The Economics of the Public Sector Second Half

1
The Economics of the Public Sector Second Half

• Topic 6 Analysis of Human Capital Policies
• School Quality

2

• SCHOOL QUALITY
• There is a large debate about the importance of
  school resources on student outcomes. The most
  commonly suggested school quality reforms are
  class size reductions, institution of summer
  school programs, and increases in teacher
  salaries and per student expenditures.

3

• Usually it is hard to find large effects of
  school resources on students outcomes, at least
  large enough to compensate the costs of such
  investments. However, this is a controversial
  area of research, both in the US and the UK. For
  example, Dearden, Ferri and Meghir (2002) find no
  effects of school pupil teacher ratios in adult
  mens schooling and earnings in the UK.

4
(No Transcript)
5
(No Transcript)
6

• However, they do find a very large effect of the
  pupil teacher ratio at 16 on females wages at 33.

7

• It may be that pupil teacher ratios are already
  small to start with and further reductions may
  not have much of an effect. Policies should
  always be analyzed relative to the current state
  of affairs. The same policy may have very
  different impacts in different countries.

8

• Using the same dataset, Dustmann, Rajah and
  Soest (2003) argue that there is a strong effect
  of class size on post-compulsory schooling
  attendance, which then has an important effect on
  wages.

9
(No Transcript)
10

• Which paper shows the most accurate estimates?
  It is hard to say. However, even if Dustmann et
  al (2003) are correct one needs a careful cost
  benefit analysis of the policy (costs include
  things such as additional teachers and
  classrooms). Dustmann et al (2003) in that
  reducing class size may be a good investment

11

• The debate on the effects of class size in the
  US as much or even more controversial, although
  there is a clean experiment that can be analyzed
  STAR.
• Project STAR was an experiment in which 11600
  students and their teachers were randomly
  assigned to small- and regular-size classes
  during the first four years of school. This
  program was evaluated by many researchers,
  notably Alan Krueger.

12

• Krueger (2002) presents a summary of his
  findings on the STAR experiment.

13

• Krueger (2003) argues that if we assume that a
  standard deviation increase in test scores
  increases earnings by 8 we obtain reasonable
  rates of return for reduction in class size from
  22 to 15 in grades K-3.

14

• At face value, these findings suggest that
  reductions in class size is a potentially
  effective policy. However, generalization of
  these results requires some care, as does the
  generalization of results of any experiment. The
  reduction in class size in STAR was very large
  and only done over a few grades. Furthermore, the
  link between test scores and earnings may be much
  weaker than the one assumed by Krueger.

15

• On the other end of the debate, Hanushek (2003)
  argues that increases in school resources are not
  an effective way to improve school quality,
  especially if incentives in schools are
  neglected. When he analyzes different studies of
  the effect of school quality he fails to find a
  consistent pattern showing that school resources
  improve student outcomes. Krueger (2003), on the
  other end, claims that Hanusheks methodology is
  not valid. The debate is far from settled.

16

• Hanushek (1999) also shows that even though
  school resources have improved dramatically in
  the US the performance of students has not
  changed substantially.

17
(No Transcript)
18
(No Transcript)
19
(No Transcript)
20

• Furthermore, across countries he fails to find a
  systematic relationship between investments in
  schools and student performance in international
  tests.

21

• Carneiro and Heckman (2003) take some of the
  most optimistic estimates of the effect of school
  resources on labour market outcomes of
  individuals, from a study by Card and Krueger
  (1992), and compute the net present value of
  school resources policies under different
  assumptions.
• Card and Krueger (1992) estimate a 1-4 increase
  in earnings from a decrease in the pupil-teacher
  by 5 pupils per teacher.

22

• Only if we take very high-end estimates of the
  effect of schooling quality on earnings and
  discount costs by a very low rate do we find any
  sizeable positive effect of schooling quality on
  future earnings.

23

• This debate is far from settled. Both in the US
  and the UK there is some controversy surrounding
  very basic questions
• Are there strong effects of school resources on
  student achievement?
• Are there strong effects of school resources on
  labour market outcomes of individuals?
• Can we say that policies that increase school
  resources are cost effective?

24

• It seems odd to say that school quality has no
  effect on students outcomes. In fact,
  researchers agree that one component of school
  quality, TEACHERS, matter dramatically for
  students learning outcomes (Hanushek, Kain and
  Rivkin, 1998).
• Unfortunately there is not a clear idea of what
  makes a good teacher, or how can we design
  policies that create good teachers.

25

• Evidence for Developing Countries
• Conditional Cash Transfers Progresa (Mexico)
• If children attend school and health clinigs
  their families are entitled to cash grants.
  School enrollment increases by 3.4. Similar
  programs now exist in several other countries.
• Free School Meals Kenya Preschools
• Enrollment increased by 30.
• UniformsFree BooksClassroom Construction
  Kenya
• The only component with any impact on attendance
  was provision of uniforms.
• Deworming Kenya
• School participation increased for treated
  students, but also for their classmates in the
  same school, and even other students in nearby
  schools.

26

• Flip Charts Kenya
• No impact.
• Remedial Teaching India
• Hire women from community to teach basic
  literacy and numeracy to children with low
  achievement levels. Large test score gains.
• Teacher Incentives India
• Large reductions in teacher absence
• Computer Assisted Learning India
• Large increase in math knowledge

27

• There are two important lessons from the
  experiments in the developing world.
• 1) Costs Matter Reducing the cost of school by
  providing a conditional cash transfer, or by
  providing uniforms, has important effects on
  attainment. However, the effects on achievement
  (learning) are often quite low
• 2) Perhaps the reason why learning is so slow is
  because teachers have low quality. In fact, two
  interventions that improved teaching quality had
  large effects on achievement.

28

• Carneiro and Heckman reading of the debate (for
  the US) is the following
• Although the effects of schooling quality vary
  across environments and additional funding for
  some schools may be justified, marginal
  improvements in school quality are likely to be
  ineffective in raising lifetime earnings and more
  fundamental changes are required if we hope to
  see a significant improvement in our educational
  system.
• So what are these fundamental changes?
  Economists have argued that these are
  fundamentally changes in incentives of teachers
  and principals.

29

• SCHOOL CHOICE
• In most countries state schools are local
  monopolies with few competitors, since
  individuals have to attend the state school of
  their area of residence. Even though there may
  exist many good professionals in all these
  schools, the incentives of many principals and
  teachers to produce knowledge are weak. They are
  not accountable to anyone because it is not easy
  to monitor them.

30

• One valuable source of information parental
  and student perception of qualities of teachers
  and schools is rarely used to punish poor
  teaching.
• It is possible to opt out of the state school
  system by choosing to attend a private school,
  but this often comes at a high cost (payment of
  tuition fees).

31

• In the US, inner-city schools present the most
  difficult problems. Students in these schools
  show low achievement.
• It is hard to attract teachers to these schools
  where students are low achieving, unmotivated,
  and the environment is poor and often violent.

32

• Derek Neal (1997) studies the effects of
  Catholic schools on the achievement of
  individuals in the US. He finds that there is no
  difference in the achievement of suburban
  students attending Catholic and state schools.
  However, there are large difference between
  inner-city African-Americans and Hispanics
  attending Catholic and state schools. Opting out
  of the state schooling system only has an effect
  where this system is providing very low quality
  services inner-city schools.

33
Lack of school choice seems to hurt mostly
children from inner-city schools who cannot opt
out from the inner-city state schools.
34

• School choice has been advocated as a reform to
  improve the quality of educational services for
  students. Proponents of school choice argue that
  competition among schools to attract students
  will force schools to decrease costs and increase
  the quality of services provided. Additionally,
  by having parents actively choose the schools
  attended by their children, school choice systems
  would likely increase the degree of parental
  involvement in childrens schooling.

35

• Opponents of school choice argue that increased
  competition among schools will lead to increased
  stratification and inequality among students as
  well as a dilution of basic schooling standards
  and that poor parents lack the information and
  the ability to make informed decisions for their
  children. Hence, school choice systems would be
  most beneficial to those already able to exercise
  choice in the current system (the richer
  families).
• However, Derek Neal (2002) argues that it is
  possible to design school choice systems with
  different features that will lead to very
  different outcomes. The above debate may be too
  simplistic.

36

• What is the evidence?
• The evidence on this topic is growing, but it is
  controversial, and the debate is often more
  political than scientific.
• One of the most influential papers in this area
  is by Hoxby (2000), who claims that schools in
  districts with more competition perform better.

37
(No Transcript)
38
Gibbons, Machin and Silva find small effects of
competition among English Schools, except for
students in faith schools. Students from
non-faith schools do not benefit from larger
competition from other schools. Students from
faith schools benefit from competition from other
faith schools.
39

• Cullen, Jacob and Levitt (2000) find that
  increased choice in Chicago led to a dramatic
  increase in sorting of students by ability across
  schools. Motivated and high achieving students
  were able to choose the school they wanted to
  attend. Less motivated students did not take
  advantage of this choice.
• Bayer and McMillan analyze the effects of
  competition in the San Francisco Area and they
  find strong and large effects.

40

• One could argue that choice does not have strong
  effects because parents do not value school
  effectiveness. Indeed, that is precisely the
  finding of Rothstein parents value peers rather
  than effectiveness of schools.
• Sandra Black, and Gibbons and Machin find that
  parents are willing to pay a premium for living
  close to a good school, although this premium is
  not very high 1 standard deviation increase in
  test scores is associated with a 2 increase in
  house prices.

41

• The longest school choice program implemented at
  a national scale is in Chile. However, it is hard
  to evaluate the effect of such a program in
  school performance in Chile because there is no
  available data on school performance for the
  years prior to the implementation of the program.
• In Chile there are tuition charging private
  schools, but there are also free private schools
  (subsidized by the government) that compete with
  state schools, which also compete amongst
  themselves.

42

• Test scores at the school level are publicized
  so that parents can compare schools when deciding
  which one to choose. Parents can place their
  children in any school in Chile regardless of
  place of residence, as long as there are enough
  vacancies.

43

• Chile has seen a dramatic increase in private
  schooling and today about 50 of the students
  attend private schools. Annual spending on
  education as a percentage of GDP is as high as in
  any developed country.
• Nevertheless, a convincing evaluation of the
  program is still to be done.
• Recently, voucher experiments have been
  conducted in several different cities of the US,
  including Milwaukee, Cleveland, Minneapolis,
  Washington, Dayton and New York. Some of them
  have been evaluated.

44

• Petersen, Wolf, Howell and Campbell (2002)
  analyze the experiments in Dayton, Washington and
  New York. They only find consistent positive
  effects of the program on achievement for
  African-Americans in New York.

45

• Krueger and Zhu (2002) reanalyze this data and
  show that it has severe attrition problems and
  does not correctly account for family background
  variables. Their findings, which correct for
  these problems, show no significant impact of the
  program on achievement

46
(No Transcript)
47

• However, a robust finding reported by Peterson,
  Wolf, Howell and Campbell (2002) is that choice
  improves parental satisfaction with several
  aspects of the school, as well as school quality
  in different dimensions, and parent-school
  communication.
• Whether these are indicators of improved school
  quality or not with impact on students outcomes
  is still subject to scrutiny.

48
(No Transcript)
49
(No Transcript)
50
(No Transcript)
51
(No Transcript)
52

• There is also a voucher program in Colombia,
  analyzed by Angrist et al.
• The authors find strong effects on achievement of
  access to vouchers.

53

• In summary, some state schools seem to perform
  poorly. These tend to be located in the
  inner-city and serve the poorest children in the
  population. State schools are basically local
  monopolies that face little competition and weak
  incentives to attract students and teach them
  useful skills.

54

• Furthermore, as in most public bureaucracies,
  teachers pay is not based on performance or on
  their ability to teach skills that are on high
  demand. Competition across schools is likely to
  provide better incentives for both teachers and
  principals, and improve school quality.

55

• However, there does not exist (yet) decisive
  evidence that competition improves school quality
  and learning outcomes of students, at least on
  observed measures such as test scores, although
  an option out of the state inner-city school
  system seems to be of very large value. The
  stronger conclusion of the voucher experiments
  seems to be that parents with access to choice
  show a higher degree of satisfaction with their
  schools than parents without access to choice.
• The evidence on student sorting is also mixed.

56

• Furthermore, as Neal (2002) argues, it can be
  very difficult to provide incentives for good
  performance in schools (and in other public
  bureaucracies). The reason is that there are
  multiple dimensions people care about and at the
  same time most measurements are in only one of
  these dimensions test scores. Moreover, these
  measurements are often very imperfect (are these
  test scores adequate measures of cognitive
  achievement?).

57

• Performance-pay based of teachers based on test
  scores can create bad incentives
• - Teaching to the Test
• - Teacher Cheating
• Jacob and Levitt (2002) find that teachers in
  4-5 of Chicago elementary school classrooms
  change their students test answers to mask a
  poor performance of their classroom. Pay based on
  class performance increases the incentives to
  teach better, but also increases the incentives
  to teach. Therefore reforms in incentive
  structures in schools should be paired with
  increased monitoring of teachers behavior.

58
(No Transcript)
59
(No Transcript)