Labour Supply in a High Unemployment Economy: Theory
Conventional theory views the labour supply as determined by numerous individual and household level decisions on both participation and hours of work. The typical model used is the static neoclassical model of labour-leisure choice, which examines the effect of wage and non-labour incomes on the decision to supply labour to the market. Killingsworth and Heckman, Pencavel and Blundell and Macurdy provide a review of different approaches of modelling labour supply. This simple static neoclassical framework is often extended to include other household members into the utility function as labour supply decisions are typically made in the context of decisions taken by other household members (as explained by the income-pooling hypothesis whereby all household members share household incomes). Data limitations often cause the static model to be favoured in empirical investigation of the determinants of labour supply. Below critically assess the conventional empirical approaches used to estimate such a labour function and present some related evidence.
Differences in the approaches to estimating the static model of labour supply seem to reflect differences in data availability. Our literature review identified three main approaches to estimating the labour supply function: (i) estimations regarding hours of work decisions, (ii) multiple discrete choices of labour supply decisions (where choices represent the categories of working hours) and (iii) the binary choice approach where labour supply choice is defined as participation versus non-participation.
Studies that employ the first approach usually control for non-participation (i.e. corner solution of h=0), that is in line with Heckman’s argument on the importance of the theoretical distinction of labour supply choices (participation versus non-participation) and hours of work supplied.
Heckman argues that when deciding to supply labour to the market, an individual makes two decisions: (i) participation that determines whether we observe h=0 and (ii) hours of work that is the size of ht given that h>0. In the Tobit framework, the direction of the effect of any explanatory variable on ht is the same with regard to Pr(h>0 | x) and E(h | Ы>0, x,,). This is a restrictive assumption, since a given explanatory variable may have an opposite effect on the probability of observing ht>0 compared to its effect on the size of ht when ht>0. This may be the case when the participation decision is independent of how many hours to work, suggesting that the Heckman sample selection model is indeed the appropriate model for estimating labour supply elasticities.
In the second approach of multiple discrete choice of labour supply, multinomial models are utilised, whereby hours of work are reported in categories. The use of this approach is justified on the ground that there is little variations due also to demand-side restrictions in the observed hours of work variable even when it is reported as a continuous variable.