Several studies investigate the effects of school resources on student performance. However, these studies tend to focus more on intervention effect sizes than on their cost-effectiveness. New research investigate the effects on upper secondary school admission of a school intervention that introduces centralised, structured and professionalised student career guidance in lower secondary school. Disregarding the sunk cost of implementation, the reform was cost-neutral. Hence, an increased admission rate due to the reform entails improved cost-effectiveness of the career guidance system.
In a Difference-in-Differences (DiD) framework, we find that the reform increases admission to upper secondary school between 4.0 and 6.3 percentage points for immigrants, but shows at best small improvements for non-immigrant students. These findings support one of the main objectives of the reform; to allow for the student career counsellors to focus face-to-face guidance on students with few resources.
The Danish Guidance System and the Danish Guidance Reform (DGR) in 2004
The DGR constitutes a change from a laissez-faire guidance system to a centralised, structured and highly professionalised system. Furthermore, the new system allows for counsellors to focus their guidance resources on students who are at risk of leaving the educational system. Before 2004 school principals were responsible for the career guidance offered at lower secondary schools. Class teachers who had attended one or more short-term training courses in counselling performed the actual guidance activities. Before the DGR was introduced, no national or regional requirements existed in terms of the content of the career guidance or the level of counselling qualifications. The annual grants from the municipalities to the schools, including funding for career guidance, were not directly earmarked for counselling. After the DGR about 50 new regional centres are responsible for the career guidance at primary and upper secondary schools. The guidance counsellors work full-time and have completes a six-month full-time programme (courses at the tertiary level). In cooperation with the municipalities, the centres determine the regional guidance activities and, although no specific national requirements exist, the centres are required to document admission, drop-out rates, etc. As the centres have the authority to reallocate their resources amongst the schools and the students, the guidance targets mainly students at risk of leaving the educational system after compulsory schooling.
Identification strategy, results and robustness
Until 2007, the DGR did not include private schools. Given this natural experiment setting, we use a DiD strategy in which public schools are the treatment group and private schools are the control group. In Denmark only minor differences exist between public and private schools both with regard to administration and funding, as well as student performance when controlling for the SES of the parents. Hence, we identify the difference in the causal effects of structured and unstructured student guidance on admission to upper secondary school. For both non-immigrants and immigrants, we find that the DGR has a positive significant effect on admission to upper secondary school. However, for non-immigrants, the result is small (from no effect to 1 percentage point), whereas for the immigrants the effect is larger (4.0–6.3 percentage points). We perform several sensitivity checks. First, we find no changes in the enrolment pattern into private and public schools in the period when the reform was introduced. Second, we find only minor changes in the covariates concurrent with the reform, except for the grade point average (GPA). Because the GPA changes negatively with the reform, we estimate the model both with and without GPA. Third, we estimate how sensitive our results are to changes in time trend using a classic prereform falsification test and by including school-specific time trends. Fourth, we test how sensitive our results are to potential treatment of the untreated. We conclude that the main result for immigrants is fairly robust, except for the inclusion of school-specific time trends. For non-immigrants, the result is sensitive to even small changes in the model specification.