People do not adapt to income changes: A re-evaluation of the dynamic effects of (reference) income on life satisfaction with GSOEP and UKHLS data

Do people adapt to changes in income? This paper shows that there is no evidence of adaptation to income in GSOEP (1984-2015) and UKHLS (1996-2015) data. Following the empirical approach of Vendrik (2013), I arrive at this surprising answer by estimating (dynamic) life satisfaction equations, in which I simultaneously enter contemporaneous and lagged terms for a respondent's own house- hold income and their estimated reference income. Additionally, I instrument for own income and include lags of a large set of controls. Furthermore, I find that people also do not adapt to changes in reference income. Instead, reference income e ects may be subject to reinforcement over time. To explain my findings, a comprehensive account of the puzzling and often divergent results of Ferrer-i Carbonell and Van Praag (2008), Binder and Coad (2010), Di Tella et al. (2010), and Pfa (2013) is given. What was found to be adaptation to raw household income in these studies turns out to have been driven by reinforcement of an initially small negative e ect of household size that grows large over time. Implications of this result for the estimation of equivalence scales with subjective data are discussed.