Can cities be held responsible for early school leaving? Evidence from the Netherlands

Abstract

This paper examines if ‘naming and shaming’ is an effective tool to increase accountability in school dropout for cities with disadvantaged student populations. It argues that a comparison with other cities might be unfair if regional and population characteristics differ. It discusses the example of two Dutch new towns. The new town policy deliberately attracted low- and medium-income households in the past, such that today the population of those cities differs from other cities. We use a matching analysis to account for observed differences in population and regional characteristics. The results point out that ‘naming and shaming’ may be a dangerous policy to increase accountability: early school leaving differences are driven, to a large extent, by observed differences in population and regional characteristics.

Keywords:

• early school leaving
• naming and shaming
• matching analysis
• housing policy
• selective migration

Acknowledgments

We are grateful to the municipality of Almere, Lelystad, the province of Flevoland, and Platform 31 for financial support and useful comments on an earlier version of this paper. The usual caveat applies.

Notes on contributors

Kristof De Witte is an associate professor at ‘Leuven Economics of Education Review’ at University of Leuven, and at the ‘Top Institute for Evidence Based Education Research’ (TIER) at Maastricht University. He is particularly interested in public economics and in the measurement of efficiency of entities. Chris Van Klaveren is an assistant professor at TIER – Maastricht University. His interests comprise education economics. Anton J.H. Smets is a senior policy advisor for the Province of Flevoland. His Ph.D. research dealt with urban economics.

Notes

1. Note that, for an individual school, an early school leaver induces an additional loss of 6000 euro of enrollment fees paid by the government.

2. Due to data limitations this paper focuses on nonvocational tracks.

3. For simplicity we assume that average dropout level of A is compared to average dropout level of one particular reference city R (although extension to multiple reference cities is straightforward).

4. For a detailed discussion of the iterative matching procedure we refer to De Witte and Van Klaveren (2012). It should be noted that the results presented in this study rely on Mahalanobis distances. The results are very similar if other distance measures are used.

5. Six reference cities are selected. A discussion on more cities would reduce the focus of the paper and is beyond its scope. A discussion on less cities would reduce its insights.

6. A robustness analysis with alternative and additional characteristics delivered similar outcomes.

7. Note that thanks to the selection of comparable reference units, the common support assumption of the matching estimator holds.

8. The latter is to account for the fact that a municipality is still responsible for dropout students, even if they left education during previous academic years. In this sense, this approaches to the definition of Eurostat which defines the percentage of early school leavers as all students without diploma relative to all students younger than 23. Due to data limitations, we only consider the last three academic years.