Please use this identifier to cite or link to this item:
Pursiainen, Heikki
Kortelainen, Mika
Pääkkönen, Jenni
Year of Publication: 
Series/Report no.: 
54th Congress of the European Regional Science Association: "Regional development & globalisation: Best practices", 26-29 August 2014, St. Petersburg, Russia
We analyze differences in school quality using a comprehensive panel data set covering all upper secondary school graduates in Finland during the years 2002-2013. School quality is defined as the effect of the school on matriculation exam results controlling for quality of student intake. In other words, the quality difference between two schools is the expected difference in exam results for a randomly chosen student switching schools. Using methods similar to Chetty, Friedman and Rockoff (2013) we are able to measure both cross-sectional differences in school quality and the persistence of these differences over time. We also control for the uncertainty inherent in assessing the quality of smaller schools with a relatively low number of graduates. We use each pupil's comprehensive school grades to control for previous education / pupil quality. Also, comprehensive school fixed effects are used to control for differences in comprehensive school grading as well as unobserved socioeconomic factors. The method is potentially sensitive to bias induced by school selection. To assess the potential bias we partially match our student sample to a spatial database by home address and use this to assess bias. We find no evidence of significant bias. Our first result is that there are significant cross-sectional differences in school quality even after controlling for student intake quality. The quality difference between the top schools and bottom schools each year measured in average matriculation score points is around one grade point in a scale of 1 to 7. In Finland university entry is partly controlled by these matriculation exam results. A one-point difference in grade averages will significantly affect the chances of entry into the most competitive university curricula. This result must, however, be qualified in a number of ways. First, large differences are observed only between the very top and bottom institutions. Most schools are much closer to each other in quality: the interquartile range each year is only about a fifth of a grade average point. Most schools are thus clustered quite close to each other in quality. Also, while there is persistence over time in school quality, this is far from complete. This means that the ranking of the middling-quality majority of schools is highly unstable over time, making any yearly league tables highly suspect. There is more persistence in the very top and bottom institutions, which are roughly the same during the whole period under consideration. Finally, school quality seems to be for the most part evenly distributed regionally. While there are certainly good schools in the largest cities, the success of the most selective institutions is mostly explained by quality of intake rather than teaching.
school ranking
regional differences in education provision
Document Type: 
Conference Paper

Files in This Item:

Items in EconStor are protected by copyright, with all rights reserved, unless otherwise indicated.