positively related to the uncertainty in the labor market: If a bad state of the labor market
occurs, the individual can continue at school, but if a good state occurs, she can always switch
to the labor market. As increased variability of the state of the labor market increases the
upside payoff more than the downside payoff, the expected return to education investment
rises. Jacobs (2007) reaches the same conclusion in a somewhat different model, in which he
assumes that one can always leave the labor market again for education. Then increased
variability in the return to education increases the probability that education investment is
profitable at one point in time.
In this paper, termination of education is not regarded as a choice variable, as we analyze
determinants of education outcome at the compulsory level of schooling. For this reason, the
mechanisms described by Hogan and Walker (2007) and Jacobs (2007), which rely on
duration of education as an individual choice variable, are not directly applicable. However,
the probability to enroll in post-compulsory education and the range of higher education
institutions an individual student can choose among is typically related to preceding
performance at the compulsory level. Thus, low effort in compulsory schooling reduces the
probability to attain the real option inherent in higher education.
Parents and teachers may influence their childrens’ and students’ effort levels through
instructing and bargaining. This is an important aspect because parents and teachers are
concerned about student effort at school, which may come about by altruistic, dynastic
preferences of parents, and a mission or wage maximization of teachers. For reasons of
simplicity, our model abstracts from these additional influences. In principle, however, the
optimal effort level from the viewpoint of parents and teachers may be derived in the same
way as the optimal student effort above, albeit some parameters of the model may be
different. For example, children are likely to be more short-sighted than adults. In our model,
their discount rate r may be lower, perhaps because of more pronounced hyperbolic
discounting (see for example Laibson, 1997, or O’Donoghue and Rabin, 1999). Then it
follows from equation (3) that the optimal effort level is higher for parents and teachers than
for students, which results in attempts of the first to positively influence effort of the latter. In
addition, it follows from equation (4) that the response to the welfare state is positively related
to r, making the optimal effort level from the parental and teacher point of view more
responsive to income redistribution than the optimal effort level from the student’s point of
view. Finally, the general comprehension in society of the importance of skills and
knowledge, and the societal degree of risk aversion, also affect the parameters of the objective
function. Thus, the actually observed student effort is likely to be a result of not only students’
behavior, but also parental and teacher behavior, and country-specific factors such as culture
and population risk aversion. In our econometric analysis, we do not aim at determining
specific parameters of the model above. Instead, we are the first make an attempt to