Economic Classroom Experiments/Currency Attack

Investors attack a currency. Whether the attack is successful or not depends on the number of attackers. There are two pure strategy Nash equilibria in this macro coordination game, everyone attacks or nobody attacks. Two versions of the game are run simultaneously. Almost invariably, students coordinate on the "attack" equilibrium in one, but not the other version.

Level
Any level

Prerequisite knowledge
None

Suitable modules
Macroeconomics, any module discussing monetary policy

Intended learning outcomes
Participants in the experiment will


 * 1) Illustrate the relevance of coordination for macroeconomics
 * 2) Illustrate multiplicity of equilibria.
 * 3) Illustrate what determines equilibrium selection.

Description
Two versions of a coordination problem for macroeconomics are run simultaneously in five rounds. Each student in a group of seven must decide simultaneously whether to attack a currency or not. The payoff from attacking is increasing in the number of attackers. It pays to attack only if there are sufficiently many attackers. The game has two pure strategy Nash equilibria, everyone attacks or nobody attacks. The "white" version of the experiment has a lower payoff than the "red" version from non-attacking. Almost invariably, students coordinate on the "attack" equilibrium in the "white" version and on the "no attack" equilibrium in the red version.

Running the experiment
We describe the procedure with which we ran the experiment. This procedure is hence tested. Obviously, each instructor is free to make any adjustments she or he considers suitable.

Timing
The experiment takes up to 30 minutes overall. Additional time may be allocated to more class discussion, possibly in a different lecture. This is particularly important if students are given the data to evaluate.

Preparation
Students should be given instruction sheets (web link) at least a day before the actual experiment. Preparation of [Decision Sheets.] The students should be divided into two roughly equally sized groups. Each group should get differently colored decision sheets (we used white and red). If help is available, the decision sheets can be cut into the five decision strips and stapled. Otherwise the students have to tear off the decision strips during the experiments. To be planned in advance:
 * How are you going to split the students into two groups? Male/ female left/ right half of the room.
 * Does the room design provide a natural way to do the split.
 * Use two students, one for each version (white/red) to collect the decision strips.
 * Will you give the students the data to evaluate the experiment? How?
 * If you want to give money / prizes decide how. (In one version we had 14 candies, in another £ 14 available which we would use to pay one randomly selected student according to his gains in one randomly selected round.)
 * The design below, where each student in, say, the white group plays against the decisions of six randomly selected members of the white group is made for large classes. In smaller classes one may count how all students in the group decided and let each student play against that statistic. (Notice that some students play against themselves in our design, but this does not affect the Nash equilibria). If you have fewer than 14 students in your class, further adjustments of the design are needed, which we leave to the instructor

Procedure
Choose two students to help you in running the experiment, one for each color. Give them the differently colored decision sheets doc and explain them how to distribute them. While the students distribute the sheets, draw the [table] for the results at the board. Summarize the main points from the [instruction sheets.] Give students three minutes (later 1 minute) to make their choice and tear of the decision strips. Decision strips must contain round, name and decision (A or N). Advice students to write down their own decisions for themselves on a separate piece of paper. Let the two students select the decision strips for each group separately. Suppose the decisions of the white group are selected first. The select randomly six decision strips from the white group. Write on the board how many of these students chose N and how many A. Given these six choices state the payoff of any student that chose N and that of any student who chose A. Do the same for the red group. Repeat the above for rounds 2, 3, 4 and 5. Hand the data collected to small groups of students for writing summary evaluations. (see homework sheet on web)

Results
Students were either Red or White. N is Not attack, A is Attack.

Payoffs for not attacking were 			5 for white	11 for Red. Payoffs for attacking depended upon number attacking.

Results from Macroeconomics at U. of Exeter Dec. 2005

Computerized Version
There is a computerized version of this experiment available on the Exeter games site.

You may find the sample instructions helpful.