The economic literature on training the unemployed or disadvantaged suggests very modest impacts. Nevertheless, policy makers seem as interested as ever in retraining as a solution to unemployment and associated social problems. Part of this optimism seems tied to plans (including loans
and voucher schemes, and tax favored savings accounts) in which potential trainees will purchase their own training, giving them much more discretion over the amount and nature of training they receive. It is hoped that such schemes will both increase the amount of training that occurs and lead
to better training outcomes.
We examine data on training among job losers in the 1995 Canadian Out of Employment Panel (COEP) and find that about a quarter of workers get some training after job loss. Moreover, almost as many people are found to self-finance their own training as train in government-sponsored programs. This is interesting for several reasons. First, we take this “self-financed” training as prima facie evidence of a substantial payoff from training, for at least for some part of the population. Second, this experience with self-financed training can shed light on the possible results of the “choice-based” training schemes currently under discussion in a number of countries.
The case for loan, voucher or account based schemes, seems to rest on several premises. First, credit market failures are an important impediment to training and retraining. Second, individuals have private information that would lead them to make different - and superior - training choices. In this paper we assess the evidence for these premises that the COEP data provide. Thought the data are not experimental, they are rich and longitudinal. This allows us to make several carefully controlled contrasts.
Our principal findings are as follows:
(1) Among job losers, self-financed training is typically less intense, but of longer duration, than “assisted” training (paid for by a government, union, or a former employer). Self-financed training is also more likely to occur at a community college or university. Training choices appear to vary with the method of finance, even after controlling for a rich set of individual characteristics.
(2) We find a robust correlation between liquid assets at the time of job loss – “cash-on-hand” - and subsequent self-financed training. This correlation remains after controlling for past wages, previous education and variety of other variables. If cash on had were capturing variation in discount rates or rates of return to education, we would expect to observe this correlation for assisted training
as well, but we do not. Finally, when non-trainees are asked why they did not train, respondents with no liquid assets at job loss are much more likely to cite lack of funds as the main reason. All of this suggests that credit constraints do play a role in training take-up.
(3) About 11% of non-trainees cite lack of funds as the main reason they did not train. A much larger fraction (40%) reported that they didn’t need it or that it wouldn’t change their job prospects. Interestingly, even after controlling for a large number of observable characteristics, this group had better outcomes than either trainees or those who did not train for other reasons.
(4) Taking those who didn’t train because of a lack of funds or because they were refused or rejected from an assisted training program as the comparison group, we find positive but statistically insignificant impacts of both assisted and self-financed training on employment and earnings outcomes.