Abstract
The past decade has seen an increasing body of evidence emerge that subjective wellbeing can be empirically measured through relatively straightforward questions on self-reported happiness or life satisfaction carried in sample surveys. This paper analyses the determinants of subjective wellbeing using data from the 2008 New Zealand General Social Survey. The paper first provides a brief summary of what is known about the determinants of subjective wellbeing from the international literature. A simple model of the determinants of subjective wellbeing in New Zealand is described, and coefficients for the model are estimated using ordinary least squares. The model is re-estimated using an ordered Probit to confirm that the results are not biased due to the ordinal nature of the data. The paper finds that demographic factors are largely not significant drivers of wellbeing in New Zealand and conforms to the international literature in identifying income, unemployment, health status, and social contact as the four main factors affecting subjective wellbeing.
Keywords:
Notes
1. Life satisfaction is an overall assessment of how people evaluate their life at a particular point in time, ranging from positive to negative. It is a cognitive assessment rather than a statement of a person's current emotional state.
2. D includes all of the variables listed under ‘Demographics’ in .
3. X includes all the remaining variables listed in , i.e. the variables listed under ‘Education and health’, ‘Economic and work’, ‘Social life and community relationships’ and ‘Safety and security’.
4. There is one notable exception to the view that the error terms of X and D are likely to be uncorrelated with E. This lies in the fact that responses to subjective questions included in X are likely to be correlated to some degree with personality type (an element of E). This may result in subjective measures impacting on the coefficients of objective measures included in the analysis. To reduce the impact of this effect, the regressions were run first with objective measures of X only, and then with subjective measures of X included, The inclusion of the subjective variables used in this analysis does not significantly alter the coefficients of the objective variables.
5. The weights used in the model are the survey weights, which control for the multi-stage sample design.
6. In practice, there is strong evidence that treating life satisfaction data as if they were a cardinal measure does not unduly bias the results of analysis (Ferrer-i-Carbonell & Frijters, 2004).
7. Correlation coefficients were produced across all independent variables and we do not suspect multi-collinearity issues with the analysis.