A Lecture on Experimental Game Theory
This lecture was written for one of the final lectures of the third year undergraduate course 'The Economics of Competitive Strategy'. In this course students are taught how to apply game theoretic techniques to aid understanding of the strategic decisions that firms face. In recent years industrial economists have set up experiments to test whether the assumptions of game theory and the predictions of games can be assumed to be realised in real markets. This lecture seeks to illustrate some of the experiments performed by making the students re-enact experiments described in the literature.
The students are told in the preceding lecture that they will be asked to play experiments in the following lecture. Hence, on entering the lecture room they are not surprised when asked to sit at a table facing another student, and instructed not to speak throughout the course of the experiments.
Each student is given an instruction card that explains how to play a game and the choices to be made. Choices are written on a game results card and given to me, acting as experiment co-ordinator. Three games are played in close succession. Players alternate between being player one and player two. In each game played, students can win pennies.
- Game one: the 'Dictator Game'. This game tests whether economic agents maximise payoffs or act altruistically (i.e. in a manner not predicted by economic theory).
- Game two: the 'Ultimatum Game'. This tests whether economic agents seek revenge on uncharitable opponents, even if this implies forgoing some profit.
- Game three: a bargaining game. It tests whether students act in accordance with the predictions of a relatively complicated, extensive form, game. If they find it difficult to identify the equilibrium solution of the game, then concerns are raised about whether real economic agents can be expected to act in ways predicted by economic theory.
After all three games are played I explain
- the reason for playing each game;
- the outcome of the games as predicted by economic theory.
I discuss the results that emerge from the experiments played in the lecture, and compare these results with those that have been reported in the literature.
The first year I gave this lecture I discovered that the students did not want to leave the lecture with a hand full of small change. Consequently, in subsequent years the number of pennies won by each student has been tallied on a scoreboard and converted into chocolates using an appropriate exchange rate.
In the following seminar, discussion initially centres on whether the results of experiments can be affected by the conditions under which games are played. For example, are the results affected by opponents being able to see each other when making strategic choices. Clearly, at an increasing number of universities it would now be possible to ask the students to play games in computer rooms developed primarily for the collection of experimental data. Nevertheless, by playing games in the usual lecture context, discussion about the experiment design factors that may bias results is encouraged.
Every year the students have responded very positively to being taught about experimental game theory and experiment design issues through actively participating in experiments. Student feedback has been very positive, and their interest has been sufficiently stimulated that they are keen to research the subject more thoroughly after the lecture and seminar.