Gender Differences in Risk Attitudes: Field Experiments on the Matrilineal Mosuo and the Patriarchal Yi
Total Page:16
File Type:pdf, Size:1020Kb
Journal of Economic Behavior & Organization 83 (2012) 59–65 Contents lists available at ScienceDirect Journal of Economic Behavior & Organization j ournal homepage: www.elsevier.com/locate/jebo Gender differences in risk attitudes: Field experiments on the matrilineal Mosuo and the patriarchal Yi a,∗ b,1 Binglin Gong , Chun-Lei Yang a School of Management, University of Fudan, Shanghai 200433, China b Research Center for Humanities and Social Sciences (RCHSS), Academia Sinica, Taipei 115, Taiwan a r t i c l e i n f o a b s t r a c t Available online 22 June 2011 We conduct experiments on two different risk tasks with subjects from two neighboring ethnic groups, the matrilineal Mosuo and the patriarchal Yi in China. Women are more risk averse than men at both tasks within both ethnic groups. However, the gender gap is smaller JEL classification: in the Mosuo. Regressions show that socio-economic factors such as family size, family head, C93 education, age, and income also have significant effects on subject’s risk choices. D81 © 2011 Elsevier B.V. All rights reserved. J15 J16 Keywords: Risk attitude Gender difference Matrilineal society Patriarchal society Field experiment 1. Introduction Do men and women systematically behave differently in social and economic decision making? Stylized facts from evidence gained in modern cultures paint a general picture of women as the more risk-averse, more socially oriented, 2 more selfless, and less competitively inclined gender. There also has been considerable debate on whether these gender differences are attributable to nature or nurture, or some combination of both. In other words, are the differences primarily attributable to the natural differences in genes between the sexes? As an alternative hypothesis, is it sensible to argue that gender differences are culture specific and determined by the different social and economic functions men and women fill 3 in a society? However, modern societies may not display enough variations in the relative role differences between men and women to solve this debate convincingly, in spite of otherwise huge differences in cultural and political characteristics and economic conditions. Against this background, Gneezy et al. (2009) find that the gender difference in competitive inclination is reversed in a matrilineal society, the Khasi in India, compared to the Maasai in Tanzania, a traditional patriarchal society. Subjects in their study were given a choice to either partake in a ball-throwing game without competition, with 10 attempts and each basket scored yielding one dollar, or to compete with another anonymous person from the same village playing the same ∗ Corresponding author. Tel.: +86 21 55664201. E-mail addresses: [email protected] (B. Gong), [email protected] (C.-L. Yang). 1 Tel.: +886 2 2789 8161; fax: +886 2 2785 4160. 2 See Croson and Gneezy (2009) and Eckel and Grossman (2008) for thorough surveys from an experimental economist’s viewpoint. 3 Gneezy et al. (2009), p. 1644ff, has an excellent discussion on this nature-nurture debate, in association with the discussion of their data. 0167-2681/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jebo.2011.06.010 60 B. Gong, C.-L. Yang / Journal of Economic Behavior & Organization 83 (2012) 59–65 game, which yields three dollars for each scored basket in a win and no money at all in a loss. While men are more inclined to compete in Maasai like in most other modern patriarchal Western societies, Khasi women are more competitively inclined than men and are even weakly more so than Maasai men. Recently, Gong et al. (2010) have also found a reversed gender difference in the dictator decision in the Mosuo in China, one of the few remaining matrilineal and matrilocal societies in the world. Mosuo men exhibited more selfless behavior by giving more than twice the amount Mosuo women do to unknown others. All these suggest that the gender differences cannot be unequivocally attributed to nature in general. Does the same apply to other aspects in gender differences, for example, risk attitude? Since many relevant economic and strategic decisions have a risk aspect, it has wide applications in various scenarios. In the current study, we investigate gender difference in risk attitudes in two neighboring traditional societies in southwestern China, the matrilineal Mosuo and the patriarchal Yi. The main risk test we use is the standard investment risk (IR) problem as introduced by Gneezy and Potters (1997), where subjects choose what portion of their endowment of 100 experimental dollars to bet in a lottery, which returns three times the bet with a 50 percent chance and nothing otherwise. Charness and Gneezy (2012) find that the stereotype gender difference has been consistently observed in all studies conducting IR among Western subjects, but also conclude that more research into the boundaries of these findings is needed. Gneezy et al. (2009) ran the IR experiment as control to their competition game and found no significant gender differences, though there is a significant ethnicity difference, with 4 the matrilineal Khasi being less risk averse. Using the risk measure developed by Holt and Laury (2002), Cárdenas et al. (2012) find a smaller gender gap regarding risk attitude in Sweden than in Columbia, which suggests that gender inequality 5 in social and economic status may be a crucial determinant for the stereotype gender difference. Besides the investment risk decision, our subjects also have to make a choice in the compound risk (CR) problem, which is structurally congruent to the competition decision in Gneezy et al. (2009). Instead of throwing 10 balls in the competition game, which requires some skill and effort in addition to luck, subjects have to make 10 independent random draws from 4 cards of different suits. They decide whether to get paid on the number of diamond drawn or to compete with the experimenter at the same task. In competition, they get nothing in case of losing, but three times the simple-gamble payoff in case of winning, and the same amount in case of a tie. Due to design, CR is best suited to separating risk-attitude effect from the competitive personality effect in the original competition game, against the objections that the competition decision 6 may merely reflect one’s risk preference. After both IR and CR decisions are made, a fair coin flip determines which of them will be carried out for each subject. The chosen gamble will be performed according to the subject’s choice. We find that women are more risk averse than men in both risk tasks within both the matrilineal and patriarchal societies. This is consistent with findings in most studies, as surveyed by Charness and Gneezy (2012), Croson and Gneezy (2009) and 7 Eckel and Grossman (2008). Notably, the gender gap among the Mosuo appears to be smaller than among the Yi, which is consistent with findings in Cárdenas et al. (2012). Women from the matrilineal society of the Mosuo are more likely to compete than women from the Yi. For men no such difference exists. We also ran regressions to control for different additional socio-economic variables based on the extensive exit polls collected. We find that factors like family size, being family head, education, income and age have significant effects on risk attitude and can substitute for the gender and ethnicity effects to certain extent. In Section 2, we summarize the ethnic background of our field study and a demographic survey for participants. Our exper- imental design and implementation details will be described in Section 3. Data analysis is in Section 4. Further discussions and interpretations can be found in the last section. 2. Our subject pool – Mosuo and Yi people We conducted our experiments in two neighboring societies, the matrilineal Mosuo and the patriarchal Yi, in Yunnan 8 9 province of southwestern China. It is advantageous that these peoples are genetically closely related, plus they live in separate villages near each other and thus are comparable in terms of their income level, economic activities, geographical and cultural environment. Moreover, there are almost no cross-marriages between these two ethnic groups, primarily due to the huge differences in their cultures based on inheritance principles. Mosuo and Yi children do not attend the same elementary schools, but possibly the same secondary schools if they stay in school this far. 4 See Croson and Gneezy (2009) and Eckel and Grossman (2008) for surveys on gender differences in a wider range of risk experiments. Majority of studies have results consistent with the stereotype of women being more risk averse. However, Holt and Laury (2002), Moore and Eckel (2003), and Schubert et al. (1999) discuss cases where this is not true. 5 As to exploring the boundaries by varying ethnicities, Henrich and McElreath (2002) and Kuznar (2001) are further cross-ethnicity studies on risk attitude. 6 For this function, we also called it the competition risk problem in a previous version. 7 It is clearly different to Gneezy et al. (2009) where no gender differences are observed in IR. However, we confirm their observation that the matrilineal subjects are less risk averse than the patriarchal ones in IR. 8 Our Mosuo subjects are from Yongningin Ninglang County and our Yi subjects are from villages near Yongning. We also intentionally chose to conduct the experiments in remote areas away from the tourist places to minimize influences of the Han people, and of the modern society in general. 9 See Wen et al. (2004) for a comprehensive genetic-marker study regarding the mtDNA characteristics, with further discussions of ethnic migrations. See Gong et al. (2010) for further discussions and references on their common historical and anthropological background.