Irrational Economist_ Making Decisions in a Dangerous World - Erwann Michel-Kerjan [29]
The vigorous debates about how widely accepted various “self-evident” principles of rational choice theory are raised two challenges for decision analysis. First, if not everyone buys into them, how rational can they really be? Second, if decision analysts use their authority, intellect, and confidence to persuade clients to revise their basic preferences and beliefs to comply with the dictates of their model, what do these revised opinions represent? Do such artificially constructed measures really capture people’s true values and beliefs about risk and return? Indeed, do they measure anything at all, other than compliance? For example, we can counsel people not to pay attention to feelings of regret, either before or after a decision, because it is irrational to worry about things you can’t control. But if these sentiments run deep, they cannot and should not be ignored. As Pascal warned, the mind can never fully know what stirs the heart, and yet disregarding this advice is precisely what decision analysis seeks to do. Telling a child not to be afraid after seeing a scary movie does not always work, however rational the advice may be. A better, more psychologically rooted approach would be to tell the child a happy story so as to shift attention away from the scary one. Overly rational models do not always offer the best solution.
WHAT IS BEHAVIORAL DECISION THEORY?
Behavioral decision research is the study of how people make decisions in actuality, be it in controlled laboratory settings or in the messiness of the real world. This descriptive research draws mostly on cognitive psychology and suggests that people have the greatest trouble making good decisions when dealing with losses and/or ambiguous probabilities.4 Howard Kunreuther was a pioneer in showing that we often make suboptimal decisions about insurance, especially when they involve very low probabilities. For example, many people in high-risk areas decline flood and earthquake insurance, even when such policies are heavily subsidized by the federal government. And yet, these same people will gladly buy very expensive air-accident insurance (from machines placed at airports) or opt for costly medical insurance policies with low deductibles. How can people be saved from their own folly? Is it really just a matter of sitting down with a decision analyst (in the form of either a person or a software package), constructing a utility function, and calculating expected utilities? Unfortunately, it is not quite that simple. For the rational model to work its magic, decision makers must be able to articulate their basic taste and preferences and make sure they do not violate basic axioms of rational choice. As noted above, that is a real problem.
Decision analysis and other forms of decision support try to help overcome people’s inherent information processing limitations. However, behavioral research very much undermines some core premises of the prescriptive model, namely that people have coherent beliefs and preferences (and wish to obey the axioms). Also, the behavioral research calls into question the validity of various methods used to assess utilities as well as probabilities. This “double-edged sword” has not been as fully appreciated in the decision theory literature as it should be, especially among decision analysts.
LIMITS OF THE RATIONAL MODEL
I briefly address here why and where behavioral perspectives need more consideration in decision analysis, starting with the fundamental decomposition and aggregation assumptions underlying the classic normative model. Decomposing a decision into components