A methodological challenge in testing Easterlin's paradox, which states that increasing income fails to boost happiness beyond a satiation point, lies in the determination of this threshold if it exists. In the existing literature, various levels of GDP per capita have been chosen based upon visual inspection of the scatter plots of the data. The estimated income-well-being gradients are sensitive to the choice of such kink points. We first replicate the results in Stevenson and Wolfers (American Economic Review, Papers and Proceedings 2013; 103(3): 598-604), then apply the methods proposed in Hansen (Journal of Business and Economic Statistics 2016), which estimate the regression kink model with an unknown threshold, to the data in Stevenson and Wolfers (2013) to re-examine the two modified versions of Easterlin's hypothesis. In most cases, we do not find the existence of any kink point. In the few cases where a kink point is detected, the estimated kink point thresholds are different from those chosen in the previous literature. Overall, we find no evidence of a satiation point, which is in support of Stevenson and Wolfers (2013).
