Mikko Salonen HECER, University of Helsinki [email protected] SCHOOL CHOICE IN FINLAND Part of a research project 'School choice in Finland' led by Hannu Vartiainen, financed by Academy of Finland
INTRODUCTION Right now there are large scale reforms that Education in Finland * are going to take place on fall 2013 for the In Finland every year approximately 60 000 Academic Vocational Typical ages secondary schools and fall 2014 for the degrees degrees students take part in a national system universities. Previously the school selection doctor where new students are allocated to mechanism for the univerisities was not master polytechnic + 2-3 secondary schools. The allocation is done centralized. Our goal is to study how the bachelor polytechnic + 3-4 based on student preferences and criteria of current system works (for secondary upper 18-19 each school. The students can apply to no secondary vocational schools, universities and polytechnics) and school 17-18 more than five different schools. Yearly school (voluntary) estimate its welfare losses compared to an (voluntary) 16-17 approximately 4 000 to 5 000 students are alternative mechanism. After the school 15-16 not admitted to any school. Of those reform we plan to estimate it's welfare 14-15 students who are admitted to a vocational effects as well. 13-14 upper secondary school 25% drop out 12-13 during the first year. Student drop out rate We have data on admittance to comprehensive school polytechnic universities for the newly (compulsory) 11-12 for general upper secondary school is 10 %. 10-11 These young people who have dropped out graduated secondary school applicants and on secondary school selection. We 9-10 or otherwise are without a secondary school 8-9 hope to attain the same data after the placement have a higher probability of 7-8 becoming marginalized. reform has taken place. pre-school 6-7 *table modified from http://en.wikipedia.org/wiki/Education_in_Finland (on 6/5/2013) THE CURRENT SYSTEM STEPS OF THE ALGORITHM school gets admitted to school k the student is removed from the other institutions quota The mechanism for secondary schools Step 0: Applicants list at most five different as well as from all the waitlists that are less works as a schoolproposing deferred schools according to their preferences. Sc preferred to k. The institution that now has acceptance (DA) algorithm with certain hools receive applications and arrange them free quota of one is given the turn next step. restrictions. Students are allowed to apply according to their own criteria. Schools now to no more than five different schools, admit top students until their quota is full Preference lists EXAMPLE 1 a1: S2,S1 schools have very heterogenous prefe and put all other students to waitlist. a2: S1,S2 rences, one might get priority points for Step k: For a school k the algorithm goes Each school S has quota of a3: S1,S2,S3 : a , a , a gender and listing a program first (vocatio through each applicants removing those stu one. The preferences and S1 1 2 3 : a , a , a nal school). We are still pending confirma dents from the quota and waiting list that different algorithms are S2 2 1 3 S3: a3 tion on the specific details how the have been accepted to a more preferred represented below. mechanism works from the national board school. Students from the waitlist replace Step 0: the initial al S1 S2 S3 of education. The universities have no joint those students in the quota that have been location a1 a2 a3 application system in use prior to 2014 re removed previously. When a student that al Step k: no applicant a2 a1 form. ready has a placement in a less preferred gets to two schools a3 a3 RESEARCH QUESTIONS EXAMPLE 2 Step 1 How does the current Finnish system per Same preferences as in example 1 but now S1 S2 S3 Step 2 form? school 2 has quota of 2 instead of 1. a1 a2 a3 Which algorithm would be most suitable Step 0: Initial allocation S1 S2 S3 for the Finnish system? Step 1: School one selects applicant 1 a2 a1 a1 a2 a3 Is there a need to make any modifications Step 2: School two selects applicants 1 and a3 a3 to the algorithms to make them more sui 2. Since for applicant 1 school 2 is preferred a2 a1 table for the Finnish system? she is removed from the quota of school 1. Step 3 a3 a3
How does the performance of the chosen Step 3: School 1 now selects applicant 2 S1 S2 S3 algorithm compare to the system now in and she therefore removed from the quota a2 a2 a3 use? of school 2. Step 4 How big are the possible welfare gains Step 4: School 2 now selects applicant 3. a3 a1 S1 S2 S3 from changing the mechanism? Step 5: Since applicant 3 is selected by a a3 more preferred school 2, school 3 is left a2 a1 a3
without a student. a3 a3 EXPECTED RESULTS proofness of the students. Below there are illustrations of different allocations of schools and students using different algorithms for Since the current mechanism seems to behave like a school example 1. proposing DAalgorithm (pending confirmation from the ministry) it is therefore stable, strategyproof for the schools and REFERENCES asymptotically strategyproof for the students in the absence of Abdulkadiroglu, Atila, and Tayfun Sönmez (2003), School Choice: A Mechanism preference list restrictions. However, some priority points can be Design Approach, American Economic Review, 93–3: 729747 given to students and schools preferences are highly heterogenic Abdulkadiroglu, Atila, Parag Pathak and Alvin Roth (2005), The New York City high which might cause prob school match. American Economic Review 95: 364–67 School optimal DA Student optimal DA Abdulkadiroglu, Atila, Parga Pathak, Alvin Roth and Tayfun Sönmez (2005), The (The current system) (e.g. Hungary's system) lems especially on strategy Boston public school match. American Economic Review 95: 368–71 Top Trading Cycles (TTC) Gale, David and Lloyd Shapley (1962), College admissions and the stability of S1 S1 marriage. American Mathematical Monthly 69: 9–15 a a 1 1 The Ministry of Employment and the Economy (2012), Nuorten yhteiskuntatakuu a1 S2 a1 S2 2013 Työ ja elinkeinoministeriö. TEM raportteja 8. (1/12/2013) a2 S2 a2 S2 S http://www.tem.fi/files/32290/TEMrap_8_2012.pdf S 1 a2 1 a2 a3 a3 a3 S3 a3 S 3 S S3 S3 3 a3