Chivalry And Solidarity In Ultimatum Games

Economic Inquiry, April, 2001 by Catherine C. Eckel, Philip J. Grossman

PHILIP J. GROSSMAN [*]

We report the results of ultimatum game experiments designed to test for differences in the behavior of women and men. Women's proposals are on average more generous than men's, regardless of the sex of the partner, and women respondents are more likely to accept an offer of a given amount. A given offer is more likely to be accepted if it comes from a woman; we term this result chivalry. Women paired with women almost never fail to reach an agreement; we term this result solidarity. Age, earnings, and race also significantly affect proposals and the rates of rejection. (JEL C78, C92, J16)

Economics has done very badly [explaining] large differences among ethnic groups. This is important ... also for gender differences. Some of this can be due to individual differences, but some of it clearly must be social.

Ken Arrow [1]

I. INTRODUCTION

A question that has occupied social psychologists for many years, and one that economists recently have come to address, is whether the decision-making calculus of individuals differs according to their sex. Evidence from social psychology suggests that, whether by nature or nurture, women behave differently from men in many arenas. [2] While economists naturally tend to focus their attention on economic parameters, economic models of behavior potentially could be expanded to incorporate systematic effects of such characteristics as sex or race. If these characteristics are associated (on average) with differences in the decision-making calculus of individuals, then simpler models that abstract from them will describe and predict less accurately.

We consider differences between women and men in bargaining behavior and address two questions: (1) Does the strategy adopted or the offer made or accepted differ systematically by the sex of the decision-maker? (2) Does the sex of the opponent influence a player's strategy choice? In a study of race and gender discrimination in bargaining for a new car, Ayres and Siegelman (1995) address the first question and find significantly different negotiated prices, depending on the gender of the bargainers. They note, "Dealers quoted significantly lower prices to white males than to black or female test buyers using scripted bargaining strategies" (304). Although they find no significant gender-pair effects, in an earlier study Ayres (1991) reports that women receive worse deals from women sellers.

Our laboratory study differs from field studies of bargaining behavior in three significant ways. First, the bargaining environment is much simpler: The available bargaining strategies are limited, and the heterogeneity of the bargainers is reduced. This enables us to more clearly observe the strategies chosen by each pair. Second, we can test directly for the effect of gender pairing; a given subject faces all decision environments. Finally, we observe rejection behavior by the bargainers. This allows us to distinguish between payoff-maximizing differences in offers that anticipate rejection rates and costly discrimination.

We study the simplest of bargaining games, the ultimatum game, in which two players split a fixed amount of money according to specific rules [3]. The advantage of the ultimatum game over other simple decision environments, such as a prisoners' dilemma or a public good game, is that the ultimatum game allows us to study gender differences from two perspectives. First, the proposer stage of the game is a strategic environment; subjects' payoffs are interdependent, and an unequal proposal carries the risk of rejection. Second, the second (accept-or-reject) stage of the game is a nonstrategic, riskless environment. Our results indicate differences in the behavior of men and women, but the differences are subtle and complex. Our research also provides some preliminary evidence on the effect of other cultural correlates.

II. BACKGROUND

The conclusion of much psychological research on sex differences is that men tend to be more competitive while women are more cooperative. [4] Gilligan (1982) suggests that women differ considerably from men in moral decision making. Women tend to stress the interests of the group and the avoidance of harm to others--"a morality of responsibility"--and men stress individual rights and justice--"a morality of rights" (22). Economists' interest is in whether this difference holds up in economic decision making, when money is at stake.

Attempts to test this competitive/cooperative hypothesis using social dilemma experiments (e.g., prisoners' dilemma and public goods games) have generated mixed results. In a social dilemma, cooperation means choosing a strategy that contributes to group earnings at the expense of individual earnings; competition is individual earnings maximization. Some studies find greater cooperation by women, as in Aranoff and Tedeschi (1968), Meux (1973), and Ortmann and Tichy (1999). Others find men more cooperative, as in Rapoport and Chammah (1965), Kahn et al. (1971), and Mack et al. (1971). Still others find inconsistent or insignificant differences between the sexes, as in Dawes et al. (1977), Stockard et al. (1988), and Orbell et al. (1994). Among economics experiments, Mason et al. (1991) find no sex difference in a duopoly game. In public goods experiments, Brown-Kruse and Hummels (1993) find men to be more cooperative than women, while Nowell and Tinkler (1994), Seguino et al. (1996), and Cadsby and Maynes (19 98) report evidence that women may be more cooperative. Bolton and Katok (1995) find no significant differences between the play of men and women in dictator games; in somewhat different dictator game environments, Selten and Ockenfels (1998) and Eckel and Grossman (1996, 1998) find women to be significantly less selfish than men. In a series of dictator games with different prices of giving, Andreoni and Vesterlund (2001) find that giving by men is more price-elastic. While women are more generous at a 1/1 trade-off, men are more generous when the price of giving to the other person is lower. In a study similar to our own, Solnick (2001) tests for sex differences in an ultimatum game. We return to a discussion of her article, and a comparison with our results in the concluding section. Eckel and Grossman (2001) provide a survey of ultimatum, dictator, and public goods games.