Education Flashcards
(91 cards)
1
Q
- Authors of empirical meta-analysis of returns to education
- Time period covered
- High income country average private rate of return to education?
A
- Psacharopoulous and Patrinos (2018)
- 1950-2014
- 8.2%
2
Q
Why might class size decisions be endogenous to outcomes (hence can’t run simple regression of the 2)?
A
- Low ability often put into smaller classes to receive extra support (but this doesn’t mean that small classes cause low achievement)
3
Q
Most important externality of education?
A
Tax benefit
4
Q
Countries in which explicit voucher systems been used?
A
Chile, Sweden
5
Q
Leckie and Goldstein (2009)
A
LEAGUE TABLE DATA = MISLEADING
- Shows performance of kids 3/5 cohorts ahead of new joiners
- E.g. new, effective headteacher may have since joined, not reflected in exam data
6
Q
Evidence of differing priorities of parents of high/low SES when choosing school for children?
A
Institute for Government (2012)
- High socioeconomic status parents place most weight on exam results
- Low socioeconomic status parents place greater weight on probability of admission and travel distance
7
Q
What do high socioeconomic status parents place most weight on when choosing school for children? (Institute for Government 2012)
A
Exam results
8
Q
What do low socioeconomic status parents place most weight on when choosing school for children? (Institute for Government 2012)
A
- Probability of admission
2. Travel distance
9
Q
Evidence of importance of peer effects
A
Rothstein (2006)
10
Q
Bradley and Taylor (2002)
A
- Analysed effect of competition on secondary exam performance in 90s
- Increased competition between schools led to significant improvement in exam results
- Size of effect much greater w/larger no. schools within 1km radius
11
Q
Gibbons, Machin and Silva (2006)
A
- No association between changes in competition between primary schools and exam performance
2a. Result could be due to focus on primary education
2b. Young kids less independent, so travel distance more important vs educational quality when choosing school
12
Q
Conclusion of literature review on impact of school vouchers on educational outcomes?
A
Barrow (2009)
- Most studies report small positive effect on achievement
- Reasonable no. studies not statistically significant
13
Q
Conclusion of literature review on impact of school competition on educational outcomes?
A
Belfield and Levin (2002)
- Reasonably consistent positive association
- Size of effects modest
- > 1/3rd estimates not statistically significant
14
Q
Evidence that design of school voucher schemes important?
A
Epple et al (2015)
- Initial studies showed no impact of voucher schemes in Chile and NZ
- Much more favourable effects reported following reforms to admissions policy legislation that prevents cream-skimming
15
Q
Association (pure correlation) between schooling and wages implies an x% increase in wages per year of schooling
A
10% (Angrist and Pischke)
16
Q
Explain ability bias in education
A
- High ability people likely to seek more education
2. These people also have higher wage-earning ability
17
Q
Different methods for identifying causal effect of education?
A
- RCT
- Natural experiment
- Instrumental variable
- Conditional independence assumption
18
Q
Angrist and Krueger (1991)
A
ESTIMATES RETURN TO SCHOOLING
- IV = quarter of birth
- Minimum schooling defined in terms of age, so those born earlier in year have lower level of minimal schooling
- Estimated return per additional year of schooling = 7.5%
19
Q
Evaluate quarter of birth as an instrumental variable for schooling (Angrist and Krueger 1991)
A
- Randomly assigned (not completely though, some aim for September baby)
- Relevant - only influences wages through years of schooling
- Weak instrument - effect is small and there is only variation of a few months
- External validity limited (LATE) – only concerns pupils who leave school at 16/17, a small fraction of people and effect might be different across rest of distribution (e.g. for ‘always-takers’ who continue to university)
20
Q
Which study used quarter of birth as an instrumental variable for estimating returns to education?
A
Angrist and Krueger (1991)
21
Q
Evaluate use of sample of identical twins to control for ability bias and estimate returns to education (Ashenfelter and Rouse 1998)
A
- Good control for innate ability
- Don’t know why twins differ in schooling (hence don’t know whether thing that influenced education choices also influenced wages)
- Twins more likely in IV births (expensive treatment, so may be associated w/more affluent parents)
- Necessarily small sample size (not may identical twins)
22
Q
Which study used a sample of identical twins to control for ability bias and estimate returns to education?
A
Ashenfelter and Rouse (1998)
23
Q
Evidence of signalling effect in education
A
- Harmon et al (2004)
A. Wages of educated employees vs self-employed (for whom signalling doesn’t matter)
B. Returns to schooling indicate small signalling effect:
(i) Employees = 6%
(ii) Self-employed = 5%
- Last year of school has particularly high returns
24
Q
Evidence (meta-analysis) of impact on class size on educational outcomes
A
Hanushek (2002)
- Little evidence that class size influences outcomes
- 72% studies = statistically insignificant
25
Efficiency reasons for state financing of education?
1. Positive externalities
2. Competition (minimum efficient scale for schools/universities and travel/switching costs lead to local monopolies)
3. Information (parents/pupils not well-informed about quality)
4. Imperfect capital markets (inability to borrow against human capital - no collateral - leads to socially and individually sub-optimal investment in human capital)
26
Examples of positive externalities of education
1. increased tax revenue
2. knowledge spillovers (Moretti 2004)
3. reduced crime (Machin et al 2011)
4. more inventions
5. social cohesion/cultural values
6. better health (Heckman et al 2016)
27
Why is investment in early years education more cost effective?
Heckman (2008)
Early investment in skills and education a virtuous cycle, so more cost effective
28
UK evidence that differences in early child test scores important for achievement later in life
Feinstein (2003)
1. Statistically significant relationship between test scores at 22 months and later achievement in life (e.g. probability of doing A-levels)
29
Evidence that effective early years education generates life-long gains in earnings?
Chetty et al (2011)
1. RCT
2. Above-average kindergarten teacher generates ~$320,000 more in total earnings vs below-average (for class of 20 students)
30
1. Evidence of impact of high-quality 'classroom environment' in early years education?
2. Causal mechanism?
Chetty et al (2011)
1. High quality classroom includes teacher quality, peer effects, class chemistry (cannot isolate individual factors)
2. Random assignment to higher quality classroom:
(i) Important predictor of kindergarten test scores (though effect faded in later school years)
(ii) Strongly related to adult outcomes (e.g. earnings, college attendance, marriage rates etc)
3. Soft/non-cognitive skills = likely causal mechanism (explains why effect fades in later school years but re-emerges in adulthood)
31
Why might parents under-invest in education of children?
1. Positive externalities – reduced crime, social cohesion etc (private benefit < social benefit due to external benefit)
2. (Lack of) altruism – parents may put lower weight on children’s development than social planner
3. Imperfect information – parents lack information on importance of investments in education/development and optimal way to invest
4. Credit constraints
32
How can government overcome reasons for parents' under-investment in children's education?
1. Reduce cost (subsidies)
| 2. Increase perception of benefits (e.g. some may not know benefits of reading to child every night)
33
RCT evidence of impact of parenting intervention on young children
Jamaica RCT (Walker et al 2005)
1. Weekly home visits over 2 year period for children aged 2-9
2. Taught mothers how best to stimulate child
3. Significant impact on range of outcomes, even into adulthood (incl. 25% higher wages)
34
Walker et al (2005) ran an RCT in Jamaica w/weekly home visits over 2 year period for children aged 2-9, which taught mothers how best to stimulate child.
What was impact on wages into adulthood?
25% higher wages
35
Evidence that children from lower socioeconomic backgrounds receive less investment in their early development?
Hart and Risley (2003)
1. Children from lower income households hear up to 30M fewer words before age 3
36
Reasons for EARLY investment in childhood development?
1. Human capital formation:
(i) Psychological literature suggests marginal investment in early childhood has greater impact on development outcomes
2. Skill dynamics:
(i) Early investments create virtuous cycle
37
Why might regressions over-estimate the returns to education?
High ability people seek more education and they had pre-existing higher potential wages
38
Why might regressions under-estimate the returns to education?
Low-ability people may want to compensate by getting more education
39
Example of IV research design used to estimate returns to schooling
Angrist and Krueger (1991)
1. IV = quarter of birth
2. Minimum schooling defined in terms of age, so those born earlier in year have lower level of minimal schooling
3. Estimated return per additional year of schooling = 7.5%
40
..... and .....
Association (pure correlation) between schooling and wages implies a .....% increase in wages per year of schooling
Angrist and Pischke
Association (pure correlation) between schooling and wages implies a 10% increase in wages per year of schooling
41
Oreopolous (2006)
ESTIMATES RETURN TO SCHOOLING
1. Exploits UK increase in school leaving age from 14 to 15
2. Almost ½ of all pupils affected
3. Estimated return per additional year of schooling = 10%
42
Example of natural experiment used to estimate returns to schooling
Oreopolous (2006)
1. Exploits UK increase in school leaving age from 14 to 15
2. Almost ½ of all pupils affected
3. Estimated return per additional year of schooling = 10%
43
..... (.....)
1. Exploits UK increase in school leaving age from ..... to .....
2. ...... of all pupils affected
3. Estimated return per additional year of schooling = .....%
Oreopolous (2006)
1. Exploits UK increase in school leaving age from 14 to 15
2. Almost ½ of all pupils affected
3. Estimated return per additional year of schooling = 10%
44
Ashenfelter and Rouse (1998)
ESTIMATES RETURN TO SCHOOLING
1. Sample of identical twins
2. Estimated return per additional year of schooling = 9%
45
Different IV approaches to estimating returns to schooling
1. QOB
2. Distance to next college
3. Opening of new colleges
4. Differences in tuition fees
46
Elements of the private returns to education
1. Higher wages
2. Better working conditions
3. More interesting work
4. Higher status
5. Better health (Currie & Moretti 2003)
6. Value of knowledge
7. Higher-earning spouse
47
Harmon et al (2004)
EVIDENCE OF SMALL SIGNALLING EFFECT
1. Wages of educated employees vs self-employed (for whom signalling doesn't matter)
2. Returns to schooling indicate small signalling effect:
(i) Employees = 6%
(ii) Self-employed = 5%
48
Hanushek (2002)
EDUCATION META-ANALYSIS
1. Little evidence that class size influences outcomes (72% of studies = statistically insignificant)
2. Spending per pupil matters v. little (66% of studies = statistically insignificant)
3. Teacher quality promising
4. Many successful PISA countries recruit teachers from top universities
49
Rothstein (2006)
1. Parents value best peer group more than most effective school
2. Large peer effect on test scores found
50
Evidence that parents value best peer group more than most effective school
Rothstein (2006)
51
Chetty et al (2011)
EDUCATION RCT
1. Above-average kindergarten teacher generates ~$320,000 more in total earnings vs below-average (for class of 20 students)
2. Random assignment to higher quality classroom:
(i) Important predictor of kindergarten test scores (though effect faded in later school years)
(ii) Strongly related to adult outcomes (e.g. earnings, college attendance, marriage rates etc)
52
Why is league table data misleading?
Leckie and Goldstein (2009)
1. Shows performance of kids 3/5 cohorts ahead of new joiners
2. E.g. new, effective headteacher may have since joined, not reflected in exam data
53
Institute for Government (2012)
DIFFERING PRIORITIES IN SCHOOL CHOICE BY SES OF PARENTS
1. High socioeconomic status parents place most weight on exam results
2. Low socioeconomic status parents place greater weight on probability of admission and travel distance
54
Barrow (2009)
LITERATURE REVIEW - IMPACT OF SCHOOL VOUCHERS
1. Most studies report small positive effect on achievement
2. Reasonable no. studies not statistically significant
55
Belfield and Levin (2002)
LITERATURE REVIEW - IMPACT OF SCHOOL COMPETITION
1. Reasonably consistent positive association
2. Size of effects modest
3. >1/3rd estimates not statistically significant
56
Epple et al (2015)
SCHOOL VOUCHER DESIGN
1. Initial studies showed no impact of voucher schemes in Chile and NZ
2. Much more favourable effects reported following reforms to admissions policy legislation that prevents cream-skimming
57
Heckman (2008)
Early investment in skills and education a virtuous cycle, so more cost effective
58
Feinstein (2003)
UK - EARLY CHILD TEST SCORES PREDICT ACHIEVEMENT IN LATER-LIFE
Statistically significant relationship between test scores at 22 months and later achievement in life (e.g. probability of doing A-levels)
59
Walker et al (2005)
Jamaica RCT
1. Weekly home visits over 2 year period for children aged 2-9
2. Taught mothers how best to stimulate child
3. Significant impact on range of outcomes, even into adulthood (incl. 25% higher wages)
60
..... (.....)
- Jamaica RCT
1. Weekly home visits over ..... year period for children aged .....
2. Taught mothers how best to stimulate child
3. Significant impact on range of outcomes, even into adulthood (incl. .....% higher wages)
Walker et al (2005)
- Jamaica RCT
1. Weekly home visits over 2 year period for children aged 2-9
2. Taught mothers how best to stimulate child
3. Significant impact on range of outcomes, even into adulthood (incl. 25% higher wages)
61
Hart and Risley (2003)
Children from lower income households hear up to 30M fewer words before age 3
62
Children from lower income households hear up to 30M fewer words before age 3
Hart and Risley (2003)
63
..... and ..... (.....)
Children from lower income households hear up to ..... fewer words before age .....
Hart and Risley (2003)
Children from lower income households hear up to 30M fewer words before age 3
64
What higher education funding system does Barr (2003) argue for?
1. Variable fees
2. All students entitled to full loan to cover fees + living costs, w/income-contingent repayments
3. Promote access, incl. extra financial help for those from poor households
65
Brighouse (2000)
REASONS FOR PARENTAL CHOICE IN EDUCATION
1. Inherently desirable
(i) Parents want it
2. Equitable – equitable to offer choice to all, not just those who can afford it
3. Efficient – good for schools because competition encourages efficiency + innovation
66
Necessary conditions for school competition to improve educational outcomes?
1. Schools incentivised to respond to changes in demand
2. Schools able to respond to changes in demand
3. Parents well-informed
4. Parents able to exercise choice
5. Parents care about ‘good’ schools (rather than, say, peer effects)
67
Why does government regulate and subsidise higher education, given v. high private returns?
1. Credit market failures/constraints
2. Asymmetric information (returns to education positive on average, but w/high variance)
3. Positive externalities
4. Incomplete information
68
How does the government intervene in higher education market?
1. Income-contingent student loans
(i) Alleviates credit constraints
2. Subsidise HE via teaching grants
(i) Targeted at subjects w/high social returns (e.g. medicine/STEM)
3. Information provision (e.g. Unistats)
(i) Tries to enable students to make better-informed decisions
69
Explanation for socioeconomic gaps in university attendance?
1. Funding – not main reason
(i) Large gaps in UK when HE was ‘free’ and participation gap actually fallen as tuition fees risen
2. Prior attainment – secondary school = vital period for intervention to improve access to HE
70
Evidence that there is a role for government intervention in higher education market via information provision
Letters sent by Department of Education on true cost of studying at high-status universities in UK resulted in significantly higher probability of applying to Russell Group uni
71
Evidence that providing excess capacity in schools expensive and usually resisted
Waslander and van der Weide (2010)
72
Waslander and van der Weide (2010)
Providing excess capacity in schools expensive and usually resisted
73
Ahlin (2003)
EVALUATION OF SWEDISH SCHOOL VOUCHER SCHEME
Increased school competition (voucher scheme) led to significant improvements in maths, but not English/Swedish
74
1. No association between changes in competition between primary schools and exam performance
2a. Result could be due to focus on primary education
2b. Young kids less independent, so travel distance more important vs educational quality when choosing school
Gibbons, Machin and Silva (2006)
75
1. Analysed effect of competition on secondary exam performance in 90s
2. Increased competition between schools led to significant improvement in exam results
3. Size of effect much greater w/larger no. schools within 1km radius
Bradley and Taylor (2002)
76
Parents of low SES less likely to:
1. Attend parents evening
2. Cultivate high educational and labour market aspirations
3. Be able to help w/reading and homework
Laureau (1987)
77
Laureau (1987)
Parents of low SES less likely to:
1. Attend parents evening
2. Cultivate high educational and labour market aspirations
3. Be able to help w/reading and homework
78
Evidence that low SES of parents impacts education of children
1. Laureau (1987)
A. Parents of low SES less likely to:
(i) Attend parents evening
(ii) Cultivate high educational and labour market aspirations
(iii) Be able to help w/reading and homework
2. Hart and Risley (2003)
(i) Children from lower income households hear up to 30M fewer words before age 3
79
Evidence of association between poverty and low educational attainment
Alcock et al (2012)
~1/2 the proportion of children eligible for free school meals obtained 5/more GCSE A*-C, compared to those not eligible
80
Alcock et al (2012)
~1/2 the proportion of children eligible for free school meals obtained 5/more GCSE A*-C, compared to those not eligible
81
~1/2 the proportion of children eligible for free school meals obtained 5/more GCSE A*-C, compared to those not eligible
Alcock et al (2012)
82
Empirical evidence that competition improves school exam results
1. Bradley and Taylor (2002) – increased competition between schools led to significant improvement in exam results
2. Belfield and Levin (2002) - literature review found reasonably consistent positive effect of competition (though effects modest and over 1/3 estimates not statistically significant)
83
EVALUATION OF SWEDISH SCHOOL VOUCHER SCHEME
Increased school competition (voucher scheme) led to significant improvements in maths, but not English/Swedish
Ahlin (2003)
84
SCHOOL VOUCHER DESIGN
1. Initial studies showed no impact of voucher schemes in Chile and NZ
2. Much more favourable effects reported following reforms to admissions policy legislation that prevents cream-skimming
Epple et al (2015)
85
UK - EARLY CHILD TEST SCORES PREDICT ACHIEVEMENT IN LATER-LIFE
Statistically significant relationship between test scores at 22 months and later achievement in life (e.g. probability of doing A-levels)
Feinstein (2003)
86
EDUCATION RCT
1. Above-average kindergarten teacher generates ~$320,000 more in total earnings vs below-average (for class of 20 students)
2. Random assignment to higher quality classroom:
(i) Important predictor of kindergarten test scores (though effect faded in later school years)
(ii) Strongly related to adult outcomes (e.g. earnings, college attendance, marriage rates etc)
Chetty et al (2011)
87
ESTIMATES RETURN TO SCHOOLING
1. IV = quarter of birth
2. Minimum schooling defined in terms of age, so those born earlier in year have lower level of minimal schooling
3. Estimated return per additional year of schooling = 7.5%
Angrist and Krueger (1991)
88
EDUCATION META-ANALYSIS
1. Little evidence that class size influences outcomes (72% of studies = statistically insignificant)
2. Spending per pupil matters v. little (66% of studies = statistically insignificant)
3. Teacher quality promising
4. Many successful PISA countries recruit teachers from top universities
Hanushek (2002)
89
LITERATURE REVIEW - IMPACT OF SCHOOL VOUCHERS
1. Most studies report small positive effect on achievement
2. Reasonable no. studies not statistically significant
Barrow (2009)
90
LITERATURE REVIEW - IMPACT OF SCHOOL COMPETITION
1. Reasonably consistent positive association
2. Size of effects modest
3. >1/3rd estimates not statistically significant
Belfield and Levin (2002)
91
Early investment in skills and education a virtuous cycle, so more cost effective
Heckman (2008)
