Subjective Probabilities in Choice Experiments' Design: Three Essays

This dissertation which consists of three essays investigates the influence of subjective probabilities on decision making processes under conditions of risk. In particular, it examines whether subjects adjust new risk information on their prior subjective estimates, and, to what extent this adjustment affects their choices. In the first essay, by using an artefactual field experiment, I examine the potential correlation between incentive compatibility and validity of subjective probabilities elicited via the Exchangeability Method, an innovative elicitation mechanism which consists of several chained questions. Here, validity is investigated using de Finettiâs notion of coherence under which subjective probabilities are coherent if and only if they obey all axioms and theorems of probability theory. Experimental results suggest that subjects provided with monetary incentives and randomized questions more likely express valid subjective probabilities than others because they are not aware of the chaining which undermines the incentive compatibility of the Exchangeability Method. In the second essay, by using the same experimental data, I show that valid subjective probabilities do not significantly diverge from invalid ones, indicative of little effect of internal validity on the actual magnitude of subjective probabilities. In the third essay, by using a field Choice Experiment, I investigate to what extent subjects adjust risk information given in the status quo alternative on their subjective probability estimates. An innovative two-stage approach that incorporates subjective probabilities into Choice Experimentsâ design is developed to investigate this phenomenon, known as the scenario adjustment. In the first stage, subjective probabilities that given outcomes will occur are elicited using the Exchangeability Method. In the second stage, two treatment groups are designed: in the first group, each subject is presented with a status quo alternative which incorporates her/his subjective probabilities, and, hence, no adjustment is required; in the second group, each subject faces a status quo alternative where the presented risk is not consistent with her/his probability estimates, and, hence, a mental adjustment to the scenario might take place. By comparing willingness to pay across the treatment groups, my results suggest that, when subjects are provided with SQ alternatives in which the risk is lower than the perceived one, the mental adjustment takes place, but, when subjects are provided with SQ alternatives in which the risk is higher than their own estimates, these subjects appear to make irrational choices