شالوده شکنی تردمیل با لذت : آیا شادی خودبازگشت است؟

Affective habituation is well-documented in social sciences: people seem to adapt to many life events, ranging from lottery windfalls to terminal illnesses. A group of studies have tried to measure habituation by seeing how lagged values of life events affect present happiness. We propose an additional adaptation channel: current happiness may depend directly on past happiness, which amounts to assessing whether happiness is autoregressive. We run dynamic happiness regressions using individual-level panel data from the German Socio-Economic Panel Study, the Japanese Panel Survey of Consumers, the British Household Panel Survey and the Swiss Household Panel. As in previous studies, the coefficients on lagged events (e.g., becoming unemployed, getting married) suggest strong habituation. However, all the econometric models suggest that the coefficient on lagged happiness is positive and significant. We discuss whether this may be evidence of happiness having an inertial force (besides the usual habituation channel).

Brickman and Campbell (1971) coined the term “hedonic treadmill” to describe how people tend to adapt to good and bad events and then return to the same baseline level of happiness. Since then a number of papers in the fields of psychology and economics have shown some evidence that people adapt to particular life events, ranging from lottery windfalls (Brickman et al., 1978) to losing a limb to cancer (Tyc, 1992). A group of papers have used survey data to measure habituation through the lagged effects of life events: e.g., Lucas et al., 2003, Diener et al., 2006, Di Tella et al., 2007 and Clark et al., 2008. We propose an additional channel for adaptation: current happiness may depend directly on past happiness, which amounts to assessing whether happiness is autoregressive. We call this the “general habituation” channel: i.e. having experienced moments of happiness (unhappiness) in the present will directly make people prone to feelings of unhappiness (happiness) in the future, regardless of whether the original increase (decrease) in happiness was due to changes in income, health or love partners. As a matter of fact, in the psychology literature there are similar distinctions between the different adaptation channels: e.g., Kahneman (2000) distinguishes between the “hedonic treadmill” and the “satisfaction treadmill.”1 To the best of our knowledge, we are the first to run dynamic happiness regressions with individual-level data. We exploit data from the German Socio-Economic Panel Study, the Japanese Panel Survey of Consumers, the British Household Panel Survey and the Swiss Household Panel. We propose a variety of econometric models to overcome many identification challenges. In accordance with the existing results in the literature, we find that happiness increases one year after negative events like becoming unemployed or widowed, and it decreases one year after positive events like getting married or having children. However, the coefficient on lagged happiness is positive and statistically relevant. We discuss whether, instead of habituation, this may imply that past feelings of happiness have an inertial effect on contemporary happiness (besides the usual habituation channel). Section 2 briefly summarizes the literature on hedonic adaptation and discusses the difference between the general- and specific-habituation channels. Section 3 presents the econometric results. The final section concludes.

Motivated by a conceptual distinction in the interpretation of the hedonic adaptation hypothesis, we tried to assess whether happiness is autoregressive. Firstly, we should ask whether using dynamic instead of static regressions modifies the estimates on the (static and dynamic) coefficients of the most common variables in the happiness literature. As shown in Table 2, the coefficients remain roughly the same, especially if we take into consideration the great loss in efficiency and sample size implied by the dynamic econometric models. Secondly, we found that the coefficient on lagged happiness is positive and statistically significant. Taken at face value, that would suggest that happiness is inertial: i.e., ceteris paribus, having greater feelings of happiness in the past directly increases the probability of feeling happy in the present. This channel would work independently from the specific habituation channel: i.e., as shown in Table 2, it is still true that there is specific-habituation to many stimuli, like getting married or becoming unemployed. Notwithstanding, the positive autoregressive coefficient may have alternative interpretations in other model specifications (e.g., Wunder and Schwarze, 2010). We leave this as an open question for future research.