Journal of Economic Literature 2009, 47:2, 1–27 http:www.aeaweb.org/articles.php?doi=10.1257/jel.47.2.1

Gender Differences in Preferences

Rachel Croson and Uri Gneezy*

This paper reviews the literature on gender differences in economic experiments. In the three main sections, we identify robust differences in risk preferences, social (other-regarding) preferences, and competitive preferences. We also speculate on the source of these differences, as well as on their implications. Our hope is that this article will serve as a resource for those seeking to understand gender differences and to use as a starting point to illuminate the debate on gender-specific outcomes in the labor and goods markets.

1. Introduction The main source of data used in the cur- rent article is economics experiments. In conomists and policymakers have ob- the experiments we review, the decisions Eserved gender differences in a number that individuals make allow the researcher of different domains, including consumption, to isolate one factor of a decision (e.g., risk investment and, perhaps of most concern, in preferences) and study it in isolation from the labor market (see Francine D. Blau and other factors (e.g., altruism). Experiments Lawrence M. Kahn 2000 for a review). It is are also replicable, so the same experiment often hypothesized that these differences are can be conducted multiple times with dif- caused by preference differences between ferent individuals with diverse backgrounds the genders. and demographics. This allows us to test the In this article, we review experimental impact of various parameters, such as self- evidence on preference differences between selection and learning, on men and women. men and women, focusing on three factors We also include some data from naturally that have been extensively studied: risk pref- occurring markets (e.g., portfolio selection) erences, social preferences, and reaction to when relevant. competition.1 We find that women are indeed more risk- averse than men. We find that the social preferences of women are more situation- ally specific than those of men; women are * Croson: University of Texas, Dallas. Gneezy: Uni- neither more nor less socially oriented, but versity of California, San Diego. their social preferences are more malleable. 1 Another type of preference difference relates, for example, to family–career trade-offs. We do not explore Finally, we find that women are more averse this issue in the current survey. This does not mean that to competition than are men. we believe that these issues are of less importance or rel- evance, only that experimental methods cannot illuminate A number of previous papers review experi- them as clearly. mental psychology studies on the impact of

1

03-Croson-472.indd 1 5/18/09 4:42:05 PM 2 Journal of Economic Literature, Vol. XLVII (June 2009)

gender.2 We hope that this article serves a using both real and hypothetical gambles. similar purpose in economics; as a resource The robust finding is that men are more risk- for those seeking to understand gender differ- prone than are women. Previous surveys of ences and (perhaps) to use as a starting point economics (Catherine C. Eckel and Philip to illuminate the debate on gender-specific J. Grossman 2008c) and psychology (James outcomes in the labor and goods markets. P. Byrnes, David C. Miller, and William D. The remainder of this article is divided Schafer 1999) report the same conclusions: into three topics. Section 2 reviews evidence women are more risk averse than men in the on gender differences in risk preferences. vast majority of environments and tasks. This Section 3 reviews evidence on gender differ- table (and future tables as well) also note ences in social preferences. Section 4 reviews whether the authors included controls other evidence on gender differences in competi- than gender in their analyses (e.g., year in tive preferences. The final section provides a school, age, major, country of origin, race, etc). conclusion and discussion. The inclusion of controls, and exactly which were included, varies by paper. There are two notable and interest- 2. Risk Preferences ing papers in this table. First, Melissa L. Many of the decisions people make involve Finucane et al. (2000) find a gender differ- risk.3 In this section, we review the experi- ence among whites, but not among any other mental economics literature examining gen- ethnic group. They term this “the white male der differences in risk preferences. effect.” This is important because it implies there may be cultural biases causing gender 2.1 Objective Probability Lotteries differences in risk taking. This topic of cul- ture will reemerge in the section on compe- To set the stage, we begin by discussing risk- tition below. The second paper is by Renate taking in what we call objective probability Schubert et al. (1999) who find one situa- lotteries, with known probabilities and dollar tion in which men are more risk averse than outcomes. Table 1 lists ten papers investigat- women: when lotteries are framed as losses ing gender differences in risk preferences rather than gains.4

2 Meta-analyses have been published in examin- 3 We use “risk” and “uncertainty” interchangeably ing the impact of gender on intelligence testing (e.g., throughout the paper. We do not use Knight’s (1921) dis- Marise Born, Nico Bleichrodt and Henk van der Flier tinction by which risk refers to situations where one knows 1987); cognitive ability including mathematical, ver- the probabilities and uncertainty refers to situations when bal, and spatial ability (e.g., Janet S. Hyde, Elizabeth this randomness cannot be expressed in terms of specific Fennema and Susan J. Lamon 1990); personality devel- probabilities. This is in line with the approach that, even opment (e.g., Alan Feingold 1994); conformity and under uncertainty, one can assign subjective probabilities social influence (e.g., Blair T. Johnson and Alice H. to outcomes. It is interesting to note that, while most real Eagly 1989); self-disclosure (e.g., Kathryn Dindia and life situations involve Knight’s uncertainty, laboratory Michael Allen 1992); leadership style, evaluation, and experiments are more focused on decisions under risk in effectiveness (e.g., Eagly, Steven J. Karau, and Mona G. which probabilities are known. Makhijani 1995); aggressive behavior (e.g., Eagly and 4 One paper not included in the table, Tomomi Tanaka, Valerie J. Steffen 1986); and social behavior (e.g., Eagly Colin F. Camerer, and Quang Nguyen (forthcoming), finds and Wendy Wood 1991). In an excellent overall review, no significant risk differences in estimations of prospect- Eagly (1995) describes over twenty-five years of psy- theory preferences (no gender differences in loss aversion chological gender research (see also the heated debate or in the curvature of the value function). However, they in the February 1996 issue of American Psychologist do not report gender differences in risk aversion param- that followed). eters from traditional expected utility models.

03-Croson-472.indd 2 5/18/09 4:42:05 PM Croson and Gneezy: Gender Differences in Preferences 3

Table 1

Experimental Gain/ Controls details Pay loss Summary Risk taking included?

Holt and Students Yes Gain Choice between lotteries Low payoffs: Yes Laury according to mean–variance. M F > (2002) Varied also the level of pay High payoffs: M F = Hartog, Mail survey No Gain Willingness to pay for high-stakes M F Yes > Ferrer-I- and Dutch lotteries. Gender difference in risk Carbonell, and newspaper aversion parameter is estimated at Jonker (2002) 10 to 30 percent

Dohmen et Rep. sample real Both Survey instrument is validated in M F Yes > al. (2005) of German and experiments. Survey questions population hyp predicted behavior well and students

Powell and Students Yes Both Choice of insurance cover in one M F No > Ansic (1997) treatment and an unfamiliar finan- cial decision about gains in another Eckel and Students Yes Both Choice between lotteries according M F Yes > Grossman to mean–variance. Frame (gain/ (2002a) loss) changed between treatment Eckel and Students Yes Both Choice between lotteries according M F Yes > Grossman to mean–variance. Lotteries and (2002b) investment frames with the possibil- ity of loss, and a lottery frame with no loss

Fehr-Duda, Students Yes Both Gender differences depend on the M F Yes > Gennaro, size of the probabilities for the lot- and Schubert teries’ larger outcomes (2006)

Levin, Students No Both Half of the subjects were given the M F No > Snyder, and “chance of winning” each gamble, Chapman and half were given the “chance of (1988) losing” each lottery

Finucane Phone survey No Both Ethnically diverse group of partici- M F Yes > et al. (2000) pants. White males were more risk taking than all other groups Schubert Students Yes Both Choice between certain payoffs Gains: No et al. (1999) and lotteries in abstract and contex- M F > tual frames Losses: M F > Contextual: M F =

03-Croson-472.indd 3 5/18/09 4:42:05 PM 4 Journal of Economic Literature, Vol. XLVII (June 2009)

2.2 Portfolio Selection: High Stakes holds. Bernasek and Stephanie Shwiff (2001) Decisions ­overcome this by obtaining detailed infor- mation about the gender of the household’s In economics, the highest-stakes deci- decision maker and the household financial sions made by individuals, for themselves decision-making process. Using a survey on or as agents working for others, are often pension investments of universities’ faculty of special interest. It is an open question employees, they again show that women tend whether laboratory experiments with small to be more risk averse. stakes will yield conclusions that generalize In summary, we find that women are more to these high-stakes settings. One approach risk averse than men in lab settings as well is to conduct experiments with high stakes as in investment decisions in the field. While when possible (e.g., in poor countries where gender differences in risk preferences are modest payments by Western standards relatively consistent, very few explanations have high purchasing power). Most of the are offered for the observed differences. In comparisons between high- and low-stakes the remainder of this section, we identify data have shown that conclusions driven some potential explanations and discuss the from modest stakes do generalize. However, evidence supporting each. We also identify in the domain of financial risk taking, we exceptions to the general result in particular can often generate direct evidence. There tasks and by special subject pools. are several studies directly comparing high- stakes decisions of men and women, and this 2.3 Explanations for the Gender Difference literature demonstrates strong gender dif- in Risk Taking ferences, consistent with the results found in the lab. 2.3.1 Emotions For example, Annika E. Sunden and Brian J. Surette’s (1998) investigation of the alloca- The first explanation offered for gender dif- tion of defined contribution plan assets finds ferences in risk taking is based on differences that sex is significantly related to asset alloca- in emotional reactions to risky situations. In tion. Single women were less risk prone than an influential paper, George F. Loewenstein single men, consistent with the lab evidence et al. (2001) develop what they call “risk above (see also Finucane et al. 2000; Nancy as feelings” (see also the discussion of the Ammon Jianakoplos and Alexandra Bernasek “affect heuristic” in Paul Slovic et al. 2002); 1998). Similarly, Richard P. Hinz, David D. referring to our fast, instinctive and intui- McCarthy, and John A. Turner (1997) used tive reactions to risk. These affective reac- data on participants in the federal govern- tions are often better predictors of what we ment’s Thrift Savings Plan and found that do when facing a risky choice than the more women invest their pension assets more con- cognitive, reasoned approaches. We believe servatively than men. A large percentage of that this framework is crucial in understand- women invested in the minimum-risk port- ing gender differences in risk preferences. folio available to them. Married women also We look at the gender-specific influence of invest less in common stock than married emotions on outcomes and probabilities. men (see also Vickie L. Bajtelsmit and Jack Previous research from psychology indi- L. VanDerhei 1997). cates that women experience emotions more A potential problem with these studies is strongly than men (see the review in R. A. the inability to find out who makes invest- Harshman and A. Paivio 1987). A stronger ment decisions in married couple house- emotional experience can affect the utility of

03-Croson-472.indd 4 5/18/09 4:42:05 PM Croson and Gneezy: Gender Differences in Preferences 5

a risky choice. In particular, women report probabilities is found by Helga Fehr-Duda, more intense nervousness and fear than Manuele de Gennaro, and Schubert (2006). men in anticipation of negative outcomes In their experiment, risk taking depends on (e.g., Leslie R. Brody 1993; Frank Fujita, the size of the probabilities for the lotteries’ Ed Diener, and Ed Sandvik 1991). If nega- larger outcomes. Women are more risk averse tive outcomes are experienced as worse by in decisions with large probabilities in the women than by men, they will naturally be gain domain and in decisions with small and more risk averse when facing a risky situa- medium probabilities in the loss domain. The tion. Thus gender differences in emotional relation between the size of the probability experiences of outcomes, especially lower and t he emot iona l react ion is yet an open ques- utility resulting from bad outcomes, is one tion in the literature. Yuval Rottenstreich and explanation of increased risk aversion. Christopher K. Hsee (2001) demonstrate that Emotions also affect the perceptions of individuals use different probability weights probability. Previous research demonstrates for high-affect and low-affect gambles, but no that, in identical situations, women tend to gender differences have been demonstrated feel fear and men tend to feel anger (Michele in this probability weighting function. Grossman and Wood 1993). There is also 2.3.2 Overconfidence evidence that, when individuals are angry, they evaluate a given gamble as less risky A second reason for gender differences in than they do when they are afraid (Jennifer risk attitudes and in the evaluation of risk may S. Lerner et al. 2003). Thus if women are relate to confidence. The literature finds that more likely to be afraid of losing (e.g., to both men and women are often overconfident, overweight the probability of a loss), relative with men being more overconfident in their to men, they will evaluate a given gamble as success in uncertain situations than women being more risky, and will act in a more risk- (Sarah Lichtenstein, Baruch Fischhoff, and averse way. Lawrence D. Phillips 1982; Kay Deaux and A recent demonstration provides an ele- Elizabeth Farris 1977; Mary A. Lundeberg, gant test of the different influence of fear and Paul W. Fox, and Judith Punccohar 1994). For anger on estimation of probabilities and the example, Ralph Estes and Jinoos Hosseini resulting risk-taking behavior. Lerner et al. (1988) investigate the effects of selected vari- (2003) study the emotional reactions that fol- ables on investor confidence. Subjects were lowed September 11th by surveying a nation- asked to examine the financial statements of ally representative sample of Americans on a hypothetical company and then decide how September 20, 2001. They find that experi- much to invest in it. Next, the subjects were encing more anger in response to September asked to assess their confidence in the cor- 11th (men experienced more anger) triggered rectness of this investment decision.5 Women more optimistic beliefs about future gambles, were substantially less confident than men in while experiencing more fear in response to their investment decisions. In Jack B. Soll and September 11th (women experienced more Joshua Klayman (2004), participants were fear) triggered greater pessimism. Across asked to provide high and low estimates such all risks, males expressed lower perceptions of risk than did females, and differences in 5 Note that this measure of overconfidence (how sure reported emotions explained a large part of the individual is in their decision) is different than the the variance. question of misestimation of probabilities. The latter involves estimating the likelihood of an event occurring in An interesting aspect of gender differ- the world, while the former involves estimating the likeli- ences in the assessment of risk for different hood that one’s own estimate is likely to be correct.

03-Croson-472.indd 5 5/18/09 4:42:05 PM 6 Journal of Economic Literature, Vol. XLVII (June 2009)

that they were X percent sure that the cor- 2.4 Exceptions to the Rule: Managers and rect answer for a given question lay between Professional Populations them. Participants exhibited substantial over- confidence: The correct answer fell inside Many of the studies discussed above selected their intervals much less than X percent of the members of the general population (or the time. Both men and women were overconfi- convenient university population). However, dent, but men were more biased than women some studies have focused on a subsample of (for women, .58X percent of the answers fell the population; managers and professionals. within the stated range in experiment 1 and Among this population, gender differences in .60X percent in experiment 3, compared with financial risk preferences are smaller than in a performance for men of .40X percent in the general population and often nonexistent. experiment 1 and .51X percent in experiment For example, Stanley M. Atkinson, 3). Samantha Boyce Baird, and Melissa B. Muriel Niederle and Lise Vesterlund (2007) Frye (2003) compared the performance find that men are substantially more overcon- and investment behavior of male and female fident about their relative performance in a fixed-income mutual fund managers. They task (solving mathematical problems) than find that the way male and female managed women, and that the beliefs on relative perfor- funds do not differ significantly in terms of mance help predict entry decisions into com- performance, risk, and other fund character- petition (see the competition section below). istics. Their results suggest that differences If men are more confident of their likelihood in investment behavior often attributed to of coming out ahead in the gamble, they will gender may be related to investment knowl- be more likely to accept it than are women. edge and wealth constraints. J. E. V. Johnson and P. L. Powell (1994) 2.3.3 Risk as Challenge or Threats compare decision-making characteristics of A final explanation that we present for males and females in a “nonmanagerial” pop- the observed risk preference difference is ulation (in which the majority of individuals the interpretation of the risky situation. For have not undergone formal management example, Elizabeth Arch (1993) offers an education), with those of a “managerial” explanation of the gender difference in risk population of potential and actual manag- taking on the basis of the believed appropri- ers who have undertaken such education. ate response. Males are more likely to see a The managerial subpopulation males and risky situation as a challenge that calls for females display similar risk propensity and participation, while females interpret risky make decisions of equal quality, while in the situations as threats that encourage avoid- nonmanagerial subpopulation women are ance. This theme will reappear in the section more risk averse than men. Similar results on competitive behavior as well. are reported by Robert Master and Robert Arch argues that differences in risk behav- Meier (1988) with participants who owned a ior do not result from differences in ability, small business or managed one and by Sue persistence, or eagerness to perform a task Birley (1989), who studies entrepreneurs. well. Rather, the differences result from a The conclusion is that gender differences in different motivation between genders. Men risk preferences among the general population are more stimulated by challengeing, ego- do not extend to managers. This could be the involving situations; women are not stimu- result of selection; people that are more risk lated by the same factors, and may even be taking tend to choose managerial positions. impaired by them (Jeanne H. Block 1983). While fewer women select these positions,

03-Croson-472.indd 6 5/18/09 4:42:05 PM Croson and Gneezy: Gender Differences in Preferences 7

those that do choose them have similar risk Future research should try to disentangle preferences as men. This result could also be the two possible driving forces behind this an adaptive behavior to the requirements of exception to the rule: selection (more risk the job. In any case, the evidence suggests that taking people choose and remain in profes- managers and professional business persons sional careers) and learning (people learn present an important exception to the rule from their professional environment). that women are more risk averse than men. A nice piece of evidence that ties together 3. Differences in Social Preferences this exception to the general rule is presented by Peggy D. Dwyer, James H. Gilkeson, and When individuals exhibit a social prefer- John A. List (2002) who use data from nearly ence, others’ payoffs (or utilities) enter into 2,000 mutual fund investors and find that their utility function. Social preferences are women take less risk than men in their mutual modeled in the economic literature in the fund investments. However, the observed dif- form of altruism (e.g., Gary S. Becker 1974; ference in risk taking is significantly attenu- James Andreoni 1989), envy (e.g., Vai-Lam ated when a financial investment knowledge Mui 1995), inequality aversion (e.g., Gary E. control variable is included in the regres- Bolton and Axel Ockenfels 2000; Ernst Fehr sion model (see Matthias Gysler, Kruse, and and Klaus M. Schmidt 1999), or reciprocity Schubert 2002 for similar results in the lab). (e.g., Matthew Rabin 1993; Gary Charness and Rabin 2002; Armin Falk and Urs 2.5 Conclusion Fischbacher 2006; Martin Dufwenberg and A large literature documents gender dif- Georg Kirchsteiger 2004). While all these ferences in risk taking; women are more risk models describe how an individual may be averse than men. We highlight some of the other-regarding, the extent and form of the factors that we believe cause this gender dif- social preferences may also differ across the ference. One major factor is the affective genders. reaction to risk. Men and women differ in In this section, we discuss a number of their emotional reaction to uncertain situa- studies that demonstrate how strongly (and tions and this differential emotional reaction in what direction) social preferences mani- results in differences in risk taking. Emotions fest themselves in men and in women. We affect the evaluation of outcomes as well as include evidence on altruism and inequality- the evaluation of probabilities. However, aversion from ultimatum and dictator game emotions are not the only reason for gender studies. We also include evidence on reci- differences in risk preferences. Men are also procity from studies using trust and related more confident than women and, as a result, games. Finally, we briefly mention a large may have a different perception of the prob- number of older studies using the Prisoners’ ability distribution underlying a given risk. Dilemma game and discuss in more detail Men also tend to view risky situations as chal- recent studies using social dilemmas and/or lenges, as opposed to threats, which leads to public goods provision games.6 increased risk tolerance. Those differences are found in most 6 In addition, we identified four studies investigat- domains of risk taking. It is interesting to ing the impact of gender on coordination (Charles Bram note that these differences are attenuated Cadsby and Elizabeth Maynes 1998, Cadsby et al. 2007, by experience and profession. For example, Hakan J. Holm 2000, and Rachel Croson, Melanie Marks, and Jessica Snyder 2008). Since these studies speak only studies with managers and entrepreneurs weakly to the question of other-regarding preferences, find no gender differences in risk­preferences. they are not reviewed here.

03-Croson-472.indd 7 5/18/09 4:42:05 PM 8 Journal of Economic Literature, Vol. XLVII (June 2009)

Results on gender differences vary in these receives zero.7 The earliest ultimatum exper- studies. For example, sometimes women are iment was Werner Guth, Rolf Schmittberger, more trusting than men and sometimes less and Bernd Schwarze (1982). so. We believe that this variance is explained Although this game has a continuum of by a differential sensitivity of men and Nash equilibria, there is a unique subgame women to the social conditions in the experi- per fect equ i l ibr ium (a ssuming sel fish players) ment. Research from psychology suggests in which the proposer offers the responder , ε that women are more sensitive to social cues and the responder accepts. Deviations from in determining appropriate behavior than are this equilibrium on the responder’s side (that men (Carol Gilligan 1982). Small differences is, the rejection of positive offers) have been in experimental design and implementation interpreted as inequality-aversion, negative can affect these social cues, leading women to reciprocity, or punishment. Deviations from appear more other-regarding in some experi- this equilibrium on the proposer’s side (that ments and less other-regarding in others. is, the making of positive offers) have been Throughout this section, we provide two interpreted as inequality-aversion, altruism, types of data to support our explanation. and (occasionally) risk-aversion. First, we look within experiments that have Two lab experiments examine gender demonstrated gender differences for evidence effects in ultimatum settings: Eckel and that women are more responsive than men Grossman (2001) and Sara J. Solnick (2001). to the conditions of the experiment. Second, Both find that men and women offer the we look between studies and compare the same amounts, and that offers made to differences in male and female behavior. If men are higher than offers made to women. our explanation is correct, we will see more However, these studies differ in their charac- variability in female behavior across related terization of responder behavior (Eckel and studies than in male behavior. This evidence Grossman 2008a). is summarized in section 3.4. Eckel and Grossman find that women are As with risk preferences, psychologists more likely to accept lower offers than men. have also studied social preferences of the In contrast, Solnick found that women were genders. Meta-analyses of gender differ- more demanding than men. These differ- ences in social loafing, which maps to pub- ences may be attributable to differences in lic goods contributions and social dilemma the conditions of the experiment. In Eckel games (Karau and Kipling D. Williams and Grossman (2001), participants are 1993), and helping behavior, which maps paired with a responder randomly chosen into altruism (Eagly and Maureen Crowley from a group of four counterparts sitting 1986), are both useful sources for the inter- across an aisle, who were either all female, ested reader. all male, or of mixed genders. Proposers made offers that were communicated to 3.1 Ultimatum Games responders who accepted or rejected them. In the ultimatum game, two players are In Solnick (2001), ­participants sat on oppo- allocated a sum of money (the pie) that can be site sides of a curtain and had no face-to- divided between them. The proposer makes face contact. Her study used the strategy an offer to the responder of how the money will be divided, which the responder accepts or rejects. If the offer is accepted, each 7 Note that the ultimatum game is a simplified form of alternating-offer bargaining (also called Stahl-Rubinstein party receives the amount that the proposer bargaining). While many experiments have been run in suggested. If the offer is rejected, each party the latter paradigm, none have examined gender effects.

03-Croson-472.indd 8 5/18/09 4:42:06 PM Croson and Gneezy: Gender Differences in Preferences 9

Table 2 Rejection Rates in Ultimatum Games

Eckel and Grossman Solnick |Difference|

Male Responders M to M 18.8% 4.5% 14.3% F to M 9.4% 6.3% 3.2% |Difference| 9.4% 1.7% 8.7% Average

Female Responders M to F 17.2% 0.0% 17.2% F to F 3.1% 23.1% 20.0% |Difference| 14.1% 23.1% 18.6% Average

F – M 4.7% 21.4% Controls included? Yes No

method, where responders indicated their percent difference in Eckel and Grossman). minimum ­willingness to accept. Gender In contrast, women’s rejection rates are quite was communicated by the first name of the sensitive to the gender of their counterpart (a counterpart (a practice which Holm 2000 23.1 percent difference in Solnick and a 14.1 suggests yields the same results as inform- percent difference in Eckel and Grossman). ing the participant “your counterpart is a (fe) These comparisons, and similar analyses male student”; see also Chaim Fershtman below, support our organizing explanation of and Uri Gneezy 2001). greater context sensitivity of women. Table 2 shows rejection rates in comparable In an ultimatum field experiment, Guth, conditions to enable a comparison between Carsten Schmidt, and Matthias Sutter (2007) the studies. When men are responders, their asked readers of a weekly news magazine to rejection rates differ by an average of 8.7 per- propose (and respond to) offers in a three- cent between the two studies. When women party ultimatum game. In this game, the pro- are responders, their rejection rates differ poser makes an offer to split a pie between by an average of 18.6 percent between the himself, the responder (who can accept or two studies. This suggests that behavior of reject as usual), and a dummy player who has female responders is more sensitive to the no decision authority. They find that female experimental context (face-to-face, strategy participants are significantly more likely to vs. game methods) than is the behavior of propose a three-way equal split than are male responders. men, and suggest it is due to altruism or Comparing rejection rates within the inequality aversion. studies provides further evidence of greater However, given the ultimatum game struc- context-sensitivity by women. In both stud- ture, these behavioral differences could also ies, men’s rejection rates are not very sen- be due to risk aversion (see previous section). sitive to the gender of their proposer (a Dictator games allow us to tease apart these 1.8 percent difference in Solnick and a 9.4 competing motivations.

03-Croson-472.indd 9 5/18/09 4:42:06 PM 10 Journal of Economic Literature, Vol. XLVII (June 2009)

3.2 Dictator Games making the allocations is a disinterested third party (rather than a self-interested dictator), In the dictator (Robert Forsythe et al. and find the same results. Reinhard Selten 1994) game, the proposer again has a pie of and Ockenfels (1998) use a variant of the dic- money to divide between himself and the tator game called the solidarity game, where recipient. But the recipient has no decision to participants can offer “conditional gifts” to make; she can only accept the offer. Thus the insure each other against losses, and again dictator game is really an allocation problem. find that women are more inequality-averse Proposer decisions can be caused by inequal- than men. Dufwenberg and Astri Muren ity aversion or altruism, but strategic or risk- (2006a) look at gender effects in a team dic- related concerns are not relevant here. tator game (originally studied by Timothy Two studies use a simple dictator setting N. Cason and Mui 1997), where groups of to investigate gender effects. In Eckel and three divide money between themselves and Grossman (1998), participants play a double- a fourth recipient. The researchers find that blind dictator game with a $10 pie. They find female majority groups give the fourth party that in conditions of anonymity, women give significantly more than male majority groups, almost twice as much as men to their paired and are more likely to implement equal splits, recipient (on average women give $1.60 and again supporting the notion that women are men give $0.82). In Bolton and Elena Katok more inequality-averse than men. (1995), a less anonymous design is used in A number of studies go beyond identify- which participants again divide $10. The ing the main effects of gender to look at the options facing the participants are less con- interaction of the genders of the proposer tinuous, and no subject is permitted to offer and recipient in two-party dictator games. more than $5. They again find that women In Dufwenberg and Muren (2006b), par- give slightly more than men, but this differ- ticipants are told that their counterpart is a ence is not close to statistically significant “randomly selected (fe)male student in the (on average women give $1.23 and men give course.” This experiment involves almost no $1.13). anonymity and, consistent with Bolton and However, note again the comparison be- Katok, they find no significant differences tween these two studies. As the social con- between male and female giving. ditions of the experiment changed, male In contrast, Avner Ben-Ner, Fanmin Kong, giving changed by $0.31 while female giving and Louis Putterman (2004) run dictator changed by $0.37. This again suggests that games with male, female, and partners of the behavior of women (at least somewhat) is unknown gender. They find no gender differ- more sensitive to the conditions of the experi- ences in giving when the gender of the recip- ment than the behavior of men. ient is unknown (women give 3.29 out of 10, Four papers find that women are more men give 3.41) or male (women give 3.81, inequality averse in their dictator giving. men give 3.50). However, women give signif- Andreoni and Vesterlund (2001) manipu- icantly less to other women (2.185) than they late the cost–benefit ratio of giving money do to men (3.81) or to persons of unknown to the recipient. They find that women are gender (3.29). A similar manipulation was run more concerned with equalizing earnings in which the recipient was described as being between the parties, while men are more “from Minnesota” (the home state of most of concerned with maximizing efficiency. David the participants) or “not from Minnesota.” L. Dickinson and Jill Tiefenthaler (2002) run This distinction was relevant for women, who similar experiments, except that the party sent less to out-of-staters than they did to

03-Croson-472.indd 10 5/18/09 4:42:06 PM Croson and Gneezy: Gender Differences in Preferences 11

fellow Minnesota residents, but not for men. continuous versions were introduced by This study thus provides additional evidence Joyce Berg, John W. Dickhaut, and Kevin that women are sensitive to the social context A. McCabe (1995) and John B. Van Huyck, of the experiment (the gender or home state Raymond C. Battalio, and Mary F. Walters of the recipient) in ways that men are not. (1995). In these games, player one can send Paralleling these results, Daniel Houser all, some, or none of his endowment to player and Daniel Schunk (2007) run dictator games two (in the Kreps version, the decision is with schoolchildren between 8 and 10 years binary; send all or send none). The amount old. Children allocated 20 M&Ms between sent is multiplied, usually by 3 (occasionally themselves and another child. They also find by 2), and received by player two. Player two that girls’ giving was sensitive to the gender can then return as much or as little of the of their counterpart, girls offer more to boys money in her possession (sometimes includ- (9.8) than to other girls (7.9), and this differ- ing her initial endowment) to player one ence is statistically significant; p .05. In (in the Kreps version the decision is again < contrast, boys’ offers are not statistically dif- binary; return half or none). Note that this ferent depending on whether they’re offering second stage exactly mirrors a dictator game to boys (6.7) or to girls (4.6); p .1 (Houser as described above; player two is a dictator > and Schunk 2007, p. 10). toward player one. However, the motivations In summary, these studies find that men for returning behavior may be different; here choose efficient allocations while women are the pie which player two divides is created more inequality averse. However, compari- by the trusting actions and vulnerability of sons between the first two studies (Eckel and player one. In this section, we distinguish the Grossman and Bolton and Katok), and within two behaviors: trust (the sending of resources the final two studies (Ben-Ner et al., Houser to player two) and reciprocity or trustworthi- and Schunk), suggest that women’s decisions ness (the returning of resources to player are more context-specific than men’s. one). Table 3 describes a number of studies 3.3 Trust and Reciprocity examining gender in trust and trust-related Another series of experiments examine games. social preferences like trust and reciprocity. 3.3.1. Trusting Behavior What differentiates these games from those above is that they are typically positive-sum, The amount sent (or likelihood of send- involving a multiplier for money passed to ing in discrete games) is usually used as a a second party. They also explicitly test for measure of trusting behavior. Unfortunately, second-mover behaviors that are conditional. like the first move in an ultimatum game, Reciprocity, also called conditional altruism, this decision confounds trust and risk pref- describes behavior in which one party’s pref- erences. Thus while a series of studies finds erences over another party’s consumption that women send the same or less than men are conditional on the other party’s actions. I in this setting, this can be attributed either to act altruistically toward you if and only if you lower trust or to risk aversion. have been generous with me in the past. A number of studies find no gender dif- Many of the studies below rely on the ferences in sending behavior (Croson and trust game paradigm. A discrete version of Nancy R. Buchan 1999; Kenneth Clark the trust game was introduced by David M. and Martin Sefton 2001; James C. Cox Kreps (1990) and first experimentally tested and Cary A. Deck 2006; Iris Bohnet 2007; by Camerer and Keith Weigelt (1988). More Christiane Schwieren and Sutter 2008;

03-Croson-472.indd 11 5/18/09 4:42:06 PM 12 Journal of Economic Literature, Vol. XLVII (June 2009)

Table 3 Trust Games

Controls Study Experimental details Trust Reciprocity included?

Croson and Buchan (1999) Continuous game M F M F Yes = < U.S., China, Japan, Korea

Schwieren and Sutter (2004) Continuous game M F M F No = < trust in behavior versus ability in behavior in behavior

Clark and Sefton (2001) Sequential PD M F M F Yes = = trust 1st, reciprocity 2nd = =

Cox and Deck (2004) Discrete game M F M F No = = vary size of pie, single/double blind, response

Bohnet (2006) Continuous game (study 1) M F M F Yes = =

Ashraf et al. (2006) Continuous game M F M F Yes = = U.S., Russia, South Africa, strategy method

Eckel and Wilson (2004a) Discrete game M F M F Yes > = choice of partners (represented by icons)

Migheli (2006) Continuous game M F M F Yes > =

Innocenti and Pazienga (2006) Continuous game M F M F No > = double blind, gender communicated man/woman

Slonim (2004) Mostly continuous game M F M F Yes > = partner selection (gender, age known) no selection no selection

Kanagaretnam et al. (2006) Continuous game M F M F Yes > = multiple rounds, repaired, switch roles

Snijders and Keren (2004) Discrete game M F M F Yes > < subjects play both roles (strategy method)

Chaudhuri and Continuous game M F M F No > < Gangadharan (2004) subjects play both roles (strategy method)

Buchan et al. (2004) Continuous game M F M F No > < interaction of gender by first name, F, M or unknown

Slonim and Garbarino (2006) Mostly continuous game M F na Yes > online panel, strategy method, within subject

Bellemare and Kroger (2005) Continuous game M F M F Yes < > Dutch panel of Ss, strategy method

Eckel and Wilson (2004b) Discrete game M F written M F Yes > = written info or photo of partner M F photo <

Ben-Ner et al. (2004) Sequential dictator, same or different pairings na M F Yes < double-blind

Eckel and Grossman (1996) Sequential dictator na M F Yes <

Bohnet et al. (2008) Betrayal aversion game M F na No = Kuwait M F >

03-Croson-472.indd 12 5/18/09 4:42:06 PM Croson and Gneezy: Gender Differences in Preferences 13

Bohnet, Benedikt Hermann, and Richard sent by men is .46, but by women is .60. Zeckhauser forthcoming). Other studies Women thus appear more responsive to the find that men are more trusting than women conditions of the experiment, especially to (Eckel and Rick K. Wilson 2004b; Chris knowing the gender of their counterpart (and Snijders and Gideon Keren 2001; Ananish the realization of what that gender is) than Chaudhuri and Lata Gangadharan 2007; men, similar to the results of Ben-Ner et al. Buchan, Croson, and Solnick 2008; Matteo and Houser and Schunk in dictator games Migheli 2007; Alessandro Innocenti and described above. Maria Grazia Pazienza 2006; Robert Slonim Finally, in Eckel and Wilson (2004a), par- 2006; Ellen Garbarino and Slonim 2009). ticipants are either told information about Only a very few studies find women more their counterpart or see their picture. The trusting than men (Charles Bellemare and results indicate that women trust less than Sabine Kröger 2003; Bohnet, Hermann, and men when they have only written informa- Zeckhauser forthcoming in Kuwait only). We tion about their counterpart, but more than believe that these inconsistent gender differ- men when they have a photo. Again, women’s ences are caused by greater responsiveness behavior is more variable than men’s behav- of women to conditions of the experiment. ior. There is a 19 percentage point differ- Three within-study comparisons provide ence between the male trusting rates in the direct evidence for our explanation. two conditions (92 percent versus 73 per- In Cox and Deck (2006), the authors cent), and a 24 percentage point difference vary the size of the pie available, the social between the female trusting rates in the two distance of the experiment (single versus conditions (64 percent versus 88 percent). double-blind), and the ability of the second Anna Dreber and Johannesson (2008) player to respond. The proportion of women compared trusting behavior between men who send varies from 64 percent to 32 per- and women using a different experimental cent with the conditions for a range of 32 setting introduced by Gneezy (2005). The percentage points. In contrast, the propor- setting consists of a sender–receiver game in tion of men who send varies from 55 percent which the sender has a monetary incentive to 35 percent for a range of only 20 percent- to send a deceptive message to the receiver, age points. A probit model in table 4 of their and the receiver can either act according to paper reports that the decisions of men are the message or not, indicating distrust. They not statistically sensitive to the treatments, found no difference in trusting behavior but that the decisions of women are. The between men and women, as indicated by authors write “ . . . depending on the deci- receivers acting in accordance with the mes- sion context, women may appear to be more sage sent. They did, however, find that male or less generous than men because men are senders were more likely to send a deceptive relatively less responsive . . . ” (p. 597). message. In Buchan, Croson, and Solnick (2008), In summary, a number of studies have the authors look at the interaction of the demonstrated that women trust less than or two genders; participants in this study either the same as men in these settings. But wom- know (or do not know) the gender-specific en’s trust levels are more context-sensitive first name of their counterpart in a continu- than those of men. ous trust game. The range of amounts (max 3.3.2. Reciprocal Behavior minus min) that men send is $1.22, while the range of amounts that women send is $1.47. While some studies have found no gen- The standard deviation of average amounts der differences in reciprocity (Clark and

03-Croson-472.indd 13 5/18/09 4:42:06 PM 14 Journal of Economic Literature, Vol. XLVII (June 2009)

Sefton 2001; Cox and Deck 2006; Eckel matter of principle: one is, or is not, fair . . . . and Wilson 2004b; Eckel and Wilson 2004a; For women, fairness does not appear to be Bohnet 2007; Migheli 2007; Innocenti and a moral imperative. Choices are made with Pazienza 2006; Slonim 2006), others have greater consideration of the circumstances found that women are more reciprocal than surrounding the decision . . . . Women are men (Croson and Buchan 1999; Chaudhuri less likely to be driven by a rigid ethical and Gangadharan 2007; Snijders and Keren code” (pp. 153–54, italics ours). We find this 2001; Buchan, Croson, and Solnick 2008; explanation compelling, and have provided Schwieren and Sutter 2008; Ben-Ner et al. further evidence throughout this section 2004; Eckel and Grossman 1996). One study, (summarized below) that the increased sensi- Bellemare and Kroger (2007), finds that men tivity of women to the context of the situation are more reciprocal than women. is the cause of inconsistent gender differences Two experiments demonstrate the in-­ in social preferences. creased responsiveness of women to con- text in this setting. Ben-Ner et al. (2004) 3.4 The PD, Social Dilemmas, and Public use a two-stage dictator game with roles Goods Provision being switched and pairs being either kept A great many studies from psychology have together (specific reciprocity) or reshuffled examined gender differences in the prisoners’ (generalized reciprocity). The authors find dilemma setting. In an early study, Anatol that women are influenced by the amount Rapoport and Albert M. Chammah (1965) they received in the first round more strongly show that men cooperate significantly more than are men. Thus the link between the than women, as do a series of later studies amount received and the amount returned (e.g., Arnold Kahn, Joe Hottes, and William is significantly stronger for women than for L. Davis 1971; David Mack, Paula N. Auburn, men; further supporting the conclusion that and George P. Knight 1971). However, other female behavior is more sensitive to context studies have shown that women are more than is male behavior. cooperative than men (e.g., S. Sibley, S. Senn, In Eckel and Grossman (1996), partici- and A. Epanchin 1968; J. T. Tedeschi, D. pants chose to be dictators with a large pie Hiester, and J. Gahagan 1969), while others and a counterpart who had previously acted have shown no significant differences (e.g., unfairly toward a third party, or with a small Robyn M. Dawes, Jeanne McTavish, and pie and a counterpart who had previously Harriet Shaklee 1977; John Orbell, Dawes, acted fairly. They find that women are more and Peregrine Schwartz-Shea 1994). likely to both reward and to punish than In economics experiments, Robert H. are men. The authors also find that female Frank, Thomas Gilovich, and Dennis T. punishment behavior is sensitive to the cost Regan (1993) finds that women are signifi- of punishment, while male behavior is not. cantly more cooperative than men in prison- Women punish 64 percent of the time when er’s dilemma games. Andreas Ortmann and it is cheap, and 32.7 percent of the time when Lisa K. Tichy (1999) reports the same result in it is expensive, while men punish 39.3 per- the first round of a repeated experiment, but cent of the time when it is cheap and 40.8 that gender differences disappear over time. percent of the time when it is expensive. Additionally, male subjects acted the same in The authors argue that “[t]he results are mixed groups and all male groups (cooper- consistent with Gilligan’s (1982) claims about ating 27 percent of the time and 38 percent male and female differences. As she argues, of the time respectively). Females, however, for men, fairness is more of an absolute, a are significantly more cooperative in the

03-Croson-472.indd 14 5/18/09 4:42:06 PM Croson and Gneezy: Gender Differences in Preferences 15

Table 4 Public Goods Social Dilemmas / Contribution rates Significantly Controls Study details Males Females different? included? Solow and Kirkwood n 5, continuous, identity 66% 60% No = (2002) manipulated (strangers, MGP, band) Cadsby and Maynes n 4, discrete, all M/F groups, 67% 60% No = (1998) manipulate MPCR, anonymity Sell et al. (1993) n 4, continuous, 57% 52% No = all M/F/mixed/unknown groups Andreoni and n 5, continuous, photos of 47% 41% No = Petrie (2007) counterparts Brown-Kruse and n 4, discrete, all M/F groups, 68% 56% M F No = > Hummels (1993) manipulate MPCR, comm. Sell and Wilson (1991) n 4, continuous, 51% 37% M F No = > full, total or no feedback Seguino et al. (1996) n 5 to 52, continuous game 49% 66% F M Yes = >

Range of contributions 21% 30%

mixed-sex groups than in all-female groups altruistic. An analysis of a large-scale VCM (cooperating 65 percent of the time and 50 dataset exploring gender differences is cur- percent of the time respectively). Again, this rently underway in Simon Gachter and Eva experiment provides some support for our Poen (2004). conjecture that women are more sensitive to Early VCM experiments find compet- the context of the experiment than are men. ing results. Jamie Brown-Kruse and David Economists have spent more energy inves- Hummels (1993), Jane Sell and Wilson (1991), tigating continuous versions of dilemma and John L. Solow and Nicole Kirkwood games in the field of public goods provision. (2002) find that men contribute more A series of experiments investigates gender toward the public good than women. In con- differences in the voluntary contribution trast, Stephanie Seguino, Thomas Stevens, mechanism (VCM). In this game, intro- and Mark A. Lutz (1996) find that women duced by Gerald Marwell and Ruth E. Ames ­contribute more toward the public good (1981), individuals have resources they can than men. Finally, Sell, W. I. Griffith, and allocate toward their private consumption or Wilson (1993), Cadsby and Maynes (1998), the group’s public consumption. Resources and Andreoni and Ragan Petrie (2008) find are worth more to the individual when pri- no significant differences. vately consumed, but generate more social As above, these studies have significant value when used to provide public goods. methodological differences, as described in Equilibrium contributions toward the public table 4. However, when comparing between good in these settings are zero, and devia- studies, we find that male contributions are tions from that benchmark are considered more stable (with a range of 21 percent),

03-Croson-472.indd 15 5/18/09 4:42:06 PM 16 Journal of Economic Literature, Vol. XLVII (June 2009)

than female contributions (with a range of 30 differences and look for evidence that women percent). are more responsive than men to the condi- Finally, Janie M. Chermak and Kate Krause tions of the experiment. We find such evi- (2002) examine the effect of gender in a dif- dence in a wide variety of settings. ferent public goods game, one modeling In ultimatum games, women’s accept– common pool resources. They find that gen- reject decisions vary more with the gender der matters when individuals know the roles of their partner than do men’s (Eckel and they are to play. In those treatments women Grossman; Solnick). In dictator games, we are more generous (take less) than men. find that women’s decisions are sensitive to However, when individuals do not know their the gender (and home state) of their counter- roles, there are no gender differences. The part while men’s are not (Ben-Ner, Kong, and authors conclude (as do we) that “ . . . gender Putterman; Houser and Schunk). effects . . . are sensitive to protocol and con- In trust decisions, we find that the amounts text” (p. 61). women send varies more than the amounts men send with the identification (and gen- 3.5 Organizing Explanation der) of their counterpart (Buchan, Croson, A large body of work identifies gender and Solnick 2008), and with the existence differences in other-regarding preferences. of a picture of their counterpart (Eckel and However, many of the results are contradic- Wilson). Similarly, female trust is sensitive tory. In some experiments, women are more to the social distance in the experiment and altruistic, inequality averse, reciprocal, and the ability of the second player to respond, cooperative than men, and in others they are while male trust is not sensitive to these fac- less so. tors (Cox and Deck). We believe that the cause of these con- In reciprocal decisions, we again find that flicting results is that women are more women are more sensitive to what happens in sensitive to cues in the experimental con- the experiment. Men are less likely to pun- text than are men. Research from psy- ish (reward) a partner who had previously chology suggests that women are more been unfair (fair) than are women (Eckel sensitive to social cues in determining and Grossman). Women are influenced appropriate behavior (Kahn, Hottes, and more strongly than men by the first-mov- Davis 1971). Small differences in experi- er’s decision in sequential dictator games as mental design and implementation will well (Ben-Ner et al.). And women are more thus have larger impacts on female partici- reciprocal in trust games than men (Croson pants than on male participants. Some and Buchan; Buchan, Croson, and Solnick; examples of these design and implementa- Chaudhuri and Gangadharan; Snijders and tion differences include ­economic ­variables Keren; Schwieren and Sutter). like the size of the payoffs, the price of Second, we look between studies and altruismm, or the repetition of the game, compare the differences in male and and psychological variables like the amount female behavior. Between-study compari- of anonymity between counterparts, the sons of levels is always tricky, thus we are amount of anonymity between the partici- more careful in our interpretations here. pant and the experimenter, and the way that If our explanation is correct, we will see the situation is described. more variability in female behavior across We provide two types of analyses to sup- related studies than in male behavior. We port our explanation. First, we identify find between-study evidence for our expla- experiments that have demonstrated gender nation as well.

03-Croson-472.indd 16 5/18/09 4:42:07 PM Croson and Gneezy: Gender Differences in Preferences 17

In responder behavior in ultimatum reluctant than men to engage in competitive games, we compare the Eckel–Grossman interactions like tournaments, bargaining and and Solnick papers and find that rejection auctions. Additionally, men’s performance, rates by women differ by 18.6 percent while relative to women’s, is improved under com- rejection rates by men differ by only 8.7 petition. Thus as the competitiveness of an percent. In ­dictator giving, we compare the environment increases, the performance and Eckel and Grossman and Bolton and Katok participation of men increase relative to that papers and find that male giving differed by of women. $0.31 while female giving differed by $0.37 4.1 Reacting to Competition between the two studies. Finally, compar- ing seven VCM experiments, we find that What happens when people find them- female’s contributions changed by 30 per- selves in competition? Do men and women centage points, while male’s contributions react differently to the competitive incen- changed by only 21 percentage points. tives? Recent findings suggest that men’s We believe, as suggested by Gilligan (1982), performance is more affected by the com- that men’s decisions are less context-specific petitiveness of the environment than wom- than women’s. Participants of both genders en’s performance. We demonstrate this with are likely maximizing an underlying utility two studies. function, but the function that men use is In the first demonstration in the lab, less sensitive to the conditions of the experi- Gneezy, Niederle, and Aldo Rustichini (2003) ment, information about the other party, asked men and women to solve mazes on a and (even) the other party’s actions, than the computer for fifteen minutes. In a between- function that women use. This causes what subjects design, participants were paid either appear to be inconsistent results; sometimes according to a piece rate (a dollar amount per men appear more altruistic than women maze solved) or according to a winner-take- and other times, women appear more other- all tournament. Under the piece rate, men regarding than men. But primarily what we performed slightly (but not statistically sig- see is women’s behavior is more context- nificantly) better than women, solving 11.2 dependent than that of men. mazes on average, compared with 9.7 for We conclude this section with a recent women. However, when participants were field experiment that demonstrates this dif- paid on a competitive basis, males’ mean per- ference in sensitivity directly. Carl Mellström formance increased significantly to 15, while and Johannesson (2007) test whether pay- that of the female subjects remained statisti- ing people to donate blood will crowd-out cally the same at 10.8. The main finding is their intrinsic motivation to do so. They that in competitive situations where only the find a strong gender difference. While men’s best person in the group is rewarded, males donation behavior was not affected by the react with extra effort, while females do not. ­availability of payment, donations by women In a field study, Gneezy and Rustichini were negatively affected. (2004b) tested this conjecture in a physical education class. In a within-subject design, children ran twice over a short track with the 4. Competitive Behavior teacher measuring their speed. First they ran In this section, we look at a third gender alone, and then in pairs with different gender difference identified in experiments: dif- compositions. When the children ran alone, ferences in attitudes toward competition. there was no gender difference in perfor- Recent findings suggest that women are more mance. In competition, boys’ time improved

03-Croson-472.indd 17 5/18/09 4:42:07 PM 18 Journal of Economic Literature, Vol. XLVII (June 2009)

by .163 seconds, but girls’ ran .015 seconds tives, while the majority of females (65 per- slower than when they ran alone. cent) request the piece-rate compensation. It is tempting to generalize from those two When controlling for individual ability, it is studies and conclude that “men are more evident that while many well-performing responsive to competition.” However, there females hurt themselves financially by shy- are still many open questions. For example, ing away from competition, poorly perform- it is hard to know how sensitive the results ing males also hurt themselves by embracing are to the task used. Another unanswered it. Note that those results are related to the question regards the gender composition of findings regarding overconfidence discussed the group. In the maze study, women did in the risk section above. react to the competitive incentives in single Gneezy and Rustichini (2004a) used two sex groups, but not in mixed groups. In the tasks: one that favored men (shooting bas- race study, however, the gender composition kets) and one that favored women (solving of the group did not affect the results, and in anagrams). When solving anagrams, 40 per- Nabanita Datta Gupta, Anders Poulsen, and cent of the men and 25 percent of the women Marie-Claire Villeval (2005) men competed chose to compete; in shooting baskets the more against men than against women. numbers were 53 percent and 15 percent, Future research is needed to answer these respectively. That is, more men than women questions. chose the competitive environment in both tasks, but the gap in choice was smaller with 4.2 Self-Selection the task that favored women. The maze and the race studies concen- These and other findings (e.g., Donald trated on gender differences in reactions to Vandegrift and Paul Brown 2005; Datta competition. But what if participants could Gupta, Poulsen, and Villeval 2005) suggest choose the incentive scheme? If men and that women are less likely to choose to com- women rationally anticipate the gender dif- pete than men. Yet, women who choose com- ferences observed, they may very well choose petitive environments perform just as well as different environments. Several papers have men in those settings. investigated gender differences in the choice 4.3 Bargaining of incentives. In these studies, participants in lab experiments had the option of choosing One area in which avoiding competi- their own compensation scheme: piece rate tion can have a strong impact is bargain- or a winner-take-all tournament. ing. Competitiveness in this literature is Niederle and Vesterlund (2007) examine ­measured indirectly by inference from the compensation choice for addition prob- strategies. Competitiveness is associated lems, where there are no gender differences with negotiators who make large demands in performance under either the piece rate of their opponents or use distributive, win– or the tournament compensation. They have lose tactics like making threats, insults, groups of two women and two men who first and firm positional commitments. In other experience both compensation schemes with words, competitiveness involves concerns feedback about their own performance, and about one’s own outcomes in a conflict, then choose the incentive scheme for the while cooperativeness is characterized by next task. Despite the equality in perfor- a concern for the outcomes of the other mance they find that most males (73 per- party (cooperativeness thus implies social cent) request that their performance be preferences of some sort, as discussed above). compensated under the tournament incen- This definition is somewhat problematic

03-Croson-472.indd 18 5/18/09 4:42:07 PM Croson and Gneezy: Gender Differences in Preferences 19

because it ignores the possibility that these a negotiation, but significant in their propen- motivations are not mutually exclusive; sity to engage in a negotiation at all. many interactions involve elements of both motivations. 4.4 Why are Men More Competitive than Many studies in psychology documented Women? an economically small but significant gender effect in negotiation performance (see the Why do we see this genger difference meta-analyses in Amy E. Walters, Alice F. in attitudes and behavior? One suggested Stuhlmacher, and Lia L. Meyer 1998; Stuhl- explanation is backlash: It might be rational macher and Walters 1999; Joyce Neu, John L. for women to avoid negotiating in some situ- Graham, and Mary C. Gilly 1988; and D. F. ations. Bowles, Babcock, and Lei Lai (2007) Womak 1987). However, recent research sug- show experimentally that participants penal- gests that studies miss an important part of ize female job candidates more than male the process: The decision whether to initiate/ candidates for assertive negotiation behav- take part in negotiation (that is, the selection ior (see also Eckel and Grossman 1996). issue). Note that this question is related to the This explanation is related to the findings above discussion of selecting into more or less in the discrimination literature regarding competitive settings. incentives to underinvest in education, for In a recent book on gender and negotiation, example, because the expected rewards are Linda Babcock and Sara Laschever (2003) lower for women than for men in equilibrium claimed that women avoid competitive nego- (Becker 1965). tiation situations relative to men. For example, An additional set of data comes from exper- in a laboratory study participants were told iments with children. William T. Harbaugh, that they would be paid between $3 and $10 Krause and Steven G. Liday (2002), for for their participation. After each participant example, show that younger boys and girls finished, an experimenter thanked them and (second, fourth, and fifth grades) make the said “Here’s $3. Is $3 OK?” Only 2.5 percent same dictator offers as each other, but that of the female participants but 23 percent of older boys (ninth and twelfth grades) make the male participants requested more money lower dictator offers than do girls (boys aver- (Deborah A. Small et al. 2007). Babcock age 0.97 token out of 10, while girls average (2002) reports that average starting salaries 2.12 tokens out of 10). The fact that gender of male MBAs graduating from Carnegie differences exhibit only later in life suggests Mellon were 7.6 percent higher than those of an environmental cause. females. This difference is attributed to the Gneezy, Kenneth L. Leonard, and List observation that only 7 percent of the women (2006) use an experimental task to explore attempted to negotiate their salary offer, whether there are gender differences in while 57 percent of their male counterparts selecting into competitive environments negotiated (see also Hannah Riley Bowles, across cultures, examining a patriarchal soci- Babcock, and Kathleen L. McGinn 2005; ety (the Maasai in Tanzania) and a matri- Barry Gerhart and Sara Rynes 1991; Laura lineal society (the Khasi in India). Similar J. Kray, Leigh Thompson, and Adam D. to the evidence from the West discussed Galinsky 2001; Kray, Galinsky, and Thompson above, Maasai men opt to compete at twice 2002; Stuhlmacher and Walters 1999). the rate as Maasai women (50 percent ver- Thus in bargaining situations, women are sus 25 percent, respectively). However, this less likely to exhibit competitive preferences result is reversed amongst the Khasi, where than men, slightly in their reactions once in women choose the competitive environment

03-Croson-472.indd 19 5/18/09 4:42:07 PM 20 Journal of Economic Literature, Vol. XLVII (June 2009)

considerably more often than Khasi men relates with prenatal testosterone, positively (men chose to compete 39 percent of the correlates with prenatal estradial, and is fixed time whereas women chose to compete 54 early in life (Matthew H. McIntyre 2006). percent of the time). These results provide An interesting example of the role of bio- further support for the argument that societal logical measurements in the auction litera- ­structure is crucially linked to the observed ture is Yan Chen, Peter Katuscak, and Emre gender differences in competitiveness, and Ozdenoren (2009) who find that women’s thus, that “nurture matters.” competitiveness depends on menstruation An opposing view, that differences between and contraceptive pill usage. In first-price men and women are based on genetic differ- auctions, while women bid significantly ences, argues that “nature” is important as higher than men do in all phases of the well. From Charles Darwin through today, cycle, they find a sine-like pattern of bid- many evolutionary biologists and psycholo- ding throughout the menstrual cycle, with gists hold that the basic structure of the higher bidding in the follicular phase and brain is genetically determined.8 In this view, lower in the luteal phase. The studies dem- the regularities of human behavior as well onstrate, just as convincingly, that “nature as consistent differences between male and matters” as well. female psychology could be inherited char- We conclude from those findings that both acteristics. Under this nature explanation, nature and nurture are responsible for the at some point in human history men and gender differences in competition. The inter- women evolved different strategies to maxi- esting question is thus the weight of each fac- mize the fitness of their genes. For example, tor and, more interestingly, the interaction of genetic or hormonal differences could cause the two forces. Further research is clearly women to be less competitive than men (e.g., needed. Stephen Colarelli, Jennifer L. Spranger, and M. Regina Hechanova 2006). 5. Summary and Discussion Support for this explanation can be found in studies of the effect of biological measure- This article has reviewed the experimen- ments on behavior. For example, ­testosterone tal literature on gender differences in risk (and other hormones, such as cortisol) are preferences, social preferences, and competi- known to be correlated with aggression and tive preferences. In general, this literature are different between genders. There is a has documented fundamental differences large literature documenting the role of tes- between men and women (with exceptions tosterone in competitiveness (for a review, noted in the text). see Helen S. Bateup et al. 2002). Prenatal Most lab and field studies indicate that hormone exposure is thought to correlate women are more risk averse than men (sec- with sexually dimorphic behaviors as well tion 2), with important exceptions for manage- (John T. Manning and Rogan P. Taylor 2001). rial populations. We suggest a list of possible Dreber and Moshe Hoffman (2007) recently mechanisms behind these findings, including found that financial risk aversion correlates emotions, overconfidence, and framing. with a proxy for prenatal hormone exposure, A number of studies also indicate that namely the ratio between the second and women’s social preferences are different than fourth fingers. This measure negatively cor- men’s (section 3), although the results of these studies are varied. We suggest an organizing explanation that relies on the observation 8 See Darwin (1871), A. J. Bateman (1948), and Robert L. Trivers (1972). that women are more sensitive to social cues

03-Croson-472.indd 20 5/18/09 4:42:07 PM Croson and Gneezy: Gender Differences in Preferences 21

than are men. This leads to higher variabil- correlated with other X variables either miss- ity in women’s behavior than in men’s, which ing or observed. In the case of social prefer- we observe both within experimental studies ences, we argue for an interaction between T and between studies. (the experimental context) and X (the gender Finally, a third stream of literature sug- of the participant). gests that women’s preferences for competi- In this sense, we do not really summarize tive situations are lower than men’s, both experimentation in the classic physical sci- in purely competitive situations and in bar- ences sense—i.e., studies that use random- gaining settings (section 4). One important ization to achieve identification. In particular, and interesting question about these differ- gender is not randomly assigned. We believe ences is whether they are ingrained (nature) that more assumptions may be needed to or taught (nurture). We present evidence in infer what we would like to infer from these favor of both explanations, and suggest that experimental studies, and more research is the research question going forward should needed in this direction. be the relative weights of these two factors Second, an important bias in the litera- and their interaction. ture on gender differences is that journals In summary, we have identified three types are more likely to publish papers that find a of preferences which differ between men and gender difference than papers that do not. women. Each of these has implications for Moreover, this publication bias may cause the economic decisions that men and women researchers to invest more effort into finding make in labor and product markets. differences than to finding no difference. In We wish to end with three methodological the current article, we devote much attention notes. First, one way to organize our discus- to including studies that do not find gender sion is using the following simple model of differences, even when they are unpublished, the world (see List 2006): in our attempt to counteract this bias. Going forward, we urge researchers to routinely Y X T , record the gender of the participants when = β + τ + η possible (as is the case in the psychology lit- where Y is the outcome of interest (risk pos- erature). This will greatly expand our under- ture, social preference behavior, ­competitive standing of gender differences and avoid the spirit), X is a vector of person-specific vari- publication bias that is currently in place. ables (including gender), T is a binary treat- In all inference from a sample of individu- ment variable (experimental treatments als, one is concerned about whether the par- controlled by the researcher), is the error ticipants in the sample are self-selected. In η component, and and are estimated the field, the degree of self-selection must β τ parameters. often be inferred or measured indirectly. In In the typical case, to estimate the ana- the lab, it can often be controlled (e.g., using τ lyst simply needs proper randomization when students in a class who are required to par- using controlled experimental methods. Here ticipate, or paid at such a high rate that virtu- we are using T primarily as an explanatory ally all volunteer), or measured (comparing variable for our most interesting estimate, traits of volunteers and nonvolunteers). For that of on the gender term. This “treatment example, we discussed above some findings β effect” is of course not randomly determined showing that women experience increases by the researchers of the different studies, but in auction bids near the time of ovulation. instead selected to illuminate their research Interestingly, Richard L. Doty and Colin question of interest. T can therefore be Silverthorne (1975) find that menstrual

03-Croson-472.indd 21 5/18/09 4:42:07 PM 22 Journal of Economic Literature, Vol. XLVII (June 2009)

cycles affect volunteering behavior; most of Bellemare, Charles, and Sabine Kröger. 2003. “On the female volunteers for their experiment Representative Trust.” Tilburg University CentER Discussion Paper 2003-47. were in the ovulatory phase, whereas most Bellemare, Charles, and Sabine Kröger. 2007. “On of the female nonparticipants were in the Representative Social Capital.” European Economic postovulatory, premenstrual, and menstrual Review, 51(1): 183–202. Ben-Ner, Avner, Fanmin Kong, and Louis Putterman. phases. When data is collected in classes in 2004. “Share and Share Alike? Gender-Pairing, which all participants take part in the experi- Personality, and Cognitive Ability as Determinants ment, this bias should not affect the results. of Giving.” Journal of Economic Psychology, 25(5): 581–89. But further research is needed to investigate Ben-Ner, Avner, Louis Putterman, Fanmin Kong, and the effect of such selection biases in labora- Dan Magan. 2004. “Reciprocity in a Two-Part Dic- tory experiments. tator Game.” Journal of Economic Behavior and Organization, 53(3): 333–52. Berg, Joyce, John W. Dickhaut, and Kevin A. McCabe. References 1995. “Trust, Reciprocity, and Social History.” Games and Economic Behavior, 10(1): 122–42. Andreoni, James. 1989. “Giving with Impure Altruism: Bernasek, Alexandra, and Stephanie Shwiff. 2001. Applications to Charity and Ricardian Equivalence.” “Gender, Risk, and Retirement.” Journal of Eco- Journal of Political Economy, 97(6): 1447–58. nomic Issues, 35(2): 345–56. Andreoni, James, and Ragan Petrie. 2008. “Beauty, Birley, Sue. 1989. “Female Entrepreneurs: Are They Gender and Stereotypes: Evidence from Laboratory Really Any Different?” Journal of Small Business Experiments.” Journal of Economic Psychology, Management, 27: 32–37. 29(1): 73–93. Blau, Francine D., and Lawrence M. Kahn. 2000. Andreoni, James, and Lise Vesterlund. 2001. “Which “Gender Differences in Pay.” Journal of Economic Is the Fair Sex? Gender Differences in Altruism.” Perspectives, 14(4): 75–99. Quarterly Journal of Economics, 116(1): 293–312. Block, Jeanne H. 1983. “Differential Premises Arising Arch, Elizabeth. 1993. “Risk-Taking: A Motivational from Differential Socialization of the Sexes: Some Basis for Sex Differences.” Psychological Reports, Conjectures.” Child Development, 54(6): 1335–54. 73(3): 6–11. Bohnet, Iris. 2007. “Why Women and Men Trust Oth- Ashraf, Nava, Iris Bohnet, and Nikita Piankov. 2006. ers.” In Economics and Psychology: A Promising “Decomposing Trust and Trustworthiness.” Experi- New Cross-Disciplinary Field, ed. Bruno S. Frey mental Economics, 9(3): 193–208. and Alois Stutzer, 89–110. Cambridge and London: Atkinson, Stanley M., Samantha Boyce Baird, and MIT Press. Melissa B. Frye. 2003. “Do Female Mutual Fund Bohnet, Iris, Benedikt Hermann, and Richard Zeck- Managers Manage Differently?” Journal of Finan- hauser. Forthcoming. “The Requirements for Trust cial Research, 26(1): 1–18. in Gulf and Western Countries.” Quarterly Journal Babcock, Linda. 2002. “Do Graduate Students Negoti- of Economics. ate Their Job Offers?” Unpublished. Bolton, Gary E., and Elena Katok. 1995. “An Experi- Babcock, Linda, and Sara Laschever. 2003. Women mental Test for Gender Differences in Beneficent Don’t Ask: Negotiation and the Gender Divide. Behavior.” Economics Letters, 48(3–4): 287–92. Princeton and Oxford: Princeton University Press. Bolton, Gary E., and Axel Ockenfels. 2000. “ERC: A Bajtelsmit, Vickie L., and Jack L. VanDerhei. 1997. Theory of Equity, Reciprocity, and Competition.” “Risk Aversion and Pension Investment Choices.” American Economic Review, 90(1): 166–93. In Positioning Pensions for the Twenty-First Cen- Born, Marise, Nico Bleichrodt, and Henk van der tury, ed. Michael S. Gordon, Olivia S. Mitchell, and Flier. 1987. “Cross-Cultural Comparison of Sex- Marc M. Twinney, 45–66. Philadelphia: University Related Differences on Intelligence Tests: A Meta- of Pennsylvania Press. analysis.” Journal of Cross-Cultural Psychology, Bateman, A. J. 1948. “Intra-sexual Selection in Droso- 18(3): 283–314. phila.” Heredity, 2: 349–68. Bowles, Hannah Riley, Linda Babcock, and Lei Lai. Bateup, Helen S., Alan Booth, Elizabeth A. Shirt- 2007. “Social Incentives for Gender Differences in cliff, and Douglas A. Granger. 2002. “Testosterone, the Propensity to Initiate Negotiations: Sometimes Cortisol, and Women’s Competition.” Evolution and It Does Hurt to Ask.” Organizational Behavior and Human Behavior, 23(3): 181–92. Human Decision Processes, 103(1): 84–103. Becker, Gary S. 1965. Human Capital: A Theoretical Bowles, Hannah Riley, Linda Babcock, and Kath- and Empirical Analysis, with Special Reference to leen L. McGinn. 2005. “Constraints and Triggers: Education. New York: National Bureau of Economic Situational Mechanics of Gender in Negotiation.” Research. Journal of Personality and Social Psychology, 89(6): Becker, Gary S. 1974. “A Theory of Social Interactions.” 951–65. Journal of Political Economy, 82(6): 1063–93. Brody, Leslie R. 1993. “On Understanding Gender

03-Croson-472.indd 22 5/18/09 4:42:07 PM Croson and Gneezy: Gender Differences in Preferences 23

Differences in the Expression of Emotion.” In and Culture: International Experimental Evidence Human Feelings: Explorations in Affect Develop- from Trust Games.” American Economic Review, ment and Meaning, ed. Steven L. Ablon, Daniel 89(2): 386–91. Brown, Edward J. Khantzian, and John E. Mack, Croson, Rachel, Melanie Marks, and Jessica Snyder. 87-121. Hillsdale, N.J.: Analytic Press. 2008. “Groups Work for Women: Gender and Group Brown-Kruse, Jamie, and David Hummels. 1993. Identity in Social Dilemmas.” Negotiation Journal, “Gender Effects in Laboratory Public Goods Con- 24(4): 411–27. tribution: Do Individuals Put Their Money Where Darwin, Charles. 1871. The Descent of Man, and Selec- Their Mouth Is?” Journal of Economic Behavior tion in Relation to Sex. London: John Murray. and Organization, 22(3): 255–67. Datta Gupta, Nabanita, Anders Poulsen, and Marie- Buchan, Nancy R., Rachel Croson, and Sara J. Sol- Claire Villeval. 2005. “Male and Female Competi- nick. 2008. “Trust and Gender: An Examination tive Behavior: Experimental Evidence.” Institute for of Behavior and Beliefs in the Investment Game.” the Study of Labor Discussion Paper 1833. Journal of Economic Behavior and Organization, Dawes, Robyn M., Jeanne McTavish, and Harriet Shak- 68(3–4): 466–76. lee. 1977. “Behavior, Communication, and Assump- Byrnes, James P., David C. Miller, and William D. tions about Other People’s Behavior in a Commons Schafer. 1999. “Gender Differences in Risk-Taking: Dilemma Situation.” Journal of Personality and A Meta-analysis.” Psychological Bulletin, 125(3): Social Psychology, 35(1): 1–11. 367–83. Deaux, Kay, and Elizabeth Farris. 1977. “Attributing Cadsby, Charles Bram, Yasuyo Hamaguchi, Toshiji Causes for One’s Own Performance: The Effects of Kawagoe, Elizabeth Maynes, and Fei Song. 2007. Sex, Norms, and Outcome.” Journal of Research in “Cross-National Gender Differences in Behavior Personality, 11(1): 59–72. in a Threshold Public Goods Game: Japan versus Dickinson, David L., and Jill Tiefenthaler. 2002. Canada.” Journal of Economic Psychology, 28(2): “What Is Fair? Experimental Evidence.” Southern 242–60. Economic Journal, 69(2): 414–28. Cadsby, Charles Bram, and Elizabeth Maynes. 1998. Dindia, Kathryn, and Michael Allen. 1992. “Sex Dif- “Gender and Free Riding in a Threshold Public ferences in Self-Disclosure: A Meta-analysis.” Psy- Goods Game: Experimental Evidence.” Journal chological Bulletin, 112(1): 106–24. of Economic Behavior and Organization, 34(4): Dohmen, Thomas, Armin Falk, David Huffman, Uwe 603–20. Sunde, Jürgen Schupp, and Gert G. Wagner. 2005. Camerer, Colin F., and Keith Weigelt. 1988. “Experi- “Individual Risk Attitudes: New Evidence from a mental Tests of a Sequential Equilibrium Reputation Large, Representative, Experimentally-Validated Model.” Econometrica, 56(1): 1–36. Survey.” Institute for the Study of Labor Discussion Cason, Timothy N., and Vai-Lam Mui. 1997. “A Labo- Paper 1730. ratory Study of Group Polarisation in the Team Dic- Doty, Richard L., and Colin Silverthorne. 1975. “Influ- tator Game.” Economic Journal, 107(444): 1465–83. ence of Menstrual Cycle on Volunteering Behav- Charness, Gary, and Matthew Rabin. 2002. “Under- iour.” Nature, 254(5496): 139–40. standing Social Preferences with Simple Tests.” Dreber, Anna, and Moshe Hoffman. 2007. “Portfolio Quarterly Journal of Economics, 117(3): 817–69. Selection in Utero.” Unpublished. Chaudhuri, Ananish, and Lata Gangadharan. 2007. Dreber, Anna, and Magnus Johannesson. 2008. “Gen- “An Experimental Analysis of Trust and Trustworthi- der Differences in Deception.” Economics Letters, ness.” Southern Economic Journal, 73(4): 959–85. 99(1): 197–99. Chen, Yan, Peter Katuscak, and Emre Ozdenoren. Dufwenberg, Martin, and Georg Kirchsteiger. 2004. 2009. “Why Can’t a Woman Bid More Like a Man?” “A Theory of Sequential Reciprocity.” Games and Unpublished. Economic Behavior, 47(2): 268–98. Chermak, Janie M., and Kate Krause. 2002. “Individ- Dufwenberg, Martin, and Astri Muren. 2006a. “Gen- ual Response, Information, and Intergenerational der Composition in Teams.” Journal of Economic Common Pool Problems.” Journal of Environmental Behavior and Organization, 61(1): 50–54. Economics and Management, 43(1): 47–70. Dufwenberg, Martin, and Astri Muren. 2006b. “Gen- Clark, Kenneth, and Martin Sefton. 2001. “The erosity, Anonymity, Gender.” Journal of Economic Sequential Prisoner’s Dilemma: Evidence on Recip- Behavior and Organization, 61(1): 42–49. rocation.” Economic Journal, 111(468): 51–68. Dwyer, Peggy D., James H. Gilkeson, and John A. List. Colarelli, Stephen, Jennifer L. Spranger, and M. 2002. “Gender Differences in Revealed Risk Taking: Regina Hechanova. 2006. “Women, Power, and Sex Evidence from Mutual Fund Investors.” Economics Composition in Small Groups: An Evolutionary Per- Letters, 76(2): 151–58. spective.” Journal of Organizational Behavior, 27(2): Eagly, Alice H. 1995. “The Science and Politics of 163–84. Comparing Women and Men.” American Psycholo- Cox, James C., and Cary A. Deck 2006. “When Are gist, 50(3): 145–58. Women More Generous than Men?” Economic Eagly, Alice H., and Maureen Crowley. 1986. “Gender Inquiry, 44(4): 587–98. and Helping Behavior: A Meta-analytic Review of Croson, Rachel, and Nancy R. Buchan. 1999. “Gender the Social Psychological Literature.” Psychological

03-Croson-472.indd 23 5/18/09 4:42:07 PM 24 Journal of Economic Literature, Vol. XLVII (June 2009)

Bulletin, 100(3): 283–308. Schubert. 2006. “Genders, Financial Risk, and Prob- Eagly, Alice H., Steven J. Karau, and Mona G. Makhi- ability Weights.” Theory and Decesion, 60(2–3): jani. 1995. “Gender and the Effectiveness of Lead- 283–313. ers: A Meta-analysis.” Psychological Bulletin, 117(1): Feingold, Alan. 1994. “Gender Differences in Per- 125–45. sonality: A Meta-analysis.” Psychological Bulletin, Eagly, Alice H., and Valerie J. Steffen. 1986. “Gender 116(3): 429–56. and Aggressive Behavior: A Meta-analytic Review of Fershtman, Chaim, and Uri Gneezy. 2001. “Discrimi- the Social Psychological Literature.” Psychological nation in a Segmented Society: An Experimental Bulletin, 100(3): 309–30. Approach.” Quarterly Journal of Economics, 116(1): Eagly, Alice H., and Wendy Wood. 1991. “Explaining 351–77. Sex Differences in Social Behavior: A Meta-analytic Finucane, Melissa L., Paul Slovic, C. K. Mertz, James Perspective.” Personality and Social Psychology Flynn, and Theresa A. Satterfield. 2000. “Gender, Bulletin, 17(3): 306–15. Race, and Perceived Risk: The ‘White Male’ Effect.” Eckel, Catherine C., and Philip J. Grossman. 1996. Health, Risk and Society, 2(2): 159–72. “The Relative Price of Fairness: Gender Differ- Forsythe, Robert, Joel L. Horowitz, N. E. Savin, and ences in a Punishment Game.” Journal of Economic Martin Sefton. 1994. “Fairness in Simple Bargain- Behavior and Organization, 30(2): 143–58. ing Experiments.” Games and Economic Behavior, Eckel, Catherine C., and Philip J. Grossman. 1998. 6(3): 347–69. “Are Women Less Selfish than Men? Evidence from Frank, Robert H., Thomas Gilovich, and Dennis T. Dictator Experiments.” Economic Journal, 108(448): Regan. 1993. “Does Studying Economics Inhibit 726–35. Cooperation?” Journal of Economic Perspectives, Eckel, Catherine C., and Philip J. Grossman. 2001. 7(2): 159–71. “Chivalry and Solidarity in Ultimatum Games.” Eco- Fujita, Frank, Ed Diener, and Ed Sandvik. 1991. “Gen- nomic Inquiry, 39(2): 171–88. der Differences in Negative Affect and Well-Being: Eckel, Catherine C., and Philip J. Grossman. 2002. The Case for Emotional Intensity.” Journal of Per- “Sex Differences and Statistical Stereotyping in Atti- sonality and Social Psychology, 61(3): 427–34. tudes toward Financial Risk.” Evolution and Human Gachter, Simon, and Eva Poen. 2004. “Are Females Behavior, 23(4): 281–95. More Cooperative than Males?” Unpublished. Eckel, Catherine C., and Philip J. Grossman. 2008a. Garbarino, Ellen, and Robert Slonim. 2009. “The “Differences in the Economic Decisions of Men and Robustness of Trust and Reciprocity across a Het- Women: Experimental Evidence.” In Handbook of erogeneous U.S. Population.” Journal of Economic Experimental Economics Results, Volume 1, ed. Behavior and Organization, 69(3): 226–40. Charles Plott and Vernon Smith, 509–19. New York: Gerhart, Barry, and Sara Rynes. 1991. “Determinants Elsevier. and Consequences of Salary Negotiations by Male Eckel, Catherine C., and Philip J. Grossman. 2008b. and Female MBA Graduates.” Journal of Applied “Forecasting Risk Attitudes: An Experimental Study Psychology, 76(2): 256–62. Using Actual and Forecast Gamble Choices.” Jour- Gilligan, Carol. 1982. In a Different Voice: Psychologi- nal of Economic Behavior and Organization, 68(1): cal Theory and Women’s Development. Cambridge 1–17. and London: Harvard University Press. Eckel, Catherine C., and Philip J. Grossman. 2008c. Gneezy, Uri. 2005. “Deception: The Role of Con- “Men, Women and Risk Aversion: Experimental sequences.” American Economic Review, 95(1): Evidence.” In Handbook of Experimental Econom- 384–94. ics Results, Volume 1, ed. Charles Plott and Vernon Gneezy, Uri, Kenneth L. Leonard, and John A. List. Smith, 1061–73. New York Elsevier. 2006. “Gender Differences in Competition: The Eckel, Catherine C., and Rick K. Wilson. 2004a. “Con- Role of Socialization.” Unpublished. ditional Trust: Sex, Race and Facial Expressions in a Gneezy, Uri, Muriel Niederle, and Aldo Rustichini. Trust Game.” Unpublished. 2003. “Performance in Competitive Environments: Eckel, Catherine C., and Rick K. Wilson. 2004b. Gender Differences.” Quarterly Journal of Econom- “Whom to Trust? Choice of Partner in a Trust ics, 118(3): 1049–74. Game.” Unpublished. Gneezy, Uri, and Aldo Rustichini. 2004a. “Executives Estes, Ralph, and Jinoos Hosseini. 1988. “The Gender versus Teachers.” Unpublished. Gap on Wall Street: An Empirical Analysis of Con- Gneezy, Uri, and Aldo Rustichini. 2004b. “Gender and fidence in Investment Decision Making.” Journal of Competition at a Young Age.” American Economic Psychology, 122(6): 577–90. Review, 94(2): 377–81. Falk, Armin, and Urs Fischbacher. 2006. “A Theory of Grossman, Michele, and Wendy Wood. 1993. “Sex Dif- Reciprocity.” Games and Economic Behavior, 54(2): ferences in Intensity of Emotional Experience: A 293–315. Social Role Interpretation.” Journal of Personality Fehr, Ernst, and Klaus M. Schmidt. 1999. “A Theory of and Social Psychology, 65(5): 1010–22. Fairness, Competition, and Cooperation.” Quarterly Guth, Werner, Carsten Schmidt, and Matthias Sutter. Journal of Economics, 114(3): 817–68. 2007. “Bargaining Outside the Lab—A Newspaper Fehr-Duda, Helga, Manuele de Gennaro, and Renate Experiment of a Three-Person Ultimatum Game.”

03-Croson-472.indd 24 5/18/09 4:42:07 PM Croson and Gneezy: Gender Differences in Preferences 25

Economic Journal, 117(518): 449–69. Trust and Reciprocity.” http://socserv.mcmaster.ca/ Guth, Werner, Rolf Schmittberger, and Bernd econ/mceel/papers/TrustKMNS.pdf. Schwarze. 1982. “An Experimental Analysis of Ulti- Karau, Steven J., and Kipling D. Williams. 1993. “Social matum Bargaining.” Journal of Economic Behavior Loafing: A Meta-analytic Review and Theoretical and Organization, 3(4): 367–88. Integration.” Journal of Personality and Social Psy- Gysler, Matthias, Jamie Brown-Kruse, and Renate chology, 65(4): 681–706. Schubert. 2002. “Ambiguity and Gender Differ- Knight, Frank H. 1921. Risk, Uncertainty and Profit. ences in Financial Decision Making: An Experi- Boston: Hart, Schaffner and Marx; Boston and New mental Examination of Competence and Confidence York: Houghton Mifflin Company. Effects.” Swiss Federal Institute of Technology Cen- Kray, Laura J., Adam D. Galinsky, and Leigh Thomp- ter of Economic Research Working Paper 02-23. son. 2002. “Reversing the Gender Gap in Negotia- Harbaugh, William T., Kate Krause, and Steven G. tions: An Exploration of Stereotype Regeneration.” Liday. 2002. “Bargaining by Children.” University Organizational Behavior and Human Decision Pro- of Oregon Economics Department Working Paper cesses, 87(2): 386–410. 2002-4. Kray, Laura J., Leigh Thompson, and Adam D. Gal- Harshman, R. A., and A. Paivio. 1987. “‘Paradoxical’ insky. 2001. “Battle of the Sexes: Gender Stereo- Sex Differences in Self-Reported Imagery.” Cana- type Confirmation and Reactance in Negotiations.” dian Journal of Psychology, 41: 287–302. Journal of Personality and Social Psychology, 80(6): Hartog, Joop, Ada Ferrer-i-Carbonell, and Nicole 942–58. Jonker. 2002. “Linking Measured Risk Aversion to Kreps, David M. 1990. “Corporate Culture and Eco- Individual Characteristics.” Kyklos, 55(1): 3–26. nomic Theory.” In Perspectives on Positive Political Hinz, Richard P., David D. McCarthy, and John A. Economy, ed. James E. Alt and Kenneth A. Shep- Turner. 1997. “Are Women Conservative Inves- sle, 90–143. Cambridge; New York and Melbourne: tors? Gender Differences in Participant-Directed Cambridge University Press. Pension Investments.” In Positioning Pensions for Lerner, Jennifer S., Roxana M. Gonzalez, Deborah the Twenty-First Century, ed. Michael S. Gordon, A. Small, and Baruch Fischhoff. 2003. “Effects of Olivia S. Mitchell and Marc M. Twinney, 91–103. Fear and Anger on Perceived Risks of Terrorism: A Philadelphia: University of Pennsylvania Press. National Field Experiment.” Psychological Science, Holm, Hakan J. 2000. “Gender-Based Focal Points.” 14(2): 144–50. Games and Economic Behavior, 32(2): 292–314. Levin, I. P., M. A. Snyder, and D. P. Chapman. 1988. Holt, Charles A., and Susan K. Laury. 2002. “Risk “The Interaction of Experimental and Situational Aversion and Incentive Effects.” American Eco- Factors and Gender in a Simulated Ricky Deci- nomic Review, 92(5): 1644–55. sion-Making Task.” Journal of Psychology, 122(2): Houser, Daniel, and Daniel Schunk. 2007. “Fairness, 173–81. Competition and Gender: Evidence from German Levitt, Steven D., and John A. List. 2007. “What Do Schoolchildren.” Unpublished. Laboratory Experiments Measuring Social Prefer- Hyde, Janet S., Elizabeth Fennema, and Susan J. ences Reveal About the Real World?” Journal of Lamon. 1990. “Gender Differences in Mathematics Economic Perspectives, 21(2): 153–74. Performance: A Meta-analysis.” Psychological Bul- Lichtenstein, Sarah, Baruch Fischhoff, and Lawrence letin, 107(2): 139–55. D. Phillips. 1982. “Calibration of Probabilities: The Innocenti, Alessandro, and Maria Grazia Pazienza. State of the Art to 1980.” In Judgment under Uncer- 2006. “Altruism and Gender in the Trust Game.” tainty: Heuristics and Biases, ed. Daniel Kahneman, University of Sienna Department of Economics Paul Slovic, and Amos Tversky, 306–34. Cambridge Working Paper 5/2006. and New York: Cambridge University Press. Jianakoplos, Nancy Ammon, and Alexandra Bernasek. List, John A. 2006. “Field Experiments: A Bridge 1998. “Are Women More Risk Averse?” Economic between Lab and Naturally Occurring Data.” Inquiry, 36(4): 620–30. Advances in Economic Analysis and Policy, 6(2), Johnson, Blair T., and Alice H. Eagly. 1989. “Effects Article 8. of Involvement on Persuasion: A Meta-analysis.” Psy- Loewenstein, George F., Elke U. Weber, Christopher chological Bulletin, 106(2): 290–314. K. Hsee, and Ned Welch. 2001. “Risk as Feelings.” Johnson, J. E. V., and P. L. Powell. 1994. “Decision Psychological Bulletin, 127(2): 267–86. Making, Risk and Gender: Are Managers Difffer- Lundeberg, Mary A., Paul W. Fox, and Judith Punc- ent?” British Journal of Management, 5(2): 123–38. cohar. 1994. “Highly Confident but Wrong: Gender Kahn, Arnold, Joe Hottes, and William L. Davis. 1971. Differences and Similarities in Confidence Judg- “Cooperation and Optimal Responding in the Pris- ments.” Journal of Educational Psychology, 86(1): oner’s Dilemma Game: Effects of Sex and Physical 114–21. Attractiveness.” Journal of Personality and Social Mack, David, Paula N. Auburn, and George P. Knight. Psychology, 17(3): 267–79. 1971. “Sex Role Identification and Behavior in a Kanagaretnam, Kiridaran, Stuart Mestelman, Khalid Reiterated Prisoner’s Dilemma Game.” Psychonomic Nainar, and Mohamed Shehata. 2006. “The Impact Science, 24: 280–82. of Sex, Value Orientations and Risk Attitudes on Manning, John T., and Rogan P. Taylor. 2001. “Second

03-Croson-472.indd 25 5/18/09 4:42:08 PM 26 Journal of Economic Literature, Vol. XLVII (June 2009)

to Fourth Digit Ratio and Male Ability in Sport: Study on Gender Differences.” Economics Letters, Implications for Sexual Selection in Humans.” Evo- 99(3): 494–97. lution and Human Behavior, 22(1): 61–69. Seguino, Stephanie, Thomas Stevens, and Mark A. Marwell, Gerald, and Ruth E. Ames. 1981. “Econo- Lutz. 1996. “Gender and Cooperative Behavior: mists Free Ride, Does Anyone Else?: Experiments Economic Man Rides Alone.” Feminist Economics, on the Provision of Public Goods, IV.” Journal of 2(1): 1–21. Public Economics, 15(3): 295–310. Sell, Jane, W. I. Griffith, and Rick K. Wilson. 1993. Master, Robert, and Robert Meier. 1988. “Sex Differ- “Are Women More Cooperative than Men in Social ences and Risk Taking Propensity of Entrepreneurs.” Dilemmas?” Social Psychology Quarterly, 56(3): Journal of Small Business Management, 26: 31–35. 211–22. McIntyre, Matthew H. 2006. “The Use of Digit Ratios Sell, Jane, and Rick K. Wilson. 1991. “Levels of Infor- as Markers for Perinatal Androgen Action.” Repro- mation and Contributions to Public Goods.” Social ductive Biology and Endocronology, 4: 10. Forces, 70(1): 107–24. Mellström, Carl, and Magnus Johannesson. 2008. Selten, Reinhard, and Axel Ockenfels. 1998. “An “Crowding Out in Blood Donation: Was Titmuss Experimental Solidarity Game.” Journal of Eco- Right?” http://www.ne.su.se/research/enter/pdf/ nomic Behavior and Organization, 34(4): 517–39. johannesson.pdf. Sibley, S., S. Senn, and A. Epanchin. 1968. “Race and Migheli, Matteo. 2007. “Trust, Gender, and Social Sex of Adolescents and Cooperation in a Mixed- Capital: Experimental Evidence from Three West- Motive Game.” Psychonomic Science, 13: 123–24. ern European Countries.” Unpublished. Slonim, Robert. 2006. “Gender Selection Discrimina- Mui, Vai-Lam. 1995. “The Economics of Envy.” Jour- tion: Evidence from a Trust Game.” http://wsom- nal of Economic Behavior and Organization, 26(3): faculty.cwru.edu/slonim/Research%20August%20 311–36. 2005/Selection%20Discrimination%20Revision%20 Neu, Joyce, John L. Graham, and Mary C. Gilly. 1988. Slonim%20Sept%2027.pdf. “The Influence of Gender on Behaviors and Out- Slovic, Paul, Melissa L. Finucane, Ellen Peters, and comes in a Retail Buyer–Seller Negotiation Simula- Donald G. MacGregor. 2002. “The Affect Heuris- tion.” Journal of Retailing, 64(4): 427–51. tic.” In Heuristics and Biases: The Psychology of Niederle, Muriel, and Lise Vesterlund. 2007. “Do Intuitive Judgment, ed. Thomas Gilovich, Dale Grif- Women Shy Away from Competition? Do Men Com- fin, and Daniel Kahneman, 397–420. Cambridge pete Too Much?” Quarterly Journal of Economics, and New York: Cambridge University Press. 122(3): 1067–1101. Small, Deborah A., Michele Gelfand, Linda Babcock, Orbell, John, Robyn M. Dawes, and Peregrine and Hilary Gettman. 2007. “Who Goes to the Bar- Schwartz-Shea. 1994. “Trust, Social Categories, and gaining Table? The Influence of Gender and Fram- Individuals: The Case of Gender.” Motivation and ing on the Initiation of Negotiation.” Journal of Emotion, 18(2): 109–28. Personality and Social Psychology, 93(4): 600–613. Ortmann, Andreas, and Lisa K. Tichy. 1999. “Gen- Snijders, Chris, and Gideon Keren. 2001. “Do You der Differences in the Laboratory: Evidence from Trust? Whom Do You Trust? When Do You Trust?” Prisoner’s Dilemma Games.” Journal of Economic Advances in Group Processes, 18: 129–60. Behavior and Organization, 39(3): 327–39. Soll, Jack B., and Joshua Klayman. 2004. “Overconfi- Powell, Melanie, and David Ansic. 1997. “Gender Dif- dence in Interval Estimates.” Journal of Experimen- ferences in Risk Behaviour in Financial Decision- tal Psychology: Learning, Memory, and Cognition, Making: An Experimental Analysis.” Journal of 30(2): 299–314. Economic Psychology, 18(6): 605–28. Solnick, Sara J. 2001. “Gender Differences in the Ulti- Rabin, Matthew. 1993. “Incorporating Fairness into matum Game.” Economic Inquiry, 39(2): 189–200. Game Theory and Economics.” American Economic Solow, John L., and Nicole Kirkwood. 2002. “Group Review, 83(5): 1281–1302. Identity and Gender in Public Goods Experiments.” Rapoport, Anatol, and Albert M. Chammah. 1965. Journal of Economic Behavior and Organization, “Sex Differences in Factors Contributing to the 48(4): 403–12. Level of Cooperation in the Prisoner’s Dilemma Stuhlmacher, Alice F., and Amy E. Walters. 1999. “Gen- Game.” Journal of Personality and Social Psychol- der Differences in Negotiation Outcome: A Meta- ogy, 2(6): 831–38. analysis.” Personnel Psychology, 52(3): 653–77. Rottenstreich, Yuval, and Christopher K. Hsee. 2001. Sunden, Annika E., and Brian J. Surette. 1998. “Gen- “Money, Kisses, and Electric Shocks: On the Affec- der Differences in the Allocation of Assets in Retire- tive Psychology of Risk.” Psychological Science, ment Savings Plans.” American Economic Review, 12(3): 185–90. 88(2): 207–11. Schubert, Renate, Martin Brown, Matthias Gysler, and Tanaka, Tomomi, Colin F. Camerer, and Quang Hans Wolfgang Brachinger. 1999. “Financial Deci- Nguyen. Forthcoming. “Risk and Time Preferences: sion-Making: Are Women Really More Risk-Averse?” Experimental and Household Survey Data from American Economic Review, 89(2): 381–85. Vietnam.” American Economic Review. Schwieren, Christiane, and Matthias Sutter. 2008. Tedeschi, J. T., D. Hiester, and J. Gahagan. 1969. “Trust in Cooperation or Ability? An Experimental “Trust and the Prisoner’s Dilemma Game.” Journal

03-Croson-472.indd 26 5/18/09 4:42:08 PM Croson and Gneezy: Gender Differences in Preferences 27

of Social Psychology, 79(40): 43–50. in Tournament Competition.” Journal of Socio-Eco- Trivers, Robert L. 1972. “Parental Investment and Sex- nomics, 34(6): 834–49. ual Selection.” In Sexual Selection and the Descent Walters, Amy E., Alice F. Stuhlmacher, and Lia L. of Man, ed. Bernard Campbell, 136–79. Chicago: Meyer. 1998. “Gender and Negotiator Competitive- Aldine. ness: A Meta-analysis.” Organizational Behavior Van Huyck, John B., Raymond C. Battalio, and Mary and Human Decision Processes, 76(1): 1–29. F. Walters. 1995. “Commitment versus Discretion in Womak, D. F. 1987. “Cooperative Behavior by Female the Peasant–Dictator Game.” Games and Economic Negotiators: Experts or Masochists?” In Advances in Behavior, 10(1): 143–70. Gender Communication Research, ed. L. B. Nadler, Vandegrift, Donald, and Paul Brown. 2005. “Gender M. K. Nadler, and W. R. Mancillas, 219–40. New Differences in the Use of High-Variance Strategies York: University Press of America.

03-Croson-472.indd 27 5/18/09 4:42:08 PM