Abstract
This thesis contributes to the rapidly growing field of the economics of wellbeing. Economic research in this field overwhelmingly relies on well- being scores, that are self-reported and ordinal in nature. The focus of this thesis is the econometric analysis of subjective wellbeing data.
Chapter 1 looks at the comparison of ordinal scales between groups. Comparison of means, that is routinely performed in the happiness lit- erature, is only possible under strong assumptions that are rejected by data. Chapter 1 suggests using the median instead of the mean for the comparison, and shows how this can be easily implemented in empirical work.
Chapters 2 and 3 look at the effect of reporting behavior on the results of econometric analysis. Reported wellbeing scales can be influenced by external factors, that are not relevant for global wellbeing. Chapter 2 in- terprets these effects as measurement error and suggests a framework to explicitly account for them in models used in empirical work. Chapter 2 applies this framework to reaccess the relationship between wellbeing and its main drivers. Chapter 3 applies the same framework to look at the evolution of wellbeing over the lifespan.
Chapter 4 studies the evolution of female happiness in the US over the last decades. Average female happiness declined during this period, despite improvements of the position of women in multiple measurable outcomes, e.g. earnings and employment. The literature refers to this finding as The Paradox of Declining Female Happiness. Chapter 4 uses a novel machine learning technique to show that the decline can be ex- plained by changes in individual characteristics.