Round 1

Reviewer 1 Report

The use of "international dollars" should be better explained. Are these PPP (parity purchase power) adjusted incomes? if yes, why are the standard errors presented in Table 1 so high? It needs to be explained and discussed since it potentially erodes the entire results.

For Table 2 I would strongly recommend instead of coefficients, using marginal effects (derivatives) to better interpret the likelihood associated with the effects of each of the variables. Especially for these of the dummies used.

For the figures used in the Appendix (page 13), I would recommend using box plots instead of bar graphs to capture the life satisfaction distribution.

Actually, it would be better to add second y axes and add box plots for the income distribution. And in the appendix also redo the picture for each of the income classes.

Author Response

Reviewer # 1

Comments to the Author

• The use of "international dollars" should be better explained. Are these PPP (parity purchase power) adjusted incomes? if yes, why are the standard errors presented in Table 1 so high? It needs to be explained and discussed since it potentially erodes the entire results.

We thank reviewer 1 for raising this concern. We have amended the manuscript to include further details on how the conversion from local currency to international dollars is conducted by the Gallup World Poll. Please refer to a newly added footnoted in track changes format.

The manuscript in its current shape highlights on the high skewness in individual income, as documented by the high standard deviations from the reported means of income in Table 1. This motivates the use of a (natural) logarithmic transformation of income in our analysis.

• For Table 2, I would strongly recommend instead of coefficients, using marginal effects (derivatives) to better interpret the likelihood associated with the effects of each of the variables. Especially for these of the dummies used.

We thank reviewer 1 for this suggestion and we agree with the concern raised regarding the validity of relying on Table 2 coefficients for examining the likelihood analysis. However, the manuscript as it currently stands, refrains from interpreting the likelihood associated with the effects for each of the variables as well as the magnitude of the effects using Table 2 coefficients, that are obtained from an ordered response logit modelling estimation. As discussed, in text, we have standardized the dependent variable across all respondents within a year—country specific combination, to have a mean of zero and a standard deviation of one for life satisfaction responses. We have then conducted an ordinary least squares estimation modelling to examine the size of the effects. Further robustness and sensitivity analyses have also been undertaken to check the validity of our results.

Alternatively, we could report marginal effects as the reviewer suggests. We have now modified the manuscript to discuss, in writing, the marginal effects for our preferred specification (i.e., column (4) of Table 2). See in track changes (Section 5.2).

• For the figures used in the Appendix (page 13), I would recommend using box plots instead of bar graphs to capture the life satisfaction distribution. Actually, it would be better to add second y axes and add box plots for the income distribution. And in the appendix also redo the picture for each of the income classes.

Following other reviewers’ comments to substantially reduce the annexes, and to cut this specific section (previously noted Appendix Section 7.3), we have decided to omit the discussion on the income distribution from the Appendix and therefore figures associated to it. We have also moved many parts from the Appendix to in-text. All these changes are marked in track format.

We however remain open to re-incorporating those in future rounds of revision, if reviewer 1 suggests that this is a key part of the analysis and should be further explored using the alternative specification that is mentioned in here. As it currently stands, reviewers 2 and 3 comments suggest undertaking a significant re-structuring of the main text/appendix. We have done so accordingly for this round of revisions.

Reviewer 2 Report

The article "Does Income Class Affect Life Satisfaction? New Evidence from 2 Cross-country Microdata" seems interesting to me and I consider that the methodology used is correct

I will continue to make some comments:

The structure of the paper seems confusing to me. Parts of the text are taken to the appendices that, due to their relevance, should be part of the main text, on the contrary, I think other appendices are left over or should be reduced in length.

Motivate the importance of work and its interest in science

Page 2, line 82 et seq comment a few lines from the bibliography.

Move from appendix 7.1 the bibliographic references to section 2 "Subjective wellbeing and inequality: Theory and mechanism" which has been too short.

Move from appendix 7.2 the bibliographical references to section 2 Subjective wellbeing and inequality: Theory and mechanism" which has become very short.

Would it be possible to use another measure of well-being that was not subjective? In my opinion, objective measurements are generally more accurate than subjective ones.

I think that the section referring to Income and other explanatory variables (3.2) could be presented in a table, which would improve clarity

The authors have assessed the application of a panel model with fixed/random effects. Why do you think it is more appropriate to use an ordered logit?

Appendices 7.3 and 7.4 I think can be reduced or even deleted.

Table 1, I think, can be taken to the appendix, since it is of less interest.

Place annex 7.7, summarized, in the main text (section 5.2). Explain and motivate the use of OLS

The conclusions should be expanded

Author Response

Reviewer # 2

Comments to the Author

The article "Does Income Class Affect Life Satisfaction? New Evidence from 2 Cross-country Microdata" seems interesting to me and I consider that the methodology used is correct.

We thank reviewer 2 for the positive feedback regarding our manuscript.

I will continue to make some comments:

• The structure of the paper seems confusing to me. Parts of the text are taken to the appendices that, due to their relevance, should be part of the main text, on the contrary, I think other appendices are left over or should be reduced in length.

We apologize for any confusion induced by the structuring of the manuscript in its current format. We have now amended the manuscript to be more comprehensive of many parts incorporated in the appendix. We have also omitted many parts to cut the length of the annexes. To do so, we have followed the comments and suggestions of all three reviewers. Changes are in track format.

• Motivate the importance of work and its interest in science

We thank the reviewer for this. We have now amended the manuscript to add a section on the general importance and relevance of this analysis. These complements previous sections in-text that thoroughly discuss the contribution and significance of this work.

• Page 2, line 82 et seq comment a few lines from the bibliography.

We thank the reviewer for this suggestion. We have now modified the manuscript to comment some of the papers we cite from the literature on the impact of income inequality on wellbeing. We have chosen to describe papers that have found controversial links.  

• Move from appendix 7.1 the bibliographic references to section 2 "Subjective wellbeing and inequality: Theory and mechanism" which has been too short.

We thank the reviewer for this suggestion. We have now modified Section 2, to further include the additional theories and mechanisms discussed in (previously noted) Appendix Section 7.1. This substantially enhances and enriches the discussion on the alternative theories and mechanisms linking income, income inequality and subjective wellbeing.

• Move from appendix 7.2 the bibliographical references to section 2 Subjective wellbeing and inequality: Theory and mechanism" which has become very short.

We thank you for this comment. As per our previous response, we have amended the paper to move the content of (previously noted) Appendix Section 7.2 to Sections 2 and  3.

• Would it be possible to use another measure of well-being that was not subjective? In my opinion, objective measurements are generally more accurate than subjective ones.

While we appreciate the reviewer’s concern, we believe that one of the key contributions of our manuscript is to further enrich the literature on subjective wellbeing. Subjective measurements of wellbeing have been documented to be crucial for the overall welfare analysis. Moving to objective measurements of wellbeing diverts us from the main objective of our analysis.

This is particularly important given that the literature on income and its consequences for human welfare with a focus on subject wellbeing have yielded controversial results thus far, with a lack of consensus on whether individuals residing in highly income dispersed places have less social wellbeing in comparison to more equally income distributed places. This is especially relevant given the various theories and mechanisms that exist and that link income, income inequality and subjective wellbeing.

The novelty of our analysis (among others) remains in the large sample used, as well as the methodology incorporated to tackle this question.

• I think that the section referring to Income and other explanatory variables (3.2) could be presented in a table, which would improve clarity

In Appendix Section 7.5 (in the revised version of the manuscript), we report a detailed table (Appendix Table A4) displaying Gallup World Poll variables used in this paper, including income and other explanatory controls. We therefore refrain from adding an additional table to avoid redundancy. We are however willing to moving the table in-text upon reviewer’s concern. For this round of revisions, we have referred in-text, in subsection 3.2, to this appendix table.

• The authors have assessed the application of a panel model with fixed/random effects. Why do you think it is more appropriate to use an ordered logit?

Given the nature of the dependent variable of interest, our measure of global life evaluation,  from the World Gallup Poll data, the Cantril Ladder of Life, which is recorded on a 0–10 scale with end points labelled “Worst possible life for you” and “Best possible life for you”, it remains crucial to undertake the ordered logit analysis. This has been done using the panel data with both fixed and random effects. However, as previously mentioned, we have also relied on an ordinary least squares estimation with standardized (year and country specific) answers of respondents to further assess the magnitude of the effects.

We have also undertaken sensitivity and robustness analysis, including the ordinary least squares with a continuous 0—10 scaling of the dependent variable without standardization, as well as the one way fixed effects modelling.

• Appendices 7.3 and 7.4 I think can be reduced or even deleted.

We thank the reviewer for this suggestion. We have now deleted annex 7.3 from the analysis. However, we keep the discussion on the goodness of fit of our analysis (previously Appendix Section 7.4) as this constitutes a crucial critique to the Latent Class Analysis modelling.

• Table 1, I think, can be taken to the appendix, since it is of less interest.

We understand the reviewer’s concern, of plausibly distracting the reader from the main results. However, we believe that it is crucial to present the characteristics of various respondents belonging to different hypothesized income classes for a number of reasons.

First, the variables displayed have been documented in the literature to be determining factors of individual’s wellbeing. It is thus important to present summary statistics of these variables.

Second, the various income classes are hypothesized using the predicted posterior probability methodologies. We thus feel the need to present a balance test prior to undertaking the analysis on linking income classes and life satisfaction.

Lastly, and given the comments we received in this round of revision, especially given that we have now omitted (as per the previous comment from this report) the discussion on general sample descriptive statistics (socio-demographic characteristics) and subjective wellbeing plots, it remains crucial to have an overview of these statistics.

• Place annex 7.7, summarized, in the main text (section 5.2). Explain and motivate the use of OLS

We thank the reviewer for this comment. We have moved Appendix Section 7.7 in text, particularly to Section 5.2. We have also emphasized on the need to check the validity of our ordered probit response modelling results using an OLS estimation and report these in-text.

• The conclusions should be expanded

In this round of revisions, we have modified the manuscript to further expand the conclusions as suggested by reviewer 2.

Reviewer 3 Report

The paper analyzes the impact of income class on subjective wellbeing. The Authors used rich data from the Gallup World Poll. The topic is very interesting and the research results should be presented in high-quality Journal.

Strengths of the article:

-      -  The scope of the research – the  sample included 1,294,943 respondents in 160 countries between 2009—2017.

-      -  Contribution to the literature that the Authors present in line 80-88. The article on this quality has to have a strong impact to the current state of knowledge. And this one has.

-             Very interesting and significant empirical analysis, that was conducted appropriately.

Weaknesses of the paper:

-             The literature review has to be improved.  The Authors should significantly extend this part by providing in-depth literature studies which will present who and when previously did similar research in this field. They said that “The closest methodological analogy to our paper is a work by Anderson et al. (2016)” but who and when previously explored the relation between income class and subjective wellbeing. It was explored for shure. What did they obtain?

-             The paper should be re-structured. The most interesting analysis and data are presented in the appendix. The reader has an impression that the appendix is a part of the main text and at the end there is no conclusions and discussion part, because it was presented in the middle of the article. It may be confused because of that. The appendix may includes data but not with the most important analysis of them. It should be presented in the main part "Results".

Summarizing, the article should be assess high, but it needs some improvements before publication.

Author Response

Reviewer # 3

Comments to the Author

The paper analyzes the impact of income class on subjective wellbeing. The Authors used rich data from the Gallup World Poll. The topic is very interesting and the research results should be presented in high-quality Journal.

Strengths of the article:

     -  The scope of the research – the  sample included 1,294,943 respondents in 160 countries between 2009—2017.

    -  Contribution to the literature that the Authors present in line 80-88. The article on this quality has to have a strong impact to the current state of knowledge. And this one has.

-             Very interesting and significant empirical analysis, that was conducted appropriately.

We thank and appreciate the positive feedback from reviewer 3.

Weaknesses of the paper:

• The literature review has to be improved.  The Authors should significantly extend this part by providing in-depth literature studies which will present who and when previously did similar research in this field. They said that “The closest methodological analogy to our paper is a work by Anderson et al. (2016)” but who and when previously explored the relation between income class and subjective wellbeing. It was explored for shure. What did they obtain?

In the present format of the manuscript, we thoroughly discuss previous research in the Introduction section. We also provide a detailed overview in Section 2 of various findings connecting to existing theories and mechanisms which highlight the controversial findings linking income, income inequality and subjective wellbeing. For this round of revisions, we have now further modified Section 2 based on additional comments from reviewers, to enrich this discussion.

We have also modified the Introduction Section to add more recent findings linking income class and subjective wellbeing. We are also willing, if the reviewer suggests, to further append Section 2 by reviewing and restating the findings from the literature that are included in the Introduction. This allows the reader to follow a clear flow of literature reviewing.

It must be noted that we specifically highlight on Anderson et al. (2016) paper as it directly links to our manuscript with respect to the novel methodology used (Latent Class Analysis) to explore the link between income, income inequality and subjective wellbeing that is treated in-depth in the literature as the reviewer mentions. This does not in any way ignore other findings or other papers in the literature.

2)            The paper should be re-structured. The most interesting analysis and data are presented in the appendix. The reader has an impression that the appendix is a part of the main text and at the end there is no conclusions and discussion part, because it was presented in the middle of the article. It may be confused because of that. The appendix may includes data but not with the most important analysis of them. It should be presented in the main part "Results".

Based on all three reviewers’ comments and suggestions, we have now significantly amended the manuscript to reduce the length of the appendix by moving many parts in-text and omitting others. This avoids any confusion to the reader and highlights some of the crucial findings that were otherwise presented in various annexes. All of these changes are in track format.

Summarizing, the article should be assess high, but it needs some improvements before publication.