there is little knowledge about if and how the additional selection mechanisms
have influenced the results of AE evaluations.
Studies of AE in Sweden have exclusively considered data collected from
participation at komvux. Ekstrom (2003) focused on participants at upper
secondary level komvux from 1988 to 1993, with annual wage earnings in
2000 as outcome variable. The regression results implied negative effects of
AE for men born in Sweden but positive effects for female immigrants. No ef-
fects were found for females born in Sweden and male immigrants. The nega-
tive effects for men were 6 per cent for those aged 43-55 and 3 per cent for
men aged 25-42. It is the only study so far not based on the mainly unem-
ployed participants in the Adult Education Initiative (AEI), introduced in 1997.
Axelsson and Westerlund (2005) found the AEI participants to increase annual
earnings by SEK 12,800, roughly corresponding to an effect of 10 per cent.3
Albrecht et al. (2004) instead reported no significant effects of the AEI on
wage earnings but a higher probability for men aged 25-40 to find employ-
ment.4 Stenberg and Westerlund (2007) studied long-term unemployed in the
AEI and their results showed positive effects on annual wage earnings but no
effects for those that were registered in the AEI during only one semester or for
males with the maximum four semesters of studies. Stenberg (2007) and
Axelsson and Westerlund (2005) compared enrolees in 1997 in the AEI and
Labour Market Training (LMT), which was mainly vocational. Both studies re-
ported LMT to have more beneficial effects on wage earnings. Stenberg (2005)
made a similar comparison with LMT but found ambiguous effects on unem-
ployment immediately following program.
In the US, there is a substantial literature evaluating the effects on wages of
community college studies (Grubb 2002 provides a survey). The courses are
often vocational and/or at post-secondary level but some of the evaluation
studies are reminiscent of the present one as they estimate returns to a year’s
worth of credits to generate results comparable to the returns to schooling lit-
erature. Jacobson et al. (2005) had access to data on a large sample of individu-
als aged 20 to 59 in Washington State that were displaced by their employers in
3 The value of € 1 varied from SEK 8 to SEK 9 throughout the period covered by this report.
4 Their criteria to identify AEI participants did not include the special grant, UBS, or records of
unemployment, meaning that regular komvux and the AEI were not necessarily separated.
IFAU - Does adult education at upper secondary level influence annual wage earnings?