|
MR KEYNES BEAUTY CONTESTS: HOW BAD WE ARE AT
GUESSING GAMES
How deep is our thinking when we try and calculate how other people
will behave, for example when we make investments in financial markets?
Not very, according to Professors John Duffy and Rosemarie Nagel
writing in the November 1997 issue of the Economic Journal. Their
experimental research into how individuals form expectations of
other peoples expectations uses a game called the beauty
contest. The results suggest that we tend to use a very low
order degree of reasoning about other peoples behaviour even
when the game is repeated for many rounds.
Duffy and Nagels experiment draws on John Maynard Keynes
insight that the behaviour of investors in financial markets is
like newspaper beauty contests in which readers were asked to choose
the six prettiest faces from 100 photographs. The winner was the
person whose preferences were closest to the average preferences
of all participants. Keynes reasoned that contest participants,
like financial investors, do not choose faces (investments) that
they personally find the most attractive but are instead guided
by their expectations of other peoples expectations.
In the beauty contest game, named in honour of Keynes,
a group of subjects is asked to guess a number from 0 to 100. The
winner is the person(s) whose guess is closest to half of either
the median, the mean, or the maximum number chosen by all players.
The winners are paid $20. The game is repeated several times, and
subjects are informed of the results of each game. Regardless of
the statistic used to determine the winning number, economic theory
predicts that all players should announce zero so that everybody
is a winner.
The objective of the experiment is to elicit the degree of reasoning
that individuals use in forming expectations of others. Suppose,
for example, that half of the mean is used to determine the winning
number. An individual might reason that the mean could be no higher
than 100. Therefore, the winning number could be no greater than
half of 100, or 50. If this individual believed that all the others
would choose 50, it would be rational for her to choose 25, approximately
half of the mean. This analysis can be continued further to the
point where if all individuals used an infinite degree of reasoning,
they would arrive at the equilibrium where all announce zero.
Duffy and Nagel find that players tend to use a very low order
degree of reasoning when forming expectations of others' expectations.
For example, in the first round of the half-median game,
many subjects choose 25 or 12.5 while no one chooses 0. In the second
round, after seeing that the median was say 25, many subjects will
choose numbers close to 12.5 or 6.25. If the game is repeated for
many rounds, the numbers chosen get closer to zero, but most players
continue to use a very low order degree of reasoning.
Duffy and Nagel also find that the statistic that determines the
winning number matters for how quickly players move toward the equilibrium
where all announce zero. Convergence to the equilibrium is slowest
in the half-maximum game and fastest in the half-median
game.
The half-maximum game can be thought of as a financial
market where the actions of one big investor (the person choosing
the highest number) create uncertainty for the other smaller investors.
The half-median game can be thought of as a market where
all investors have equal influence. In forming expectations of others'
expectations, players come closest to coordinating on the market
fundamentals equilibrium where all announce zero in the half-median
game, and are furthest from such coordination in the half-maximum
game.
Note: On the Robustness of Behaviour in Experimental "Beauty
Contest" Games by John Duffy and Rosemarie Nagel is published
in the November 1997 issue of the Economic Journal. Duffy is Professor
of Economics at the University of Pittsburgh, PA 15260, USA; Nagel
is Professor of Economics at Universitat Pompeu Fabra, 132 Balmes,
Barcelona 08008, Spain.
For Further information: contact RES/ESRC Media Consultant for
Economics Romesh Vaitilingam on 0171-878-2919, 0117-983-9770 or
mobile 0468-661095; Rosemarie Nagel on 00-34-3-542-2739 (email:
nagel@upf.es); or John Duffy on 001-412-648-1733 (email: jduffy+@pitt.edu).
|