Pricing and Bundling Electronic Access to Information Jeffrey

Publishing 1996 US GDP 1997 Sales l Reed Elsevier ($5. 75 B) l Gannett

Selected Publishing M&A Activity Harcourt General acquires Times Mirror’s Mosby Inc. ($415 MM) (9 Oct 98) 2. Bertlesmann AG to buy 50% ($200 MM) in Barnes and Noble online (7 Oct 98) 3. Penton Media to buy Mecklermedia ($200 MM) (8 Oct 98) 4. Wolters-Kluwer acquires Ovid Technologies (30 Sept 98) 5. Microsoft in discussions with Reed Elsevier? (24 Sept 98, Assoc Press) 6. Reed Elsevier acquires Matthew Bender and Shephard’s ($1. 65 B) (27 Apr 98) 7. Individual Inc. and Desktop Data and ADP/ISS (24 Feb 98) 8. Washington Post and Newsbytes News Network (18 Dec 97) 9. Reed Elsevier and Kluwer propose merger (13 Oct 97; abandoned) 10. Reed Elsevier acquires Chilton Business Group ($447 MM) (23 June 97) 11. Reed Elsevier acquires MDL Information Systems ($320 MM) (24 Mar 97) © 1998, Jeffrey K. Mac. Kie-Mason 1. 3

Problems and Opportuntities l Problems: – – l High first copy costs Threat from “bypass publishing” Incremental costs from new delivery media Development risk for new value-added services Opportunities with electronic pricing: – – Revenue from new service provision Extracting more value from heterogeneous users © 1998, Jeffrey K. Mac. Kie-Mason 4

What is known about bundling? v 2 Bob Alice v 1 © 1998, Jeffrey

What is known? v 2 p 2 Bob Alice p 1 © 1998, Jeffrey K. Mac. Kie-Mason v 1 Unbundled: Sell to those with few extreme values, not those with several low values Alice buys good 1 (v 1 > p 1) Bob buys nothing 6

What is known? v 2 p. B Bundle: Average across values. Bob Pick up

What is known? v 2 p. B Bob Pure bundling Alice p. B v

What is known? v 2 Mixed bundling Buy #2 p 2 Both buy. Buy both Bob Alice p 1 Buy #1 v 1 Offering choice of two price schemes rather than one increases sales But not a general solution! © 1998, Jeffrey K. Mac. Kie-Mason 9

Early literature l Adams and Yellen (QJE 1976); Mc. Afee, Mc. Millan and Whinston (QJE 1989); Salinger (J Bus 1995) l Limited to 2 goods – l drastically limits expression of customer heterogeneity and innovative bundling Didn’t examine behavior as marginal cost Ô 0 – defining feature of information goods © 1998, Jeffrey K. Mac. Kie-Mason 10

Recent efforts l Chuang & Sirbu (1997), Bakos & Brynjolfsson (1997) l Explore N goods, but still very limited bundling possibilities – – l Fully bundled tends to be profit maximizing when – – – l unbundled, fully bundled and self-selection between the two don’t consider partial bundles, or user-chooses bundles consumers have similar “average intensity” economies of scale in distribution of info goods not always beneficial to readers Outstanding issues: – – – heterogeneity dimensionality of product space publisher competition © 1998, Jeffrey K. Mac. Kie-Mason 11

Info goods are infinitely configurable l Word processing – – l Java component “home” config “pro” config suite bundle – – – l © 1998, Jeffrey K. Mac. Kie-Mason Stock data by the. . . quote day exchange With or without access to analyst reports, technical charts, &c. 12

Rebundling is opportunity and curse l Possibilities for 3 items: {A}, {B}, {C}, {A, B}, {A, C}, {B, C}, {A, B, C} l Complexity grows rapidly: l Need principled approach to exploring the design space © 1998, Jeffrey K. Mac. Kie-Mason 13

Our agenda l New mode of bundling for heterogeneity: generalized subscriptions (w/Riveros) l Competition when firms bundle (w/Fay) l Two-sided learning in differentiated product (bundle) space (w/Kephart et al. ) – – l consumers learn about changing price/bundle offerings providers learn about consumer tastes and competitor strategies Field research: PEAK © 1998, Jeffrey K. Mac. Kie-Mason 14

Simple bundling l If users have similar average values for info goods, offer large bundle: “seller chooses” l If users have different average values, let them select individual components: “buyer chooses” © 1998, Jeffrey K. Mac. Kie-Mason 15

What we know Users similar Users different Costs Seller chooses: Buyer chooses: low bundle

What we need Seller choose Buyer choose Real world (e. g. , buyer chooses

Consumer preferences l How do consumers value articles from a collection? – For one consumer, each article can have different value Value of best article = w 0 N articles; k is fraction user values > 0 k. N © 1998, Jeffrey K. Mac. Kie-Mason article n 18

Consumer heterogeneity l Between two consumers, ranking and values can be different Let w 0 and k vary across users – – w 0: most valued article k : fraction with value > 0 k. N © 1998, Jeffrey K. Mac. Kie-Mason article n 19

Bundle options l Fully bundled: N=100 articles at p. B l Unbundled: each article at pu l Generalized subscription (“user chooses sub-bundle”): NG=10 articles at p. G © 1998, Jeffrey K. Mac. Kie-Mason 20

Bakos & Brynjolfsson l B&B assume (main results): – – – l p all individuals draw article values from same distribution as each other all articles drawn from same distribution (LLN holds: sample average converges to distribution mean) so bundling is highly favored Aggregate demand for bundle of size N: N=1 articles sold © 1998, Jeffrey K. Mac. Kie-Mason p N=100 articles sold 21

Comparison to B&B l l Homogeneous consumers Two consumer types (different max value: w.

More heterogeneity: Chuang & Sirbu l Heterogeneity: – – l different intensity: number of articles with value > 0 random: article values ~ U[0, 1] for first ni ~ exp(13. 9) articles heterogeneity not averaged out as N increases Profits increase by 10% when customers offered selfselection from menu of <GS, unbundled, bundled> © 1998, Jeffrey K. Mac. Kie-Mason 23

Next challenge: Competition l Prior lit has examined bundling by monopoly sellers; what happens with some competition? – l Exceptions: Fishburn, Odlyzko, and Siders (Tech Rpt 1998); Matutes and Regibeau (JIE 1992) Some questions: – – – How much does competition reduce extraction of surplus? Can publishers in competition use bundling to recover fixed costs (is equilibrium sustainable)? What effect on incentives to create new content © 1998, Jeffrey K. Mac. Kie-Mason 24

Choice with competing bundles l Assume articles randomly divided between two publishers l Consumers: – – – all articles are ex ante identical in value based on expectations, risk-neutral consumers decide among offerings: bundle, per item, or nothing -- want at most one of any article after purchase, values are revealed and consumers decide which items to read P 2 Both firms offer bundles only: V 2 Buy collection 1 only VB-V 1 © 1998, Jeffrey K. Mac. Kie-Mason Buy both VB-V 2 Buy neither Buy collection 2 only V 1 P 1 25

Competing articles l Suppose firms only offer to sell by the article l (One firm bundling, one offering articles lies in-between) P 2 Both firms offer articles only: w 0 Buy articles from collection 1 only Buy some articles from both Buy neither Buy articles from collection 2 only w 0 © 1998, Jeffrey K. Mac. Kie-Mason P 1 26

Results: Competition with homogeneous consumers l If both firms bundle: – – efficient: all articles purchased (same as monopoly) competition leaves much more consumer’s surplus l – only 65% of monopoly profits even when competitor has only 20% share in a two-stage game with a fixed cost Fi to create Ni articles, incentives to invest efficiently are preserved in the duopoly © 1998, Jeffrey K. Mac. Kie-Mason 27

Competition with homogeneous consumers (cont. ) l If neither firm bundles: – – Inefficient outcome: P > MC Lower: l l profits (13%) consumer’s surplus (9%) welfare (12%) But if one bundles, one does not: – no pure strategy Bertrand (price competition) equilibrium © 1998, Jeffrey K. Mac. Kie-Mason 28

Heterogeneous customers: Monopoly l When customers were homogeneous (B&B), monopolist always prefers bundling l When heterogeneous (different values of k), offering selfselection mixture of bundles and articles dominates (divide and conquer) l Simple sub-bundling is inferior – – since articles are substitutes, creating sub-bundles creates artificial competition this ignores possibility of clustering by customer types to sort (different journals) © 1998, Jeffrey K. Mac. Kie-Mason 29

Heterogeneous customers: Competition l If both firms bundle: pure strategy Bertrand does not exist l If one or both sell unbundled, pure strategy equilibrium does exist (similar results with Stackelberg) – l both have (much) higher social welfare than monopoly But bundling is strategically advantageous: – – – firm that bundles earns more per item when both mix bundles and articles, nearly all revenue from the bundle larger firms can extract more surplus per item © 1998, Jeffrey K. Mac. Kie-Mason 30

Competition Extensions l Endogenize the bundling strategy – – – l any outcome can result bundling is not dominant when mixtures are offered, most of the revenue comes from the bundle sales Endogenize the collection size – – monopoly bundling is inefficient, but captures more of the total surplus former effect dominates: the duopoly’s greater allocative efficiency makes duopoly investment incentives closer to efficient level than monopoly © 1998, Jeffrey K. Mac. Kie-Mason 31

PEAK Project l Network access to 3. 5 years of all 1100 Elsevier journals l Large scale field trial – – 12 university, research lab and technical college libraries over 100, 000 authorized users full text searching high-resolution screen or print viewing l Over $325, 000 in up-front payments l Experimental variation in choices available: – – – unbundled user chooses sub-bundles © 1998, Jeffrey K. Mac. Kie-Mason 32

PEAK Bundles l Traditional subs: $6 / issue l Unbundled articles: $7 l Buyer

Clients subscribed to fraction of all print titles l © 1998, Jeffrey K. Mac. Kie-Mason But every authorized individual now has immediate desktop access to every page of every journal (at varying prices) 34

Summary l Low transactions costs enables plethora of disaggregated and reaggregated products l Workability needs more limited design space l We are extending the 2 -extremes theory of bundling l We are studying competitive strategy when firms bundle l In PEAK we introduce user-chooses sub-bundles and test the ideas in the field © 1998, Jeffrey K. Mac. Kie-Mason 39