How stable are economic preferences? Evidence from survey-based measures

An important assumption in most economic models is that preferences of individuals are exogenously given, and hence are not affected by experiences and/or background. In recent years, this neo-classical view has been challenged by findings from the literature on behavioral economics and finance. In this relatively new field of inquiry, individual preferences are assumed to be shaped and influenced by their economic, social, and political environment, and hence in this sense are endogenous. This view also accords an important role to cultural norms and religious worldviews in shaping the economic preferences of individuals. In this paper, we exploit a unique panel of data on Japanese households and compute measures of important economic preferences, namely, risk, time, and altruistic preferences. The survey we use in this study is the Preference Parameters Study (PPS, henceforth), a household survey carried out by the Osaka University in Japan. In addition to providing information that allows measurement of the aforementioned preferences, this survey also contains detailed information on household composition, consumption, income, and socio-demographic characteristics such as gender, race, religion etc. In this study, we exploit the panel structure of this data to for the stability of these preferences using data for five consecutive waves from 2005-2009. An important contribution of our paper is to filter out real change in these preferences from random noise that can be attributed to measurement error, among other things. For this purpose, we use two alternative econometric approaches. First, is the latent Markov model where true responses are modeled as latent variables that follow a first-order Markov chain. Second, is the correlated random effect models with measurement error. Preliminary results indicate substantial stability in these preferences over our sample period.