Abstract/Summary

Attitudes to risk have generated a lot of attention over the years due to its vital importance in decision-making processes that are necessary for life and livelihoods. Attitudes towards uncertainty have received less attention even though arguably most important decisions are under uncertainty rather than risk. In addition, many studies modelling attitudes to risk have adopted experiments that place significant cognitive burden on respondents. Crucially, they are also framed in a way that do not reflect everyday problems. Specifically, the most common way of eliciting attitudes is to ask decision makers to choose between discrete monetary lotteries with known probabilities attached to the payoffs. Yet, arguably, the vast majority of choices that people make in their day-to-day lives are with respect to continuous non-monetary outcomes. To address these gaps, this thesis investigates responses to continuous ‘prospects’ across different conditions (risk & uncertainty), contexts (monetary & time) and content domains (gain, loss & mixed). Further, this thesis examines the link between attitudes to risk/uncertainty and mental health related factors and the effect of attitudes to risk and uncertainty on farmers’ decisions both for themselves and for others. This thesis uses both non-parametric methods - relating to the patterns that characterise participants’ choices and their determinants; and parametric models – based upon cumulative prospect theory (CPT) as it extends to continuous prospects. The data were gathered using lab-in-field experiments in which Nigerian farmer’s chose between pairs of prospects with continuous distributions, which were not exclusively monetary in nature. Attitudes towards risk, as opposed to uncertainty were elicited by specifying that all outcomes over the specified interval were ‘equally likely’ (thus a uniform probability density). Uncertainty was specified by indicating to farmers that one outcome within the specified interval would be realised but without the specification of an associated probability density. Key findings are that attitudes differ under different conditions, contexts and content domains. Using continuous prospects, respondents did not treat equally likely outcomes as ‘equally likely’ and appear to demonstrate cumulative probability distribution warping consistent with the CPT. However, there were behaviours that are difficult to reconcile with CPT such as the preferences of many respondents could only be modelled using “extreme curvature” of the value function. This was induced by what we term negligible gain avoidance (i.e. avoiding prospects with zero lower bound in the gain domain) or negligible loss seeking (i.e. preferring prospects with zero upper bound in the loss domain) behaviours. CPT, Salience theory, Heuristics and other theories examined in this study could not alone explain these behaviours. Results from investigating the effect of bipolar disorder tendencies (BD) on risk attitudes show that BD significantly affects the shape of the value and probability weighting functions; and farmers that have BD are more likely to make random choices. Other results show that risk aversion for losses increases participation in off-farm income generating activities; and that farmers’ likelihood to engage in specific types of offfarm activities is determined by their risk and uncertainty attitudes.