RELIABILITY OF STATED PREFERENCE METHODS

Many times governments and policy makers have to choose among different projects or policies to implement. In principle, the best choice is the one which maximizes the social welfare that, in turn, depends on individual preferences. But very often preferences are unknown and even not observable. In practice, a common procedure is to directly ask a sample of individuals about their preferences, which are therefore stated by agents rather than revealed by their behaviour. Methods for preference revelation can be classified into two broad families. The first one involves the case in which respondents are asked to simulate their market behaviour in a fictitious context designed by the researcher. The final goal of these studies is the estimation of willingness to pay (WTP), or willingness to accept (WTA), for changes in provision of non-market goods. A large literature investigates both theoretical issues connected with these procedures (Bates, 1988) and empirical results from country experiences (Mackie at al., 2003). The second family of surveys are commonly employed in public opinion analysis. In this case respondents are asked to reveal their current attitudes, whilst in some circumstances they are required to state their satisfaction with a certain policy or service. In the last decades the interest towards such analysis largely increased and a broad amount of surveys have been systematically collected (Rabin, 2002). Whatever the kind of analysis, when individuals correctly report the behaviour they would keep in a real context, or honestly admit their attitudes and perceptions, the target of the policy maker is reached. Hence, the issue of reliability of stated preferences becomes crucial in order to understand what we can learn from surveys and how SP analysis can be exploited by policy makers. Our research question is simply the following one: can we trust in SP methods? In order to answer this question the work is organised in three sections. The first one is devoted to the definition of the concept of “reliability”. In the first place, the latter depends on the family of SP methods we are dealing with. When individuals are required to replicate their market behaviour in a fictitious scenario, two perspectives can be applied: the first one based on mainstream economic theory (Hicks and Allen, 1934) and the other one in accordance to the so called behavioural programme (Sunstein and Thaler, 2008). Both approaches are discussed, pointing out the problematic issues which characterise each methodology and trying to propose a definition for the concept of reliability. The second family of surveys can be classified into two sub-groups, based on the object of the analysis. The first group includes all situations where agents are required to reveal their actual behaviour (Bertrand and Mullainathan, 2001) while the second one is composed by those studies in which agents are asked to express their feelings or perceptions about a certain aspect of their life (McFadden et al. 2005). Again, the concept of reliability has been investigated for each group of surveys. The second and the third sections are devoted to empirical works which try, recalling the definition of reliability suggested in the first chapter, to apply this concept to empirical studies.