Gender differences in risk aversion:

A developing nations case ∗

Binay Kumar Adhikari
PhD student
The University of Alabama
Tuscaloosa, AL
bkadhikari@ua.edu
(205-348-7592)

Virginia E. O’Leary
Professor emerita
Auburn University
olearvi@auburn.edu

January 01, 2011

∗
Acknowledgements: Previous version of this paper was distributed as Kathmandu
University School of Management (KUSOM) working paper. The authors want to thank
KUSOM for partially funding this study and James Ligon, Michael Finke, the editor and
an anonymous referee for useful comments. Remaining errors are our own.

Electronic copy available at: http://ssrn.com/abstract=1522974

 Abstract

This study used Hanna and Lindamood (2004)’s graphic-based sur-

vey instrument to examine whether women who are employed in the
Nepalese banking sector show more risk aversion than men. Women
indeed reported their intention to take less risk and invested less of
their wealth in risky assets than men. However, the difference dis-
appeared after controlling for other relevant variables, notably their
perceived knowledge of financial markets. Our analyses suggested that
women demonstrated more risk aversion than men because they con-
sidered themselves to be less knowledgeable about financial markets.
Our findings support the need to educate female investors to increase
their confidence in their abilities to succeed in the world of finance.
Keywords: Gender Differences, Risk Aversion, Investment Risk

Electronic copy available at: http://ssrn.com/abstract=1522974

1 Introduction
Risk aversion is a concept in economics, finance, and psychology related to the
behavior of consumers and investors faced with uncertainty. In the context
of personal financial decisions, Dalton and Dalton (2008, p. 898, as cited
in Hanna, Waller & Finke, 2008) define risk tolerance as “the level of risk
exposure with which an individual is comfortable; an estimate of the level of
risk an investor is willing to accept in his or her investment portfolio.” Risk
tolerance is the inverse of risk aversion (i.e. a higher risk aversion implies
lower risk tolerance). Extensive empirical evidence suggests that women are
more risk averse than men. One study conducted by Weber, Blais and Betz
(2002) involved a survey to quantify five distinct risk domains: financial risks,
health and safety risks, recreational, ethical and social risks. Their results
indicated that women were more risk-averse in all domains except social risk.
Hinz, McCarthy and Turner (1997), Sunden and Surette (1998), and Olsen
and Cox (2001) found that increased risk aversion affects the investment
choices of women.
In this paper we examine gender differences in financial risk aversion
among the employees of the Nepalese banking sector using Hanna and Lin-
damood (2004)’s graphic-based survey instrument. To our knowledge, ours
is the first study of this kind in the context of a developing country. We
find that knowledge of financial markets and products explains the difference
in risk aversion more than the participants’ gender. Evidence of seemingly
higher risk aversion exhibited by women than men is muted once the par-
ticipants’ perceived level of knowledge of financial markets and products is
introduced as an explanatory variable.

2 Literature Review
Several studies have sought to determine whether the difference in risk pref-
erences between men and women found in the psychological literature trans-
lates into a difference in investment choices. Two of the early empirical
studies to identify women as more risk averse than men when making in-
vestment choices were conducted by Cohn, Lewellen, Lease and Schlarbaum
(1975) who obtained data from a mail questionnaire survey among customers
of a large retail brokerage firm, and Riley and Chow (1992) who used data

3
on actual asset allocation of a random sample of the US population. Later
studies by Hinz et al. (1997) and Bajtelsmit and Vanderhei (1997) specifi-
cally examined the pension choices of U.S. investors and concluded that, even
after controlling for income and age, women generally chose less risky pen-
sion fund options. Similarly, Sunden and Surette (1998) used household data
from Survey of Consumer Finances (SCF) of 1992 and 1995 and reported that
even after marital status was added to their list of control variables women
still chose less risky pension funds.
Using the 1983 SCF data, Hawley and Fujii (1993) found that education,
income and debt were positively related to risk tolerance. Married couples
and households headed by a single male were more risk tolerant than other-
wise similar households headed by a single female. Age was not statistically
significant in the analysis. Warner and Cramer (1995) explored the saving
behaviors of baby boomers and found similar results. Using the 1983 SCF
data, Sung and Hanna (1996) found that income and education were posi-
tively related to risk. Generally, risk tolerance decreased with age after 45.
Self-employed persons were significantly more willing to take financial risks
than those employed by others. In a theoretical paper Jaggia and Thosar
(2000) noted an inverse relationship between age and risk taking. However,
Sunden and Surette (1998) did not find any impact of age on risk preferences
in their empirical research.
A number of studies have investigated the role of knowledge/education
on risk taking. In a study using data from a national survey of nearly 2000
mutual fund investors’ investing behavior, Dwyer, Gilkeson and List (2002)
found the impact of gender on risk taking is significantly weakened when
investor knowledge of financial markets and investments is statistically con-
trolled. Other studies that find significant role of knowledge/education in-
clude those by Hawley and Fujii (1993), Sung and Hanna (1996) and Haliassos
and Bertaut (1995). For instance, Haliassos and Bertaut (1995) showed that
those who have not attended college are significantly less likely to hold stock
than those with at least a college degree. However, using US sample data of
household holding of risky assets, Jianakoplos and Bernasek (1998) found no
relationship between knowledge and risk taking.
Daly and Wilson (2001) suggested that the increased responsibilities ac-
companying marriage and children will make a man less tolerant of risk.
Supportive of this theory is the finding by Sunden and Surette (1998) that

4
marriage makes both men and women more risk averse in their choice of
pension plans. Similarly, Yao and Hanna (2005) found that risk tolerance
was highest for single males, followed by married males, then unmarried fe-
males, then married females. Säve − S öderbergh (2003) argued, however,
that marriage might encourage a couple to invest in riskier assets because
each person now has a second income stream insuring against the loss of
his or her own income. Furthermore, Säve − S öderbergh suggested that
marriage has the potential to affect the risk preferences of men and women
differently. Jianakoplos and Bernasek (1998) also found that the presence of
children significantly increases the risk tolerance of married couples in their
investments but significantly decreases the risk tolerance of single women.
Therefore, the literature indicates that marriage may significantly affect the
risk preferences of individuals, even though there are potentially offsetting
theories explaining the direction of such effects.
In summary, age, income, knowledge, and marital status are the four
variables most commonly controlled for in the literature concerned with the
relative risk preferences of men and women. Other variables, of course, may
also be influencing the results. For example, race has been shown to influence
risk perceptions (Flynn et al. 1994). Finucane et al. (2000) found white men
to be the most risk tolerant. Moreover, Riley and Chow (1992) noted that the
geographical location of an investor might influence risk preferences, although
they suggest that income is probably driving this result.

2.1 Rationale for Studying Risk Aversion in a Devel-

oping Country like Nepal
Studies related to gender and risk in the Western world are abundant. It is of
interest to explore whether the gender differences observed in the developed
world relate to the developing world. This becomes especially interesting
since studies including Asian population suggest that there exists an evident
cultural difference in overconfidence. For example, Hsee and Weber (1999)
found that Chinese were significantly more risk-seeking than Americans in the
investment domain. They explained this phenomenon in terms of “cushion
hypothesis” which suggests people in a collectivist society such as China
are more likely to receive financial help if they are in need, and they are
consequently less risk averse than those in an individualistic society such as

5
the USA. A few other studies led by Yates and his associates (e.g. Yates
et al., 1996) indicated that respondents in Asian cultures (e.g. in China
and Taiwan) exhibited markedly higher degrees of overconfidence than did
respondents in Western cultures (e.g. in the United States). Consequently,
the question that whether there are cultural differences in the structure of risk
aversion (such as gender difference) is pertinent. Despite Asia’s increasing
influence in the world’s trade and financial markets, very few studies on
risk aversion have been conducted in Asian countries and especially in south
Asia. In fact, Yao (2008) noted that risk aversion studies that include Asian
population are very limited. This paper attempts to fill this gap in the
literature.
Nepal is a developing country situated between the emerging economic
powers, India and China. All of these countries have an increasing presence of
woman in the workforce. Women, because they usually have lower working-
life incomes than men (Bajtelsmit & Bernasek, 1996), are likely to have less
wealth than men when they retire. The potential effect of this difference on
the retirement benefits of men and women is compounded by the fact that
women typically have longer life spans over which their retirement benefits
must be allocated and they also tend to retire earlier than men (Blondal
& Scarpetta 1998 as cited in Watson & McNaughton 2007). If women are
indeed more risk averse in their investment choices, this characteristic will
magnify the problems associated with their lower work-life incomes, lower
retirement ages, and longer life expectancies.
The importance of women as financial decision makers is highlighted by
a World Bank (2001)’s report that focused on gender issues and their broad
economic and social implications. This study concluded, among other things,
that it is critically important to take gender into account in the field of
social protection and the design of public programs. The positive effect of
increased household income on the childrens welfare - their education, health,
and nutrition - is stronger if that increase is controlled by, or channeled
through, the mother. There may be a case, from a development effectiveness
perspective, for targeting larger funds to women or designing investment
products and schemes targeted to the attitudes and values of women, who
are likely to be more productive mobilizers of their households’ resources
than men.
The worldwide literature has repeatedly found evidence for the greater fi-

6
nancial conservatism of women compared to men. A conservative investment
strategy results in less income on average than a more aggressive strategy
since higher returns generally come at the cost of higher risk. This relation-
ship between risks and returns implies that women, who choose to bear lower
risk, will generally earn lower returns - in the long run. Life expectancy at
birth has been increasing for both males and females in Nepal. It has in-
creased from 42 years for males and 40 years for females in 19711 to 62 years
for both males and females in 2005. Women’s life expectancy is projected
to exceed that of the men in the near future (World Health Report, 2005
and population projection for Nepal 2001-2021). As put by Bajtelsmit and
Bernasek (1996, p. 1) “Women’s greater longevity implies that, even with the
same investment strategy and pension accumulation, retirement wealth must
support a longer period of retirement. Women have lower lifetime earnings,
less growth in earnings, less wealth, less pension coverage and lower pension
participation rates.” To date, no empirical study of gender differences in
financial risk preference in Nepal has been conducted. This research was
undertaken to provide baseline information of value to those working to es-
tablish forward looking policies and procedures to accommodate the financial
interests and needs of working women and men.

2.2 Hypothesis
We hypothesize that after other factors (age, marital status, income, in-
vestment knowledge) known to influence an individual’s risk preferences are
controlled, Nepalese women will demonstrate, on average, more risk aversion
than Nepalese men.

3 Method
Hanna, Gutter and Fan (2001) observed that there are at least four methods
of measuring risk tolerance: asking about investment choices, asking a com-
bination of investment and subjective questions, assessing actual behavior,
and asking questions based on hypothetical scenarios. They argue that infer-
1
Source: Country Health System Profile, Nepal http://
www.searo.who.int/EN/Section313/Section1523 6868.htm

7
ring risk aversion based on observing actual portfolio allocations has many
limitations, including the fact that many households have no portfolio to
allocate so that nothing can be inferred about their risk aversion from their
allocation. We employed hypothetical questions because it has been shown
to be the firmest link to the theoretical concept of risk aversion (see e.g.
Hanna et al., 2001). Barsky, Juster and Shapiro (1997) showed that if the
respondent chooses to take an uncertain risk that could result in a decrease
in income (or a significant gain) instead of one that is certain although less
advantageous, the expected utility of the income resulting from the riskier
choice exceeds the expected utility of having the current income stream with
certainty.2

3.1 Instrument
We used a modified version of the Hanna et al. (2001) pension choice mea-
sure of risk aversion which is an improvement over Barsky et al. (1997) that
follows the expected utility model. The modified pension choice questions
(Hanna & Lindamood, 2004) include graphical illustrations to represent the
quantity of the change in the pension. These illustrations increase the re-
spondents’ understanding of the consequences of the hypothetical alternative
outcomes and thus more accurately estimate their true risk level.3 In addi-
tion to the series of pension choice questions, the survey also included the
2
Let U be the utility function and C be permanent consumption. An expected utility
maximizer will choose the 50-50 gamble of doubling lifetime income as opposed to having
it fall by factor 1 − λ if: .5U (2C) + .5U (λC) ≥ U (C)
Assuming the utility function is constant relative risk aversion (CRRA), the following
shows the relation between the Arrow-Pratt measure of relative risk aversion A and λ :
λ = (2−2(1−A) )[1/(1−A)] . This equation holds if A 6= 1, and λ = .5 when A = 1. Therefore,
by asking questions with different levels of λ, the Arrow-Pratt coefficient of relative risk
aversion can be directly estimated.
3
Illustrative of the questions contained in the questionnaire is the following:
Suppose that you are about to retire, and have two choices for a pension. Pension A gives
you an income equal to your pre-retirement income. Pension B has a 50% chance your
income will be double your pre-retirement income, and a 50% chance that your income will
be 20% less than your pre-retirement income. You will have no other source of income
during retirement, no chance of employment, and no other family income ever in the
future. All incomes are after-tax. Which pension would you choose?
Subsequent questions pose different percentage reductions in income. There were six
questions out of which the respondent is required to answer a maximum of four questions.
The respondent who accepts the possibility of largest cut in income for a possibility of

8
SCF Investment Risk question for comparison purposes.4

3.2 Participants
Two hundred and six employees of 12 banks and financial institutions in
Nepal participated in this study. The sample size reported varies in different
tables since participants were free to skip any questions they chose not to
answer. Employees of the banking sector were chosen because the questions
demanded a certain level of financial knowledge and a sense of probability
and risk. Participants could choose either Nepali or English version of the
questionnaire. MBA students of Kathmandu University pretested the instru-
ment.

3.3 Model and Variable Definition

Respondents’ subjective risk tolerance ranks were derived from the tolerance
questions and assigned a numeric rank in ascending order coded 1 (extremely
low), 2 (very low), 3 (moderately low), 4 (moderate), 5 (high), 6 (very high)
and 7 (extremely high) risk tolerance. Similarly, Investment Risk Tolerance
(IRT) responses were also obtained using Survey of Consumer Finances ques-
tions. Since both dependent variables of interest, SRT and IRT, are ordered
responses, an ordered probit regression model was applied for the analysis.
The central idea of the model is that, underlying the ordered response is
doubling the income gets the higher possible point (on a scale of 1 to 7) indicating an
extremely high subjective risk tolerance (SRT).
4
The Survey of Consumer Finances (SCF) is used to gather data on assets, liabilities,
financial attitudes, and financial behaviors of individuals and families. The SCF risk
assessment item has been widely used as a proxy for financial risk tolerance. The question
appears as follows: Which of the following statements on this page comes closest to the
amount of financial risk that you are willing to take when you save or make investments?
1. Take substantial financial risk expecting to earn substantial returns
2. Take above average financial risks expecting to earn above average returns
3. Take average financial risks expecting to earn average returns
4. Not willing to take any financial risks
(Note: In this paper we reverse coded the above responses (4 being substantial financial
risk to make it comparable to SRT measure)

9
a latent, continuously distributed random variable representing relative risk
aversion. Additionally, we also employed OLS regression as follows since OLS
estimates have a more straightforward interpretation and well-known model
diagnosis tools are available for it.

SRT = α + β(F EM ALE) + γ(AGE) + δ(M ARRIED) + θ(KN OW LEDGE)+

+λ(IN COM E) + τ (AGE ∗ F EM ALE) + φ(F EM ALE ∗ M ARRIED) + 
(1)

IRT = a + b(F EM ALE) + c(AGE) + d(M ARRIED) + f (KN OW LEDGE)+

+g(IN COM E) + h(AGE ∗ F EM ALE) + j(F EM ALE ∗ M ARRIED) + µ
(2)

where SRT is the respondents subjective risk tolerance rank derived from
Hanna and Lindamoods risk tolerance questions; coded from 1 for extremely
low to 7 for extremely high risk tolerance; IRT is investment risk tolerance as
it appeared in the SCF ranging from 1 to 4 (reverse coded to so that higher
number represents higher risk tolerance); FEMALE is a dummy variable
coded 1 for women and 0 for men; AGE is the respondents age range coded
1 (24 and below), 2 (25-29), 3 (30-34), 4 (35-39), 5 (40-44), 6 (45-49), 7
(50 and above); MARRIED is a dummy variable coded 1 for married and
0 for not married; INCOME is respondents monthly income range coded 1
(Up to Rs.19,999), 2 (Rs. 20,000-39,999) , 3 (Rs. 40,000-59,999), 4 (Rs.
60,000-79,999), 5 (Rs. 80,000-99,999), 6 (Rs. 100,000-119,999) and 7 (Rs.
120,000 and above); KNOWLEDGE is the respondents self-reported rating of
knowledge of investment market and products ranging from 1 to 7. 1= very
little knowledge, 7 = very good knowledge; AGE*FEMALE is the interaction
of age and sex; FEMALE*MARRIED is the interaction of sex and marital
status;  and µ are independent random disturbance terms.
In addition, the respondents also indicated an approximate percentage of
wealth they had invested in stock and business (STOCK) which we used to
tentatively assess their actual investment risk taking behavior.

10
4 Results and Discussion
Table 1 shows the frequency distribution and the means of all variables of
interest. About 37% of our respondents were females and 70% were married.
The average age was about 30 years and most made between Rs. 20,000 to
40,000 (about US $370 to $740, not adjusted for purchasing power parity)
per month. The average IRT and SRT scores were 2.64 and 4.71 respectively
both of which are on the upper half of the respective rating scales. Most of
the respondents indicated that they had invested some share of their wealth
in risky assets (stock and business). However, there were many (30) who
had not invested anything on stock or businesses. This is not uncommon in
Nepal where the stock market is small and stocks are not popular vehicles of
investment. Table 5 shows correlations among these variables.
Table 2 presents a quick comparison of the findings of Hanna et al. (2001)
and the present study. Two obvious differences emerge. First the average
relative risk aversion is lower for both males and females in this study. This is
primarily driven by more men and women reporting extreme risk tolerance.5
Second, unlike the results of Hanna et al. the difference in risk taking propen-
sity of male and female is statistically significant (p-value of chi-squared <
.01) in this study.

5
It seems unlikely that such a large portion of respondents (24.7%) had a true relative
risk aversion (RRA) of less than 1. RRA of 1 or less suggests that the household should
invest more than their wealth in risky assets and buy on margin (see e.g. Hanna and Chen,
1997)- an implication contrary to most empirical evidence. Precise calculation of RRA is
not critical for our purpose. However, for robustness, we repeated the analyses that follow
by discarding the observations with implied RRA of less than 1. Our key findings were
not different from the ones with full data. We, therefore, decided to report the results that
come from all available data. We thank a reviewer for pointing out this important issue.

11
 Table 1: Summary Statistics
IRT Freq. SRT Freq. Sex Freq. Age (years) Freq.
1 7 1 31 Male (0) 126 24 and below (1) 20
2 81 2 5 Female (1) 75 25-29 (2) 76
3 65 3 8 30-34 (3) 46
4 29 4 31 35-39 (4) 18
5 38 40-44 (5) 24
6 33 45-49 (6) 11
7 48 50 and Above (7) 6
Total 182 Total 194 Total 201 Total 201
Mean 2.64 Mean 4.71 Mean 0.37 Mean 3.03

Marital Status Freq. Knowledge Rating Freq. Monthly Income (in Rs.) Freq. %Invest Stock and Business Freq.

12
Not Married (0) 71 1 31 Up to Rs.19,999 (1) 54 None (0) 30
Married (1) 129 2 5 20,000-39,999 (2) 70 1 - 20% (1) 58
3 8 40,000-59,999 (3) 28 21 - 40% (2) 48
4 31 60,000-79,999 (4) 11 41 - 60% (3) 31
5 38 80,000-99,999 (5) 8 61 - 80% (4) 13
6 33 100,000-119,999 (6) 6 80 -100% (5) 4
7 48 120,000 and above (7) 13
Total 190 Total 194 Total 190 Total 184
Mean 0.7 Mean 4.35 Mean 2.57 Mean 2.73
Numbers in parenthesis represent coded values for the analysis. Means should be interpreted based on the
coded values
See the text (model and variable description section) for more details about the key variables. Totals do
not match because respondents were free to skip any question they did not want to answer.
 Table 2: Comparison of Preliminary Findings with Hanna et al. (2001)
Our sample∗ Hanna et al.(2001) ∗ ∗
Male Female Male Female
Risk Tolerance n % n % n % n %
Extremely High, accept 50% cut (A < 1.0) 41 33.3% 7 9.9% 0 0.0% 2 1.4%
Very High, reject 50% cut (1.0 ≤ A < 2.0) 22 17.9% 11 15.5% 12 5.8% 6 4.1%
Moderately High, reject 33% cut (2.0 ≤ A < 3.8) 17 13.8% 21 29.6% 48 23.2% 30 20.5%
Moderate, accept 10% cut (3.8 ≤ A < 7.5) 16 13.0% 15 21.1% 94 45.4% 60 41.1%
Low, accept 8% cut (7.5 ≤ A < 9.3) 4 3.3% 4 5.6% 14 6.8% 21 14.4%
Very Low, accept 5% cut (9.3 ≤ A < 14.5) 4 3.3% 1 1.4% 18 8.7% 10 6.8%
Extremely Low, reject 5% cut (A ≥ 14.5) 19 15.4% 12 16.9% 21 10.1% 17 11.6%
Entire sample 123 100.0% 71 100.0% 207 100.0% 146 100.0%
Mean Relative Risk Aversion 4.84a 5.72a 6.55b 6.88b
*This study: n = 194. ** Web Survey: n = 353. a. Weighted average based on midpoints of each range,
except value of 0.9 used for “Extremely High” category, and 16.0 used for “Extremely Low” category.

13
Chi-square for difference in male and female significant at .01. b. Weighted average based on midpoints of
each range, except value of 0.9 used for “Extremely High” category, and 16.0 used for
“Extremely Low category”. Notice, Hanna et al. say they used 16.0 used for “Very Low category” which is
a typo (We confirmed with one of the coauthors). Chi square statistic not significant.
 Table 3 reports the results of univariate OLS regressions for both SRT
(Panel A) and IRT (Panel B). The results suggested that sex and knowl-
edge were significantly associated with the level of risk tolerance (for both
SRT and IRT). In particular, women demonstrated the intention to take less
risk than men and participants with more perceived knowledge of investment
markets tend to take more risk than those with less knowledge. However,
the multivariate analysis in table 4 shows that the gender effect disappeared
after controlling for other variables. The effect of knowledge remained sig-
nificant for both SRT and IRT (from both OLS and ordered probit models).
The marginal effect (dy/dx) columns indicate that the probability of being
in the highest risk tolerance group (SRT=7, IRT=4) increased as knowledge
increased. Age was not significant for SRT but was weakly significant for
IRT and had the expected sign for IRT suggesting older employees intended
to take less risk than younger ones. We found no significant problem of
heteroscedasticity (chi-squared statistic is not significantly large) and mul-
ticollinearity (tolerance for none of the variables is less than .10) with the
OLS.

Table 3: Univariate Analysis of Subjective Risk Tolerance (SRT) Based on

Hypothetical Pension Gambles and Investment Risk Tolerance (IRT) Based
on SCF questions
Variables Coef. Std. Error t-stat. p>t
Panel A: SRT
SRT and FEMALE -0.625 0.303 -2.062 0.041
SRT and AGE 0.11 0.097 1.139 0.256
SRT and MARRIED -0.147 0.235 -0.625 0.533
SRT and KNOWLEDGE 0.538 0.125 4.322 0
SRT and INCOME -0.024 0.087 -0.281 0.779
Panel B: IRT
IRT and FEMALE -0.243 0.121 -2.013 0.046
IRT and AGE -0.022 0.039 -0.561 0.575
IRT and MARRIED 0.038 0.093 0.407 0.685
IRT and KNOWLEDGE 0.191 0.048 3.979 0
IRT and INCOME 0.011 0.034 0.316 0.752

Although our findings mainly suggest that only knowledge has a signifi-
cant impact on risk tolerance, the univariate analysis indicates that sex also
might also have impact on risk aversion. Thus, gender affects risk taking
propensity although not in the magnitude hypothesized based on the re-

14
view of the literature. Directionality of the coefficient is consistent with the
hypothesis: men tend to be more risk tolerant than women. Perceived knowl-
edge of the investment market and financial products has been shown to have
a significant impact on risk aversion. The positive sign of the regression co-
efficient suggests that risk tolerance increases with the increase in perceived
knowledge of investment related information.

Table 4: Multivariate Regression for Subjective Risk Tolerance (SRT) and

Investment Risk Tolerance (IRT)

Panel A: SRT OLS* Ordered Probit** Marginal Effect

(dy/dx, outcome = 7)a
Coef. t p>t Coef. z p>z Coef z p > |z|
FEMALE -0.3955 -0.53 0.598 -0.3076 0.75 0.452 -0.0883 -0.75 0.451
AGE 0.0165 0.13 0.9 0.0464 0.63 0.53 0.0133 0.63 0.529
MARRIED -0.4264 -1.39 0.165 -0.2973 -1.81 0.071 -0.0853 -1.81 0.07
INCOME -0.0193 -0.21 0.834 -0.0162 -0.32 0.751 -0.0046 -0.32 0.751
KNOWLEDGE 0.4819 3.48 0.001 0.2696 3.49 0 0.0774 3.55 0
AGE*FEMALE 0.0502 0.16 0.869 -0.0155 -0.09 0.926 -0.0045 -0.09 0.926
FEMALE*MARRIED -0.1203 -0.19 0.853 0.098 0.28 0.777 0.0281 0.28 0.777
CONSTANT 3.0821 4.24 0
* n = 177, F(7, 169)=2.93, P rob > F = 0.0064, R2 = 0.1082, Adj R2 = 0.071
** n = 177, LR χ2 (7) = 24.74, P rob > χ2 = 0.0008, Log likelihood = -310.42009, Pseudo R2 = 0.0383
a Corresponds to the adjacent ordered probit model

Panel B: IRT OLS* Ordered Probit** Marginal Effect

(dy/dx, outcome = 4)a
Coef. t p>t Coef. z p>z Coef z p > |z|
FEMALE -0.1291 -0.42 0.677 -0.2098 -0.46 0.647 -0.0449 -0.46 0.647
AGE -0.0952 -1.82 0.07 -0.1467 -1.91 0.057 -0.0314 -1.89 0.059
MARRIED 0.108 0.89 0.374 0.155 0.89 0.371 0.03316 0.89 0.371
INCOME 0.0005 0.01 0.99 -0.0009 -0.02 0.986 -0.0002 -0.02 0.986
KNOWLEDGE 0.1932 3.59 0 0.2938 3.61 0 0.0629 3.45 0.001
AGE*FEMALE -0.0435 -0.35 0.726 -0.0681 -0.37 0.709 -0.0146 -0.37 0.71
FEMALE*MARRIED 0.0418 0.17 0.869 0.0777 0.21 0.834 0.01663 0.21 0.835
CONSTANT 2.065 7.24 0
* n = 169, F(7, 161)=3.19, P rob > F = 0.0034, R2 = 0.1219, Adj R2 = 0.0837
** n = 169, LR χ2 (7) = 22.51, P rob > χ2 =0.0021, Log likelihood = -181.70485, Pseudo R2 = 0.0583
a Corresponds to the adjacent ordered probit model

The cross correlations presented in table 5 explain some of the findings

further. The significant positive relation between age and knowledge in-

15
dicates that older bank employees are more knowledgeable about financial
matters than their younger colleagues. Likewise, there is a positive relation-
ship between knowledge and each measure of risk tolerance, which suggests
that people with more investment knowledge endorse more risky options.
However, the direct relationship between age and risk tolerance is at most
weakly significant both in univariate and multivariate settings.

Table 5: Cross Correlations among Subjective Risk Tolerance (SRT), Invest-

ment Risk Tolerance (IRT) and other Key Variables
IRT SRT Female Age Married Knowledge Income
SRT .337**
(180)
FEMALE -.155* -.167*
(186) (199)
AGE -.031 .092 -.361**
(186) (199) (206)
MARRIED 0.038 -0.038 -0.157* 0.519**
(185) (198) (205) (205)
KNOWLEDGE .280** .284** -.359** .277** 0.145*
(181) (192) (199) (199) (198)
INCOME .024 -.021 -.045 .291** .355** .131a
(174) (183) (190) (190) (189) (185)
STOCK .180* .126a -.159* .049 -0.013 .272** .002
(169) (181) (188) (188) (187) (182) (175)
Numbers of observations are in parentheses
**Correlation is significant at the 0.01 level (2-tailed),
* at the 0.05 level (2-tailed), a at the 0.10 level (2-tailed)

Furthermore, both reported risk tolerance measures, the IRT and the
SRT, are positively related to the proportion of wealth kept in risky assets
(STOCK) i.e. stocks and direct investment in business. Those who rate
themselves as high risk takers invest a larger portion of their wealth in risky
assets (stocks and business) than those who rate themselves as more risk
averse. Table 6 shows that there is a significant difference in the proportion
of risky investment held by men and women. Specifically, women reported
to hold smaller proportion of their wealth in risky assets than men.
Women, who generally consider themselves less knowledgeable about fi-
nance than men exhibit greater risk aversion. However, the relationship is
stronger between knowledge and risk tolerance than between gender and risk

16
tolerance.6 It is, therefore, not surprising that in a review of a wide range
of literature (see e.g. Lyons et al., 2008) conclude that the research on risk
tolerance indicating women are more risk averse than men is far from clear.
All in all, our findings are consistent with the stream of literature which
concludes that knowledge, not gender, might be the key determinant of risk-
taking propensity. Consistent with this, Elder and Rudolph (2003) employed
the “final say” question in the Health and Retirement Study (HRS) to iden-
tify the sources of decision-making power within households. They found that
decisions were more likely to be made by the spouse with greater financial
knowledge, more education, and a higher wage, irrespective of gender.
We did not obtain significant interaction effect of either sex and marital
status or sex and age on risk tolerance. One reason for the insignificant re-
sult may be the small sample size to justify the interaction terms. Another
possibility is that traditionally in Nepalese households, like in most South
Asian households, husbands and/or fathers make most of the financial deci-
sions so wives and daughters remain less knowledgeable about finances and
less active in financial decision-making before and after marriage.

6
It is credible that knowledge is endogenous since a person’s attitude towards risk may
influence the willingness to acquire knowledge about financial assets. This endogeneity
would not allow us to establish a causal relationship between knowledge and risk-taking.
We realized that it is a very important issue. However, having obtained the data from
a survey, we did not have appropriate variables to do a rigorous test of endogeneity.
Nevertheless, we did experiment with the data available to us. For the regression with SRT,
we used age as the instrumental variable for knowledge and for IRT, used marital status
as the instrument. These variables were carefully chosen so that we have an instrument
which is correlated with knowledge but is not likely to be correlated with error terms and
does not have any independent explanatory power on the dependent variables (SRT and
IRT). Marital status is likely to be endogenous in western cultural context but unlikely to
be so in Nepalese context where most marriages are arranged and happen within a certain
age range. The Hausman test obtained no evidence of endogenity for our model.

17
Table 6: Independent Samples Test for Differences in Age and Stock Holdings
of Men and Women
Levene’s Test for t-test for Equality
Equality of Variances of Means
F Sig. t df Sig.
(2-tailed)
AGE Equal variances assumed 27.64 0.000 5.533 204 0.000
Not assumed 6.133 203 0.000
STOCK Equal variances assumed 2.27 0.000 2.127 182 0.0348
Not assumed 2.315 182 0.0217

5 Conclusion
The major finding of the current study, that women exhibit less financial risk
tolerance than men is apparently caused by a disparity in perceived knowl-
edge about investments. Our results are consistent with the stream of exist-
ing literature that argue that the difference in risk aversion is attributable to
task familiarly and knowledge rather than gender. In fact, Deaux and Em-
swiller (1994) and Beyer and Bowden (1997) have noted that women are less
confident and more risk averse in domains considered masculine, regardless
of their (equal) ability to perform. Men are represented in greater numbers
in financial markets than women (Merrill Lynch, 1996). In addition, women
are likely to be perceived as more conservative investors and therefore offered
less risky investments by brokers (Wang, 1994).
Psychological research has shown that men tend to be more overconfident
than women in areas like finance. Theory predicts that overconfident people
trade excessively. Barber and Odean (2001) tested this prediction using
account data from a discount brokerage firm and found that men traded
45% more than women. Trading reduced men’s (risky) net returns by 2.65
percentage points a year compared to 1.72 percentage points for women. Our
results from a developing nation reinforce the need of specific and targeted
financial education to help men and women with their investment decisions.
Men may need to be cautioned about the pitfalls of overconfidence, while
women may need guidance on how to make investment choices that carry a
calculated risk to obtain adequate growth.

18
 References

1. Bajtelsmit, V. L. & Bernasek, A. (1996). Why do women invest differ-

ently than men? Financial Counseling and Planning, 7, 1-10.

2. Bajtelsmit, V.L., & Vanderhei, J.L. (1997). Risk aversion and pension
investment choices. In M.S. Gordon, O.S. Mitchell, & M. M. Twinney
(Eds.), Positioning pensions for the twenty-first century (pp. 55-66).
Philadelphia: University of Pennsylvania Press.

3. Barber, B. & Odean, T. (1999). Boys will be boys: Gender overconfi-

dence and common stock investment. Quarterly Journal of Economics,
116, 261-292

4. Barsky, R.B., Juster, F.T., Kimball, M.S. & Shapiro, M.D. (1997).
Preference parameters and behavioral heterogeneity: An experimental
approach in the health and retirement study. Quarterly Journal of
Economics, 112 (2), 537-579.

5. Bernasek, A., & Shwiff S. (2001). Gender, risk, and retirement. Journal
of Economic Issues, 35(2), 345-356.

6. Beyer, S & Bowden, E. (1997). Gender differences in self-perceptions:

Convergent evidence from three measures of accuracy and bias. Per-
sonality and Social Psychology Bulletin, 23, 157-172.

7. Bruce, A. & Johnson, J. (1994). Male and female betting behavior:

new perspectives. Journal of Gambling Studies, 10, 183-98.

8. Byrnes, J.P., Miller, D.C. & Schfer, W.D. (1999). Gender differences in
risk taking: A meta- analysis. Psychological Bulletin, 125(3), 367-383.

9. Cohn, R. A., Lewellen, W.G., Lease, R.C., & Schlarbaum, G.G. (1975).
Individual investor risk aversion and investment portfolio composition.
Journal of Finance, 30(2), 605-620.

10. Daly, M., & Wilson, M. (2001). Risk-taking, intra-sexual competition,

and homicide. Nebraska Symposium on Motivation, 47, 1-36.

11. De Bondt, W.F.M. (1998). A portrait of the individual investor. Eu-

ropean Economic Review, 42(3-5), 831-844.

19
12. Deaux, K., & Emswiller, T. (1974). Explanations of successful perfor-
mance on sex-linked tasks: What is skill for the male is luck for the
female. Journal of Personality and Social Psychology, XXIX, 8085.

13. Dwyer, P.D., Gilkeson, J.H., & List, J.A. (2002). Gender differences in
revealed risk taking: Evidence from mutual fund investors. Economics
Letters, 76( 2), 151-158.

14. Elder, H. W., & Rudolph, P. M. (2003). Who makes the financial
decisions in the households of older Americans? Financial Services
Review, 12(4), 293-308.

15. Finucane, M.L., Flynn, J., Mertz, C.K., Satterfield, T.A. & Slovic, P.
(2000). Gender, race, and perceived risk: The ‘White Male’ effect.
Health Risk and Society, 2(2), 159-1 72.

16. Flynn, J., Slovic, P. & Mertz, C.K. (1994). Gender, race and perception
of environmental health risks. Risk Analysis, 14(6 ), 1101-1108.

17. Goldsmith, E. B., & Goldsmith, R. E. (1997). Gender differences in

perceived and real knowledge of financial investments. Psychological
Reports, 80(11), 236-238.

18. Goldsmith, E. B., Goldsmith, R. E., & Heaney, J.-G. (1997). Sex
differences in financial knowledge: A replication and extension. Psy-
chological Reports, 81, 1169-1170.

19. Grable, J. E., & Lytton, R. H. (1999). Financial risk tolerance revisited:
The development of a risk assessment instrument. Financial Services
Review, 8, 163-181

20. Hanna, S. & Chen, P. (1997). Subjective and objective risk tolerance:
Implications for optimal portfolios. Financial Counseling and Plan-
ning, 8(2), 17-25.

21. Hanna, S. D. & Lindamood, S. (2004). An improved measure of risk

aversion. Financial Counseling and Planning, 15 (2), 27-38.

22. Hanna, S.D., Gutter, M.S., & Fan, J. X. (2001). A measure of risk tol-
erance based on economic theory. Financial Counseling and Planning,
12 (2), 53-60.

20
23. Hanna, S. D., Waller, W., & Finke, M. (2008). The concept of risk
tolerance in personal financial planning. Journal of Personal Finance,
7 (1), 96-108

24. Haliassos, M. & Bertaut, Carol C, 1995. Why Do So Few Hold Stocks?,
Economic Journal, Royal Economic Society, 105(432). 1110-29

25. Hawley, C.B. & Fujii, E.T. (1993). An empirical analysis of preferences
for financial risk: Further evidence on the Friedman-Savage model.
Journal of Post Keynesian Economics, 16(2). 197-204.

26. Hersch, J. (1996). Smoking, seat belts, and other risky consumer deci-
sions: Differences by gender and race. Managerial and Decision Eco-
nomics, 17(5), 471-481.

27. Hinz, R. P., McCarthy, D. D., & Turner, J. A. (1997). Are women con-
servative investors? Gender differences in participant-directed pension
investments. In Mitchell, O. (Ed.) Positioning Pensions for the Year
2000 (pp. 91103). Philadelphia: University of Pennsylvania Press.

28. Hsee, C. K. & Weber, E. U. (1999). Cross-national differences in risk

preferences and lay predictions for the differences. Journal of Behav-
ioral Decision Making, 12, 165-179

29. Jaggia, S., & Thosar, S. (2000). Risk aversion and the investment
horizon: A new perspective on the time diversification debate. Journal
of Psychology and Financial Markets, 1(3-4), 211-215.

30. Jianakoplos, N.A., & Bernasek, A. (1998). Are women more risk
averse? Economic Inquiry, 36(4), 620-630.

31. Johnson, J. E.V. & Powell, P. L., (1994). Decision making, risk and
gender: Are managers different? British Journal of Management 5 (2),
123-38.

32. Lewellen, W., Lease, R., Schlarbaum, G. (1977). Patterns of investment

strategy and behavior among individual investors, Journal of Business,
50(3), 296-333.

33. Lyons, A. C., Neelakantan, U. & Scherpf, E. (2008). Gender and mar-
ital differences in wealth and investment decisions: Implications for

21
 researchers, financial professionals, and educators. Networks Financial
Institute Working Paper No. 2008-WP-02.

34. Merril Lynch (1996). You and your money. A financial handbook for
women investors, New York.

35. Olsen, R.A., & Cox, C.M. (2001). The influence of gender on the
perception and response to investment risk: The case of professional
investors. Journal of Psychology and Financial Markets, 2(1), 29-36.

36. Powell, M. & Ansic, D. (1997). Gender differences in risk behaviour

in financial decision-making: An experimental analysis. Journal of
Economic Psychology. 18, 605 628.

37. Riley, W., & Chow, K.V. (1992). Asset allocation and individual risk
aversion. Financial Analysts Journal, 48(6), 32-37.

38. Säve−S öderbergh, J. (2003). Essays on gender differences in economic

decision-making. PhD thesis, Stockholm University.

39. Sunden, A.E., & Surette, B.J. (1998). Gender differences in the alloca-
tion of assets in retirement savings plans. American Economic Review,
88(2), 207-211.

40. Sung, J. & Hanna, S. (1996). Factors related to household risk tol-
erance: An ordered probit analysis. Proceedings of the 42nd Annual
Conference of the American Council on Consumer Interests, 221-228.

41. Wang, P. (1994). Brokers still treat men better than women. Money,
23 (6), 108-10.

42. Warner, N. & Cramer, S. (1995). Saving behaviours. Journal of Con-

sumer Studies and Home Economics, 19, 57-67.

43. Watson, J., & McNaughton, M. (2007). Gender differences in risk

aversion and expected retirement benefits. Financial Analysts Journal.
63(4)

44. Watson, J., & Robinson, S. (2003). Adjusting for risk in comparing
the performances of male- and female-controlled SMEs. Journal of
Business Venturing, 18(6), 773-788.

22
45. Weber, E.U., Blais, A.R. & Betz, N.E. (2002). A domain-specific risk-
attitude scale: Measuring risk perceptions and risk behaviors. Journal
of Behavioral Decision Making, 15, 263290

46. World Bank (2001a). Engendering Development: Through Gender

Equality in Rights, Resources, and Voice. Policy Research Report:
Oxford University Press.

47. Yao, R. (2008). Financial issues of Asian Americans. In J. J. Xiao

(Ed.) Handbook of consumer finance research (pp. 303-316). NY, NY:
Springer.

48. Yao, R. & Hanna, S.D. (2005). The effect of gender and marital status
on financial risk tolerance. Journal of Personal Finance, 4 (1), 66-85.

49. Yates, J. F., Lee J. W., & Shinotsuka, H. (1996). Beliefs about overcon-
fidence, including its cross-national variation. Organizational Behavior
and Human Decision Processes, 65,138-147.

23