1. The Market for Education in England Simon Burgess Public Organisation Conference, June 2008

2. June 2008www.bris.ac.uk/Depts/CMPO2 Education Market in England Market problem is an assignment problem: Everyone is assigned to a school, but which pupils go to which schools? Focussing on the equity implications of the market here. This talk: –mostly drawing on “School Assignment, School Choice and Social Mobility” with Adam Briggs –partly drawing on “School Quality, School Access and the Formation of Neighbourhoods” with Tomas Key

3. June 2008www.bris.ac.uk/Depts/CMPO3 Introduction Not all schools are good schools Which pupils go to the good schools? To the extent that children from poor families are allocated to worse schools, this perpetuates disadvantage, reducing social mobility Questions: –What is the extent (if any) of a differential chance of going to a good school? –How does it happen? –What would be the impact of increasing choice?

4. June 2008www.bris.ac.uk/Depts/CMPO4 School choice School choice: –Promise of a well-functioning school choice system is that it reduces role of location –Countervailing view is that a choice system without fully flexible school size will increase the role of choice by schools, and the scope for the middle class to beat the system. Relative role for location as opposed to “working the system” is important.

5. June 2008www.bris.ac.uk/Depts/CMPO5 What we do: We estimate the chances of poor and of non-poor children getting places in good schools One of the key factors is location – distance between school and home. Our dataset allows us to measure distance very precisely and characterise the pupil’s very local area: –We compare pupils living in the same place. Exploit within-street variation and also control for other personal characteristics including prior test scores. –The difference is relatively small compared to the overall difference.

6. June 2008www.bris.ac.uk/Depts/CMPO6 Results Poor children half as likely to go to good schools. Much of that, but not all, comes through location. That is, accounting fully for location, the gap is much smaller, but not zero. Controlling for location, this gap doesn’t vary much by degree of choice. Children from poor families tend not to go to a good school, even if it is their nearest. Our econometric strategy is not to identify causal relationships in this paper (future work).

7. June 2008www.bris.ac.uk/Depts/CMPO7 Modelling Framework We model the assignment of children to schools, as a function of the characteristics of the school and of the children. It’s a matching problem. The observed data on the outcome of this assignment are realisations of an underlying process, composed of two decisions: –applications by parents and children for places in particular schools (demand), –and the administrative procedures that allocate children to schools given their choices (assignment rule)

8. June 2008www.bris.ac.uk/Depts/CMPO8 Given the basic structures of the problem, parents then formulate their response strategy: –the role of location –make any implicit advantages of their children visible to the admissions authorities, “working the system” Our strategy is to isolate how much of the difference in outcomes works through location, and how much through other channels, controlling for location.

9. June 2008www.bris.ac.uk/Depts/CMPO9 Allocation Write a general model of the outcome of the allocation as: where

10. June 2008www.bris.ac.uk/Depts/CMPO10 Reverse causation? We interpret the estimated relationship between the school’s quality score q a(i, t), t-6 and a student’s personal characteristic, f it, as representing the outcome of the assignment process. Alternative: from student characteristics to the outcome score.

11. June 2008www.bris.ac.uk/Depts/CMPO11 Timing: the quality score derives from the performance of a group of children 6 years older than the current intake. But: persistence in school attendance. Two interpretations: –“Islands story”: Schools located on “islands”, with no mobility between them. All students from succeeding generations therefore go to the school on their island. –Correlation from one generation’s poverty to the next. –But: this is not what England’s schools look like Half of children do not go to local school See map of Birmingham

12. June 2008www.bris.ac.uk/Depts/CMPO12 Figure 1: School Distance Contours in Birmingham

13. June 2008www.bris.ac.uk/Depts/CMPO13 –“Dynasties”: pupils living in particular locations always go to the same school. And with persistence in area poverty, particular locations always house poor families. –poverty of succeeding generations is correlated, score of one generation of pupils drawn from that area is correlated with the poverty of the next. –Econometrically, estimating: –Will be biased because omitted variable of the nature of i’s location is correlated with f i, and with the nature of the previous cohort of pupils who generated the school quality score. –Response: control for location to remove omitted variable bias; within postcode variation.

14. June 2008www.bris.ac.uk/Depts/CMPO14 Data Data on pupils Data on schools Data on location Our sample

15. June 2008www.bris.ac.uk/Depts/CMPO15 Pupils PLASC/NPD: Census of all children in state schools in England, taken each year in January. Key variable for our purposes is an indicator of family poverty, the eligibility for Free School Meals (FSM). Gender, within-year age, ethnicity, SEN,.. Key-stage 2 test taken at age 11 as the pupils finish primary school. This is a nationally set group of tests (in English, Maths and Science), marked outside the school

16. June 2008www.bris.ac.uk/Depts/CMPO16 Schools Quality of the secondary school that each child attends: use the publicly available and widely quoted measure of the proportion of a school’s pupils which passes at least 5 GCSE exams at age 16 (repeated using value-added). Define a “good school” as a school in the top third nationally of the distribution of %5A-C scores (repeated using top third locally) Dating – we use the score for each school from the time that the cohorts made their decisions on school applications, so deriving from the results of a cohort of pupils 6 years older.

17. June 2008www.bris.ac.uk/Depts/CMPO17 Location We have access to each pupil’s full postcode. This locates them quite precisely. Also the coordinates of the school, which locates it exactly. We rely on the postal geography of the UK for this analysis. Overall, there are about 1.78m unit postcodes covering 27.5m addresses. On average, there are 15 addresses in a unit postcode, but this varies. Using pupils’ postcodes, we match in data on neighbourhoods, on two scales: postcode, and area (ward = approx 12k people).

18. June 2008www.bris.ac.uk/Depts/CMPO18 Sample We take the cohort of new entrants into secondary school from each PLASC, so pupils in their first year of secondary school. Roughly 0.5m pupils in each cohort; we use 3 cohorts so our full sample is 1.57m pupils. State schools in England; non-selective LEAs (this cuts out 13.4% of the pupil total); omit pupils from some special schools, a few pupils are omitted if they have missing data. Sample for the overall regressions is 1.24m, 91% of the available total in non-selective LEAs.

19. June 2008www.bris.ac.uk/Depts/CMPO19 Results How much of the difference in probability of attending a good school is due to location? Need to control completely for location. Interpretation: location not exogenous – estimating how important choice of location is for parents’ strategy.

20. June 2008www.bris.ac.uk/Depts/CMPO20

21. June 2008www.bris.ac.uk/Depts/CMPO21

22. June 2008www.bris.ac.uk/Depts/CMPO22

23. June 2008www.bris.ac.uk/Depts/CMPO23 Table 5: Statistics on numbers of pupils per postcode

24. June 2008www.bris.ac.uk/Depts/CMPO24 Figure 5: Differences in school quality by differences in FSM status

25. June 2008www.bris.ac.uk/Depts/CMPO25 Table 6: Postcode-cohort FE regressions of school quality

26. June 2008www.bris.ac.uk/Depts/CMPO26 Table 7: LEA FE on full sample of whether pupil attends a good school

27. June 2008www.bris.ac.uk/Depts/CMPO27 Econometric Issues Reverse causation? Unlikely. The measure of quality used is essentially unrelated to the performance of the children in the postcode: –the measure relates to a cohort of children passing through the school 6 years previously. –the focus children clearly constitute a negligible fraction of the actual attendees of the schools –the use of within-postcode variation controls for any location effects. Selection bias? Likely. Direction seems clear. Will do some analysis of possible extent.

28. June 2008www.bris.ac.uk/Depts/CMPO28 Table 9: Postcode-cohort FE on School Quality by deciles of minimum distance to three schools

29. June 2008www.bris.ac.uk/Depts/CMPO29 Results Specialise school allocation question to whether a child goes to her/his nearest school. Focus on the interaction of child characteristic (FSM) and school quality. Again control for location

30. June 2008www.bris.ac.uk/Depts/CMPO30 Figure 6: Probability of pupils attending their nearest school

31. June 2008www.bris.ac.uk/Depts/CMPO31 Summary Children from poor families half as likely to go to good schools. Much of that, but not all, comes through location. That is, accounting fully for location, the gap is a lot smaller. Children from poor families tend not to go to a good school, even if it is their nearest.

32. June 2008www.bris.ac.uk/Depts/CMPO32 School Quality and Neighbourhood Formation Some results from (as yet incomplete) follow-up project on school quality and moving. Same data source, using more cohorts, tracking families moving house over five years. Comparing poor and non-poor families. Lot of care modelling ‘default’ secondary school for any location – three ways.

33. June 2008www.bris.ac.uk/Depts/CMPO33 Who moves, impact on default school quality …

34. June 2008www.bris.ac.uk/Depts/CMPO34 Probability of Moving

35. June 2008www.bris.ac.uk/Depts/CMPO35 Results so far Moving probability for the non-poor is influenced by quality of default school. For the poor this effect completely disappears. Moving within local area ten times more sensitive to school quality than cross-labour market moves. Main econometric challenge is initial conditions problem in dynamic non-linear panel model with unobserved heterogeneity. Follow Wooldridge (control for initial and lagged move status, stock of moves, initial quality) and results remain.

36. June 2008www.bris.ac.uk/Depts/CMPO36 Conclusions On-going project to understand the education market in England. –Role of different assignment rules –Equity aspects: Analysing the chance of children from poor families going to good schools How this comes about … –Efficiency aspects too …today’s talk is dynamics from perspective of children, but static view of school. –There may be trade-offs between assignment rules good for equity and those good for efficiency.

37. June 2008www.bris.ac.uk/Depts/CMPO37 Why do FSM-eligible children have lower probabilities of attending good schools? –Where they live; –Over-subscribed schools find ways of choosing pupils according to their incentives; –middle class parents are better at working the system of school admissions; –Costs of exercising choice prohibitive.

38. June 2008www.bris.ac.uk/Depts/CMPO38 Results and choice Promise of a well-functioning school choice system is that it reduces role of location Countervailing view is that a choice system without fully flexible school size will increase the role of choice by schools, and the scope for the middle class to beat the system. Findings cast some light on this debate: –location is associated with most but not all of the differential school quality. –policy which reduced the factor contributing to the greater part of the gap at the potential expense of widening the smaller part might have some attractions

39. June 2008www.bris.ac.uk/Depts/CMPO39 Annex

40. June 2008www.bris.ac.uk/Depts/CMPO40 Notation There are S schools denoted s, and P children denoted i. A child’s poverty status is measured by her Free School Meals (FSM) eligibility, denoted f i. The school average FSM eligibility is A child’s GCSE score is q i, and prior ability is k i. The average GCSE score of school s for time/cohort t is q s,t. This generated from a production function:

41. June 2008www.bris.ac.uk/Depts/CMPO41 Location and distance A pupil’s location is L i. Denote pupil i’s nearest school as n(i). The distance between pupil i and school s is d is. Denote pupil i’s actual school attended as a(i)

42. June 2008www.bris.ac.uk/Depts/CMPO42 Quality of school assigned to pupil i Quality score for a school s at time t is the school mean GCSE score for the cohort finishing in t, q s,t School to which i is assigned is a(i, t). So quality of the school to which pupil i from cohort t is assigned as q a(i, t), t-6

43. June 2008www.bris.ac.uk/Depts/CMPO43 Figure 2: Good to total school places per LEA for Non-FSM and FSM pupils

44. June 2008www.bris.ac.uk/Depts/CMPO44 Figure 3: Good to total places ratio for FSM pupils against good to total places ratio for Non-FSM pupils

45. June 2008www.bris.ac.uk/Depts/CMPO45 Table 2: Probit of whether pupil goes to a good school

46. June 2008www.bris.ac.uk/Depts/CMPO46 Selection bias The bias can be signed: –Assume equal dwelling-specific house prices within a unit postcode. –Expect FSM-eligible households living in the same street as ineligible households to be among the better off of such households. –Similarly, FSM-ineligible households living next door to FSM-eligible families are likely to be relatively poor compared to other FSM-ineligible households. –So income differences between households of different FSM status and living in the same street are likely to be lower than unconditional income differences between households of different FSM status. –If link between FSM status and school assignment is a relationship between household income and school assignment, our estimated differences are likely to be an underestimate of the true relationship. –Similarly, we would expect the FSM-eligible households in mixed neighbourhoods to be relatively interested in education, and the FSM- ineligible households relatively less.

47. June 2008www.bris.ac.uk/Depts/CMPO47 Figure 4: FSM vs Non-FSM gaps in school quality

48. June 2008www.bris.ac.uk/Depts/CMPO48 Table 8: Role of feasibility of choice

49. June 2008www.bris.ac.uk/Depts/CMPO49 Table 10: Probits estimating the probability that a pupil attends their nearest school

50. June 2008www.bris.ac.uk/Depts/CMPO50 Figure 6c: Fitted values Based on col 3 of table 10 for a white, female pupil born in September with average KS2 mean, English as first language, no SEN, attending a school in an urban area and with the mean distance to nearest good school