Income Inequality and Happiness: A Meta-Analysis

• Název práce v češtině: Ekonomická nerovnost a percepce štěstí: Meta-analýza
• Název v anglickém jazyce: Income Inequality and Happiness: A Meta-Analysis
• Klíčová slova: meta-analýza, ekonomická nerovnost, štěstí, bayesian model averaging, publikační selektivita
• Klíčová slova anglicky: meta-analysis, income inequality, well-being, happiness, bayesian model averaging, publication bias
• Akademický rok vypsání: 2019/2020
• Typ práce: diplomová práce
• Jazyk práce: angličtina
• Ústav: Institut ekonomických studií (23-IES)
• Vedoucí / školitel: doc. PhDr. Zuzana Havránková, Ph.D.
• Řešitel: skrytý - zadáno vedoucím/školitelem
• Datum přihlášení: 23.07.2020
• Datum zadání: 23.07.2020
• Datum a čas obhajoby: 16.06.2021 09:00
• Místo konání obhajoby: Výuka probíhá online, JONLINE, Pomocná místnost pro rozvrhování výuky probíhají online
• Datum odevzdání elektronické podoby: 03.05.2021
• Datum proběhlé obhajoby: 16.06.2021
• Oponenti: doc. PhDr. Julie Chytilová, Ph.D.

Seznam odborné literatury

Alesina, A., Di Tella, R., & R. MacCulloch (2004): Inequality and happiness: are Europeans and Americans different? Journal of Public Economics, 88(9–10), 2009–2042.
Amini, S. M. & C. F. Parmeter (2012): Comparison of model averaging techniques: Assessing growth determinants. Journal of Applied Econometrics 27(5): pp. 870–876.
Andrews, I. & M. Kasy (2019): Identification of and Correction for Publication Bias. American Economic Review 109(8), 2766-2794.Deisenhammer E.A., Kemmler G. & P. Parson (2003): Association of meteorological factors with suicide. Acta Psychiatrica Scandinavica 108(6), 455–459.
Auten, G. & D. Splinter (2019): Income Inequality in the United States: Using Tax Data to Measure Long-term Trends. American Economic Review: Papers & Proceedings 109, 307–311
Benabou, R. & E.A. Ok (2001): Social mobility and the demand for redistribution: The POUM hypothesis. The Quarterly Journal of Economics 116: 447–487.
Berg, M., & R. Veenhoven (2010): Income inequality and happiness in 119 nations: in search for an optimum that does not appear to exist. In B. Greve, (Ed.), Happiness and social policy in Europe (pp. 174–194). Cheltenham: Edward Elgar.
Bjornskov, C., Dreher, A., Fischer, J.A.V., Schnellenbach, & K. Gehring (2013): Inequality and happiness: When perceived social mobility and economic reality do not match. Journal of Economic Behavior & Organization, 91, 75–92.
Blanchflower, D. G., & A. J. Oswald (2004): Well-being over time in Britain and the USA. Journal of Public Economics, 88(7–8), 1359–1386.
Bom, P.R.D. & H. Rachinger (2019): A Kinked Meta-Regression Model for Publication Bias Correction. Research Synthesis Methods (forthcoming).
Brockmann, H., Delhey, J., Welzel, Ch. & H. Yuan (2009): The China Puzzle: Falling Happiness in a Rising Economy. Journal of Happiness Studies 10: 387–405.
Clark A. E., Fleche, S. & C. Senik (2016): Economic Growth Evens Out Happiness: Evidence from Six Surveys. Review of Income and Wealth 62(3): 405–419.
Dawes, C.T., Fowler, J.H., Johnson, T., McElreath, R. & O. Smirnov (2007): Egalitarian motives in humans. Nature 446: 794–796.
Duesenberry, J.S. (1949): Income, Saving and the Theory of Consumer Behavior. Harvard University Press, Cambridge: MA.
Dynan, K. E., & E. Ravina (2007): Increasing income inequality, external habits, and self-reported happiness. American Economic Review, 97(2), 226–231.
Easterlin, R., Morgan, R., Switek, M. & F. Wang (2012): China's Life Satisfaction, 1990-2010. Proceedings of the National Academy of Sciences of the United States of America 109: 9775–9780.
Egger, M., G. D. Smith, M. Schneider, & C. Minder (1997): Bias in meta-analysis detected by a simple, graphical test. British Medical Journal 315(7109): 629–634.
Ferrer-i-Carbonell, A. & X. Ramos (2013): Inequality and Happiness. Journal of Economic Surveys 28(5): 1016–1027.
George, E. I. (2010): Dilution priors: Compensating for model space redundancy. In IMS Collections Borrowing Strength: Theory Powering Applications -A Festschrift for Lawrence D. Brown, volume 6, p. 158-165. Institute of Mathematical Statistics.
Graham, C., & A. Felton (2006). Inequality and happiness: Insights from Latin America. Journal of Economic Inequality 4: 107–122.
Furukawa, C. (2019): Publication Bias under Aggregation Frictions: Theory, Evidence, and a New Correction Method. MIT working paper, Massetschussets Institute of Technology.
Hedges, L. V. (1992): Modeling publication selection effects in meta-analysis. Statistical Science 7(2): 246–255.
Havranek, T. (2015): Measuring Intertemporal Substitution: The Importance of Method Choices and Selective Reporting. Journal of the European Economic Association 13(6), 1180–1204.
Inglehart, R. (1997). Modernization and post modernization: Cultural, economic, and political change in 43 societies. Princeton: Princeton University Press.
Ioannidis, J.P., Stanley, T.D., & H. Doucouliagos (2017): The Power of Bias in Economics Research. Economic Journal 127(605): 236–265.
Jiang, S., Lu, M., & H. Sato (2012): Identity, inequality, and happiness: Evidence from urban China. World Development 40(6),1190–1200.
Ngamaba, K., Panagioti M., & Ch. Armitage (2017): Income Inequality and Subjective Well-Being: A Systematic Review and Meta-Analysis. Quality of Life Research 27(3): 577–596.
OECD (2020): OECD (2020), Social spending (indicator). doi: 10.1787/7497563b-en (Accessed on 30 July 2020).
Oishi, S., Kesebir, S., & E. Diener (2011): Income inequality and happiness. Psychological Science, 22(9), 1095–1100.
Piketty, T., & E. Saez (2014): Inequality in the long run. Science 344 (6186): 838–843.
Rozer, J., & G. Kraaykamp (2013): Income inequality and subjective well-being: A cross-national study on the conditional effects of individual and national characteristics. Social Indicators Research, 113(3), 1009–1023.
Senik, C. (2009): Income Distribution and Subjective Happiness: A Survey. OECD, Directorate for Employment, Labour and Social Affairs, OECD Social, Employment and Migration Working Papers.
Schneider, S.M. (2016): Income Inequality and Subjective Wellbeing: Trends, Challenges, and Research Directions. Journal of Happiness Studies 17: 1719–1739.
Stanley, T. D. (2005): Beyond Publication Bias. Journal of Economic Surveys 19(3): 309–345
Stanley, T. D., Jarrell, S. B., & H. Doucouliagos (2010): Could It Be Better to Discard 90% of the Data? A Statistical Paradox. The American Statistician 64: 70–77.
van Aert, R.C.M., Wicherts, KJ.M. & M.A.L.M van Assen (2016): Conducting meta-analyses on p-values: Reservations and recommendations for applying p-uniform and p-curve. Perspectives on Psychological Science 11(5): 713-729.
Wilkinson, R., & K. Pickett (2009): The spirit level: Why more equal societies almost always do better. London: Allen Lane Penguin.
WVS (2016): World Value Survey. Household surveys (representative at the national level), longitudinal multiple-wave file covering 1984-2015, worldvaluessurvey.org (Accessed on 30 July 2020).
WB (2020): GDP (current US$). World Bank national accounts data, and OECD National Accounts data files, https://data.worldbank.org/indicator/NY.GDP.MKTP.CD (Accessed on 30 July 2020).
Zagorski, K., Evans, M.D.R., Kelley, J., & K. Piotrowska (2014): Does national income inequality affect individuals’ quality of life in Europe? Inequality, happiness, finances, and health. Social Indicators Research 117(3): 1089–1110.

Předběžná náplň práce v anglickém jazyce

Motivation.

More than 70 years ago Duesenberry (1949) argued that people do not care as much about being poor as they care about feeling poor: relative not absolute wealth matters. A number of more recent studies (Alesina et al., 2003; Benabou & Ok, 2001; Dynan & Ravina, 2007, or Oishi et al., 2011, to name but a few) show that income inequality is rather badly perceived in different contexts. It follows that any government that wants to make people happier pushes towards policies favouring social transfers, redistributing income from the rich to the poor. Aggregate datasets support part of such a story: the share of social transfers on economic output continues to increase (OECD, 2020) and people are getting richer (WB, 2020). But also, a continuously larger share of people report life satisfaction and happiness equality (WVS, 2016, Clark et al., 2016) while the trend in income inequality, whether increasing or not, seems to be strongly dependent on a chosen reporting metrics (compare Piketty & Saez, 2014, to Auten & Splinter, 2019, for example).

These crude trends in data suggest that the role of income inequality in predicting subjective well-being is controversial, to say at least. The empirical research agrees: some authors find the relationship to be negative (Oishi et al., 2011; Wilkinson & Pickett, 2009), some find it to be positive (Rozer & Kraaykamp, 2013; Berg & Veenhoven, 2010), and some do not find any (Graham & Felton, 2006; Zagorski et al., 2014). The two qualitative literature reviews written on the topic (Ferrer-i-Carbonell & Ramos, 2013, and Schneider, 2016) and the meta-analysis of Ngamaba et al. (2017) testify to a large heterogeneity present in these studies: there could be certain aspects of data and measurement choice systematically driving the resulting reported relationship between income inequality and happiness. The regional differences (Alesina et al., 2003), income mobility (Oishi et al., 2011), economic development (Berg and Veenhoven, 2010), and political orientation (Alesina et al., 2003) are just a few examples of the assumed correlates.

I find, however, several features of the meta-analysis by Ngamaba et al. (2017) worrisome and the goal of my thesis will be to carefully address these features. First, the authors work with 24 studies only (24 observations) which sheds doubt on the statistical power of their results. I plan to enlarge the dataset, especially along the lines of economics research (but will include psychology studies as well). Second, the authors do not account for publication bias, the tendency of people involved in the publication process to preferentially publish estimates that make for a good story. I hypothesize that publication bias could play a substantial role in pushing the reported estimates upward. Third, Ngamaba et al. (2017) test only for pairwise differences between certain aspects of study design, they do not perform a multivariate analysis. I plan to use state-of-the-art meta-analysis tools to address the heterogeneity (and publication bias) in the literature. I also plan to add more aspects of study design that can systematically affect the results: data aggregation, data dimension, methods used, or quality of the study. Fourth, I find the analysis scarce on economic interpretations and would like to contribute with a more thorough discussion of the heterogeneity behind the estimates. Fifth, I will attempt to construct a synthetic study that would consider the best-practice study design and I will estimate the best-practice effect.

Hypotheses.

Hypothesis #1: Publication bias is present in the literature estimating the relationship between inequality and happiness.
Hypothesis #2: Publication bias exaggerates the mean value of the relationship between inequality and happiness reported in the empirical literature.
Hypothesis #3: The estimated effects on the relationship between inequality and happiness are driven by the level of a country’s development.

Methodology.

First, I need to create a dataset from primary studies. I will start by constructing a search query for Google Scholar that is superior to all other databases because it uses a powerful full-text search and does not discriminate as to the research field. Second, I will make sure the studies already reviewed by others, including Ferrer-i-Carbonell & Ramos (2013), Schneider (2016), and Ngamaba et al. (2017), are included in my list. Moreover, since the sample used by Ngamaba (2017) ends in October 2017, I will focus my search on novel studies published since then. There are several challenges I will presumably face: first, the reported effects might come in form of a regression coefficient or correlation coefficient and I will have to standardize the effects in order to make them comparable. Second, I will need to collect some measure of the precision of the effect that will allow me to test and treat for potential publication bias (standard error, number of observations, standard deviation, confidence intervals, etc). I might also collect the effects that do not have any measure of precision reported but those effects will not be used in the quantitative treatment of publication bias. If I find the publication bias not to be present in the literature, the effects absent measure of precision could be used for the analysis of heterogeneity, as well.

To test for the presence of publication bias, I will perform a commonly used visual test called the funnel plot (Egger et al., 1997). I will, however, also perform several more rigorous alternatives, including the linear test called Funnel Asymmetry Test (Stanley, 2005) with different variants of data weighting, and some newly developed non-linear tests, including the Weighted average of adequately powered (Ioannidis et al., 2017), the selection model of Andrews & Kasy (2019), Stem-based method (Furukawa, 2019), and Endogenous kink model (Bom & Rachinger, 2019). I also plan to include p-uniform* or p-curve into my analysis (van Aert et al., 2016), since these tests are commonly used in psychology research but have not appeared in the economics research yet (to the best of my knowledge. Next, due to the fact that the collected studies will probably differ in terms of data, methods, location, age, quality as well as other contexts, I need to properly examine and address the heterogeneity of the data.

The model uncertainty is intrinsic to any meta-analysis: many different variables capturing the design of the study are used to explain the effect studied. I cannot be sure beforehand; however, which variables are important and should be included in my regressions. If unimportant variables are kept in the model, the variance of the estimated parameters is likely to increase. Model averaging, such as its Bayesian variant, is a commonly applied tool in meta-analyses to deal with the model uncertainty. In my analysis, I want to focus on dilution prior applied in Bayesian setting (due to George, 2010), which treats for potential multicollinearity in data. Additionally, I would also like to apply several robustness checks, including different choices of the model and parameter priors as well as Frequentist model averaging (Amini & Parmeter, 2012).

Expected Contribution:

The relationship between income inequality and happiness is essential for welfare policy decisions. It is, in fact, startling that there has been only one meta-analysis conducted in this area, so far (Ngamaba, 2017). In my thesis, I want to address several issues the previous meta-analysis does not touch upon. First, I would like to elaborate the literature review to a larger extent and provide an economic and econometric rationale behind several aspects of the study design. Second, I intend to enlarge the original dataset and collect more of the potentially important explanatory variables that could drive the estimated effects in the literature. Those might include different kinds of measurement approaches to income inequality and happiness, geographic differences, data aspects (frequency, dimension, length of the run, etc) and most importantly I would like to focus on the effects of country-level development. Third, I will investigate the potential presence of publication bias which was not accounted for in the previous meta-analysis. Last, I will use the novel methods to treat for model uncertainty and investigate the potential sources of heterogeneity systematically, using methods such as Bayesian model averaging or Frequentist model averaging.

Outline.

1. Introduction: I will explain my motivation, contribution, and my main results.
2. Introducing the topic of inequality and happiness: I will briefly describe the topic of inequality and happiness.
3. Data: I will explain how I will collect estimates from the studies estimating the relationship between inequality and happiness (search query, inclusion criteria, etc.), provide the basic summary statistics, and try to cherry-pick some interesting prima facie patterns in data.
4. Publication bias: I will briefly describe what publication bias is and why it could be present in this literature and use several linear and non-linear approaches to test for its presence. I will provide a short discussion based on the comparison of results from chapter 3 (simple means) and this chapter (means corrected for publication bias).
5. Heterogeneity: This chapter will have several important parts: 1) it will serve as a meticulous literature review of the topic (because in the beginning, I will describe why I chose the explanatory variables I chose and what the current academic literature tells us about these variables, 2) it will provide a short and concise introduction into the methods I am going to use to investigate the heterogeneity, and 3) it will provide a discussion of results that will come out of the examination of heterogeneity.
6. Best-practice estimate: Based on my results from chapter 5 (heterogeneity), I will analyze what the best-practice estimate should look like if researchers correct for publication bias and potential misspecifications.
7. Conclusion: I will summarize my results and provide implications for policy and future research. I will state the potential drawbacks of my analysis, if any.