The Relative Utility Hypothesis With and Without Self-reported Reference Wages

This article makes three main contributions to the economics of happiness literature. First, using a novel data set of about 90,000 Japanese workers surveyed in annual cross-sections between 1990 and 2004, it demonstrates that individuals experience strong disutility when they perceive that their coworkers earn relatively higher wages. In contrast with other tests of the relative utility hypothesis in the literature, our estimation relies on workers' self-reported beliefs of their peers' wages, which we argue are more closely aligned to the "true" reference-group benchmark than the assumed comparison income measures employed in other studies. Second, the article shows important heterogeneous effects of both absolute and relative income on happiness. In particular, workers who are better able to accurately predict their peers' wages seem to experience both greater utility of higher own income and greater disutility of higher relative income. Third, we assess the validity of different methodologies that the literature has employed to construct comparison income measures and find significant discrepancies, particularly when reference income is derived from Mincer equations--a common approach in other studies. We demonstrate that such discrepancies stem from the difficulty in finding valid exclusion restrictions that help identify the relative income effect on happiness. In the absence of self-reported reference wages, we propose a simple IV strategy that does not eliminate the lack of consistency but delivers a lower bound of the "true" effect.