Are risk preferences consistent?

Abstract

This paper examines the consistency of risk preferences across three decision domains important in most people’s lives: work, health and personal finance. We consider the degree to which the five factor model of personality and a range of factors that influence risk-related decision-making (perceived risk, framing, emotions and cost–benefit analysis) impact upon cross-domain consistency. Data were gathered from a sample of participants for whom approaches to risk were likely to vary (academics, chess players, firefighters, mountaineers and City traders). The results showed that participants could be categorised into two groups: those who were consistent in their risk preferences in three decision domains, and those who were inconsistent or domain-specific. The consistent group was significantly lower on neuroticism and higher on agreeableness and conscientiousness with a less variable approach to weighing up the costs and benefits of taking risks than the inconsistent group. The majority of the consistent group was risk averse.

When the domain-specific risk preferences of the inconsistent group were examined, data showed that different combinations of personality and decision-making factors predicted risk preferences within each domain.

Introduction

The prediction of risk taking has been the goal of a number of studies. One approach to theoretical development is through understanding the consistency of risky decision-making (e.g. Weber, Blais, & Betz, 2002; Weber & Milliman, 1997). In this paper we consider the influence of personality and decision factors, including risk perception, on domain-specific and cross-domain risk preferences. We base this proposal on the risk modelling of Sitkin and Pablo (1992), and the subsequent empirical work of Sitkin and Weingart (1995).

In contrast to much previous research we consider risk across three domains important in most people’s lives rather than just one domain. The domains were work, health and personal finance, chosen for two reasons. First, it seemed likely that these three key domains are central to many people’s lives, as indicated in previous research (e.g. Byrnes, Miller, & Schafer, 1999). Second, it was thought that these domains would be applicable to almost all respondents in the sample groups chosen.

Two key concepts in this paper are domain-specific preferences and cross-domain consistency. Domain-specific preferences refer to the decision-making and risk preferences within a particular choice domain, for example work. Cross-domain consistency refers to the degree of variation that an individual shows for the same decision-making variable, e.g. risk preference, across each of the decision domains.

Whether someone is consistent or not is significant for both theory and practice. Support for cross-domain consistency would fit the idea that there are relatively stable risk dispositions, which might have a basis in personality. Domain-specific behaviour indicates that risk taking might be influenced by domain-specific risk factors, for example through information framing (e.g. Mittal, Ross, & Tsiros, 2002; Tversky & Kahneman, 1986). In terms of human resource management, a dispositional approach to risk could be managed through selection and placement to match risk preferences to the nature of work in risk sensitive industries, whereas domain-specific risk factors could be managed through systems such as organisational culture and incentive structures.

There is evidence for both approaches. A study of reported risk taking in six decision domains (recreation, health, career, finance, safety and social) found a distinctive personality profile. Risk taking overall was associated positively with extraversion and openness, and negatively with neuroticism, agreeableness and conscientiousness (Nicholson, Soane, Fenton–O’Creevy, & Willman, 2005). Kowert and Hermann (1997) found that risk taking was associated positively with openness and negatively with conscientiousness and agreeableness. The risk and personality research has two important implications. First, risk behaviour can be understood in terms of dispositional motivations. For example, people will take or avoid risks to achieve goals that are consistent with their character, e.g. extraverts take risks because of a generalised need for sensation (Zuckerman, 1994), and sensation is the goal of their risk taking. Second, since personality is relatively stable across adulthood (McCrae & Costa, 1997), tendencies to take or avoid risk might also be robust.

Much of the research into the influence of situational factors upon risk taking has been based upon prospect theory (Kahneman & Tversky, 1979). Research by the theory’s authors (e.g. Tversky & Kahneman, 1986) has shown that decision-making is influenced by relative perceptions of loss and gain: people become risk-averse when they think they are ahead, and risk-seeking when they think they are behind. The suggested implication is that risk preferences are variable. However, some studies have found that there can be strong associations between different types of risk taking. A study of managerial risk taking found that, while there were few generalised consistencies among measured risk behaviours, some individuals had consistent responses to different measures of risk (MacCrimmon & Wehrung, 1986). However, these studies did not measure the effects of dispositions on risk taking.

A second factor, important in understanding the variability of risk taking is the approach to data analysis; notably, a focus on individual data. The research discussed above suggests that individual level risk behaviour could be influenced by a combination of personality and situational variables such as loss and gain. The combination of factors could vary on an individual-level basis and be determined by the strength of personality characteristics and psychological decision preference variables. It could be proposed that, if risk preferences are shaped by personality, people who have a particular pattern of strong personality characteristics that are associated with consistent approaches to decision-making will be consistent in their risk taking across domains. If risk preferences are shaped by decision characteristics then risk decisions are more likely to be variable and inconsistent across domain.

Three questions present themselves from considering the literature above:

• Can people be characterised as consistent in their risk preference across different areas of their lives, and if so are there differences between them and inconsistent people in terms of personality and factors associated with risk-related decision-making?

• Can personality variables predict the risk preferences of those people characterised as inconsistent?

• Are different combinations of personality and risk factors important to risk preference within different decision domains?