Reviewer 2: Estimating the 'cost' of happiness does not of itself require identification of one specific point of non-linearity

[datarichard]

Reviewer 2 (SSM0-submission):

My reaction to the approach, without knowing the arguments in favour of it, is that it is not fit for purpose and there are more direct and robust ways of answering the questions posed by the authors. I concede I may be wrong, but the authors have done very little to convince me of the merits of their methods. Estimating the 'cost' of happiness does not of itself require identification of one specific point of non-linearity (the so-called 'change point')

[datarichard]

Suggestion:

• [ ] Replace "cost of happiness" with another word to describe the marginal effect of income
• [ ] Add Figure S6 (Happiness at $50K vs $75K from 2001-2019) to the main text

[datarichard]

if the question concerns the extent to which income inequality contributes to subjective wellbeing inequality, surely it would make more sense to directly examine inequality in each in addition to the link between income and wellbeing

• [x] Add discussion of Gini coefficient. May be a duplicate of #17

[datarichard]

The Gini coefficient is discussed in Section 4.2, and in particular the inadequacy of univariate measures of inequality that the reviewer seems to refer to. Univariate distributions of happiness or income are irrelevant to their joint distribution and its implication, as captured by our model. One implication is shown in Figure S6 which visualises the increasing disparity in happiness between the rich and the poor (now mentioned in the main text of Section 4.0, paragraph 3).

An outstanding issue is whether to include this figure in the main document.