In a cohort of 369 community-based older persons free of dementia, we examined the correlates of risk aversion and tested the hypothesis that cognitive ability is related to risk aversion. Correlation analyses revealed links between sex, education, income, and cognition with risk aversion; however, the association between age and risk aversion was not significant. Further, the results of the mixed model suggested that, when considered together, only global cognitive function and sex were associated with risk aversion. Importantly, the association of cognitive ability with risk aversion was robust in that it persisted even when we considered the potential influence of persons exhibiting the earliest signs of cognitive impairment, a stage commonly referred to as MCI, and after excluding those with MCI. These findings are the first that we are aware of addressing the relation of cognitive ability with risk aversion among older persons with a wide spectrum of cognitive abilities and provide new data regarding the relative influence of cognitive ability versus related demographic and contextual factors on risk preferences. The finding that cognitive ability is related to risk aversion in advanced age may have important implications for improving financial and health decision-making, health behaviors, and ultimately well being among older persons.
In recent years, risk preferences have emerged as a topic of great interest in economics, behavioral economics, and neuroeconomics studies in part because of their strong association with real world economic and health outcomes [3
]. The literature on risk aversion in aging is limited, however, consisting mostly of studies aimed to determine whether older persons are more or less risk averse than younger persons [21
]. We are not aware of any study that has examined the relation of cognition with risk aversion among older persons or that simultaneously considered the potential influence of age, education, sex and income, despite their well known associations with both risk aversion and cognition. The lack of focus on risk aversion reflects an important gap in the aging literature, particularly given that older adults have to make several complex decisions just at a time when, for many, cognitive abilities are beginning to deteriorate. In fact, emerging data suggest that older persons often make suboptimal financial and other decisions as compared to younger persons [21
]. The reasons why remain unknown, and risk preferences may represent an understudied yet potentially important factor. The present study identified cognitive ability and female sex as correlates of risk aversion in a diverse cohort of community based older persons and showed that the association of cognitive ability with risk aversion was robust among persons with a wide spectrum of cognitive function. If risk aversion is associated with real world decision making in aging as it is at younger ages, then these findings may suggest that older persons, particularly those with lower cognitive abilities, could benefit from assistance in understanding risk/benefit ratios and considering all possible options (not just the safe choice) when making consequential decisions.
Notably, in addition to education, sex, and income, some prior studies have reported associations between age and risk aversion, and the prevailing view seems to be that older age is associated with greater risk aversion [1
]. In correlation analyses, education, sex and income were associated with risk aversion in directions consistent with findings in younger persons; however, age was not significant, possibly due to the restricted age range of our sample. Nevertheless, as we noted earlier, it has been difficult to disentangle the effect of cognition from the effects of other demographic and contextual (e.g
., income) variables, which are related to both risk aversion and cognitive ability. The results of a mixed effect model suggest that, when the relevant correlates are considered simultaneously, only cognitive ability and sex are associated with risk aversion among older persons, with a trend for income. Thus, cognitive ability appears to be a robust determinant of risk aversion even in advanced age and among community based persons with a wide spectrum of cognitive function. Future research is needed to understand the trajectory of risk aversion in advanced age and to determine whether risk aversion changes over time with advancing age.
Although the association of cognitive ability with risk aversion has been examined in studies of younger persons, it remains unclear whether this is a general or domain-specific association either in younger or older persons [23
]. In this study, performances on semantic memory, episodic memory, working memory, and perceptual speed were related to risk aversion; visuospatial abilities were not. These findings suggest that the association of cognition with risk aversion is of a fairly general nature. Further, it is noteworthy that the strongest association was observed for semantic memory, the domain of cognition associated with ideas and concepts and that represents acquired knowledge about the world. In fact, semantic memory is considered by some to serve as an indicator of crystallized cognitive ability; thus, the strong association between semantic memory and risk aversion is consistent with the results from at least one prior study that examined "intellectual ability" using two subscales from an intelligence test and reported a general association between "intelligence" and risk aversion among a large sample of German adults [5
This study has a number of strengths, including the detailed assessment of risk aversion and cognition using standardized questions in a fairly large cohort of diverse community-dwelling older adults free of dementia. Additional strengths include the ability to examine the role of other important correlates and to conduct a number of sensitivity analyses to ensure that the findings were not due to the inclusion of persons with preclinical cognitive impairment. A limitation of the study is the selected nature of this volunteer cohort, which may have restricted our range of risk aversion and may limit the generalizability of findings. Another limitation was the assessment of risk aversion at a single point in time (rather than measuring change over time). Future studies are needed to investigate potential age-related changes in risk aversion and to examine the predictive association of risk aversion with financial and health decisions and outcomes in advanced age so that we may better understand how risk preferences impact real world outcomes across the lifespan.