Empirical studies of public policies within the primary school and the sickness insurance

Empirical studies of public policies within the primary school and the sickness insurance

This thesis consists of five self-contained essays. (1 till 3 handlar om skolan)

Essay 1 (with Elly-Ann Johansson) estimates the effect of attending an MA-class during grades 4-6 on students cognitive skills. Using a unique survey with information on students, parents and teachers, we are able to control for many factors that could otherwise bias the results. We find a negative effect on the short-run skills, as measured by grade 6 cognitive tests. In contrast, attending an MA-class in grades 4 6 does not seem to affect longer run school performance as measured by grade 9 credits.

Essay 2 compares results on national tests with teachers assessment of student performance, by using Swedish data of grade 9 students (16 years old). I examine whether there are systematic differences correlated with gender and ethnic background. The results show that girls are more gen-erously rewarded in teachers assessment compared to test results in all three subjects studied. Non-native students are more generously rewarded in teachers assessment compared to test results in two out of three sub-jects studied.

Essay 3 investigates the importance of gender and ethnic interactions among teachers and students for school performance. School leaving certificates assigned by the teacher is compared with results on compre-hensive national tests. I find that a student is likely to obtain better test scores in Mathematics, when the share of teachers of the same gender as the student increases. Correspondingly, ethnic minority students, on aver-age, obtain better test scores in Mathematics, when the share of ethnic minority teachers increases. The positive same-gender effect on test scores is counteracted by a negative assessment effect. That is, condi-tional on test scores, same-gender teachers are less generous than oppo-site-gender teachers when assessing students performance. In Swedish and English no statistically significant effects are found.

Essay 4 (with Per Johansson) evaluates a multidisciplinary collaboration programme with the aim to prevent long-term sickness. The selection of eligible candidates was mainly based on register information, implying a good prerequisite of estimating the effects by using the same information. In addition, we have run a small experiment. Both evaluation approaches provide the same result: the programme prolongs rather than shortens the sickness absence spell. The hazard of leaving a sickness absence spell is reduced by on average 22 per cent. Two potential and complementary explanations for these negative results are (i) inefficiency in the organiza-tion outside the programme (i.e., in the rehabilitation process) and (ii) moral hazard in the sickness insurance.

Essay 5 (with Per Johansson) evaluates a policy to call sick-listed indi-viduals without permanent employment to information meetings about the rights and duties associated with the sickness insurance system. The evaluation is based on experimental data: a random displacement of when the call is sent out. Comparing the survival functions of the individuals who are called immediately and those whose calls are delayed (by about 30 days) makes it possible to estimate a lower bound of the effect of be-ing called on the sickness absence duration. The result suggests that call-ing a sick-listed individual without permanent employment to an informa-tion meeting reduces the sickness absence duration by at least 23 per cent on average.