The effect of youth training on labour market state transition processes, wage expectations and search elasticities :a matching approach

Abstract

During the 1980's, unemployed school leavers were encouraged to undertake a period of government sponsored training as a potential stepping stone into work. One such programme was the Youth Training Scheme (YTS), which operated from 1981 to 1994. The effectiveness of such schemes has long been questioned. This work presents an investigation into YTS scheme effectiveness in which we compare estimated labour market elasticities of people with YTS experience against those without. We acknowledge the existence of inherent differences between the two groups which could act to bias our conclusions, then attempt to account for these using matching methods. In an experiment designed to assess a treatment effect, "independent" individuals who do not receive the treatment are usually used as a control group. Econometricians seldom have such experimental data due to the interference of self selection. One way to allow for the bias introduced by self selection is to use a matching algorithm. The purpose of the matching algorithms we employ is to produce a synthetic control group of individuals who have not experienced training but who are similar to the YTS participants. Having produced various matched datasets, we recalculate the elasticities of an optimal job search model to ascertain whether there is still evidence of a treatment effect when comparing like with like. In contrast to previous research, we also make allowance for the changing nature (heterogeneity) of the YTS scheme over time. We compare different matching methods and assess their relative performance. Observations lying beyond the region of overlapping support are shown to cause a degradation in nearest neighbour matching performance. Kernel based procedures are employed using the full range of the bandwidth parameter on which they rely. In the final part of this work we widen our field of vision to include the unemployment to work transition process. We present results which support the hypothesis of different YTS treatment effects in each generation of YTS.