Early School Leaving in the Netherlands: Policy and Research

1. Early school leaving in the Netherlands Policy and research Kristof De Witte Leuven Economics of Education Research, KU Leuven Top InstituteforEvidenceBasedEducation Research, Maastricht University

2. This research Evaluation of policy interventions to reduce early school leaving  Based on research questions by the Dutch Ministry of Education  Based on research questions by individual schools  Ifpossible, we focus on causalevidence

3. Complex issue Costs of early school leaving are high • Private costs • - Higher risk for unemployment (Psacharopoulos and Layard, 1979) • - High risk for poor health (Groot and Maasen van den Brink, 2007) • - Lower education outcomes of their children (Bowles, 1972) • Costs for society • - Higher risk for criminal activities (Lochner and Moretti, 2004) • - Lower social cohesion (Milligan et al., 2004) • - Lower economic growth (Hanushek and Wössmann, 2007) • But, Early school leaving is a complex issue

4. DropoutpreventionThe literature Type Exogenous Motivation • Push factors • Interest in schooling • Opinion about teachers •  e.g., Rumberger & Thomas, 2000 • Retentions •  e.g., Ekstron et al., 1986; Grisson & Shephard, 1989; • Goldsmidt & Wang, 1999; Jimerson, 1999; Roderick et al., 2000 • Attention during classes Level • Student characteristics • - Gender • Ethnicity •  e.g., Fernandez et al., 1989; Goldschmidt & Wang, • 1999; Steinberg et al., 1984; Cairns et al., 1989 • Ability: •  e.g., Ekstron et al., 1986; Goldsmidt & Wang, 1999 • Pull factors • Truant •  e.g., Carbonaro, 1998; Rumberger; 1995; Swanson & • Schneider, 1999 • Homework •  e.g., Goldschmidt & Wang, 1999; Seltzer, 1994 • Parents • Education parents •  e.g., McNeal, 1999; Rumberger, 1995; Pong & Ju, • 2000; de Graaf et al., 2000 • Social class •  e.g., Coleman et al., 1966; Jencks et al., 1972; • Kalmijn and Kraaykamp, 2003 • Interest and aspirations parents • - Attendance parents’ evening •  e.g., Astone & McLanahan, 1991; Rumberger et al., 1990; • Rumberger, 1995 • Importance education •  Ekstrom et al., 1986; • Checking homework •  Epstein, 1990; Suichu & Willms, 1996 • School • - Location school (urbanization)  e.g., Astone & McLanahan, 1994; Haveman et al., 1991; Rumberger, 1995; Swanson & Schneider, 1999 • School specific elements (cf. unobserved heterogeneity) •  e.g., Lee, 2000; Multilevel models • Peer group effect • Class size •  e.g., McNeal, 1997; Rumberger, 1995 • Student composition •  e.g., Bryk & Thum, 1989; McNeal, 1997; Rumberger, 1995; • Rumberger and Thomas, 2000 • Ethnicity in class •  e.g., Ainsworth-Darnell, 1998; Cook & Ludwig, 1997; Gibson, • 1997 • School track •  e.g., Jacobs and Tieben, 2009

5. School dropout in the Netherlands

6. Problem statement Lisbon European Council (2000) and Horizon 2020 targets: Halve the year 2000 number of school dropouts Extensive policy in the Netherlands organized by ‘projectdirectievoortijdigschoolverlaten’ within the Ministry of Education  National target: halve the number of new early school leavers from 71.000 in 2002 to 35.000 in 2012 (and 25.000 in 2016) Note: denominator = all students in a given year  EU based target: 8% early school leavers by 2020 Note: denominator = all people younger than 23 years old  This presentation: Dutch policy on early school leaving, and its effectiveness

7. Outline • A conceptual framework for interventions B. What happened in the Netherlands? -- Policy and effectiveness 1. National registration 2. Naming and shaming 3. Regional accountability 4. School accountability (monetary incentives for schools) 5. Qualification Law (increased compulsory education age) C. Accounting for economic influences in school dropout & The costs and benefits of further policy on early school leaving

8. Problem statement Early school leaving is a complex issue  Start from a literature review  Think at a conceptual level at policy interventions

9. Conceptual framework of interventions

10. Outline • A conceptual framework for interventions B. What happened in the Netherlands? -- Policy and effectiveness 1. National registration 2. Naming and shaming 3. Regional accountability 4. School accountability (monetary incentives for schools) 5. Qualification Law (increased compulsory education age) C. Accounting for economic influences in school dropout

11. DropoutpreventionImprovedregistration How do you know whether they left school (without diploma)?  Registration of students is the start of policy making

12. DropoutpreventionImprovedregistration Basis Register OnderwijsNummer (BRON)  Data set of all Dutch students at secondary education  Started in school year 2004/2005  Includes postcode of pupil, school number (‘brin’), parental information (e.g., one-parent family), social situation (e.g., living in poor area)  Can be matched with data from Statistics Netherlands and municipal registration (‘Gemeentelijke Basis Administratie’)  Registration in BRON on October 1. Early school leaver = A student younger than 23 who does not have a higher secondary diploma and is not enrolled in school on October 1, while he/she was last year  Note: still a lot of discussion on the definition, but at least a very good start

13. DropoutpreventionNaming and shaming Using the BRON-data, the Ministry of Education applies ‘naming and shaming’ - Everyone can observe the early school leaving rate and its change in his/her municipality and even neighborhood www.aanvalopschooluitval.nl - Regions receive a letter with their (absolute and relative) performance

14. DropoutpreventionNaming and shaming Naming and shaming along several lines - public: www.aanvalopschooluitval.nl - regional - targets set by school

15. Dropout preventionNaming and shaming Source: www.voortijdigschoolverlaten.nl

16. Dropout preventionNaming and shaming

17. DropoutpreventionNaming and shaming ‘Meten is weten’ (‘Measuring is knowing’) Having good data is the very start - For policy - For schools - For policy evaluation Despite discussions on the definition and despite the absence of stopouts, a national registration is important  note that stopouts are often registered in municipal datasets along with truancy (so-called ‘absoluutverzuim’)

18. Outline • A conceptual framework for interventions B. What happened in the Netherlands? -- Policy and effectiveness 1. National registration 2. Naming and shaming 3. Regional accountability 4. School accountability (monetary incentives for schools) 5. Qualification Law (increased compulsory education age) C. Accounting for economic influences in school dropout

19. DropoutpreventionRegionalaccountability A decentralized implementation of policy -- Adapt policy to the local needs and student group -- Combined with significant accountability (naming and shaming, ‘effect rapportages’, monetary incentive) Ministry of Education - Projectdirectiekennis 39 regional dropout authorities (RMC) Municipalities Coordinating responsible for the school group Local responsible at the school

20. DropoutpreventionRegionalaccountability Dropout prevention in the Netherlands (total budget of 313 million euro in 2008) Regional accountability  39 regions to coordinate dropout prevention measures  Regions can select policy measures out of a list suggested by the Ministry of education (‘the covenant’)  Chosen ‘covenant items’ are published on the website

21. DropoutpreventionRegionalaccountability Regional accountability: the ‘convenant’

22. DropoutpreventionRegionalaccountability Which of the prevention measures go along with lower dropout?  Quantile regression controlling for regional fixed effects, a time trend, student and parental characteristics, neighborhood characteristics, and school type (note: correlations, not causal effects due to the lack of a control group) We observe that: - Truancy registration seems to work for students with few dropouts - Curative projects have an immediate effect - Mentoring and coaching, and extended school time have the largest influence  Particularly interventions which ‘activate’ students seem to work

23. Outline • A conceptual framework for interventions B. What happened in the Netherlands? -- Policy and effectiveness 1. National registration 2. Naming and shaming 3. Regional accountability 4. School accountability (monetary incentives for schools) 5. Qualification Law (increased compulsory education age) C. Accounting for economic influences in school dropout

24. Regional accountability • Regional accountability works in the Netherlands in two ways • 1. Performance subsidy of 2500 euro • 2. Naming and shaming • ’Naming and shaming’ can be effective: • e.g. Human rights enforcement (Hanfer-Burton, 2008) • Hospitals (Mason en Street, 2005) • Policy monitoring (Pawson, 2002) •  The underlying population characteristics / issues, however, have to be similar.

25. Regional accountability The example of Flevoland The educationinspectorateobserves low educationaloutcomes in the province of Flevoland Reason: 1. Duetoselectivemigration of low andmiddleincome families to Flevoland dueto low housingprices (CBS, 2010) More non-western migrantsandone family households  Have a higher risk on early school leaving (Rumberger, 1983; Olsen and Farkas, 1989; Kalmijn and Kraaykamp, 2003; Plank, DeLuca and Estacion, 2005; Bridgeland, Dilulio, Morison, 2006; Dustmann and van Soest, 2007)

26. Regional accountability The example of Flevoland Education inspectorate observes low educational outcomes in Flevoland 2. Few parents with high socio-economic status - High risk on ESL (Ekstrom, Goertz, Pollack and Rock, 1986; Lamb and Rumberger, 1998; Teese and Walstab, 2002; Entwisle, Alexander and Steffel-Olson, 2005; Dalton, Gennie and Ingels, 2009) - Peer effects on other students (Rumberger, 1983; Herbert and Reiss, 1999; Cooper et al., 2005) 3. Low expectations of teachers - Correlates negatively with early school leaving (Finn, 1989; Adams and Becker, 1990; Herbert and Reis, 1999; Blue and Cook, 2004; Dalton, Gennie and Ingels, 2009) 4. Limited resources for schools - As schools grow quickly and financing is based on the year before

27. Regional accountability The example of Flevoland Compare the performance of students in Almere / Lelystad with these of comparable cities and regions We estimate: “If a dropout student from Almere or Lelystad would have been born in another city, would he/she also have been dropped out?”  analysis of the counterfactual  Observations: Student and regional characteristics matter a lot and explain for a large part the differences in early school leaving Namingandshamingshould account forthis.

28. Dropoutpreventionschool accountability  First round: Monetary incentive for school of 2,500 euro per dropout less in comparison to base year 2005-2006 Note that the incentive is unfair if - Some schools had dropout prevention schemes before 2005 - Background characteristics of the students differ  We tested the latter for the difference in school dropout between various cities Conclusion: If not properly accounted for the student characteristics, the monetary incentives are unfair.  Second round: Monetary incentive for school of 2,500 euro per dropout less in comparison to year before  Creates an enormous attention in the schools for esl.

29. Dropout preventionTruancy

30. Dropout preventionTruancy  Truancy is problematic because it strongly correlates with - Criminality - Teen agepregnacies - Socialproblems - Early school leaving(e.g. Henry, 2007; Rumberger, 1983; DesJardins et al., 2006)

31. Dropout preventionTruancy  Truancy, truancy reporting and truancy policy (using quasi-experimental evidence) - Based on Amsterdam data: Truancy increases the probability of early school leaving by 3.9 percentage points cfr. Early school leaving percentage in the municipality of Amsterdam amounts to 7.8% (2005-2006) and 6.8% (2007- 2008) • Improved truancy reporting does induce lower dropout, but not significantly different from 0 Only for better general schools, we observe a significant effect

32. Dropout preventionTruancy - An active policy on truancy reporting (e.g., visiting the truant and his parents at home for an extensive discussion) creates a lower school dropout  An extensive project which involved both social workers, psychologists, school management and teachers  The average time spend on truants is about 5 hours  Immediate reaction towards truants and involve their parents  Cost-effective intervention

33. Outline • A conceptual framework for interventions B. What happened in the Netherlands? -- Policy and effectiveness 1. National registration 2. Naming and shaming 3. Regional accountability 4. School accountability (monetary incentives for schools) 5. Qualification Law (increased compulsory education age) C. Accounting for economic influences in school dropout

34. DropoutpreventionQualificationlaw Qualification law (2007): Students have to obtain a ‘starter qualification’ (= higher secondary diploma)  In practice: increase in compulsory education age for vwo and mbo students  ‘RMC verzuim’ = Truancy reporting for students younger than 23 who did not obtain a qualification yet  Idea: ifstudents are obligedtoattend school longer, theywill more likelygraduatewith a degree

35. DropoutpreventionQualificationlaw The effect of the qualificationlaw is mainlydrivenbystudents in the control group, whowerenotliableto the change in law. Effect mainly in The control group

36. DropoutpreventionQualificationlaw Thanks to qualification law: Decrease of early school leaving by 2.52 percentage points, but effect is mainly driven by non-liable pupils leaving school (i.e., groenpluk)  Particularly native students with a vocational education background and without retention in grade dropped out of school right before the introduction of the qualification law. Policy has adverse and unexpected effects

37. Outline • A conceptual framework for interventions B. What happened in the Netherlands? -- Policy and effectiveness 1. National registration 2. Naming and shaming 3. Regional accountability 4. School accountability (monetary incentives for schools) 5. Qualification Law (increased compulsory education age) C. Accounting for economic influences in school dropout

38. Policy versus economy A comparison based on output indicators is often too narrow ESL is strongly correlated by - Short term economic cycle (e.g. ‘Groenpluk’ in which youngsters are pull out of school in an economic boom) - Long term economic growth (e.g. increasing importance of secondary degree) - Institutional link between the labour market and education We corrected for 12 EU Member States the Eurostat numbers for those influences  Result: - Influence of the economy, institutions, … - Unexplained part attributed to policy making ESL-number:

39. Policy versus economy Chang is ESL between 2000 and 2008, corrected for all external influences

40. Costs of policy on esl Using predictions on a theoretical model, we predict the early school leaving rate from the Netherlands.  The underlying data are from 2005-06, but allow us to predict esl-rates before and after this point in time.

41. Costs of policy on esl € 574 million cost < € 4,714 billion  2,3% of GDP Program “Aanval Op Schooluitval” 13,5% Lissabon 15,4%

42. Costs of policy on esl In case of a very strict policy on early school leaving, it would cost about 0,1% of GDP to keep all youngsters at school. This huge amount is far less than the estimated costs of early school leaving: For the Netherlands: 4,714 billion euro.

43. Conclusion There is much to learn from early school leaving policy in the Netherlands Caution should be taken: - Some structural differences in educational system (e.g., two levels of three years, strong ability tracking, central exit exam). - Not all measures are effective Advice in setting-up policy: Make sure that policy can be evaluated. Do not implement a policy in all schools at the same time, but allow for an experimental and evidence based set-up!

44. ReferencesAcademic articles All of themavaibleby feb.kuleuven.be/leer • De Witte, K., Cabus, S., Thyssen, G., Groot, W. and Maassen van den Brink, H. (2013). A Critical Review of the Literature on School Dropout. Educational Research Review. In Press. • De Witte, K. and M. Csillag (2013), Does anybody notice? On the impact of improved truancy reporting on school dropout. Education Economics. In Press. • De Witte, K. and Cabus, S. (2013), Dropout prevention measures in the Netherlands, an evaluation. Educational Review 65 (2), 155-177. • De Witte, K., Nicaise, I., Lavrijsen, J., Van Landeghem, G., Lamote, C. and Van Damme, J. (2013). The impact of institutional context, education and labour market policies on early school leaving: a comparative analysis of EU countries. European Journal of Education. 330-345. • Cabus, S. and De Witte, K. (2012), Naming and shaming in a fair way. On disentangling the influence of policy in observed outcomes. Journal of Policy Modeling 34, 767-787. • De Witte, K. and Van Klaveren, C. (2012), Comparing students by a matching analysis – on early school leaving in Dutch cities. Applied Economics 44 (28), 3679-3690. • De Witte, K. and Rogge, N. (2012), Dropout from secondary education: all's well that begins well. European Journal of Education 47 (4), 1-20. • Cabus, S. and De Witte, K. (2011), Does School Time Matter? On the impact of compulsory education age on school dropout. Economics of Education Review 30, 1384-1398.

45. ReferencesSummaries for policy makers All of themavaibleby feb.kuleuven.be/leer De Witte, K. and M. Csillag (2011). De invloed van spijbelen op vroegtijdig schoolverlaten. (The influence of truancy on early school leaving). Tijdschrift voor Onderwijsrecht en Onderwijsbeleid 11-12 (2), 152-157 Cabus, S. and K. De Witte (2011). Werkt een actief verzuimbeleid? (Is anactivetruancy policy effective?). Tijdschrift voor Onderwijsrecht en Onderwijsbeleid 11-12 (1), 36-45. De Witte, K., S. Cabus, G. Thyssen, W. Groot and H. Maassen van den Brink (2010). Voortijdig schoolverlaten: Beleidsevaluatie 2010 – Deel 1 Verkenningen. NICIS Institute. Cabus, S., K. De Witte, M. Csillag, W. Groot and H. Maassen van den Brink (2011). Voortijdig schoolverlaten: Beleidsevaluatie 2011 – Deel 2 Kwalificatieplicht – Verzuimbeleid. NICIS/TIER Universiteit Maastricht. Thyssen, G., K. De Witte, W. Groot, H. Maassen van den Brink (2010). De stereotypen van voortijdigschoolverlaten. Impulsvooronderwijsbegeleiding 41 (3), 151-153. Cabus, S. and K. De Witte (2011). Invloed van leerplichtwet op voortijdig schoolverlaten (The influence of compulsoryeducation on early school leaving). Economisch Statistische Berichten 96 (4606), 166-169. Thyssen, G., K. De Witte, W. Groot, H. Maassen van den Brink (2010). De stereotypen van voortijdig schoolverlaten (The stereotypes of early school leaving). Impuls voor onderwijsbegeleiding 41 (3), 151-153.

46. Early school leaving in the Netherlands Policy and research Feb.kuleuven.be/kristof.dewitte Feb.kuleuven.be/leer Kristof.dewitte@kuleuven.be