The concept of risk propensity has been the subject of both theoretical and empirical investigation, but with little consensus about its conceptualization and measurement of risk propensity. This paper seeks to advance the field through data from a sample of 1,669 managers and professionals.
These show the internal consistency and correlates of a new scale that measures overall risk propensity, and risk taking in six different decision domains. Correlates examined are the NEO PI-R, a Five Factor personality inventory, demographic variables and biographical self-reports. There are four key conclusions. First, we find support for integrity of the risk propensity scale, with domain specific as well as general attributes. Second, the data show risk propensity to be strongly rooted in personality, with sensation seeking confirmed as a key component in most decision domains.
A strong Big Five pattern emerges for overall risk propensity comprising high E and O, and low N, A, and C. Third, risk propensity predicts career and other behaviours as predicted, lending validation support to the measure. Fourth, the validity of the measure is also supported by findings that risk propensity differs markedly in its distribution across job types and business sectors. The implications for risk research and risk management are considered.
The concept of risk propensity has important implications for the theoretical modelling of risk behaviour and for practical insights into the motives underlying individual level choices about engaging in risky behaviour. In organizational terms, a better understanding of risk behaviour could contribute significantly to risk management programs. In this paper we have three objectives.
First we seek to establish the viability of a new measure of risk propensity and to consider whether it is a construct that can be conceptualized as stable across domains and time. Second, by examining demographic and biographical correlates we seek support for its validity and practical significance. Third, our principal objective, assuming the measure is sufficiently robust, is to explore how personality dispositions underlie risk propensity. Two developments have influenced this research.
The first development is a strong revival of interest in trait psychology, with attention converging around the Big Five factorial model of personality (McCrae & Costa, 1997). The second development has been a rapid growth of attention and concern in business around the concept of risk (Bernstein, 1996). This is partly due to greater awareness and incidence of high profile accidents in operational areas and in finance, for example the collapse of Barings bank in 1996 (Fay, 1996).
The literature concerning risk propensity has two main themes. The first theme relates to prospect theory (Kahneman & Tversky, 1979), which proposes that risk taking is asymmetric about a reference point, and that people will be risk averse when they perceive themselves to be in the domain of gain, and risk seeking in the domain of loss.
Prospect theory has stimulated numerous research studies into risk preferences and risk taking. A key premise of the theory is that individual level risk taking is relatively inconsistent across situations – a person will take risk in some circumstances, and avoid risk in other circumstances. The prompt for behavioural change could be as simple as the semantic presentation of data, for example whether a choice outcome is presented as a loss or a gain.
A second theme in the research considers the individual difference factors that could influence risk taking. A significant contribution to this research is the notion that risk taking could be linked to factors that are trans-situational, such as personality – risk propensity could thus be more a characteristic of an individual than their situation.
In this area, sensation seeking has been found to be particularly important. Zuckerman pioneered the study of this concept (Zuckerman et al., 1964), and since then a stream of research has confirmed its importance as a highly consistent predictor of various kinds of risk taking, including compulsive gambling and participation in high risk activities (Zuckerman, 1974; Zuckerman & Kuhlman, 2000).
This construct has also been the subject of extensive psycho physiological investigation, linking it clearly with individual differences in cortical arousal thresholds and levels of enzymes and neurotransmitters affecting the central nervous system (Geen, 1997). Substantial heritability of the trait may also be inferred from evidence for the genetic origins of dopamine receptor levels linked with venturesome personality (Cloninger, 1996; Farde et al., 1997). An alternative but related approach in the risk literature has been to consider risk propensity in terms of the variance in within-individual measures of risk.
An example of this work is Weinstein and Martin (1969), with other studies adopting the same approach in more recent research (e.g. Salminen & Heiskanen, 1997). These empirical works focus attention on the inter-correlation of scores on a range of measures of risk taking in different decision areas. Findings have typically shown correlations between different measures of risk to be weak.
However, research on managerial decision making by MacCrimmon and Wehrung (1986), showed that this pattern of results does not preclude the possibility of strong intra-individual relationships between different measures of risk taking for some proportion of the population. They found that a small number of people showed consistent responses on different measures of risk taking, and could be classified by the authors as consistent risk seekers, or consistent risk averters.
A similar conclusion was reached in the work of Weber and Milliman (1997) who showed that underlying risk preferences tend to remain stable across situations for a significant portion of their sample. A key consideration in the risk literature is the notion of the stability of cross-domain risk preferences (e.g. Fagley & Miller, 1997; Weber & Milliman, 1997).
This is the question of whether it is possible to be risk seeking in some areas of one’s life and a risk averse in others, and the degree to which there could be said to exist trans-situational propensities for taking risks. Evidence suggests both general and domain-specific risk propensities are possible.
Our premise in this paper is that, to be viable, the construct of risk propensity should encompass several risk domains, i.e. that individuals can be validly characterized as consistent in their risk seeking or risk aversion across types of decision. People who are inconsistent in their approaches to risk can be regarded as lacking a strong propensity to either take or avoid risks.
These individuals can be seen as likely to take risks in some situations, but not others. The domains in which they take or avoid risks could vary, or could be consistent – a finance trader might take risks routinely at work, but avoid risk when making family, leisure and personal finance choices.
A number of theories and empirical studies on risk propensity have been published, the most sophisticated of which in the literature has been the modelling set out by Sitkin and Pablo (1992). In this it is suggested that the two key inputs to risk taking are risk perception and risk propensity, with risk propensity conceptualized as a confluence of dispositional tendencies, cognitive inputs and past experience.
Here, we seek to build upon this work, with some theoretical adaptation to accommodate a measure that could be widely applicable. In their discussion, Sitkin and Pablo define risk propensity as “the tendency of a decision maker either to take or to avoid risks” (p.12). It could be implied from this definition that risk orientation needs to be consistent across several decision domains, as discussed above.
The research we report here seeks to assess whether consistent cross-domain risk propensity is an empirically viable concept. A further issue of importance is whether risk propensity can be confirmed as a behavioural construct. In our research we define risk propensity as the frequency with which people do or do not take different kinds of risks, i.e. risk propensity is tested here as a summary concept for the risk taking behaviour of an individual across time and situations.
This modelling supports the idea that risk propensity will have a domain-general aspect, underpinned by stable personality dispositions, and domain-specific aspect, due to the situational inputs that evoke risk behaviours according to the match with the interests, skills and orientations of individuals.
Data on these themes could have important implications for our understanding and for the management of risk in organizations and it was with these considerations in mind that we sought to develop a scale that asked participants about their current and past risk behaviour in different domains.