C 2016 Board of Trustees of the Bulletin of Economic Research and John Wiley & Sons Ltd
Bulletin of Economic Research 69:3, 2017, 0307-3378

THE RISK PREFERENCES OF ENTREPRENEURS

IN INDONESIA

Kitae Sohn
Konkuk University

ABSTRACT

Entrepreneurship is known to be important for innovation and economic growth, but relatively little
attention has been paid to entrepreneurs in developing countries. We examined an important aspect
of entrepreneurship: risk-taking. We analysed the Indonesian Family Life Survey and compared
risk preferences between entrepreneurs and non-entrepreneurs. We found that risk tolerant workers
were more likely to be entrepreneurs than risk averse workers by about 5 percentage points, or about
20 percent of the proportion of entrepreneurs in the labour force. The results imply that not all
entrepreneurs in Indonesia are pushed into the sector; some actively take risks and seek innovations.
Keywords: risk aversion, entrepreneurship, self-employment, developing country, Indonesian
Family Life Survey
JEL classification numbers: D81, M13, O53

I. INTRODUCTION

Praise for entrepreneurship abounds in the literature. For example, Acs et al. (2008, p. 219)
claimed that entrepreneurship is the driving force behind economic development through em-
ployment, innovation, and welfare effects. Many governments, correctly or not, agree and offer
generous subsidies to entrepreneurs. The importance of entrepreneurship has compelled re-
searchers to investigate various characteristics of entrepreneurs. One such characteristic is risk
aversion. Because this characteristic is fundamental to economic decision-making, it has long
been discussed in the field of entrepreneurship. Knight (1921) extensively discussed the risk-
taking behaviour of entrepreneurs, while Schumpeter (1942) highlighted the importance of
risk-taking for innovation and economic growth. Kihlstrom and Laffont (1979) provided a the-
ory, arguing that less risk averse (or more risk tolerant) individuals select into entrepreneurship
whereas more risk averse individuals enter into riskless paid employment. While they assumed
risk preferences to be exogenous, Vereshchagina and Hopenhayn (2009) endogenized the risk
taking behaviour of entrepreneurs in a simple dynamic occupational choice model.
This relationship between risk tolerance and entrepreneurship may be generalized to the
developing world, but there is no guarantee for such a facile generalization. The theoretical

Correspondence: Department of Economics, Konkuk University, 120 Neungdong-ro, Gwangjin-gu, Seoul,

05029, South Korea. Email: [email protected]. I am grateful to Taehyun Ahn, Jungmin Lee, and the two
anonymous reviewers for helpful comments and suggestions.

271
272 Bulletin of Economic Research

works assumed that entrepreneurs set up firms to exploit perceived business opportunities
under ‘good’ institutions. And Vereshchagina and Hopenhayn (2009) assumed that relatively
poor agents become paid employees and the rich become entrepreneurs. These assumptions
may be plausible for the developed world but are not for the developing world (Parker, 2009).
In the developing world, paid employment particularly in the public sector is clearly preferred
to entrepreneurship (Banerjee and Duflo, 2011, ch. 9), and relatively poor agents become
entrepreneurs, not paid employees. Thus, workers may even take risks to become paid employees,
while the remaining workers left behind become entrepreneurs. Rent-seeking in the public sector
illustrates this point. In our country of interest, Indonesia, corruption has been rampant. Suharto,
the dictator in 1967–88, was at the centre of corruption, and his legacy of corruption in the
executive, legislature, and judiciary has stubbornly remained influential to date (Setiyono and
McLeod, 2010; Butt, 2011). When a large amount of up-front bride is required to join this
rent-seeking enterprise and the position acquired depends on the whim of the bribe-taker, risk-
takers may play this gamble. On the other hand, a small scale of entrepreneurship in Indonesia
means that a large amount of up-front investment is not required (Sohn, 2015a; Sohn and Kwon,
2016); such an investment is small enough to be said divisible. Hence, the question of whether
entrepreneurs are more risk tolerant than paid employees in the developing world is an empirical
question.
The next section explains that most empirical studies agree that entrepreneurs in the developed
world are less risk averse, thereby supporting the theoretical works. Little light, however, has
been shed on those in the developing world regarding risk-taking. This omission is unfortunate,
given that entrepreneurs in the developing world account for a substantial proportion of the
workforce. In addition, understanding their risk preferences has policy implications because
risk-taking and the following innovations are public goods to some extent. The subsidies and tax
incentives meant for entrepreneurship would be more efficiently used if the government knows
more about their incumbent and potential entrepreneurs.
We contribute to the literature on entrepreneurship by comparing risk preferences between
entrepreneurs and non-entrepreneurs (including paid employees) in Indonesia, the fourth pop-
ulous country. We also contribute to the literature on the origin of firms. Much has been said
about financial constraints in explaining the inception of firms, particularly in the settings of the
developed world (e.g., Evans and Jovanovic, 1989; Andersen and Nielsen, 2012). But relatively
little is said about the origin of firms based on individual characteristics in the developing
world. We thus considered individuals who set up firms in the developing world and focused
on their risk preferences. A further contribution is to regional studies. Although interest in
entrepreneurship in Indonesia has grown, risk – a fundamental determinant in entrepreneurship
– is missing in the studies.
We found that entrepreneurs were more risk tolerant than non-entrepreneurs; a variety of
robustness checks supported this conclusion. These results imply that entrepreneurs do not
entirely consist of workers who failed to secure government or private jobs; some entrepreneurs
actively take risks and possibly seek innovations. In addition, the inception of firms can be
partly traced to the risk preferences of individuals who set up firms.

II. LITERATURE REVIEW

Some studies in psychology have investigated this issue, but the samples used in such studies
were small, selective, and not nationally representative (e.g., Macko and Tyszka, 2009). This
section thus mainly considers studies in economics that used large and possibly nationally
representative samples. Empirical research has generally confirmed that entrepreneurs in the

C 2016 Board of Trustees of the Bulletin of Economic Research and John Wiley & Sons Ltd
 The Risk Preferences of Entrepreneurs in Indonesia 273
developed world are more risk tolerant than paid employees, regardless of the measures of risk
aversion.
Regarding European countries, Van Praag and Cramer (2001) and Cramer et al. used an
individual’s reservation price for a lottery ticket as their measure of risk aversion and found
that less risk averse individuals were more likely to become entrepreneurs in the Netherlands.
Ekelund et al. (2005) drew on psychometric data (harm avoidance) to measure risk aversion and
confirmed the theory for Finland. Caliendo et al. (2009) and Dohmen et al. (2011) differed in
that they distinguished between context free and context specific questions for measuring risk
aversion; they again found that risk tolerant individuals were more likely to become entrepreneurs
in Germany, with context specific questions having stronger relationships than context free
questions. Hvide and Panos (2014) provided an interesting case by covering all the new firms
incorporated between 2000 and 2007 in Norway and addressed selection bias. Using high
quality data, they further claimed that risk tolerant individuals were more likely to become
entrepreneurs, with risk tolerance being proxied by stock market participation and personal
leverage.
Regarding the USA, Kan and Tsai (2006) used a series of hypothetical questions from the
Panel Study of Income Dynamics and estimated risk aversion and reached the same conclusion.
Ahn (2010) used similar questions but with different data (the 1979 National Longitudinal
Survey of Youth: NLSY) and concurred with Kan and Tsai. Fairlie (2002) also employed the
NLSY but proxied risk preferences with drug dealing experiences; his main conclusion was the
same as that of Ahn. Some studies found a statistically non-significant relationship between risk
tolerance and entrepreneurship (e.g., Tucker, 1988; Barsky et al., 1997; Parker, 2008, all for the
USA), but no one, to our knowledge, has estimated that entrepreneurs were less risk tolerant
than paid employees.
Despite various efforts, little is known about the risk preferences of entrepreneurs in the
developing world. Djankov et al. (2006) found that entrepreneurs in China were more risk-
taking than non-entrepreneurs, but their entrepreneurs were a highly selective group: owner-
managers of businesses with five or more employees. And their sample was not nationally
representative, and its size (414 entrepreneurs) was small. It could be that entrepreneurs in the
developing world are also more risk tolerant than paid employees. However, entrepreneurs in the
developed and developing worlds face very different economic conditions for risk-taking such
as private insurance, credit markets, social safety nets, and labour markets. Their individual
characteristics differ as well. For example, consider education, which is critical for many labour
market outcomes, including Indonesia (Sohn, 2013). Parker (2009, pp. 117–119) argued that
entrepreneurs in the developed world were generally more educated than paid employees. Van
der Sluis (2005), however, found in their meta-analysis that the opposite was true for the
developing world. If education and risk tolerance are positively correlated (this is true in our
sample; Supplementary Table S-1), one can expect that entrepreneurs in the developed world
are more risk tolerant than paid employees but cannot expect the same for the developing
world.
At the same time, a growing number of researchers have investigated Indonesian en-
trepreneurs. We considered only studies that analysed individual level data for an entire country.
Gibson and Olivia (2010) drew on household level data for the year 2000 and investigated the
effect of infrastructure access and quality on non-farm enterprises in rural areas. Vial (2011)
investigated the influence of financial, human, and social capital on households’ participation in
entrepreneurship. Sohn and Kwon (2016) considered whether trust promoted entrepreneurship
and found a negative answer. Despite this variety of topics, risk preferences were missing in all.
Hence, we are the first to introduce risk preferences to the literature on Indonesian entrepreneurs,
using individual level data.

C 2016 Board of Trustees of the Bulletin of Economic Research and John Wiley & Sons Ltd
274 Bulletin of Economic Research

III. DATA AND METHODS

We analysed the Indonesian Family Life Survey (IFLS), an on-going longitudinal survey that
began collecting detailed individual data from over 22,000 individuals in 7,224 households in
1993–94 (IFLS1). The sample was representative of 82 percent of the Indonesian population, and
follow-up surveys were conducted in 1997 (IFLS2), 1998 (IFLS2+, restricted), 2000 (IFLS3),
and 2007 (IFLS4).
Because the questions relating to risk preferences were asked only for IFLS4, we mainly em-
ployed IFLS4 to estimate the relationship between risk tolerance and entrepreneurship (classified
as self-employment in the data). We merged IFLS2 with IFLS4, when necessary. Entrepreneurs
and non-entrepreneurs were distinct in IFLS4, and each category was further divided. Our com-
parison group consisted of paid employees and some workers other than paid employees. This is
because in the developed world, almost all workers are either entrepreneurs or paid employees,
so the comparison group is distinctly paid-employees. In the developing world, however, such a
neat distinction is difficult to make. We thus included unpaid family workers and casual workers
in the comparison group. In addition, it is difficult to consider subsistence peasants or fishermen
as entrepreneurs and their proportions are small in the developed world; for comparison, we
regarded workers in agriculture, forestry, fishing, and hunting (which comprised one category
in the survey) as non-entrepreneurs.
Because risk tolerance was the key variable, some explanations are in order. To measure
risk preferences, interviewers asked two sets of a series of hypothetical questions in IFLS4;
Supplementary Figures S-1 and S-2 present the flow of the questions. The task was very simple,
which is important considering the low levels of education of the Indonesian population (Sohn,
2013). Dave et al. (2010) recommended simple tasks to elicit risk preferences for subjects
with low numeracy. Respondents were asked to choose one of two options in each node of the
flow.
The flow first checked whether the respondents comprehended the concept of risk. The first
question in Set 1 involved choosing between (i) Rp 800 thousand per month for sure and
(ii) Rp 1.6 million or Rp 800 thousand per month with equal chance. Obviously, option (ii)
is the dominant choice. If the respondent chose (i), however, the interviewer pointed out the
unreasonableness of the choice, saying, ‘Are you sure? In option 2 you will get at least Rp
800 thousand per month and you may get Rp 1.6 million per month. In option 1 you will
always get Rp 800 thousand per month’. The respondent could then stick with option (i) or
switch to option (ii). We referred to those who still picked option (i) as risk incomprehensive
people. We included a few respondents who did not know how to choose their options to this
group. One may object that people who we considered risk incomprehensive knew the concept
of risk but simply avoided gambling, influenced by Islam. However, many people who chose
the wrong answer for the first question in Set 1 chose the right answer for Set 2 (Panel A
in Table 2). A person averse to gambling in his whole life would not suddenly abandon this
aversion after answering a series of simple questions posed by a stranger. Furthermore, Sohn
(forthcoming-b) provided evidence that choosing the wrong answer was related to low cognitive
skills.
The hypothetical stakes here were not moderate. The sure bets of Rp 800 thousand in Set 1
and Rp 4 million in Set 2 were 4.6 percent and 23.1 percent, respectively, of the GDP per capita
in 2007. Because people tend to become more risk averse with increasing stakes, the respon-
dents in this survey might be more risk averse than those in experiments with smaller stakes
(Binswanger, 1980; Yesuf and Bluffstone, 2009). Because Sets 1 and 2 yielded considerably
different proportions of risk adverse people (Panel A in Table 2), we used both to check


C 2016 Board of Trustees of the Bulletin of Economic Research and John Wiley & Sons Ltd
 The Risk Preferences of Entrepreneurs in Indonesia 275
consistency. No real stakes were involved in the survey, so the risk questions might not ac-
curately measure risk aversion. Dohmen et al. (2011), however, found that a measure of risk
aversion derived from hypothetical stakes was highly correlated with that derived from real
stakes. They also found that context specific risk measures predicted behaviours in that context
better than context free risk measures. In this sense, our measure is appealing because the
measure captured the risk aversion in finance, which is related to entrepreneurship. In addition,
Sohn (2014b) analysed IFLS4 and found that youth smoking was positively correlated with risk
taking. Given that smoking and risk taking are positively related, his study suggests that the risk
measures in IFLS4 actually measured what they were intended to measure.
The questions in IFLS4 are similar to those in Barsky et al. (1997), with some differences.
While their questions related not only to chances and payoffs but also occupational change, ours
related to only chances and payoffs. We thus did not need to consider status quo bias. Since
the questions in IFLS4 were asked for just one year with only four possible choices, it was
difficult to estimate precise relative risk aversion. Even if it were possible, the estimation would
involve many debatable assumptions such as a particular functional form of the utility function,
the appropriate parameters, and a statistical model for measurement error. For these reasons,
we employed a series of dummy variables for risk preferences. Nevertheless, each level of risk
tolerance accorded to the upper and lower bounds of relative risk aversion calculated by the
1−γ
utility function U (c) = c1−γ , where c refers to permanent consumption, and γ refers to relative
risk aversion.
In most cases, we controlled for the following basic demographics (later called X ): sex, age,
wealth,1 education, marital status, ethnicity, religion, self-reported health, urbanity, and province
fixed effects. When necessary, we added more covariates. After including respondents aged 18–
65 and excluding observations with missing values, we were left with 18,042 individuals. We
applied cross section person weights with attrition correction to make the sample representative
of the Indonesian population and clustered standard errors at the county (kecamatan) level to
account for the arbitrary correlation within counties.
In order to understand the relationship between risk tolerance and entrepreneurship, we
followed the conventional method in the literature: we regressed being an entrepreneur on the
dummy variables indicating risk preferences along with covariates. Specifically,

E i = Ri β1 + X i β2 + u i (1)

where E i refers to whether individual i was an entrepreneur or not, R to a series of dummy

variables for risk preferences, X to a vector of covariates mentioned above, β1 and β2 to
coefficient vectors to estimate, and u to a random error. The reference group for R consisted of
risk averse individuals. We employed a linear probability model for ease of interpretation, but
the results were robust to the use of similar models such as a probit or logit. As explained below,
Sets 1 and 2 captured different aspects of risk preferences; we thus entered them separately into
the specification.

1
Wealth was the sum of the values of 13 items of household assets: house and land occupied by the
household; other house building (including land); land (not used for farm or non-farm business); poultry;
livestock/fishpond; hard stem plant not used for farm or non-farm business; vehicles; household appliances;
savings/certificate of deposit/stocks; receivables; jewellery; household furniture and utensils; and other
assets. We imputed wealth for 14.5 percent of the sample with missing values for at least one item. If the
respondents confirmed that they did not have some items, we assigned a value of Rp 1 to missing items
because we used the natural log of wealth. We entered the following variables for imputation: sex, age,
education, marital status, ethnicity, religion, self-reported health, and urbanity.


C 2016 Board of Trustees of the Bulletin of Economic Research and John Wiley & Sons Ltd
276 Bulletin of Economic Research

IV. RESULTS

IV.1 Indonesian Context

The Indonesian context would help one better understand our results. It is important to know
whether the Indonesian economy was influenced by the global financial crisis when IFLS4 was
carried out. Rampini (2004) theorized that the amount of entrepreneurial activity is pro-cyclical.
If the crisis had influenced the Indonesian economy in 2007, the risk aversion of entrepreneurs
estimated in this study could relate to only a special case. The Indonesian GDP, however, grew
by 6.3 percent in 2007 and 6.0 percent in 2008; the crisis was felt only in 2009 with a GDP
growth rate of 4.6 percent.
The majority of entrepreneurs in the survey did not have bank accounts. When they were asked
whether they or any other member of their households owned savings, certificates of deposits,
or stocks, only 29.6 percent said ‘yes’. As many as 62.8 percent of entrepreneurs were engaged
in wholesale, retail, restaurants, and hotels. This composition dramatically differed from that of
paid employees in the private sector, whose corresponding figure was only 16.8 percent. The
size of entrepreneurship was very small. Of entrepreneurs, 46.4 percent worked alone, and 49.5
percent worked with unpaid family or temporary workers. Only 4.1 percent hired permanent
employees, but the median of firm size even among them was only three workers. Overall,
the context of entrepreneurship in Indonesia is similar to that in other developing countries: a
business is very small, insecure, and unprofitable (Banerjee and Duflo, 2011, ch. 9).

IV.2 Description of risk tolerance

Table 1 presents descriptive statistics, and we omitted explanations on them to save space. We,
however, explained descriptive statistics of risk preferences in Table 2 because the patterns of
risk preferences are of interest and have important implications for our results. Similar questions
regarding risk preferences have been asked in some developed countries such as Germany, but
little is known for developing countries. Incomprehension of the concept of risk is a novel
phenomenon in the literature. Since Sohn (forthcoming-b) elaborated on it, we only noted that
a non-negligible number of respondents in IFLS4 did not comprehend the concept. As Panel A
shows, 42.3 percent was risk incomprehensive according to Set 1, but only 7.8 percent according
to Set 2.
Among respondents who comprehended the concept of risk, we referred to those who always
chose the sure bet as risk averse, and the rest as risk tolerant. Depending on the subsequent
answers, we further divided risk tolerant respondents into three levels. For Set 1, 52.8 percent
always chose the sure bet (Rp 800 thousand per month), and 25.1 percent chose the most risky
option (Rp 1.6 million or Rp 200 thousand with equal chance). In contrast, for Set 2, as high
as 85.0 percent of respondents always chose the sure bet (Rp 4 million) and only 5.3 percent
chose the most risky option (Rp 16 million or – Rp 2 million with equal chance). Hence, a
substantially greater number of respondents were risk averse according to Set 2. The reason is
probably the construction of the two sets. The second options in Set 1 varied in one direction,
but for Set 2 they varied in both directions. Specifically, for Set 1, risk tolerant respondents
chose one of the followings: Rp. 1.6 million or Rp. 600 thousand (Risk Tolerance Level 1);
Rp 1.6 million or Rp 400 thousand (Risk Tolerance Level 2); and Rp 1.6 million or Rp 200
thousand (Risk Tolerance Level 3), all with equal chance. Rp 1.6 million was always fixed. For
Set 2, risk tolerant respondents chose one of the followings: Rp 8 million or Rp 2 million (Risk
Tolerance Level 1); Rp 12 million or Rp 0 million (Risk Tolerance Level 2); and Rp 16 million
or – Rp 2 million (Risk Tolerance Level 3), all with equal chance. There was no fixed payoff in

C 2016 Board of Trustees of the Bulletin of Economic Research and John Wiley & Sons Ltd
 The Risk Preferences of Entrepreneurs in Indonesia 277
TABLE 1
Descriptive statistics

Non-Entrepreneur Entrepreneur Different?a

Discrete Variable Proportion Proportion

Entrepreneur 0.232
Paid Employee 0.768
Male 0.593 0.521 Yes
Female 0.407 0.479 Yes
Elementary School or Below 0.500 0.518 No
Junior High School 0.154 0.181 Yes
Senior High School 0.231 0.233 No
College or Above 0.115 0.068 Yes
Unmarried 0.227 0.163 Yes
Married 0.773 0.837 Yes
Non-Javanese 0.468 0.478 No
Javanese 0.532 0.522 No
Non-Islam 0.075 0.070 No
Islam 0.925 0.930 No
Somewhat Unhealthy/Unhealthy 0.117 0.131 Yes
Somewhat Healthy 0.769 0.760 No
Very Healthy 0.114 0.109 No
Rural 0.587 0.475 Yes
Urban 0.413 0.525 Yes
Set 1 on Risk
Not Comprehended Risk 0.423 0.408 No
Risk Averse 0.311 0.292 Yes
Risk Tolerance Level 1 0.054 0.052 No
Risk Tolerance Level 2 0.072 0.083 Yes
Risk Tolerance Level 3 0.140 0.166 Yes
Risk Tolerance Levels 1–3 0.266 0.301 Yes
Set 2 on Risk
Not Comprehended Risk 0.078 0.080 No
Risk Averse 0.788 0.766 Yes
Risk Tolerance Level 1 0.061 0.064 No
Risk Tolerance Level 2 0.028 0.028 No
Risk Tolerance Level 3 0.045 0.062 Yes
Risk Tolerance Levels 1–3 0.134 0.154 Yes

Continuous Variable Mean (SD) Mean (SD)

Age 38.6 (11.9) 42.5 (11.1) Yes

Ln(Wealth) 17.18 (1.56) 17.45 (1.51) Yes
Height (cm) 157.3 (8.2) 156.5 (8.1) Yes
BMI (kg/m2 ) 22.2 (3.8) 23.6 (4.2) Yes
Standardized Math Scoreb 0.097 (1.01) −0.054 (0.93) Yes
Standardized Reading Scoreb 0.081 (1.00) −0.114 (0.97) Yes
N 13,912 4,130

Notes: We applied cross-section person weights with attrition correction.

a
‘Yes’ indicates that the difference was statistically significant with a p-value<0.05; otherwise, ‘No’.
b
Sample sizes were 2,924 for non-entrepreneurs and 454 for entrepreneurs.


C 2016 Board of Trustees of the Bulletin of Economic Research and John Wiley & Sons Ltd
278 Bulletin of Economic Research

TABLE 2
Descriptive statistics of risk preferences

Panel Aa All Comprehended Only

Set 1 on Risk
Not Comprehended 42.3
Risk Averse 31.1 52.8
Risk Tolerance Level 1 5.4 9.3
Risk Tolerance Level 2 7.2 12.9
Risk Tolerance Level 3 14.0 25.1
Set 2 on Risk
Not Comprehended 7.8
Risk Averse 78.8 85.0
Risk Tolerance Level 1 6.1 6.7
Risk Tolerance Level 2 2.8 3.0
Risk Tolerance Level 3 4.5 5.3

Panel B Comprehended Not Comprehended

Risk in Set 2 Risk in Set 2 N

Comprehended Risk in Set 1 9,739 1,020 10,759

Not Comprehended Risk in 6,701 582 7,283
Set 1
N 16,440 1,602 18,042

Panel Cb

Correlation Coefficient With Risk Incom- Without Risk

between Set 1 and Set 2 prehensive Incomprehensive
Respondents Respondents

Multiple Risk Preferences 0.307 0.498

Dichotomous Risk Preference 0.207 0.466

Notes: a We applied cross-section person weights with attrition correction.

b
All correlation coefficients were statistically significant.

Set 2. As the options became more complicated in the sense that both the expected mean and
variance increased, more respondents became risk averse. Yesuf and Bluffstone (2009) found
the same pattern in their study on Ethiopia.
One should not presume that both sets captured the same aspects of risk preferences. First,
recall that 85.0 percent of respondents who comprehended the concept of risk answered con-
servatively for Set 2, whereas the figure is only 52.8 percent for Set 1. Hence, risk tolerant
respondents for Set 2 were likely to be more risk tolerant than those for Set 1 given the same
level of risk tolerance; the former deviated from the absolute majority. In addition, one can con-
jecture that risk incomprehensive people were not well-educated. We verified this by regressing
a dummy for risk incomprehension on covariates in a linear probability model (Supplementary
Table S-2). Respondents with more education better comprehended the concept of risk for Sets
1 and 2; the size of the relationship was also large. In the two sets, however, the coefficients on
women and ‘somewhat healthy’ were statistically significant but opposite in sign. These results
suggest that the two sets captured different aspects of risk preferences. In addition, the systematic

C 2016 Board of Trustees of the Bulletin of Economic Research and John Wiley & Sons Ltd
 The Risk Preferences of Entrepreneurs in Indonesia 279
relationship between risk comprehension and some covariates indicate that the respondent did
not randomly answer the questions (e.g., always choosing Rp. 800 thousand per month for Set 1).
Panel B in Table 2 reinforces this argument. Of risk comprehensive respondents for Set 1, 9.5
percent did not comprehend the concept for Set 2. Of risk comprehensive respondents for Set
2, 40.8 percent did not comprehend the concept for Set 1. Panel B also disproves the argument
that fewer respondents did not comprehend the concept of risk for Set 2 because they learned
it by answering the questions in Set 1. If learning had taken place, one would not see 1,020
respondents who comprehend the concept for Set 1 but did not for Set 2 or 582 respondents
who did not comprehended for both sets.
Therefore, Sets 1 and 2 captured different aspects of risk preferences, although it is difficult
to say how they differed. The risk tolerant groups for Set 2 appear to be more extreme than
those for Set 1, but this difference does not mean that the two sets were inconsistent. Panel C
lists the correlation coefficients of risk preferences measured by Sets 1 and 2. The coefficients
were not close to one, indicating that the two sets measured somewhat different aspects of
risk preferences. At the same time, the positive coefficients indicate that some aspects of risk
preferences measured by the two sets overlap. The results remained the same whether we
included or excluded risk incomprehensive respondents and whether we used multiple (four
levels) or dichotomous risk preferences.

IV.3 Main results

Table 3 provides the results estimated by specification (1). The general message is that risk
tolerant workers were more likely to be entrepreneurs whether we employed Set 1 or Set 2
and whether we controlled for X or not. Without X , the most risk tolerant group as measured
by Set 1 was 4.3 percentage points more likely to be entrepreneurs (Column 1). With X , the
coefficients on all risk tolerance levels became larger, and the estimation precision improved
(Column 2). The coefficients on all levels of risk tolerance were jointly statistically significant
in Columns 1 and 2. Given the small cell size for each risk tolerant group, we aggregated all
risk tolerant groups, while excluding X ; the size of the relationship between risk tolerance and
entrepreneurship was 3.4 percentage points and statistically significant (Column 3). When we
controlled for X , the coefficient was more precisely estimated and larger at 5.0 percentage
points, or 21.6 percent relative to the mean (Column 4). When risk tolerance was measured by
Set 2, the general patterns were similar (Columns 5–8). In particular, when we combined the
three levels of risk tolerance, the coefficient on this variable was nearly the same as that for
Set 1 (Columns 7 and 8).
The size of the relationship between risk tolerance and entrepreneurship was not directly
comparable to those in other studies because of varying measures of risk preferences, definitions
of entrepreneurs, and covariates. When we made rough comparisons, however, our estimates did
not differ much from others. For example, Ekelund et al. (2005, p. 657) found that relative to men
with median risk aversion, those with a 25-percentile level of risk aversion were 19.6 percent
more likely to be entrepreneurs. Similarly, Ahn (2010, p. 440) argued that a one standard
deviation increase in risk tolerance would raise the probability of entering entrepreneurship by
13 percent. In Caliendo et al. (2009), among regularly employed workers, those who would
take the risk of ‘high’ investment were 14.2 percent more likely to enter entrepreneurship than
those who would take no risk2 . Dohmen et al. (2011) estimated a somewhat higher rate: a

2
Caliendo et al. (2009) did not present this rate. We first calculated the mean transition rate by dividing
the total number of transitions to entrepreneurs between 2004 and 2005 (147 transitions) by the total number
of entrepreneurs in 2004 (906 entrepreneurs). We then divided the coefficient on the risk of high investment
(2.3 percentage points) by the mean transition rate (16.2 percent).

C 2016 Board of Trustees of the Bulletin of Economic Research and John Wiley & Sons Ltd
 TABLE 3


Risk tolerance and entrepreneurship: main results
280

Dependent Variable: Being an Entrepreneur (Linear Probability Model)

1 2 3 4 5 6 7 8

Set 1 on Risk Set 2 on Risk

Did not Comprehend 0.005 0.001 0.005 0.001 0.010 0.023 0.010 0.023
Risk (0.012) (0.012) (0.012) (0.012) (0.018) (0.017) (0.018) (0.017)
Risk Averse Reference Reference Reference Reference Reference
Risk Tolerance Level 1 0.004 0.021 0.016 0.040
(0.015) (0.014) (0.018) (0.017)**
Risk Tolerance Level 2 0.037 0.044 0.001 0.018
(0.019)** (0.017)** (0.023) (0.022)
Risk Tolerance Level 3 0.043 0.065 0.066 0.088
(0.011)*** (0.010)*** (0.015)*** (0.015)***
Risk Tolerance Levels 0.034 0.050 0.031 0.052
1–3 (0.011)*** (0.009)*** (0.013)** (0.012)***
Female 0.059 0.058 0.057 0.057
(0.008)*** (0.008)*** (0.008)*** (0.008)***
Age (/10) 0.043 0.043 0.043 0.043
Bulletin of Economic Research

(0.004)*** (0.004)*** (0.004)*** (0.004)***

Ln(Wealth) 0.020 0.020 0.020 0.020
(0.003)*** (0.003)*** (0.003)*** (0.003)***

C 2016 Board of Trustees of the Bulletin of Economic Research and John Wiley & Sons Ltd
Junior High School 0.055 0.055 0.054 0.055
(0.012)*** (0.012)*** (0.012)*** (0.012)***
Senior High School 0.005 0.005 0.005 0.006
(0.010) (0.010) (0.010) (0.010)
College or Above −0.123 −0.123 −0.122 −0.121
(0.014)*** (0.014)*** (0.014)*** (0.014)***
Married 0.043 0.043 0.044 0.044
(0.008)*** (0.008)*** (0.008)*** (0.008)***

(Continued)

TABLE 3
Continued

1 2 3 4 5 6 7 8

Set 1 on Risk Set 2 on Risk

Jawa −0.026 −0.026 −0.028 −0.028

(0.018) (0.018) (0.018) (0.018)
Islam 0.021 0.020 0.020 0.020
(0.021) (0.021) (0.021) (0.021)
Somewhat Healthy −0.005 −0.004 −0.005 −0.005
(0.008) (0.008) (0.008) (0.008)
Very Healthy −0.003 −0.003 −0.003 −0.003
(0.013) (0.013) (0.013) (0.013)
Urban 0.101 0.101 0.101 0.101
(0.012)*** (0.012)*** (0.012)*** (0.012)***
Constant 0.221 −0.633 0.221 −0.632 0.376 −0.628 0.227 −0.631
(0.009)*** (0.053)*** (0.009)*** (0.053)*** (0.008)*** (0.051)*** (0.008)*** (0.051)***
Province Fixed Effects No Yes No Yes No Yes No Yes

C 2016 Board of Trustees of the Bulletin of Economic Research and John Wiley & Sons Ltd
Mean Dep. Var. 0.232 0.232 0.232 0.232 0.232 0.232 0.232 0.232
The Risk Preferences of Entrepreneurs in Indonesia

R Squared 0.001 0.054 0.001 0.054 0.001 0.054 0.001 0.053

Notes: The sample size was 18,042. We applied cross-section person weights with attrition correction. Standard errors clustered at the county level are in parentheses.
*
: p-value<0.10, ** : p-value<0.05, *** : p-value<0.01.
281
282 Bulletin of Economic Research

one standard deviation increase in willingness to take general risks was associated with a 28.6
percent increase in the likelihood of being entrepreneurs.
Dohmen et al. (2011) found that height was correlated with risk preferences, so it could be
that risk preferences in our results captured the effects of height. We experimented with not only
height but also body mass index (BMI) (for more on height and BMI in Indonesia, see Sohn,
2014a, 2014c, 2015b, 2015c, 2015d, 2015e, 2015f, 2016a, 2016b, forthcoming-a). Adding these
two variables separately to specification (1) hardly changed β1 (Columns 1–4 of Supplementary
Table S-3). Similarly, Dohmen et al. (2010) showed that cognitive skills were correlated with
risk preferences, so that risk preferences in our results might capture the effects of cognitive
skills. This bias should not be critical because we already controlled for education. That said, for
a subset of young respondents who took tests of cognitive skills in 1997 (IFLS2), we separately
added math and reading scores; β1 hardly changed (Columns 5–8).

IV.4 Extensions
Considering the great desire for government jobs and the rent-seeking enterprise, it would be of
interest to compare the risk tolerance of entrepreneurs to that of government workers. This is also
interesting because government workers are commonly considered conservative. We aggregated
the three levels of risk tolerance because the cell size of each level of risk tolerance was small
and the aggregation hardly affected the main results (Table 3). We controlled for X but did not
list its elements to save space. Columns 1 and 2 of Table 4 suggest that entrepreneurs were
more risk tolerant than government workers. Relative to the mean, the size of the relationship
between risk tolerance and entrepreneurship was roughly half of that in Table 3. The reduction
of the relationship suggests that government workers were more risk tolerant than other workers,
thereby supporting the argument that people may take risk to join the rent-seeking enterprise.
These results differed from those of Tucker (1988), who argued that the risk tolerance of
entrepreneurs and government workers in the USA was not statistically different from each
other. We attributed the inconsistency to the different measures of risk tolerance. In this light,
Dohmen et al.’s (2011) argument is revealing. They argued that although risk tolerance in
one context was correlated with that in another context, context specific risk preferences better
predicted behaviours in the concerned context. In our sample, risk tolerance concerned financial
risk tolerance, whereas Tucker’s measures concerned everyday risk tolerance such as fastening
seatbelts and smoking.
Including casual workers and unpaid family workers in the comparison group might mislead
comparisons between the developing and developed worlds because their proportion is miniscule
in the developed world. Our results were robust to the exclusion of them (Columns 3 and 4).
Workers in rural areas may need to be excluded for comparison purposes because only a small
proportion of workers in the developed world work in rural areas. The main message remained
the same with only urban workers (Columns 5 and 6). We also excluded all workers in the
primary sector because their proportion was small in the developed world. The relationship
remained positive and statistically significant (Columns 7 and 8).
Thus far, we did not distinguish between entrepreneurs and the self-employed. These groups
conceptually differ, however; entrepreneurship highlights innovation, whereas self-employment
highlights autonomy. Because it is empirically difficult to distinguish one group from the
other, empirical studies usually treat the two groups interchangeably. Although imperfect, we
defined innovation-related entrepreneurs as the self-employed workers whose wealth levels
were at or above the 75 percentile of all self-employed workers. When we compared them with
non-entrepreneurs, the relationship remained positive and statistically significant (Columns
9 and 10). To check whether the imputation of wealth exaggerated our results, we re-ran

C 2016 Board of Trustees of the Bulletin of Economic Research and John Wiley & Sons Ltd

TABLE 4
Risk tolerance and entrepreneurship: robustness checks

Dependent Variable: Being an Entrepreneur (Linear Probability Model)

1 2 3 4 5 6 7 8
Set 1 Set 2 Set 1 Set 2 Set 1 Set 2 Set 1 Set 2
on Risk on Risk on Risk on Risk on Risk on Risk on Risk on Risk

Changes in Samples Compare with Exclude Casual and Exclude workers in

Government Workers Unpaid Family Workers Urban Workers Only the primary sector

Risk Averse Reference Reference Reference Reference Reference Reference Reference Reference
Risk Tolerance Levels 1–3 0.073 0.056 0.070 0.067 0.046 0.070 0.058 0.083
(0.016)*** (0.018)*** (0.012)*** (0.013)*** (0.014)*** (0.015)*** (0.013)*** (0.014)***
R Squared 0.223 0.221 0.093 0.091 0.081 0.082 0.134 0.135
Mean Dep. Var. 0.518 0.307 0.269 0.350
N 7,970 13,441 9,309 12,569

9 10 11 12 13 14
Set 1 Set 2 Set 1 Set 2 Set 1 Set 2
on Risk on Risk on Risk on Risk on Risk on Risk

Changes in Samples Consider Rich Exclude Imputed Manufacturing

Entrepreneurs Wealth Only

C 2016 Board of Trustees of the Bulletin of Economic Research and John Wiley & Sons Ltd
The Risk Preferences of Entrepreneurs in Indonesia

Risk Averse Reference Reference Reference Reference Reference Reference

Risk Tolerance Levels 1–3 0.016 0.031 0.045 0.045 0.061 0.125
(0.006)** (0.010)*** (0.011)*** (0.013)*** (0.030)** (0.039)***
R Squared 0.140 0.140 0.052 0.051 0.202 0.208
Mean Dep. Var. 0.083 0.231 0.239
N 12,446 15,818 2,433

Notes: For all estimations, we controlled for covariates identical to those in the even columns of Table 3. We applied cross-section person weights with attrition correction.
Standard errors clustered at the county level are in parentheses. * : p-value<0.10, ** : p-value<0.05, *** : p-value<0.01.
283
284 Bulletin of Economic Research

specification (1) after excluding observations with imputed wealth. The sign and size of the
relationship remained the same as those in Table 3 (Columns 11 and 12). Wealth may not be an
accurate measure of innovation-related entrepreneurship. We now considered only workers in
manufacturing. This sector is probably more conducive to innovation in a developing country
because it is easier to copy innovation from tradables than from non-tradables and manufactured
goods are more tradable. The positive relationship remained the same, and the relationship was
particularly pronounced for Set 2 (Columns 13 and 14). In almost all cases, the size of the
relationship relative to the dependent variable was not smaller than that in Table 3, indicating
that our main results in Table 3 were conservative and robust.
We also tested possible heterogeneity in the relationship by interacting the dichotomous
dummy variable for risk tolerance with gender, marital status, ethnicity, and being Islamic, one
at a time (Supplementary Table S-4). In general, the interaction terms were not statistically
significant, indicating that the association applies in the same way as to the general population.

V. CONCLUSIONS

We started with the question of whether entrepreneurs are more risk tolerant than non-
entrepreneurs in Indonesia. We measured the risk preferences of the two groups with two
sets of a series of hypothetical questions and estimated their relationship in linear probability
models. We found that the relationship was positive as in the developed world. Although we
do not claim that this relationship is causal, this possibility is worth considering because evi-
dence suggests that risk preferences are stable (Andersen et al., 2008). An extensive array of
robustness checks increases this possibility.
The results have important implications for understanding entrepreneurs in the developing
world. Entrepreneurs in the developing world are usually regarded as workers who failed to
secure government or private jobs. However, recent efforts to integrate personality psychology
and economics present a more optimistic view of entrepreneurs in the developing world (for a
review, see Almlund et al., 2011). This research has paid much attention to the Big Five in per-
sonality psychology (openness to experience, conscientiousness, extraversion, agreeableness,
and neuroticism) and other personality traits besides cognitive skills. In light of our finding
that entrepreneurs were more risk tolerant than non-entrepreneurs in Indonesia, such research
can guide us to derive the characteristics of entrepreneurs in the developing world. Borghans
et al. (2009) estimated that risk tolerance was negatively associated with neuroticism (fear of
failure) and agreeableness but positively with ambition. Dohmen et al. (2010, online appendix)
also found that risk tolerance was negatively associated with agreeableness but positively with
openness to experience. Bibby and Ferguson (2011) showed that risk tolerance was positively as-
sociated with sensation seeking. Although these studies analysed data from developed countries,
they suggest that entrepreneurs in Indonesia may not be the workers for whom entrepreneur-
ship is the last resort. On the contrary, they are ambitious and independent; they seek novel
experiences and are not afraid of failure. They may have characteristics that could lead to
creative destruction.
We described that in Indonesia, entrepreneurs were more risk tolerant than non-entrepreneurs.
Caution, however, should be exercised in concluding from this finding that more risk tolerant
entrepreneurs perform better and their firms survive longer. Willebrands et al. (2012) argued
that conditional on being entrepreneurs, more cautious – meaning, perceiving more risks –
entrepreneurs achieved more sales. Hvide and Panos (2014) reached the same conclusions by
analysing Norwegian data. We compared imputed wealth as a rough measure of performance
in the past between risk averse and risk tolerant entrepreneurs. When we used Set 1 to identify

C 2016 Board of Trustees of the Bulletin of Economic Research and John Wiley & Sons Ltd
 The Risk Preferences of Entrepreneurs in Indonesia 285
these groups, the difference between the two was modest and not statistically significant; with
Set 2, however, the risk tolerant group was wealthier than the risk averse group by 19 percent.
Evidence is mixed, but considered together with the previous studies, these results are of interest.
Future research should provide a more comprehensive study on the relationship between risk
aversion and entrepreneurship or business performance.
We acknowledged some limitations. First, it is too early to determine whether our findings can
be generalized to other developing countries. Because the general context of entrepreneurship in
Indonesia is similar to that of other developing countries, we believe that such a generalization
is possible. Replication studies will check this possibility. Second, we lacked causality running
from risk tolerance to entrepreneurship. However, a variety of robustness checks and evidence of
stability of risk preferences suggest that the causal argument is plausible. Instrumental variables
can tease out causality. We, however, doubt that anyone can easily find them in the near future
because risk preferences comprise a fundamental characteristic of economic behaviour; any
variable correlated with risk preferences is likely to be correlated with entrepreneurship as well.
Not surprisingly, most studies settled for describing the relationship between risk preferences
and entrepreneurship; this study is no different. Third, it would be ideal to use multiple types of
measures. We, however, believe that using only gamble-based measures can still lead to useful
research, at least as a starting point.

REFERENCES

Acs, Z. J., Desai, S. and Hessels, J. (2008). ‘Entrepreneurship, economic development and
institutions’. Small Business Economics, 31, 219–34. doi:10.1007/s11187-008-9135-9
Ahn, T. (2010). ‘Attitudes toward risk and self-employment of young workers’. Labour Eco-
nomics, 17, 434–42. doi:10.1016/j.labeco.2009.06.005
Almlund, M., Duckworth, A. L., Heckman, J. and Kautz, T. (2011). ‘Personality psychology
and economics’,in Hanushek, E. A., Machin, S. and Woessmann, L. (eds.), Handbook of the
Economics of Education, Amsterdam:, 1–181.
Andersen, S., Harrison, G. W., Lau, M. I. and Elisabet Rutström, E. (2008). ‘Lost in state space:
Are preferences stable?’ International Economic Review, 49, 1091–112.
Andersen, S. and Nielsen, K. M. (2012). ‘Ability or finances as constraints on entrepreneurship?
Evidence from survival rates in a natural experiment’. Review of Financial Studies, 25, 3684–
710.
Banerjee, A. V. and Duflo, E. (2011). Poor economics: A radical rethinking of the way to
fight global poverty. New York: PublicAffairs.
Barsky, R. B., Juster, F. T., Kimball, M. S. and Shapiro, M. D. (1997). ‘Preference parameters
and behavioral heterogeneity: An experimental approach in the Health and Retirement Study’.
Quarterly Journal of Economics, 112, 537–79.
Bibby, P. A. and Ferguson, E. (2011). ‘The ability to process emotional informa-
tion predicts loss aversion’. Personality and Individual Differences, 51, 263–66.
doi:10.1016/j.paid.2010.05.001
Binswanger, H. P. (1980). ‘Attitudes toward risk: Experimental measurement in rural India’.
American Journal of Agricultural Economics, 62, 395–407.
Borghans, L., Heckman, J. J., Golsteyn, B. H. H. and Meijers, H. (2009). ‘Gender differences
in risk aversion and ambiguity aversion’. Journal of the European Economic Association,
7, 649–58.
Butt, S. (2011). ‘Anti-corruption reform in Indonesia: An obituary?’ Bulletin of Indonesian
Economic Studies, 47, 381–94.

C 2016 Board of Trustees of the Bulletin of Economic Research and John Wiley & Sons Ltd
286 Bulletin of Economic Research

Caliendo, M., Fossen, F. M. and Kritikos, A. S. (2009). ‘Risk attitudes of nascent entrepreneurs –
New evidence from an experimentally validated survey’. Small Business Economics, 32,
153–67.
Cramer, J. S., Hartog, J., Jonker, N. and Van Praag, C. M. (2002). ‘Low risk aversion encourages
the choice for entrepreneurship: An empirical test of a truism’. Journal of Economic Behavior
& Organization, 48, 29–36.
Dave, C., Eckel, C. C., Johnson, C. A. and Rojas, C. (2010). ‘Eliciting risk preferences: When
is simple better?’ Journal of Risk and Uncertainty, 41, 219–43. doi:10.1007/s11166-010-
9103-z
Djankov, S., Qian, Y., Roland, G. and Zhuravskaya, E. (2006). ‘Entrepreneurship in China and
Russia compared’. Journal of the European Economic Association, 4, 352–65.
Dohmen, T., Falk, A., Huffman, D. and Sunde, U. (2010). ‘Are risk aversion and im-
patience related to cognitive ability?’ American Economic Review, 100, 1238–60.
doi:10.1257/aer.100.3.1238
Dohmen, T., Falk, A., Huffman, D., Sunde, U., Schupp, J. and Wagner, G. G. (2011). ‘Individual
risk attitudes: Measurement, determinants, and behavioral consequences’. Journal of the
European Economic Association, 9, 522–50.
Ekelund, J., Johansson, E., Järvelin, M.-R. and Lichtermann, D. (2005). ‘Self-employment and
risk aversion – Evidence from psychological test data’. Labour Economics, 12, 649–59.
Evans, D. S. and Jovanovic, B. (1989). ‘An estimated model of entrepreneurial choice under
liquidity constraints’. Journal of Political Economy, 97, 808–27.
Fairlie, R. W. (2002). ‘Drug dealing and legitimate self-employment’. Journal of Labor Eco-
nomics, 20, 538–37.
Gibson, J. and Olivia, S. (2010). ‘The effect of infrastructure access and quality on non-farm
enterprises in rural Indonesia’. World Development, 38, 717–26.
Hvide, H. K. and Panos, G. A. (2014). ‘Risk tolerance and entrepreneurship’. Journal of
Financial Economics, 111, 200–23.
Kan, K. and Tsai, W.-D. (2006). ‘Entrepreneurship and risk aversion’. Small Business Eco-
nomics, 26, 465–74. doi:10.1007/s11187-005-5603-7
Kihlstrom, R. E. and Laffont, J.-J. (1979). ‘A general equilibrium entrepreneurial theory of firm
formation based on risk aversion’. Journal of Political Economy, 87, 719–48.
Knight, F. H. (1921). Risk, uncertainty and profit. Boston, MA: Houghton Mifflin Company.
Macko, A. and Tyszka, T. (2009). ‘Entrepreneurship and risk taking’. Applied Psychology, 58,
469–87. doi:10.1111/j.1464-0597.2009.00402.x
Parker, S. C. (2008). ‘Entrepreneurship among married couples in the United States: A simulta-
neous probit approach’. Labour Economics, 15, 459–81. doi:10.1016/j.labeco.2007.03.004
Parker, S. C. (2009). The economics of entrepreneurship. Cambridge: Cambridge University
Press.
Rampini, A. A. (2004). ‘Entrepreneurial activity, risk, and the business cycle’. Journal of
Monetary Economics, 51, 555–73. doi:10.1016/j.jmoneco.2003.06.003
Schumpeter, J. A. (1942). Capitalism, socialism and democracy. New York: Harper.
Setiyono, B. and McLeod, R. H. (2010). ‘Civil society organisations’ contribution to the anti-
corruption movement in Indonesia’. Bulletin of Indonesian Economic Studies, 46, 347–70.
Sohn, K. (2013). ‘Monetary and nonmonetary returns to education in Indonesia’. Developing
Economies, 51, 34–59. doi:10.1111/deve.12001
Sohn, K. (2014a). ‘Age and size at maturity in Indonesian women: A norm of reaction?’
American Journal of Human Biology, 26, 713–15. doi:10.1002/ajhb.22571
Sohn, K. (2014b). ‘A note on the effects of education on youth smoking in a developing country’.
Journal of the Asia Pacific Economy, 19, 66–73.
Sohn, K. (2014c). ‘Sufficiently good measures of obesity: The case of a developing country’.
Journal of Biosocial Science, 46, 797–817. doi:10.1017/S0021932013000692
Sohn, K. (2015a). ‘Gender discrimination in earnings in Indonesia: A fuller picture’. Bulletin
of Indonesian Economic Studies, 51, 95–121. doi:10.1080/00074918.2015.1016569


C 2016 Board of Trustees of the Bulletin of Economic Research and John Wiley & Sons Ltd
 The Risk Preferences of Entrepreneurs in Indonesia 287
Sohn, K. (2015b). ‘The height premium in Indonesia’. Economics & Human Biology, 16, 1–15.
doi:10.1016/j.ehb.2013.12.011
Sohn, K. (2015c). ‘The influence of birth season on height: Evidence from Indonesia’. American
Journal of Physical Anthropology, 157, 659–65.
Sohn, K. (2015d). ‘Is leg length a biomarker of early life conditions? Evidence from a historically
short population’. American Journal of Human Biology, 27, 538–45. doi:10.1002/ajhb.22682
Sohn, K. (2015e). ‘The male-taller norm: Lack of evidence from a developing country’.
HOMO – Journal of Comparative Human Biology, 66, 369–78. doi:10.1016/j.jchb.2015.
02.006
Sohn, K. (2015f). ‘The value of male height in the marriage market’. Economics & Human
Biology, 18, 110–24.
Sohn, K. (2016a). ‘Does a taller husband make his wife happier?’ Personality and Individual
Differences, 91, 14–21.
Sohn, K. (2016b). ‘Height and happiness in a developing country’. Journal of Happiness
Studies, 17, 1–23. doi:10.1007/s10902-014-9566-8
Sohn, K. (forthcoming-a). ‘Biological standards of living: Age at menarche vs. height’. Annals
of Human Biology.
Sohn, K. (forthcoming-b). ‘Risk incomprehension and its economic consequences’. Journal of
Development Studies.
Sohn, K. and Kwon, I. (2016). ‘Does trust promote entrepreneurship in a developing country’.
Singapore Economic Review. Forthcoming.
Tucker, I. B. I. (1988). ‘Entrepreneurs and public-sector employees: The role of achievement
motivation and risk in occupational choice’. Journal of Economic Education, 19, 259–68.
Van der Sluis, J. (2005). ‘Entrepreneurship selection and performance: A meta-analysis of the
impact of education in developing economies’. World Bank Economic Review, 19, 225–61.
doi:10.1093/wber/lhi013
Van Praag, C. M. and Cramer, J. S. (2001). ‘The roots of entrepreneurship and labour demand:
Individual ability and low risk aversion’. Economica, 68, 45–62.
Vereshchagina, G. and Hopenhayn, H. A. (2009). ‘Risk taking by entrepreneurs’. American
Economic Review, 99, 1808–30. doi:10.1257/aer.99.5.1808
Vial, V. (2011). ‘Micro-entrepreneurship in a hostile environment: evidence from Indonesia’.
Bulletin of Indonesian Economic Studies, 47, 233–62.
Willebrands, D., Lammers, J. and Hartog, J. (2012). ‘A successful businessman is not a gambler.
Risk attitude and business performance among small enterprises in Nigeria’. Journal of
Economic Psychology, 33, 342–54. doi:10.1016/j.joep.2011.03.006
Yesuf, M. and Bluffstone, R. A. (2009). ‘Poverty, risk aversion, and path dependence in low-
income countries: experimental evidence from Ethiopia’. American Journal of Agricultural
Economics, 91, 1022–37. doi:10.1111/j.1467-8276.2009.01307.x

SUPPORTING INFORMATION

Additional Supporting Information may be found in the online version of this article at the
publisher’s web site:
Table S-1 Factors Related to Risk Tolerance.
Table S-2 Factors Related to Risk Incomprehension.
Table S-3 Risk Tolerance and Entrepreneurship: Additional Covariates of Human Capital.
Table S-4 Risk Tolerance and Entrepreneurship: Interaction Effects.
Figure S-1 Flow Chart for Set 1 with Equal Chance for Each Second Option.
Figure S-2 Flow Chart for Set 2 with Equal Chance for Each Second Option.


C 2016 Board of Trustees of the Bulletin of Economic Research and John Wiley & Sons Ltd