Eyob Fissuh Abstract
“Three Essays on Empirical Labour Economics”
This dissertation is comprised of three essays on empirical labour economics. The unifying theme of the thesis is the econometric methodology: using the Survey of Labour and Income Dynamics (SLID), each essay employs panel data models that control for sample selection bias and individual heterogeneity.
The first essay estimates the elasticity of labour supply of men in a life-cycle setting. This paper confirms that failure to control for the individual heterogeneity and sample selection bias produces upward biased point elasticity estimates: our model that controls for sample selection bias and individual heterogeneity produces an intertemporal labour supply substitution elasticity of 0.16 and those which do not control for these problems 0.23–0.48.
The second paper estimates the impact of health on wages in Canada using Mincer type wage offer model––that corrects for sample selection and control for individual heterogeneity–– and finds that the effect of health on wages is positive, as expected, but not statistically significant. We also demonstrate that treating health as an exogenous variable may give misleading results.
The third essay examines the impact of childcare costs on maternal employment in Canada. This paper extends the existing static maternal labour supply model into a life cycle model, where a typical mother is faced with a problem of maximizing life cycle utility subject to wealth and time constraints. Our study confirms that labour supply decisions of mothers are generally less responsive to childcare price: childcare price elasticity of labour supply for single mothers is -0.068 and for married mothers -0.016. This result clearly suggests that lowering childcare price may not be a good policy instrument to increase maternal labour supply.
In addition to shedding light on a number of highly debated issues in labour economics, perhaps, the main contribution of this dissertation is that it demonstrates the importance of controlling for sample selection bias in empirical labour economics and that the selection process could mainly operate via time variant variables––hence the traditional fixed effects model does not suffice.