2 - Features of Platforms and Markets with Platforms

Question 1

Demand for network good, the utility function. What components are there?

Answer 1

The utility depends on the stand-alone benefit from using the good/technology AND the network benefit, which is a function of the expected number of users in the platform.

Question 2

How can the component ‘network benefit’ be modelled?

Answer 2

Either linear, where each new member contributes linearly to all users. Or concave where each additional user contributes lesser to the utility of the others in the network. For ex my first 50 friends on Facebook increases my utility more then the rest since the first ones are typically my closes friends.

Question 3

Why is it so hard to compute the demand for a network good?

Answer 3

Because if it is simultaneous adoption, we have coordination problems that makes it hard to compute. And if we have sequential adoption, there is typically a path-dependence that creates lock-in effects - either you lose or you will be the winner-takes-all.

Question 4

Going from the model with no network effects to adding the network effects, what changes?

Answer 4

Suddenly the decision for each individual whether to join or not is interdependent of the other’s decision. We add the term B, which is the network benefit which will only be there if the other join.

Question 5

What is Nash Equilibrium?

Answer 5

It is the outcome of the game when each player takes an action that leads to the highest payoff given the action taken by the other players (best response). = all players choose mutual best responses. In eq. no player has incentive to deviate.

Question 6

What happens in the game if the price is between the valuation of A and B?

Answer 6

It depends on the strength of the network effects. if there are weak NE, then only A will join, not B. With strong NE either both join or none join.

Question 7

What is important when we have strong network effects?

Answer 7

Expectations!! it is alla bout expectations about the platform and who will join and not. If both have optimistic expectations, both will join. It becomes a self-fulfilling prophecy.

Question 8

Two direct consequences of strong network effects:

Answer 8

1. unpredictability - impossible to predict if people will join or not as it depends on the individuals’ expectations.
2. potential inefficiency - as everyone are better off if all join, yet it could be the case that no one joins.

Question 9

What changes in the general model with continuum of consumers?

Answer 9

Now the size of the network can be infinite (not just player A and B in the model), so my decision to join the network will NOT affect the networks size, the only thing that matters is my expectation of the network size n^e (with coefficient B - linear increasing function that increases my Utility).

Question 10

Heterogenous consumers, what can be heterogenous?

Answer 10

Either how I value the stand-alone benefit of using the platform or my valuation of the network benefits.

Question 11

With heterogenous stand-alone valuations AND strong network effects, how does the demand for the network look like?

Answer 11

When B>V, we have an increasing function of demand (n) at some price leves since the strong effects outweigh law of demand. This result in a three fulfilled-expectations-equilibria when the price is in between B and V:
no consumer join (n=0), all consumers join (n=1), or only a share joins.

Question 12

With heterogenous stand-alone valuations AND weak network effects, how does the demand for the network look like?

Answer 12

It is a decreasing function of p and it only has a unique demand for each level of p. The traditional law of demand outweigh the weak network effect.

Question 13

What is the critical mass?

Answer 13

It is the mass of joiners that is needed to be passed in order to reach the large stable network. Because the critical mass is an unstable equilibrium in the increasing function of strong network effects. if the WTP<p> the network size will shrink and if the WTP>p, the n will increase like a snowball effect. </p>

Question 14

What was the crucial thing for Million Dollar Homepage to grow?

Answer 14

That some few in the beginning actually expected that people will buy ads on the site and that individuals will click into the website –> since they expected high n, they bought ads and then the snowball effect kicked in. It became a self-fulfilling prophecy. If no one believed in it at the beginning, it could easily fail too. That is why it is so hard to predict.

Question 15

Wha problem arise for consumers to choose when we have several goods that are incompatible?

Answer 15

Coordination problems. Because network effects are specific to each good and all would be better of choosing the same, but it is hard to coordinate when you have different valuations.

Question 16

Excess inertia & excess momentum

Answer 16

Inertia - users don’w switch network wen they should.
momentum - users switch when they shouldn’t.

This can happen when we have a new tech/network that is superior to the old on.

Question 17

What can we expect from the difference in the installed base (nr of users) in network A and B if there are no network effects?

Answer 17

We expect that the difference in the installed base approaches zero when more and more consumers make a decision of which network to join. “Random walk”.

Question 18

Under what two conditions are excess inertia more likely to happen?

Answer 18

• When we have incomplete information about the other user’s utility.
• When markets have indirect rather than direct network effects.
• -> chicken-egg problem that is hard to coordinate, no one wants to switch to new good bc fear of not being followed.

Question 19

Winners-take-most markets, where there is no tipping towards only one single platform. Three things contributing to this:

Answer 19

1. Differentiation - because consumer’s stand alone preferences matter.
2. Interoperability - when networks are compatible with other.
3. Multihoming - users can be active on multiple platforms at a time.

Question 20

Horizontal vs vertical differentiation -

Answer 20

Horizontal - services appeal to a particular segment of users.
Vertical - quality of services is perceived as higher by all users.

Question 21

What is the hotelling model of product differentiation?

Answer 21

It is modelled as a line with unity mass from 0 to 1. The platforms are located at each extreme/end of the line and the consumers are located somewhere on the line dependent on their unique perception of the stand-alone value.

Question 22

What is the benefits and costs to each consumer in the hotelling model?

Answer 22

The benefit is the stand alone value + the q if they choose the platform with the higher quality (vertical diff). The costs are the price (p) and the transportation cost (the distance you have to move on the line (the cost of not being at your preferred location - the ideal platform).

Question 23

What are the two main driving forces for demand in the hotelling model?

Answer 23

q and the difference in ne (expected network size).
Because high q attracts more consumers and if we have strong network effects, the expected size of the networks will also be an important determinant of the individuals’ choice.

Question 24

What two things will determine the equilibrium market structure in the hotelling model?

Answer 24

1. Network effects - drive users to join the same platform (higher network benefits).
2. Horizontal differentiation - leads consumers to join the platform that is closer to their taste (stand alone benefit).

• -> When B is strong relative to horiz.diff, more likely to have one dominating plat.
• -> When horiz.diff is relatively strong, more likely to have coexisting platforms in eq.

Question 25

When we have an equilibrium where a share of the consumers join platform 1 vs 2, the relation between thao (transportation cost) and B (network effects) become important, how?

Answer 25

Because the difference between t-B is a determinant of hoe n (network size) varies with prices.

• If t>B the law of demand applies because network effects have a weak force relative to the horiz.diff (t).
• If B>t, network effects dominate and higher fees (p) are related to higher market share.

Question 26

Defensability

Answer 26

Means that the largest platforms get some kind of protection from competitors as they have strong network effects that lock in consumers, so that they do not switch to new platform, BUT defensability is not only about strong network effects, there are other ways to defend as well

Question 27

What 4 things can make defensability less protective?

Answer 27

1 - strength of networks effects (to have network effects is not everything, the strength of it matters too).
2 - multihoming - when users can be active on more than one platform
3 - coordination of usage decisions
4 - trust and disintermediation - can users trust each other on the platform?

Question 28

How is the strength of the network effects related to differentiation? In terms of defensability.

Answer 28

Network effects are stronger when one group is differentiated in the perspective of the other group. For ex marketplaces - when suppliers are diff –> a new marginal supplier adds to the variety and the likelihood of a buyer to find something is higher –> hard for newcomers to replicate –> the strong NE offers defensability.

Question 29

Global vs local network effects

Answer 29

Local = when users only care about the size of the local network - ex Uber.
Global = when you also care about the network at a global scale - ex Airbnb.
–> Global offers more defensability because it’s harder for new competitors to set up a global platform.

Question 30

Multi-homing:

Answer 30

Ability for users to be active on more than one platform. A form of interoperability.
Ex. both riders and drivers of Uber can be riders and drivers on Lyft as well.
Also app-developers can create apps for multiple systems etc.

Question 31

How is coordination/switching costs related to multi-homing?

Answer 31

Because if it is possible to be active on multiple platforms, you want to create high switching costs or coordination costs so that your users stay and use your platform as much as possible. Ex Spotify has a subscription fee - users don’t want to pay more than one fee. AND they have learned you, your preferences and music style so if users switch to Apple, it will take time for Apple to learn the user.

Question 32

How does the competition change when we have compatible networks?

Answer 32

Instead of competing FOR the market, they can now compete IN the market, as the market’s tendency to tip for one dominant is no longer there.

Question 33

Example of partial compatibility

Answer 33

Ex Skype has partial compatibility as it offers free calls to all other Skype-users but charge a fee for calling users of other networks/those not in the Skype-network.

Question 34

What is the utility of a user in the compatible framework?

Answer 34

It is the stand alone valuation + g, which is the number of new users - price.

Question 35

What is the reaction function in the Katz-Shapiro model?

Answer 35

It is the derivative of the profit function with respect to q. So, each firm maximizes profits by choosing a an optimal capacity choice given the choice of q by the other network.

Question 36

Will the total market demand/number of users increase or decrease when compatibility increase?

Answer 36

This is calculated by taking the derivative of the total demand (qA+qB) with respect to gamma (compatibility) and this will show that the market expands. Because the derivative is positive.

Question 37

What types of firms benefit from a better perceived relative quality?

Answer 37

Those firms with lowest cost of production and those with larger installed base. because both these two increases the difference of the equilibrium difference in qA-qB.

Question 38

Quality differentiation effect:

Answer 38

This is an effect of compatibility. When compatibility increases for firms that are relatively more efficient (lower cost of production) the derivative of the difference qA-qB with respect to gamma becomes negative, which means that it is less attractive for efficient firms to become compatible. The same applies for firms that have a larger installed base.

Question 39

What are the two main effects of compatibility?

Answer 39

1. The total market demand increases, so that the total pie increases.
2. The quality differentiation reduces - so that your share of the market reduces.

Question 40

4 combinations of compatibility strategies that determine the form of competition in the standardization game:

Answer 40

1. Straightforward standardization - both choose A or B (compatibility)
2. Battle of the Sexes - one firm prefers A, the other B. They agree that standardization is best but disagree on which standard.
3. Pesky Little Brother - one firm prefers compatibility but not the other.
4. Standards War - both decides to compete with their preferred standard. No one wants compatibility.

Question 41

What must be true for the battle of sexes to ‘happen’?

Answer 41

They must rank equilibria differently, i.e., firm 1 prefers AA over BB but firm 2 prefers BB over AA. Because if they both agree on either A or B, it would be straightforward standardization.

Question 42

What is the Nash equilibrium in the pesky little brother case?

Answer 42

There is no equilibrium, because if firm 1 is the large firm and it produces A, then firm 2 also wants to produce A (compatibility), BUT this makes firm 1 unwilling to produce A and instead they want B and then the other firm wants this as well. So it is just a chasing-game where they never end up somewhere.

Question 43

Example of standards war case, who won?

Answer 43

Blu-Ray and HD-DVD both choose to compete with their own standards and not being compatible with each other. BR won because they managed to attract the best suppliers/developers which in turn attracted the consumers, even though DVD had the largest installed base of consumers from the beginning.

Question 44

How do we solve for each outcome in the standardization game?

Answer 44

We use the Katz-shapiro model and take the inverse demand (p given q) and determine the indifferent consumer to find the expected number of new consumers. So each firm maximizes profits by choosing a capacity and we take the derivative and set to 0. This gives us each firm’s reaction function - solve this system of equations to get the equilibrium q and p (and hence profits).

Question 45

What two scenarios can we consider regarding the installed base?

Answer 45

1. Asymmetry - that one firm has a larger installed base than the other. In the model we will assume that one firm has an installed base and will benefit from psy, whereas the other has 0.
2. Switching cost –> standardization: firms can benefit from a common installed base if they choose standardization. If not, they can only benefit from their prior existing users. (trade-off between compatibility and costs because it will cost you “c” to produce your least prefer good).

Question 46

What parameters does the outcome in asymmetric installed base depend on?

Answer 46

It depends on the key parameters, c (cost of not producing preferred good)and psy (installed base).

Question 47

asymmetry: Low c and small psy, what outcome?

Answer 47

Battle of sexes - both firms prefer to adopt a common standard but preferences diverge.

Question 48

asymmetry: Low c and high psy, what outcome?

Answer 48

Pesky little brother (no equilibrium) - small firms prefer compatibility more than ever due to the smallc, but larger firm preferes incompatbility due to the large difference in installed base.

Question 49

asymmetry: Middle option, when there can be a bit different sizes of c and psy, what outcome?

Answer 49

Straightforward standardization on A.

• small firm: ready to incur c to take advantage of psy.
• large firm: demand expansion effect dominates quality differentiation effect, so they are okey with compatibility with the smaller firm.

Question 50

asymmetry: c is high, but installed base can be both high and low, what outcome?

Answer 50

Standards war.
Both firms prefer to compete to establish their preferred tech as the standard. They have too divergent opinions so the high cost makes it a deal-breaker.

Question 51

What is the problem when we have collective switching costs?

Answer 51

The problem is that they are collective, i.e., all the users prefer to wait and see which platform that “takes off” in the market and therefore it may never take off, unless the producers agree on standardizing.

Question 52

When will standardization (AA or BB) be in equilibrium in the collective switching costs scenario?

Answer 52

Standardization is the nash outcome if taking advantage of the installed base compensates for the cost of standardization. otherwise, AB is the eq, because producing least preferred goods BA can never be eq.

Question 53

3 strategies companies can use in the standards war environment: recall that in a standards war, it is important to have a large installed base due to the network effects that brings. So, how can firms achieve this?

Answer 53

1. Intrinsic quality - improve the stand-alone value.
2. Backward compatibility - making new bersions of a product backward compatible with previous versions (it can hurt performance though!)
3. Expectations management - manage people’s expect through marketing. “ we have sold X amount, we have the largest installed base!”

Question 54

What is the trade-off regarding the strategy of increasing the intrinsic quality?

Answer 54

Intrinsic value evolve over time when spillovers and feedback come into play. So you might want to be the first mover to get this fast.. BUT the trade-off is that this might come at the expense of product quality, because if you wait and see what others do, you can develop the best product and launch later. Like AirBnb did.

Question 55

Trade-off with backward compatibility:

Answer 55

The benefit for the users is that they can enjoy network benefits over time, BUT it can often constrain the capacity to improve the product - some features must be fixed over time. However, not being backward compatible, means that old users are forced to buy the new version, and you never like to be forced to do something.

Question 56

Problem with handling expectations management through preannouncements:

Answer 56

That you hurt sales in the older version/existing product if you announce that a new product will be launched soon. AND it must be credible - you cannot delay it too much.