of the individual’s decision rule may change.2 One such change is that the
wage premium increases sufficiently to make re-enrolment in education a
rational decision (Weiss, 1971 and Iwahashi, 2004). Other examples include
relaxed borrowing constraints (Wallace and Ihnen, 1975), that the relative
prices of leisure and schooling change (Killingsworth, 1982) or that the in-
formation set changes, not least completion uncertainty (Altonji, 1993, Sjogren
and Sallstrom, 2004). Another possibility is that the individual’s discount rate
changes over time. An individual with a stronger preference for immediate
income will, all else the same, be more likely not to enrol in education. But if
the discount rate is liable to be reduced with age, the sign of the decision-
making equation may switch from a negative to a positive value. Warner &
Pleeter (2001), studying how individuals had chosen between different offers at
the time of the military drawdown in the US in the early 1990s, found young
individuals to have higher than average discount rates.
Empirical evaluations of education in Sweden have, until a few years ago,
only considered youth education. The syllabus of komvux at upper secondary
level was until 1994 somewhat adapted for adults but has since been the exact
same as the one for youth education. An interesting point of departure is
therefore Swedish evaluations of upper secondary education for youths which
have found payoffs in the region of 3.5 to 4.5 per cent (Isacsson 1999, Kjell-
strom 1999, Meghir and Palme 2005). It is of course far from certain that these
results can be generalized to education for adults. AE usually offers more op-
tions in terms of at what speed it is carried out and when, at what age,
education is (re-) initiated and completed. Individuals in AE also often have
work experience, possibly making their choices of study based on better in-
formation. These features not only make AE different from youth education but
also further complicate the selection mechanisms faced by the researcher. The
above mentioned studies report, like Card (1999), modest bias in conventional
OLS estimators and although in a sense encouraging, there have been no
studies of AE with a set-up resembling a social experiment. Consequently,
2 Ben-Porath (1967) is an exception to this rule. He assumed decreasing marginal productivity of
human capital, inducing individuals to spread formation over a long period of time to maintain a
high marginal payoff to effort.
IFAU - Does adult education at upper secondary level influence annual wage earnings?