5 / 20
5(20)
1
Introduction
In order to reduce the moral hazard in social insurance programs benefits
are not in general paid out unconditionally, that is, without monitoring
and screening the eligibility. Given that screening and monitoring is costly
an important empirical question is how this enforcement should be made
optimally. The theoretical literature on law enforcement agents show that
sanctions should be large when monitoring is costly. This theoretical
prediction is however of limited value for policy given that there exists type
II errors in any insurance (i.e., the situation when individuals are wrongly
condemned), together with a public opinion against too hard punishments.
In many programs, e.g., unemployment insurance (UI) and sickness
insurance (SI) programs, it is easy to show theoretically that there are
both ex ante and ex post effect of monitoring and sanctions. There is
quite strong empirical support of ex post effects of especially benefit
sanctions in the UI literature. There is a sharp increase in the exit rate
from unemployment to employment when individuals are sanctioned.
There is also, by now, plenty of evidence of the effects of monitoring in
the SI program (cf. Hägglund (2010), D’Amuri (2011), Hartman et al
(2013)). The general result is that time limits and screening/monitoring
reduce time on sickness benefits.
Empirical evidence on ex ante or deterrence
1
effects is harder to arrive at.
Boone and van Ours (2006) calibrate their theoretical model to Dutch data
and show by using simulation that the strength of the deterrence effect
depends on the monitoring intensity. If the monitoring intensity is high,
the deterrence effect could be very important. There are results of “threat
effects” of active labour market training (ALMT) programs (see Graversen
and Larsen (2013) for an updated review of the empirical literature as well
as a reanalysis of the result in Geerdsen (2006)). That is, individuals being
offered an ALMT program have a faster exit rate from unemployment
before the date when supposed to enter into the program than those not
offered a program. As the offer can be seen as screening the unemployed
individual’s motivation this result gives some support that deterrence
effects in the UI program could be important.
In the “threat effects” literature the individuals know with certainty when
to enter an ALMT program. This differs from the theoretical framework
analyzing ex ante effects in which the probability of being detected is
assumed to be known but less than one. It is thus interesting to empirically
study deterrence effects in programs with small probabilities of being
detected which is the situation studied in this paper.
1
A potential ex ante effect from monitoring is from deterring people from taking-up
the insurance, hence, the ex ante effect is in the following also denoted a
deterrence effect.