Strategy & Economics of Platforms Flashcards
Definition of Platforms
Platforms are intermediaries that facilitate and manage interactions between users. (Peitz & Belleflamme, 2017)
Digital Platforms - examples
- Airbnb
- Uber
- Ebay
- Amazon
- Netflix
- Microsoft
- BlaBlaCar
- Apple
- Kickstarter
- ClassPass
Definition Network Effects
Each user’s utility from using a good increases with every other user of the same good or with each user of compatible goods
Users of same good or compatible goods are called “the network”
Network effects are typical for platforms but also occur in other settings. (example languages)
Definition Externality
Costs or benefits of an agent’s action to other agent(s) that are not incorporated into the actions “price”
Because cost or benefits of externalities to other agents are not incorporated into price, there is usually “too much” or “too little” of that action.
This “too much” or “too little” from society’s perspective is referred to as market failure in economics: Without intervention, the market will fail to reach an “optimal state”.
Definition Network Externalities
Network effects arise from network externalities:
Definition:
The additional utility that users derive from other users of same good / platform or compatible goods / platforms.
Network externalities within the same group
Within-group network externalities
Network externalities across different groups
Across-group network externalities
Direct network effects
Users’ utility directly increases in the number of other users of the same side of the platform via positive within-group network externalities.
examples: telephony, facebook
Indirect network effects
Users’ utility increases in the number of other users on the other side of the product via positive across-group network externalities.
examples: video-game consoles, smartphone OS
(indirect network effects are most powerful if they work in both directions across groups)
Myopic expectations
ne = nt-1 (t: current period)
- consumers base their decision on past observed network size nt-1
- consumers will fail to anticipate future growth or decline of network
Fulfilled (“rational”) expectations
ne = nt
- Consumers’ expectations are always fulfilled such that excepted network size equals actual network size
- Consumers will always correctly anticipate growth or decline of network
What is the connection between myopic and fulfilled expectations?
Fulfilled expectations can be understood as an end-point / equilibrium of a dynamic process with myopic expectations.
Competing platforms: Brian Arthur Model
Idea:
- Two technologies
- Insignificant random events may matter
- Technology with more adopters improves faster either because of economies of scale or network effects
- Earlier adoption can lead to dominant position / monopoly
Questions:
- Multiple equilibria?
- Efficient outcome?
- Flexible diffusion process or lock-in?
- Small events averaged away or not, i.e. does history matter?
Model of Brian Arthur:
Base case:
- heterogeneous preferences
- increasing returns (r,s > 0)
- A random walk with absorbing barriers is observed -> lock-in
- Long-term market structure: Monopoly for A or B
Model of Brian Arthur:
Modification:
- heterogeneous preferences
- constant returns (r,s = 0)
- R-agents always choose technology A and S-agents always choose technology B
- The process is a random walk without drift (if proportions R- and S- types are the same)
- Market is shared between A and B in the long run (no lock-in)
Model of Brian Arthur:
Modification:
- heterogeneous preferences
- decreasing returns (r,s < 0)
- A random walk with reflecting barriers
- Market is shared between A and B
(hippe dingen zijn niet meer hip als iedereen ze aanneemt)
Model of Brian Arthur:
Modification:
- homogeneous preferences
- increasing returns (r,s > 0)
- Choice order (R or S) does not matter
- First agent chooses initially superior technology A. This increases the returns for adopting A and the other agents also choose A so that B is incapable of getting started.
Result:
- outcome is predictable: A corners the market and B is excluded
- outcome is path-efficient if returns accrue at the same rate, but if they increase at different rates, process may become path-inefficient
- the market may become locked-in to an inferior choice.
Three generic market structures can be observed in platform markets
- Monopoly after intitial competition:
If network effects are strong compared to idiosyncratic preferences (heterogeneity), the market will settle on a single platform - Niche survival:
If network effects are moderate compared to the degree of consumer heterogeneity, small platforms may be able to survive in the market even if there is one dominant platform - Dominant heterogeneity:
if network effects are small compared to consumer heterogeneity, no platform will gain a significant competitive advantage by building up a large network. The inherent consumer preferences will outweigh network effects, and expected market shares will closely resemble the distribution of inherent preferences
Examples for each generic market structure?
Which factors might limit network effects and tipping towards monopoly?
- Consumer heterogeneity
- Local network effects (example language)
- Decreasing strength of network effects
Strategies to induce market tipping
Battle between technologies or platforms are often characterized as “competition-for-the-market”
- “increasing installed-base”: increasing number of previous adopters
- Dynamic pricing / Free products
- Compatibility / Leveraging existing installed base
- Attracting influential adopters
- Vertical integration
- Being first mover
- Increasing strength of Network Effects: Shifting absorbing barrier
Dynamic pricing / free products
- Examples
- problems/risk
- investment in Network-size
- Pays off once platform becomes dominant and enjoys monopoly profits
examples:
- Video consoles are sold initially below cost
- Uber and Lyft offer discounts for first-time riders
Problems / Risk:
- Competitors can also price aggressively
- Dynamic pricing can therefore be very risky
(dot-com bubble)
Compatibility and Leveraging installed base
- Compatibility:
platforms can choose to be compatible with other platforms - Leveraging existing installed base:
Choosing compatibility with other products of same firm increases installed base
Example:
most versions of windows were backwards compatible with earlier versions.
Attracting Influential Adopters
In Brian arthur model, each potential user has equal importance. In reality, some adopters are more important than others.
Examples:
- early adopters
- people with large social networks (“influencers”)
- large companies
- government agencies
Vertical integration
On two-sided platforms, firms can provide complementary products on their own. Especially in early stages, good strategy to solve chicken-and-egg problem.
Examples:
- Netflix and Amazon produce their own movies and series
- Nintendo, Sony and Microsoft produce their own games
- Apple and Google provide own smartphone apps
Strategic Communication
Coordinated application of advertising and preannouncement to inform consumers and/or rival firms about (new) products
- even with fulfilled expectations, there can be low- and high-adoption equilibria
Goal of strategic communication:
- Favorably influence consumers’ expectation about network size and therefore increase (perceived) strength of network effects
- Help consumers to coordinate on high-adoption equilibria
- > Strategic communication (e.g. advertising) is also substitute for (standalone) quality.
Advertising:
- three types (and one additional type in markets with network effects)
- Persuasive Advertising:
Changes consumer preferences, i.e. consumers value the product more after advertising - Complementary Advertising:
Complement the plain good (e.g. by providing social prestige - “Veblen effect”) - Informative Advertising:
Informs about product’s existence and characteristics - Coordinative Advertising
Coordinate consumers on same product to maximize network effects
Investment effect
Lowering initial price attracts more consumers later. Platform “invests” into size.
Harvesting effect
Raising price lets platform “harvest” consumers, taking advantage of size of the platform.
Discount rate
Given a discount rate ‘R’, a firm should value future profits with a discount rate of: 1/(1+R).
Discount rate R can be interpreted as an interest rate or as reflecting general preferences for current profits over future profits.
Original definition of a Platform or “two-sided market”
We define a two-sided market as one in which the volume of transactions between end-users depends on the structure and not only on the overall level of the fees charged by the platform.
Multi-sided pricing
The market side which is valued more highly by the other side should pay less.
Reason: The other side (with the high valuation) is charged more as they also profit more from additional users on the (focal) market side that are lured by low prices.
How should a platform set prices to user groups when one-side can multi-home?
Answer:
- Platform should subsidize side that is more valuable to other side
- When one side multi-homes, access to the single-homing side is more valuable
- Platform should subsidies the single-homing side
Applications of game theory
- games such as chess, poker
- competition between firms (e.g. price-competition)
- nuclear war
- evolutionary biology
- politics
Zero-sum games
Each agents “win” equals some other agents’ “loss”
Elements of a Game (Game theory)
- Players
- Time-structure: What is the order in which the players play the game?
- Strategies
- Payoffs
Nash-Equilibrium
Set of strategies where each player’s choice is optimal given the other players’ choices (“no player has anything to gain from changing only their own strategy”)
One-shot simultaneous move game
Both firms simultaneously choose a strategy, only once
Social dilemma
Because each agent acts selfishly, everyone is in equilibrium worse of as compared to where agents coordinate on overall best outcome.
Socially optimal outcome
Outcome with highest sum of pay-offs
Excess Inertia
Outcome under selfish behavior generates lower overall payoffs than coordination.
- “Inertia” because agents stay with old technology
- “Excess” because it would be better to switch
Excess Momentum
Outcome under selfish behavior generates lower overall payoffs than coordination.
- “Momentum” because agents switch to new technology
- “Excess” because it would be better to stay with old.
System good
Many system goods can be considered as platforms.
System good = Durable good + Complements
Complements
Goods with negative cross-price elasticity of demand. (e.g. if price of good A increases, demand for good B decreases)
Examples of Systems as Platforms
- Smartphones and Apps
- Video Consoles and Games
- VCRs and Movies
- Computer and Software
What makes a good ecosystem?
- variety of complements
- high quality of complements
- exclusivity of complements (not your competitors’ complementary products)
- Steady supply of new innovative complements
Platforms have to balance two goals
- Creating value (maximizing the pie)
- Extracting value (getting a large share of the pie)
Instruments available to platforms
- Contracting
- Technical standards and interfaces
- Rules and procedures
- Information sharing
- Culture and norms
Key trade-off
Openness vs. Control
Openness
two strategies
Allow open access to platform and not regulate interactions
Two strategies:
- Full openness - Anybody can make compatible products
- Alliance - Only developing members can make compatible products
Risk: Getting small piece of large pie
Control
Control access and interactions on platform
Intermediate approaches with some openness:
- Timing
- Retain control over changes
- Charge licensing fees for access
Risk: Getting large piece of small pie
Four types of Compatibility
- Full Compatibility:
Products or platforms are fully compatible if both can use complementary products of other or connect to user-groups on other - Backward Compatibility:
Products or platforms are backward compatible if newer version is compatible with older version but not vice versa - One-way Compatibility:
Products or platforms one-way compatible if one can use complementary products of other or connect to user-groups on other but not vice-versa - Partial Compatibility:
Products or platforms are partially compatible if both can use complementary products of other or connect to user-groups on other but with some decrease in performance or some costs.
Standardization
Explicit and implicit agreements between firms to do things in an uniform way are called “standards”
Main goal of standardization is to achieve compatibility within a standard and therefore maximize network effects
Standards connect and manage user groups such as consumer and complementors, therefore they can be considered platforms.
Pros and Cons of Standardization
Pro:
- Standards lead to faster market growth
- Consumers benefit as search and switching costs are reduced
- The risk of Excess Inertia is lowered
Con:
- Standards reduce the scope for product differentiation
- Price competition can be intense among firms using the same standard
- The selected standard may not be the efficient one. Because of lock-in effects and Excess Inertia, it may become difficult to switch to a better one
Two types of standards
- De Jure standards
- De Facto standards
De Jure standards
- Established by law
- Governments want to avoid uncertainty and confusion among consumers, which may help new technologies to get adopted
- Imply “competing-in-the-market”
De Facto standards
- Standards that achieved dominant position through extensive usage rather than by approval by some organization or government
- Allow technologies to continue developing and market selection of a possibly better technology at a later stage
- Imply “competing-for-the-market”
Three forms of competition
- Battle of the sexes
- Tweedledum and Tweedledee
- Pesky little brother
Battle of the sexes
- A standard battle would be too costly for both firms
- Both prefer a joint standard but each prefers its own technology as basis
Possible “meta-strategies”:
- Commitment:
Firm convinces competitor that under all circumstances it will choose its own technology, then it become optimal for competitor to choose same technology
- Concessions:
Making it more attractive for competitor to choose own technology
Tweedledum and Tweedledee
- Firms prefer to compete to determine the industry standard
- They agree to have a standards battle
Possible meta strategies:
- Here firms would try to influence the outcome of the standards battle in their favor
- Think of strategies we discussed in lecture 1 in context of Brian Arthur model such as increasing installed-base dynamic pricing or attracting influential adopters.
Pesky little brother
- Firm A (little brother) prefers a joint standard, while Firm B (big brother) prefers two standards
Possible meta strategies:
- In this game, who ever moves first loses. Therefore, both firms would try to delay their choice as long as possible. Often, this is harder for larger firm (big brother)
- Larger firm could try to change its technology frequently to make it harder for smaller firm (little brother) to pick same technology
- Larger firm could try to prevent smaller firm to choosing same technology, e.g. by patenting or using intellectual property rights
Market failure
Without intervention market will fail to reach “optimal” or “efficient” state.
Regulation
- Public policies designed to govern economic activity and its consequences at the level of an industry
=> Industry-specific - Applies to markets where competition without regulation does not work
=> Ex-Ante State intervention
Competition Policy
- Intervention into markets to restrict firm-practices that restrict competition and to restrict permissible market structures
=> same for all industries - Applies to markets where competition without regulation does work in principle
=> Ex-Post State intervention
State intervention into platform markets can be rationalized in two ways
- Network effect necessarily lead to monopoly
=> Market-failure ex-ante and therefore need to regulate - Network effects do not lead necessarily to monopoly positions but if they do, platform monopolies are prone to abusing their market power
=> Market-failure Ex-Post and therefore need to apply competition policy
EU article: Dominant Position
Two Conditions
1) Firm (“undertaking”) has a dominant position
2) Firm engages in abuse of its dominant position
(an “undertaking” is an entity engaged in economic activity)
EU article: Dominant Position
Two Types of Abuses
- Exploitative practices (e.g. excessive pricing)
- Exclusionary practices (i.e. excluding competitors)
=> intervention in case of exploitative practices are rare
GDPR includes
- Right of access
- Right to be forgotten
- Right to data protability
Sharing economy
- Offers to temporarily use some durable asset without acquiring ownership
- Often connected with a service
- Peer-to-peer online market places governed by platforms that do not own traded assets
