary among
• Chains and non-chains
• Franchisees and company-owned

14
Compare chain and non-chain

• Standalone restaurants quality decision only
  affects self
• Chain restaurants choice affects others in chain
• If manager internalizes effects on others in
  chain, then stronger incentive for quality
• Weakened if franchisee?

15
Test

• Predictions prior to cards
• non-chains should be worse than chains
• Franchisees should be worse than company-owned
• Effects stronger for chains with many local
  outlets

16
(No Transcript)
17
(No Transcript)
18
(No Transcript)
19
Motivation

• Internet, price comparison sites reduce search
  costs
• Would competition on the Internet resemble
  perfect competition?
• There is price dispersion, but are people buying
  at the high prices?
• Need quantity data

20

• No direct quantity data
• Need clever alternative sales rank data
• Use with price data to estimate own and cross
  elasticities of demand at Amazon, BN

21
Data

• collected in 3 weeks during 2001
• All books on Publishers Weekly Bestseller list
• Books searched at Dealtime
• Random sample of books in print
• 18k had price rank data at Amazon, 13k at BN
• Ranks updated daily for bigger sellers, monthly
  for low-sellers at Amazon daily at BN
• Weeks of observation
• 1 Stable
• 2 Amazon raises price, BN responds
• Pricing not generally book-specific

22
Ranks and quantities
23
(No Transcript)
24
Estimation
25
(No Transcript)
26
CX results
Given Pareto paremeter, these imply
elasticities of 2.5 to -3
27
Panel evidence
28
Interpretation

• Amazon has small own price elast, large cross
  elast, BN reverse
• Own elast
• Amazon 0.45 (wow!)
• BN 3.5
• Amazon clear market leader

29
Smith and Brynjolfsson

• Does brand still matter?
• Look at visitors to EvenBetter.com
• Price sensitive consumers?
• Are they more likely to click thru to branded
  sellers, conditional on price?
• Yes
• Big 3 have 1.70 price advantage
• Amazon has 1.30 advantage over BN, Borders

30
Shop-bots

• For given SKU, rank sellers by price or shipping
  cost, etc
• Make search easy
• EvenBetter (DealTime)
• Covering 33 book sellers
• Default sorting ascending by price
• August-Nov, 1999

31
Results

• Large price dispersion for identical items
• Lowest is a third below mean
• Price not dispositive
• Majority of consumers dont choose lowest price
• Avg chosen price among those not choosing min
  is 20 percent above min
• Suggesting role of brand

32
Random utility model

• Goal figure out value of various brands
• Nested logit
• Results
• Price coeff -0.2, Amazon 0.48, BN 0.193,
  Borders 0.27
• How much extra are consumers willing to pay for
  Amazon?
• 0.48/0.2 2.40
• Offers from big 3 have advantage over rivals

33
Other results

• First page, first offer big effects
• First offer is lowest price
• Recall Ellison and Ellison

34
Why do consumers care about brand here?

• Some evidence
• Its the same book, but shipping time - a
  non-contractable service feature - could differ
• How does importance of brand differ for those who
  sort results on shipping time?
• Brand big 3 effect -more important
• How does importance of brand differ for
  infrequent users?
• ?

35
SB Punch line

• Brand still matters
• OK, but does brand matter less in the presence of
  more information?

36
Does Information Undermine Brand?Waldfogel and
Chen

• How do you give customers confidence in you?
• Invest in brand
• Bond promise to deliver
• Information from 3rd parties
• information intermediaries- II
• E.g. Consumer Reports
• Old question, new (online retail) context
• Branded retailers
• Amazon (90m on advertising in 994)
• II sites
• BizRate, mySimon, DealTime
• Does II use reduce tendency to patronize branded
  sites

37

• Online context auspicious
• Branded unbranded retail
• II sites
• Longitudinal data on consumer use of both
• Period of II appearance, growth
• Why Important?
• Firm strategy
• Market structure

38
Background

• Does information make markets more competitive
• Permissability of price advertising
• Stigler (1961), Benham (1972),
• Mixed results
• Mandatory information disclosure
• Devine and Marion (179) Jin and Leslie
  (forthcoming)
• Internet as reduction in search cost
• Frictionless commerce?
• Brown and Goolsbee (2002)
• Ellison and Ellison (2000)
• Bertrand Paradox
• Price vs. reliability information
• Klein and Leffler (1981)
• Brand still matters
• Smith Brynjolfson (2001)
• Goolsbee and Chevalier (2002)

39
Information Intermediaries

• Sites comparing prices and other attributes
• bots seeking prices, shipping costs
• different vendors listed
• Types of II Sites
• Price, shipping cost
• MySimon, DealTime
• Vendor reliability surveys
• BizRate (adopted by Consumer Reports)
• Business models
• Pay for click through
• Pay for placement
• Bottom line somewhere between mandatory and
  voluntary info

40
Data

• Web pages visited by 30,000 households Dec 98
  Dec 99
• Domain, sequence, hh characteristics
• Source Media Metrix
• Representative of Internet-connected
• Create a monthly panel data set
• II use (easy)
• Retail page visits (harder)
• Branded, unbranded
• II use is increasing over time
• suggests promise of within hh empirical strategy

41
(No Transcript)
42
Measuring Preference for Brandedness

• Does II use reduce preference for branded sites?
• Need choice set, Branded designation
• Complication
• Apples and oranges
• Consumers like branded orange vendors
• II covers only apples
• Preference for brand vs bundle intent
• Solution empirical choice sets
• Retail sites visited (x) times post II
• Familiar names Offline retailers, Catalog
  retailers, Manufacturers, Other known entities
• Minimal and more expansive measure of branded
  site use

43
(No Transcript)
44
(No Transcript)
45
Results

• Effect of II use on Preference for Branded
  Retailers
• Measurement setup
• Time effects HH FE
• Various measures of B
• Amazon only, all known
• Various measures of II use
• Contemporaneous
• cumulative

46
So II use depresses Amazon share 20
percent, Depresses known share 10 percent
47
Conclusion

• Firms spend a lot on brand yet info can
  undermine this
• Big testable in Internet context
• brand matters less following II use
• Possible implications
• Strategy branded firms might want to obfuscate
  information
• Market structure advertising is an endogenous
  sunk cost
• If information can undermine effect, then
  concentration may decline
• Weaknesses
• Shopping vs buying
• Non-fixed unobservables

48
(No Transcript)
49
History

• Which letters belong on the top row of a
  keyboard?
• Which are frequently used?

50
Scrabble rules

• 0 points for the empty tiles
• 1 point for the following letters E, A, I, O, R,
  N, T, L, S, U
• 2 points for the following letters D, G
• 3 points for the following letters B, C, M, P
• 4 points for the following letters F, H, V, W, Y
• 5 points for the following letter K
• 8 points for the following letters J, X
• 10 points for the following letters Q, Z

51
Standard layount
52
Dvorak keyboard
With the Dvorak keyboard, a typist can type about
400 of the English language's most common words
without ever leaving the home row. The comparable
figure on QWERTY is 100
53
"Sholes Glidden Type Writer,"
54
Sholes up-strike mechanism
55

• Patented in 1867
• non-visibility
• up-stroke Frequent jams
• Keyboard arranged to reduce jams
• Cost 125, slow adoption in economic downturn of
  1870s
• Front-stroke developments
• QWERTY no longer necessary to prevent jams
• Yet, QWERTY becomes standard

56
Why? Role of History

• Development of touch-typing
• Technical interrelatedness
• Need for compatibility between hardware and
  software
• Firm adoption of QWERTY machines increases
  benefit of learning QWERTY system

57

• Economies of scale
• Buyers care about stock of touch typists
• Typists dont care about QWERTY but do care about
  the distribution of machines
• Then tend toward standardization, not necessarily
  on best
• Historical accidents can be important
• QWERTYs initial lead slender, but

58

• Quasi-irreversibility of investment
• High costs of software conversion
• Learning and habituation

59
Fable of the Keys

• Real-world situations present opportunities for
  agents to profit from changing to a superior
  standard.
• The superiority of Dvorak is questionable.
• Ergo