By Gianluca Grimalda, Senior Researcher at the Kiel Institute for the World Economy, member of the Taskforce on Social Cohesion at the ThinkTank20 group for the G20 2021 meeting, and member of the Trustlab OECD initiative
Why are women from Western countries more in favour of income redistribution than men, and why are they voting in larger numbers for left-wing parties? For instance, 55% of women backed Joe Biden in the recent U.S. elections, while only 46% of men did, according to a nationwide poll. Are women intrinsically more generous and sensitive to social issues than men, or do they believe that voting for left-wing parties will advance their cause against discrimination and unfair treatment? In a recently published article, we argue that part of the reason for this behaviour is eminently psychological, and revolves around men’s higher degree of self-confidence than women1.
The research group of which I was privileged to be part of, assembled data from two studies that had experimentally investigated the underlying factors of preferences for redistribution. Groups of 21 participants gathered at computer rooms of eight universities around the world – three in the U.S., two in Germany and Italy, and one in Norway. Each participant was assigned a different level of initial earnings. In some of the U.S. sessions, such a distribution mirrored real-life countrywide income distribution, with the one individual at the top of the earnings scale being assigned $100 while the individual at the bottom was assigned $0.11. In other sessions, the inequality gap between the top earner and the bottom earner was of 21 to 1 and the earnings distribution was uniform. Each participant was then asked how much redistribution she would desire for the group. The redistribution rate could range from 0% – in which case everyone would take home their initial earnings- to 100%, in which case everyone would go home with the same amount as anyone else. This decision was our measure of preference for redistribution.
Our experimental methodology thus enabled us to recreate within a university room a situation of inequality mirroring those occurring in real life, and to see how demand for redistribution varied upon the variation of conditions of choice. As is standard in experimental economics, people were paid for the results of their decisions. In our case, the choice of one randomly selected “decisive individual” was applied to everyone else. Participants made several decisions under differing conditions on their earnings’ determinants. In some conditions, earnings were determined by luck, in some others by the relative performance in some tasks. Typically, demand for redistribution is lower when earnings are determined by individual merit than by luck, which demonstrates individual preferences to reward individual merit.
“A gender gap in self-confidence explained around 50% of the difference in preference for redistribution between men and women, while risk aversion could account for 12%.” #DevMattersTweet
Overall, we found that women demanded more earnings redistribution than men, in line with existing evidence. Nevertheless, the gender gap disappeared both when initial earnings were known with certainty, and also when initial earnings were assigned on the basis of luck rather than merit. In other words, gender differences in redistribution only emerged in the conditions when initial earnings were uncertain and determined by relative merit. We dug deeper into this result. We compared two possible underlying factors. One is self-confidence, that is, the belief over one’s own abilities. Another is risk aversion, that is, a preference for safe gains over uncertain ones – even when the probabilistic expected payoffs from a lottery exceed the safe gains.
Among these two factors, we found that a gender gap in self-confidence explained around 50% of the difference in preference for redistribution between men and women, while risk aversion could account for 12%. Were men justified in having higher self-confidence than women? Not really. Overall, there was no significant difference in performance in our tasks between genders. Both genders expected, on average, to perform better than what they actually did – a common finding in the experimental social sciences that is termed “over-confidence”– but men did so disproportionately more than women. Our statistical analysis controlled in any case for actual individual performance.
Our answer to the question of why women tend to demand more redistribution than men is then the following. When future income is uncertain, people use redistribution as an insurance mechanism to protect their income becoming too low if things go badly, particularly because of poor performance. Since men are more self-confident than women, they feel less need to use redistribution for this purpose. Women are more realistic in their expectations, and for this reason demand more redistribution.
Other studies found that gender gaps in self-confidence also mattered for labour market outcomes. In a study conducted by a member of our research group, female high-school students refrained from competing with other students in “winner-takes-all” tournaments, but preferred to be paid piece-rate for their effort, a result due to lack of self-confidence. Interestingly, those who opted out of competition in the experiments were also less likely to choose math and sciences-intensive university degrees. Since these degrees normally lead to higher salaries, this self-selection reverberates into final income differences. In another study, gaps in self-confidence predicted gender wage gaps even among graduates from maths and sciences-intensive degrees. Hence, gaps in self-confidence seem to predict both women’s tendency to select out of university degrees later commanding higher salaries, and the tendency to earn less even when graduating from these sectors.
“Women are more realistic in their expectations, and for this reason demand more redistribution.” #DevMattersTweet
Understanding the nature of gender gaps in the labour market is important not only for reasons of equity, but also for development, because if women’s skills and talents are not fulfilled in the labour market, the economy will not achieve its full potential. Citizens’ demand to address gender discrimination in the labour market and in society is mounting. Our findings contribute to the debate showing that such differences are partly due to basic psychological differences across genders.
Such psychological differences may in turn be rooted in gender biases in the upbringing process, or could be the result of “hard-wired” psychological attitudes inherited from our evolutionary past. It is very difficult to tell these factors apart. A pioneering study found that gender differences in competitiveness are substantially reduced in a matrilineal society compared to a patrilineal one, thus supporting the idea that “nurture” matters more than “nature” for gender differences. Nevertheless, another study did not replicate this result with respect to risk aversion in another pair of matrilineal and patrilineal societies.
Even admitting that gender differences in self-confidence were due to nurture, would then open the question how societies should try to address these differences. Since it is men who appear to be off the mark in over-estimating their capacities, a natural solution would seem to be to tone down their excessive over-confidence. But will stifling men’s overconfidence reduce entrepreneurial spirits and capacity to innovate? Some (moderate) degrees of overconfidence may be necessary to keep “animal spirits” alive in churning out inventions and innovations. Should we then try to increase women’s self-confidence, even at the cost of making their expectations less in line with reality? Or should we not interfere at all with self-confidence? These are complex issues, for which hard evidence and ethical inquiries are necessary, and public debate is needed.
1.↩ Most of the evidence in this piece was originally published in Buser, T., Grimalda, G., Putterman, L., & van der Weele, J. (2020). Overconfidence and gender gaps in redistributive preferences: Cross-Country experimental evidence. Journal of Economic Behavior & Organization, 178, 267-286