Extensions to discrete choice models of labour supply
This thesis considers extensions to discrete choice models of labour supply, primarily in the context of decisions made by families in response to policy changes to taxes and transfers. The main contributions of this thesis are: 1) introducing a non-parametric choice set sampled from similar families that includes social transfer receipt as a choice characteristic, in addition to the typical dimensions of wage rates and hours worked; 2) providing a novel estimation approach for discrete choice models of labour supply that adjusts the number of choices to allow for substitution and complementarity while preserving the Independence of Irrelevant Alternatives assumption as far as possible, leading to parameter estimates with likelihoods close to the maximum likelihood; 3) providing a method that adjusts for nuisance changes occurring between the data year and the policy implementation year through multi-stage ``chaining'' of labour supply simulations at the individual level; and 4) flexibly extending the preference estimation to allow for individual heterogeneity while enforcing the aggregate parametric distributional family for the preference estimates. Each of the 4 contributions in the main chapters are illustrated through empirical examples using synthetic household survey data for New Zealand, developed using the nearest neighbour aggregation technique described in the appendix. This approach provides high-quality synthetic data with multivariate distributions that closely match the source data across many dimensions.