retraining. Without retraining, their productivity falls to a level that makes a permanent life
on the dole attractive. This effect is particularly strong if human capital erodes rapidly.
An adverse structural change, perhaps due to a fall in international competitiveness, is mod-
elled by higher λ0 = λτ and is obtained as the sum of both effects discussed above. The
structural change will thus unambiguously increase LTU and reduce employment. The effect
on STU depends again on whether the inflow into STU dominates the outflow into LTU. A
comparison of derivatives shows that this is the case if
(1 + δT ) ∙ d<b,
i.e. if human capital erosion is relatively small compared to unemployment benefits. In this case,
incentives are high enough for skill-upgrading to be the predominant cause of STU.
4. Unemployment Trap and Present-Biased Preferences
This section relaxes the standard assumptions by allowing for present-biased preferences. Such
preferences seem to be common and imply time-inconsistent behavior, such as procrastination
(Laibson 1997 and ODonoghue and Rabin 1999). Allowing for such preferences enables us to
extend the model to account for the possibility of an unemployment trap in which people end up
on the dole permanently, against their best intentions. The unemployment trap we demonstrate
is a consequence of mistaken intertemporal optimization. It is therefore an individual-level
phenomenon, and not an equilibrium phenomenon as in other accounts of the unemployment
trap.
Reconsider the case where retraining costs are sufficiently low and people are sufficiently young
such that retraining, in principle, pays off, i.e. condition (5) holds. Diagrammatically, those with
human capital below hA stay long-term unemployed and those with human capital exceeding
hC take up employment immediately in Figure 2. The interesting group consists of those with
intermediate human capital endowment, hA < h < hC , who enter short-term unemployment
with the intention to upgrade their skills and re-enter employment later.
The duration of STU can be subdivided in τ subperiods of eligibility for unemployment
benefits, which can conveniently be thought of as weeks or months. Suppose that it takes
one subperiod to upgrade the skills, for example, by taking a computer course and suppose
that some individuals have present-biased preferences (Laibson, 1997, O’Donoghue and Rabin,
17