We examine how functional-form assumptions affect conclusions about the existence of satiation in the relationship between income and emotional well-being. Previous studies have estimated a piecewise log-linear model with one structural break in both intercept and slope, which may lead to discontinuities in the estimated relationship. We show that imposing continuity in this type of model in OLS and quantile regressions substantially alters inferences about income satiation. At commonly applied thresholds, satiation disappears. The threshold shifts upwards when determined by best statistical fit and satiation re-emerges only at very high incomes. These findings demonstrate that the presence and location of satiation are highly sensitive to modeling choices.
Derlin et al. (Wed,) studied this question.