Strategy for Information Markets/Network Externalities/Two Bandwagons

Markets with network externalities have positive feedback. On the positive side, this means that as more people join a network, everyone will find that network more attractive, and so yet more people will want to join. This is often called a Bandwagon effect.

Similar-appearing effects can arise from (at least) two different sources. These two experiments are intended to highlight that a market which appears to have network effects might be demonstrating an entirely different phenomenon.

Given some care, both experiments can be conducted with ordinary playing cards. If more than one deck of cards is required for the network externalities experiment, the class is probably too large to conduct these experiments within a reasonable time.

Herding game

The instructor will be walking around the class from student to student, so one student will be needed as an assistant to keep track of the students' actions on the board. If money or extra credit is being awarded for game play, the assistant should receive a full share.

• The instructor should create two "sets" of cards. One set contains two red cards and one black card, and is called the Red set. The other set contains two black cards and one red card, and is called the Black set.
• If working with cards all from the same deck, then the students should not be shown the cards as public knowledge, since seeing a specific card (say, the three of clubs) could give away which set is which.
• I recommend the instructor have their assistant randomly select one of the two sets, so that nobody (students, instructor, or assistant) knows which set is in use. This "double-blind" control will help prevent the instructor from giving away the correct answer.
• The assistant stays at the board to keep track of students' actions. The instructor goes from student to student.
  • The instructor offers the three cards face-down as a fan to the student, who draws one of the cards. The student should look at the card, being careful that they are the only one who can see the card. The instructor should also not see the card.
  • The student should not say anything to reveal the card they see. Instead, the student then publicly announces a guess about whether they are looking at the Red set or the Black set. The assistant records their guess on the board.
  • The instructor moves on to the next student.
• After all the students have made a guess, the instructor publicly reveals the three cards.

Network effects game

For this game, an assistant is not strictly needed, since the instructor will not need to be moving around the room. However, it is important to have an even number of students in the experiment. The assistant may be retained to help if doing so will mean an even number of students remains, or may return to take part in this second experiment.

For this game, if there are 2 n students, the instructor should shuffle n red cards and n black cards into a deck.

• The instructor hands out the cards, one to each student. The students must not reveal their cards to anyone else.
• The instructor calls on the students one at a time.
  • The student decides to choose (not guess, but choose) either Red or Black.
  • The student's payoff is (2 if they choose the same color as is on their card) plus (1 per other student who chooses the same color). The student should not reveal their card or find out their payoff until the end of the experiment.
  • The instructor or assistant records the student's choice on the board.
  • The instructor calls on the next student.
• After all the students have made their choices, count the number of Red and Black choices. If there are 14 students who chose Black, then each student who chose Black should receive 13 extra points.
• Ask each student to reveal the card they hold to decide if they receive 2 points from individual taste.

Network Effects Pertaining to Both Experiments

In reference to the "Herding Game", there are no real network effects working with the decision made by the student to whether they will choose a black set or red set. Instead, it is purely random in the sense that the majority of people will choose the first color set that they choose in the hopes that there is a similar color in that set of three cards. The results of each persons decision results in inefficiency because of "bad" guesses, which clarifies the earlier remark of there being no network externalities at work here. However, in the second experiment, we are able to see different results in the fact that there are strong positive network effects at work. It becomes the case that as students begin choosing which color card they might have, that it doesn't really matter as long as they choose the color which everyone else is choosing in order to receive the maximum amount of points. The experiment creates a bandwagon effect in which the player cannot really lose as long as they go along with the crowd. Players tend to avoid becoming an outcast and feel comfortable making the same decision as everyone else. Through this we are able to see the strong effects that occur with "positive" network effects in relation to a product or market.

Discussion

• Information cascade
• Network effects with individual lock-in