Abstract
This thesis considers selected prevailing issues and puzzles in the micro-economic analysis of happiness and satisfaction. The first part of the thesis considers issues of measurement, metrics, and method, with particular focus on the issue of cardinality. The second part of the thesis addresses two apparent puzzles identified in the literature, concerning subjective wellbeing across the lifecycle and across education levels. Panel data from the Household Income and Labour Dynamics in Australia (HILDA) survey are used in all empirical analyses.
Chapters 2 and 3 of the thesis evaluate the case for cardinal interpersonal and intrapersonal comparison of life satisfaction scores. The assumption of cardinality is desirable because it allows for more powerful statistical tools of analysis to be used, the benefits of which are well established. A second possible implication is that estimated subjective wellbeing model parameters may be interpreted directly as measuring marginal utilities, which is not otherwise possible. However, no consensus exists for whether or not this assumption is justified. Chapter 2 presents a review of this debate, concluding that cardinality is intuitively reasonable, and that even if cardinality cannot be assured or rejected with certainty we both can and should use available alternative information about wellbeing to learn more about the informational content of commonly used subjective wellbeing scales. Subsequently, chapter 3 presents an empirical analysis which applies the principle of simultaneous conjoint measurement to compare life satisfaction scores with available data on mental health. The resulting evidence provides strong support for both of ordinal and cardinal comparison of life satisfactions scores, both across and within individuals.
Chapter 4 considers the specification, estimation and interpretation of subjective wellbeing models using panel data. A key aspect of this chapter is the evaluation of the shape and curvature of the relationship between subjective wellbeing and income, as well as wealth. The results favour the implied log-normal specification, where subjective wellbeing is linearly associated with position in the distribution of income and wealth. Contrary to the more common lin-log specification this specification is not sensitive to the inclusion or omission of low values; fitted values match the data well across the entire income and wealth distributions; and it allows for a weakly S-shaped utility or wellbeing function, which is consistent with recent discoveries in behavioural studies and also observed in the data. Further, a general analysis of panel models demonstrates that standard linear pooled regression models yield estimates which tend to be robust, consistent and efficient compared to alternative models and methods of estimation.
Chapter 5 examines the extent to which associations between economic circumstances and financial satisfaction are age-dependent. This analysis is motivated by the commonly observed, though contested, U-shape in subjective wellbeing across the lifecycle. This chapter therefore provides a brief examination of the relationship between age and life satisfaction using HILDA data, which supports a robust U-shape. Subsequently, the core analysis demonstrates that half of the observed age effect is explained by changes in financial satisfaction. Further, associations between income and wealth and financial satisfaction, and between financial satisfaction and life satisfaction, are strongest in midlife and abate thereafter. Consequently, changes in material concerns represent a key explanation for why we might observe a U-shape in happiness across the lifecycle.
Chapter 6 examines the association between education and subjective wellbeing. The evidence on this relationship is scarce, inconsistent and poorly understood, and reports of a negative association, which appear counter-intuitive, are common. The analysis presented here explores two possible explanations for this puzzle: over-allocation of time toward work, and higher expectations and aspirations. Evidence presented here supports the latter explanation, but not the former. People with higher levels of education exhibit higher working hours, but also higher optimal working hours. However, wellbeing benefits of higher education from improved circumstances are neutralised by rightward drifts in the life satisfaction function: People with higher levels of education require better circumstances in order to reach a given level of life satisfaction, and as these higher expectations tend to be met, the overall effect of education on life satisfaction is neutral.
In sum, the thesis makes original contributions both in the areas of measurement and method and in the form of a greater understanding of what drives known (and less known) patterns previously identified in the subjective wellbeing literature.
Chapters 2 and 3 of the thesis evaluate the case for cardinal interpersonal and intrapersonal comparison of life satisfaction scores. The assumption of cardinality is desirable because it allows for more powerful statistical tools of analysis to be used, the benefits of which are well established. A second possible implication is that estimated subjective wellbeing model parameters may be interpreted directly as measuring marginal utilities, which is not otherwise possible. However, no consensus exists for whether or not this assumption is justified. Chapter 2 presents a review of this debate, concluding that cardinality is intuitively reasonable, and that even if cardinality cannot be assured or rejected with certainty we both can and should use available alternative information about wellbeing to learn more about the informational content of commonly used subjective wellbeing scales. Subsequently, chapter 3 presents an empirical analysis which applies the principle of simultaneous conjoint measurement to compare life satisfaction scores with available data on mental health. The resulting evidence provides strong support for both of ordinal and cardinal comparison of life satisfactions scores, both across and within individuals.
Chapter 4 considers the specification, estimation and interpretation of subjective wellbeing models using panel data. A key aspect of this chapter is the evaluation of the shape and curvature of the relationship between subjective wellbeing and income, as well as wealth. The results favour the implied log-normal specification, where subjective wellbeing is linearly associated with position in the distribution of income and wealth. Contrary to the more common lin-log specification this specification is not sensitive to the inclusion or omission of low values; fitted values match the data well across the entire income and wealth distributions; and it allows for a weakly S-shaped utility or wellbeing function, which is consistent with recent discoveries in behavioural studies and also observed in the data. Further, a general analysis of panel models demonstrates that standard linear pooled regression models yield estimates which tend to be robust, consistent and efficient compared to alternative models and methods of estimation.
Chapter 5 examines the extent to which associations between economic circumstances and financial satisfaction are age-dependent. This analysis is motivated by the commonly observed, though contested, U-shape in subjective wellbeing across the lifecycle. This chapter therefore provides a brief examination of the relationship between age and life satisfaction using HILDA data, which supports a robust U-shape. Subsequently, the core analysis demonstrates that half of the observed age effect is explained by changes in financial satisfaction. Further, associations between income and wealth and financial satisfaction, and between financial satisfaction and life satisfaction, are strongest in midlife and abate thereafter. Consequently, changes in material concerns represent a key explanation for why we might observe a U-shape in happiness across the lifecycle.
Chapter 6 examines the association between education and subjective wellbeing. The evidence on this relationship is scarce, inconsistent and poorly understood, and reports of a negative association, which appear counter-intuitive, are common. The analysis presented here explores two possible explanations for this puzzle: over-allocation of time toward work, and higher expectations and aspirations. Evidence presented here supports the latter explanation, but not the former. People with higher levels of education exhibit higher working hours, but also higher optimal working hours. However, wellbeing benefits of higher education from improved circumstances are neutralised by rightward drifts in the life satisfaction function: People with higher levels of education require better circumstances in order to reach a given level of life satisfaction, and as these higher expectations tend to be met, the overall effect of education on life satisfaction is neutral.
In sum, the thesis makes original contributions both in the areas of measurement and method and in the form of a greater understanding of what drives known (and less known) patterns previously identified in the subjective wellbeing literature.
Original language | English |
---|---|
Qualification | Doctor of Philosophy |
Supervisors/Advisors |
|
Publication status | Unpublished - 2015 |