ESTIMATING LABOUR SUPPLY FUNCTIONS IN A LINEAR EXPENDITURE SYSTEM FRAMEWORK.

The statistical analysis of individual level labour market data on hours of work and wage rates within a model of utility maximization is well established. Using these demand relationships as estimating equations gives a strong link to economic theory which has two major advantages for the researcher. Firstly, the parameter values which define the structure can be given an economic interpretation. Secondly, the underlying economic framework can be used to formulate more meaningful hypotheses about parameter values. However, the estimator is not a standard maximum likelihood estimator as it is not the root of the appropriate likelihood equation. The estimator will not possess the usual property of asymptotic normality and the standard errors cannot, therefore, be conventionally obtained. The article discusses a set of methods termed "Bootstrap" methods which can be used in such situations. The basic idea of these methods is to use the estimated parameters and random drawings from the estimated residuals to create pseudo, or bootstrap samples which can then be used to re-estimate the parameters. This procedure can be repeated a large number of times and a distribution for the parameter of interest can be built up.