The economics of addiction: an experimental investigation

Doctoral Thesis


Permanent link to this Item
Journal Title
Link to Journal
Journal ISSN
Volume Title
Addiction is an ideal puzzle for economic theory: why do most addicts expend resources to acquire their targets of addiction but then incur real costs to try and reduce or limit their consumption of these goods? Furthermore, why is the typical course of addiction characterised by repeated unsuccessful attempts to quit prior to final abstention? From the standpoint of standard consumer theory in economics these patterns of behaviour are difficult to rationalise. There is a rich theoretical literature in economics which models habit-forming behaviours, of which addiction is the exemplar, but there is a paucity of experimental economic studies eliciting and comparing the preferences – specifically, risk and time preferences – that economic theory suggests may differ between addicts and non-addicts. The experimental research that has been conducted has been dominated by psychologists, and some economists have begun to follow their methodological lead. However, detailed reviews of the experimental literature on addiction highlight a number of methodological and statistical flaws in the ways these data have been collected and analysed. This thesis is primarily concerned with methodological and statistical issues at the boundary between economics and psychology as these bear upon developing a general, consistent explanation of addiction. An incentive-compatible experimental design is formulated which lends itself to the estimation of several different theories of choice under risk and over time. In addition, a full information maximum likelihood statistical framework, which is consistent with the data generating processes proposed by structural theories and accounts for subject errors in decision making, is used to explore the relationship between risk preferences, time preferences and addiction. This thesis challenges some of the maintained assumptions in the addiction literature; e.g., that the probability discounting and hyperbolic discounting models best characterise choice under risk and over time, respectively. But it also replicates a previous finding; i.e., that smokers tend to discount the future more heavily than non-smokers. It shows, therefore, that some results withstand careful methodological and statistical scrutiny, whereas other results do not. Ultimately, this thesis argues that tools from experimental economics and econometrics, which have been under-used in addiction studies, contribute to a more accurate and reliable characterisation of this phenomenon.