Reliability and Validity of the Happiness Approach to Measuring Preferences

While the use of happiness data to estimate “utility” functions has some interesting advantages over stated and revealed preferences methods and is growing in popularity, evidence on the reliability and validity of the happiness approach to measuring preferences is lacking. Moving beyond the intuitive appeal of estimating happiness functions, I draw on the literature in psychology on so-called psychometric quality to examine the following two features of the happiness approach to measuring preferences: (i) do repeated samples and different measures of happiness or subjective well-being (SWB) render similar preferences (what is called reliability)?; and (ii) do SWB-based preference measures relate to other measures that capture similar constructs in a logical way (what is called construct validity)? Empirical evidence indicates that SWB-based preferences exhibit high intertemporal, test-retest stability and are highly consistent when measured using alternative indicators of SWB (reliability). Similarly, SWB-based preferences relate to stated and revealed preferences measures of similar constructs in expected ways (construct validity). Overall, I conclude that estimating happiness (“utility”) functions provides a reliable and valid means for measuring people’s preferences.