The second claim is that most ‘second chance’ adult remedial training programs fail to deliver any positive results (see Heckman, LaLonde, and Smith, 1999). It is this claim that commentators usually refer to when they assert that ‘second chance’ training is inevitably an inefficient investment. Heckman and colleagues based their conclusions on the findings of several rigorous evaluations of government sponsored training programs. While these evaluations cast doubt on the efficacy of government sponsored training programs, there are several reasons why we should be cautious about generalizing these findings to all adult skills upgrading programs.

First, as even Heckman (2000) acknowledges, the latest research shows that there is substantial heterogeneity in the impacts of government sponsored training programs across demographic and skill groups. For some groups these types of programs appear to generate significant benefits both to the participants and to society. For example, Heckman, LaLonde and Smith (1999) conclude that for economically disadvantaged adult women residing in the US, a case can be made that these programs have consistently been a productive social investment, whose returns are larger than those from formal schooling. Moreover, research has shown that the ‘quality’ of the intervention matters a lot. Programs with strong links to the local labour market are considerably more effective than programs without these links (Poppe, Strawn, and Martison, 2003).

Second, Heckman’s work has focused on one type of skills upgrading: government sponsored programs. There are good reasons to believe that these types of programs differ significantly from other forms of training, such as employer sponsored training and formal education (Ahlstrand et al., 2003). Government programs tend to be narrowly targeted towards welfare recipients and the long-term unemployed. In contrast, other types of programs such as college diploma programs attract motivated learners with a wide range of labour market experience. Another difference is that government sponsored training programs are often based on curriculum that is not connected to existing certificate or degree programs. Thus ‘graduates’ of these programs are typically left without an easily recognized credential that they can use as a signal to potential employers. This puts the ‘graduates’ of these programs at a serious disadvantage, especially if the program disappears once the pilot or demonstration is complete (Poppe, Strawn, and Martison, 2003).

Third, in recent years, new institutional arrangements for skills development have emerged at the local level that involve employers, educators, community organizations and learners in unique partnerships (Bernhardt, Dresser, and Hatton 2003). While few of these initiatives have been systematically investigated, preliminary evidence suggests that these innovative partnerships have the potential to positively impact the labour market outcomes of the least educated (Poppe, Strawn, and Martison, 2003).

There is a large literature on investment returns to employer sponsored training (see Lynch, 1997 for an excellent review of this literature), but these studies rarely report their findings by initial education groups. Thus, we have a sense of ‘average’ returns to employer sponsored training but we do not know if the returns for less-educated workers are higher or lower than this average.

There is also a large literature on returns to post-secondary education, but these studies tend to focus on returns for individuals who are completing their initial education rather than on returns for individuals who return to school later in life. In fact much of the research on post-secondary education assumes an age-graded life course with a fixed sequence of educational completion and an orderly transition from post-secondary education to the workplace (Elman and O’Rand, 2004). As a result, most studies focus on the determinants of skills and knowledge in adolescence and we know little about the types of learning that work best for learning that takes place later in the lifecycle.