Dynamic Consistency in Denmark: A Longitudinal Field Experiment

Abstract. Evidence that individuals have dynamically consistent preferences is usually generated by studying the discount rates of the individual over different horizons, but where those rates are elicited at a single point in time. If these elicited discount rates vary by horizon the individual is typically claimed to have preferences that imply a dynamic inconsistency, although this inference requires additional assumptions such as intertemporal separability. However, what one really wants to know is if the same subject has the same discount rate function when that individual is asked at a later point in time. Such panel tests then require than one allow for possible changes in the states of nature that the subject faces, since they may confound any in-sample comparisons of discount rate functions at different points in time. We report the results of a large-scale panel experiment undertaken in the field that allows us to examine this issue. In June 2003 we elicited subjective discount rates from 253 subjects, representative of the adult Danish population. Between September 2003 and November 2004 we re-visited 97 of these subjects and repeated these tasks. In each visit we also elicited information on their individual characteristics, as well as their expectations about the state of their own economic situation and macroeconomic variables. We find evidence in favor of dynamic consistency