Gender and Risk-Taking

Economics, Evidence, and Why the Answer Matters

Julie A. Nelson

• Publisher: Routledge
• Year: 2017
• ISBN:
  9781351980401
• Edition: 1
• Topic: Business
• Subtopic: Business General
• Index: Business

The belief that men and women have fundamentally distinct natures, resulting in divergent preferences and behaviours, is widespread. Recently, economists have also engaged in the search for gender differences, with a number claiming to find fundamental gender differences regarding risk-taking, altruism, and competition. In particular, the idea that "women are more risk-averse than men" has become accepted as a truism. But is it true? And what are its causes and consequences?

Gender and Risk Taking makes three contributions. First, it asks whether the belief that men and women have distinct risk preferences is backed up by high quality empirical evidence. The answer turns out to be "no." This leads to a second question: Why, then, does so much of the literature claim to find evidence of "difference"? This, it will be shown, can be attributed to biases arising from too-easy categorical thinking, widespread stereotyping, and a tendency to prefer results that are publishable and that fit one's prior beliefs. Third, the book explores the economic implications of the conventional association of risk-taking with masculinity and risk-aversion with femininity. Not only fairness in employment, but also the health of the financial sector and national responses to climate change, this book argues, are being compromised.

This volume will be eye-opening for anyone interested in gender, decision-making, cognition, and/or risk, especially in areas relating to employment, finance, management, or public policy.

PART I
To understand the answer, you first have to have a clear question

1
THE BETTER QUESTION

How much different and how much similar

Consider these two statements:

A. “Women are more risk-averse than men.”

B. “In our sample, we found a statistically significant difference in mean risk-aversion between men and women, with women on average being more risk-averse.”

Which statement is, on the face of it, simpler and more appealing? The first, of course. But it may be profoundly misleading.

Categorization and the gender binary

Our brains like to categorize. We like to sort things, and judge things. We have to sort things, and judge things, to get through the day! We sort food from non-food, poison from non-poison. We judge good versus bad, less versus more.

When we think of things as different from each other, we tend to slot them into opposing, non-overlapping categories. Meanwhile, we have a tendency to downplay the differences within each category, finding it easy to slip into thinking of all of a category’s members as more similar than perhaps they are. (Sometimes being poisonous, for example, is a matter of degree.) And we have a tendency to gravitate toward thinking that there is something essentially, naturally, fundamentally correct about what we are doing. We are shocked, for example, when we find another culture thinking of something we classify as clearly “non-food,” such as dogs, as “food.” But such habits of thinking, while helpful for making snap judgments, can also get in the way of making wise and informed judgments.

Sex and gender categories are some of our most basic: Children learn at a young age to classify people as male or female. But these categories have also been topics of much contention lately. As I write, newspapers in the United States are full of stories about battles over which restrooms transgender people should use. Many citizens, activists, and scholars across the humanities, sciences, and social sciences have urged greater attention to the experience of individuals who do not ascribe to the male/female category they were assigned at birth, but rather identify as gender-queer, transsexual, transgender, or simply none-of-the-above. Many are becoming increasingly aware that the identities of cis male and cis female (living in the same gender identity as one was born with) do not exhaust the possibilities. The accusation from conservative communities that other ways of being are “unnatural” only confirms the fervor with which the two-way, male/female, gender categorization is often held.

This book, however, while not denying the significance of these issues, attempts to expand thinking about sex and gender in a different way. In the academic literature that will be reviewed in this book, the researchers ask people who volunteer to take a survey or participate in an economic experiment to identify themselves as either male (or man, or boy) or female (or woman, or girl). A research participant who refused to self-identify as male or female would simply be dropped from the investigation. The researchers also tend to assume that self-reported sex is the same as the sex assigned at birth, and that this lines up with gonadal (penis versus ovaries) and chromosomal (XX versus XY) sex. In order to question the point that this literature has been making, I will, in this book, also use language that assumes an uncomplicated male/female binary at the level of genes, sex organs, and self-identification. Suppose we go along with the idea of looking only at people for whom this is true. Would it mean that we need to also accept a simple male/female binary at the level of behavior? Do men face risks one way, while women treat them differently?

Going back to the two statements above, the easy Statement A, “Women are more risk-averse than men,” is how social science research on gender differences in risk preferences gets summarized in the media, and often also by the researchers themselves. But the more difficult, wordy, and technical Statement B represents what (some) research actually finds. One problem that arises is that, while Statement A might be intended as a simple shortcut version of Statement B, what we tend to think and communicate when we use an expression like Statement A implies something quite different. Another problem is that Statement B, while longer and more detailed, still really doesn’t answer the interesting questions we’d like to ask about men, women, and their similarities and differences.

This chapter explains, first, what it means to talk about categorical versus “on average” differences. With this background, we will be prepared to explore what Statement B means, and why it is misleading in the absence of more information about how much different, and how much similar are the genders. In the next chapter, literatures from cognitive psychology, philosophy, and linguistics will be called on to explain how easy statements like Statement A are generally understood, and why—in spite of clear evidence to the contrary—they can be so much more appealing.

Distinguishing between individual-level and group-level “difference”

Let’s let “gender difference at the individual level” (henceforth, GDI) mean a difference that categorically distinguishes between men and women. All men, and only men, share a particular characteristic. All women, and only women, share some different characteristic. The trait “number of Y chromosomes per cell,” for example, could be taken as one such categorical classifier, if we assume a dichotomous, genetic definition of sex. Men have one Y chromosome per cell and women have zero.¹ Figure 1.1 illustrates this GDI. The statement “men have a higher number of Y chromosomes than women” is perfectly (and tautologically) valid in this case. Furthermore, any pairwise comparison of randomly selected individual men and women will confirm the statement. A pure GDI that is unequivocally sex-based must hold for all males and females from the entire human population.

Aside from chromosomes and genitalia, such distinctions might be thought to be rare. However, the focus of this book is on what can be seen in empirical data, and a GDI at times may also be clearly statistically identifiable even if it is not universal. This is because cultural groups tend to adhere to particular sex norms and put children through sex-differentiated socialization. For example, in a highly gender-differentiated society, one may be able to categorize an adult individual by sex, with an extremely high degree of reliability, merely by observing variables such as hair length, type of apparel, or type of work (e.g., paid occupation or homemaker)—or conversely, accurately predict an individual’s hair length, clothing, or type of work based on the individual’s sex. For example, Figure 1.1 would work in some business settings if instead of asking about chromosomes we asked about neckties: We would observe that men wear one and women wear zero. So even if statistical GDI at the broadest human species level may be a relatively special case, at the level of specific societies and contexts it can be far from trivial.²

FIGURE 1.1 An Example of GDI (Gender Difference at the Individual Level).

Let’s let “gender difference at the aggregate level” (henceforth, GDA) mean a difference that can only be found at the aggregate, group, or distributional level. These are differences that won’t be always or immediately confirmed by a pairwise comparison of a man and a woman. For example, you might know that more of your women friends than your men friends like to dance, but you also have some male friends who like dancing. Or you know that men are on average taller than women, even though you can think of pairs of your friends where this is not true.

Suppose you do a survey of your friends, asking them if they like to dance. Suppose 70% of the women answer that they like dancing, while the rest say they do not. Suppose 40% of the men say they like dancing, and the rest say they do not. Figure 1.2(a) illustrates how these answers can be used to make a simple chart. Note that the chart is a bit like Figure 1.1 in that there are only two possible answers: “no” or “yes” in this chart, similar to “0” or “1” in the earlier figure. But notice that the men’s and women’s bars no longer reach 100%. Figure 1.2(b) illustrates another case of GDA, but one where there is a greater variety of possible answers. Let’s suppose you collect data on the heights of twenty of your friends, and draw a chart putting an X ...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Dedication
• Table of Contents
• List of figures
• List of tables
• Acknowledgments
• Introduction
• Part I To understand the answer, you first have to have a clear question
• Part II Evidence about risk behavior: little difference, much similarity
• Part III Evidence about stereotyping and confirmation bias: rampant
• Part IV Why it matters
• Conclusion
• Index