Show all articles in this section

Let the games begin

AUTHOR: Robert Marks   DATE: 30.08.01   ISSUE 2, 2001

How can I tell my wife that I’m learning to conceal information, to bluff, and even to lie?”asked a student taking my game theory course a few years ago. Actually, he didn’t need to. Game theory teaches you that often it’s better to lay all your cards on the table, to reveal all, especially if in doing so others alter their actions as their expectations of your actions change, because of your disclosure.

Game theory provides a framework for analysing strategic interactions – where what you do affects your ‘competitor’ and vice versa. It is also excellent for analysing situations such as employment contracts, supply contracts or franchising contracts, where there is ‘asymmetric information’ (where the parties to the contract know different things relevant to their performance). Because of this, lawyers are increasingly using game theory to analyse aspects of the law,¹ particularly, but not exclusively, contract theory. A newer area of application is personnel economics,² which uses game theory frameworks to analyse the employment relationship, including compensation and incentives schemes for CEOs. These applications of game theory can be thought of as ‘economising’, of using insights into incentives and behaviour to design new sets of rules (such as contracts) governing the interaction of the principal and the agent.

{

"Game theory provides a framework for analysing strategic interactions - where what you do affects your "competitor" and vice versa."

}

ILLUSTRATION: Bobbi Hitchcock

As its name suggests, however, game theory has for the past 50 years been used to analyse, to understand and to guide interactions between superpowers in the cold war, between contestants in particular games, and between rivals in the market. As the principles of the discipline have become more widely known, executives have increasingly used such tools from game theory as game trees and payoff tables in helping to make their own strategic decisions.³

Over the past few years I have been following a line of research which started when a computer program of mine won the Massachusetts Institute of Technology Competitive Strategy Tournament in the 1980s. When asked what my particular insight had been to allow my program to win, I was not sure how to explain, and continued to think about game-playing routines. A happy coincidence led me to the emerging field of genetic algorithms, a form of machine (or computer) learning which simulates Darwinian evolution. Trial solutions are tested, and the best of these are used to develop a new generation of trial solutions, using techniques analogous to sexual reproduction. Eventually, natural selection (or here, artificial selection) results in solutions that are pretty good.

With former AGSM professor David Midgley (now at INSEAD in Paris) and Lee Cooper (UCLA), we applied game theory to the rivalry in an historical oligopoly, or market with few sellers. We chose the market for cans of ground coffee in a mid-west US city. (Rather like the man looking for his car keys under a streetlight, we selected this market because its supermarket scanner data were readily available.)

We modelled the rivals as stimulus-response players in a repeated game. The stimulus is what all rivals did last week, and the response is what each player should do this week – where each player has some choice over the price, the advertising, the display and coupons, and other discounts associated with its brand of coffee. We found that relatively simple players (with, for instance, shorter memories of past weeks, or smaller sets of actions to choose from) could do as well, even better, in terms of weekly profit, than the historical players.⁴ We are continuing this line of research to examine an additional player, the supermarket, which mediates among the brand managers in order to improve the store’s profits, rather than those of the brands.

We can extend this line of research to other repeated interactions, such as bidding in the repeated auctions for electricity that deregulation in Australia and internationally has established. Indeed, the auctions used to allocate broadcasting spectrum to media players and telcos here and in North America and Europe were designed by game theory economists, who have also advised bidders on strategies for these novel auctions.⁵

A further line of research I am pursuing is the significance, in terms of profits gained or lost, of the adequacy of information about rivals’ prices: this can be thought of as the coarseness or fineness of the perceived partition of market information. At what point, for instance, is the value of increased fineness of the perception of others’ price shifts offset by the increased cost of processing this information?⁶ This harks back to the work, more than 50 years ago, by the late Claude Shannon at the legendary Bell Labs, and is a further application of game theory to managerial issues.

Exploration of game theory and its use in learning to behave more effectively in strategic interactions is part of an AGSM executive program,‘Thinking Strategically’, to be held on 27–28 September.

*Robert Marks is associate professor of economics at the AGSM.

Further reading
1. D.G. Baird, R.H. Gertner and R.C. Picker, Game Theory and the Law, Cambridge, Harvard U.P., 1994.

2. E. Lazear, Personnel Economics, Cambridge, MIT Press, 1995.

3. A. Dixit and B. Nalebuff, Thinking Strategically: the Competitive Edge in Business, Politics, and Everyday Life, New York, Norton 1991.

4. D.F. Midgley, R.E. Marks and L.G. Cooper, ‘Breeding competitive strategies’, Management Science, 43(3), pp. 257—275, March 1997.

5. R.P. McAfee and J. McMillan, ‘Analyzing the airwaves auction’, Journal of Economic Perspectives, 10(1), pp. 159—175, 1996.

6. R.E. Marks, ‘Evolved perception and behaviour in oligopolies’, Journal of Economic Dynamics and Control, 22 (8—9), pp. 1209—1233, July 1998.