Compulsory Schooling in a Decentralized Setting: Studies of the Swedish Case

Sammanfattning: Essay 1 (with Eva Johansson) investigates the demand for local public schooling using survey data, a method previously never applied to Sweden. Estimating a demand specification corresponding to that in U.S. studies, more elastic demand than in the U.S. is found. In an alternative specification, a more inelastic demand is indicated, in line with earlier findings on demand for total local spending in Sweden. Support for the hypothesis that municipal employees tend to have higher preferences for local public school spending than other employees is found. The differences in elasticities are, however, not statistically significant.Essay 2 investigates school competition and its implications for student performance. Using a value added specification, increased local school competition is shown to have positive effects on average performance in mathematics, but no significant effects in English or Swedish test performance. Immigrant pupils and those in need of special education tend to gain more from increased school competition, while adverse effects on students from low education families are found in terms of English and Swedish performance. Quantile regressions indicate homogeneous effects on low and high performing students.Essay 3 (with Eva Mörk) investigates the extent to which grants, local tax base, preferences and structural characteristics affected local school resources in connection with the decentralization of compulsory schooling in Sweden, using municipal panel data covering 1989-1995. The main arguments against decentralization are not supported by our findings. First, school spending as well as teacher density is found to be more equally distributed across municipalities following decentralization. Second, local tax capacity is not found to influence local school resources more in the decentralized than in the centralized regime. Furthermore, targeted grants have a significant impact on resources while general grants have not.Essay 4 examines teacher mobility in Sweden. Particular attention is paid to the role of working conditions. Various school attributes are used to proxy for teachers’ working conditions, including the average student performance, school resources and student composition. The results indicate a link, in the expected direction, between school attributes and the probability that a teacher will leave her teaching position in favour of another school. Also, teachers with higher wages tend to have a lower propensity of leaving their school. In contrast to previous US findings, the estimated impact of wages are not sensitive to whether or not controls for school attributes are included.