Abstract
This paper addresses the well-known question of what drives people's well-being using two alternative measures of subjective well-being and comparing two econometric approaches, thus providing results robust to the recent critique by Bond and Lang (2019). The classical OLS and ordered probit analysis of self-reported life satisfaction of employees from 32 European countries show results consistent with the previous literature.
Analysis of the happiness index - a measure of hedonic well-being defined as frequency of experiencing specific emotions - provides similar results, with some exceptions. Most importantly, we show that the observed income effect on subjective well-being is much weaker for the happiness index than for life satisfaction, especially when controlling for satisfaction of basic needs.
Quantile regression analysis brings additional insights: (1) median estimates are equivalent to mean estimates obtained by OLS (2) the correlates of subjective well-being are not stable over the whole distribution with most of the coefficients being the largest in their absolute value at low quantiles (3) the relationship between income and the happiness index is weak and stable over the whole distribution when basic needs satisfaction variables are included in the model.