IFSThe Role of Credit Constraints
in Educational Choices
Lorraine Dearden, Leslie McGranahan and Barbara Sianesi
IFS
Evidence from two British cohorts
© Institute for Fiscal Studies, 2005
Research questions
1. Extent to which short-term ‘credit constraints’ affect individual educational choices– Staying on in FT education past 16– Completing HE
2. Has this changed over time?– 1958 cohort (NCDS) – 1970 cohort (BCS)
© Institute for Fiscal Studies, 2005
Who stays on in school?
Proportion staying on by parental income quartiles – BCS70
Females Males
© Institute for Fiscal Studies, 2005
Who completes HE?
Females Males
Proportion achieving HE by parental income quartiles – BCS70
© Institute for Fiscal Studies, 2005
Is this evidence of credit constraints?
Family Income
Education
© Institute for Fiscal Studies, 2005
Is this evidence of credit constraints?
Family Income
Education
Credit Constraints
© Institute for Fiscal Studies, 2005
Is this evidence of credit constraints?
Family Income
Education
Credit Constraints
Cognitive abilityNon-cognitive ability
ExpectationsTastes …
Early + long-term factors- Family inputs- Environmental inputs
© Institute for Fiscal Studies, 2005
Is this evidence of credit constraints?
• Observed correlation between family income and educational outcomes could be due to:
a) short-run credit constraints
b) long-run family background and environmental effects correlated with family income and educational outcomes
• Our aim is to single out a)
© Institute for Fiscal Studies, 2005
How do we do this?
• Apply methodology of Carneiro and Heckman (2003) to the UK
• To estimate the share of individuals affected by short-term ‘credit constraints’
© Institute for Fiscal Studies, 2005
Operational definition of ‘credit constrained’
• Individuals from the top quartile of the income. distribution are not, by assumption, credit constrained
• All others are potentially credit constrained.
• Share who is credit constrained = Any residual gap in educational attainment between top income children and all other children with the same ability and the same early family and environmental factors
• NB: if we don’t manage to capture all family effects, estimates will be an upper bound.
© Institute for Fiscal Studies, 2005
Approach
• Split family income at 16 into quartiles• Split ability at 10/11 into tertiles
– math, verbal and non-cognitive measures
• Within each ability group work out the proportion of ‘credit constrained’ individuals after controlling for long-run family background characteristics: – mother’s and father’s education, family size and
structure, father’s social status at 16, race and region of residence at 16
© Institute for Fiscal Studies, 2005
Results: Staying On – BCS
Males
Raw 22.0%
© Institute for Fiscal Studies, 2005
Results: Staying On – BCS
Males
Raw 22.0%
+ Ability 13.1%
© Institute for Fiscal Studies, 2005
Stay On Rates: Males – BCS70
Males
0.1
.2.3
.4.5
.6.7
.8.9
1
Bottom Middle Top
Low 2nd 3rd Top Low 2nd 3rd Top Low 2nd 3rd Top
Unadjusted
0.1
.2.3
.4.5
.6.7
.8.9
1
Bottom Middle Top
Low 2nd 3rd Top Low 2nd 3rd Top Low 2nd 3rd Top
Adjusted
© Institute for Fiscal Studies, 2005
Results: Staying On – BCS
Males
Raw 22.0%
+ Ability 13.1%
+ Fam. background, region 7.2%
Only stat. significant gaps 7.0%
© Institute for Fiscal Studies, 2005
Results: Staying On – BCS
Males Females
Raw 22.0% 19.8%
© Institute for Fiscal Studies, 2005
Results: Staying On – BCS
Males Females
Raw 22.0% 19.8%
+ Ability 13.1% 14.0%
© Institute for Fiscal Studies, 2005
Stay On Rates: Females – BCS70
Females
0.1
.2.3
.4.5
.6.7
.8.9
1
Bottom Middle Top
Low 2nd 3rd Top Low 2nd 3rd Top Low 2nd 3rd Top
Unadjusted
0.1
.2.3
.4.5
.6.7
.8.9
1
Bottom Middle Top
Low 2nd 3rd Top Low 2nd 3rd Top Low 2nd 3rd Top
Adjusted
© Institute for Fiscal Studies, 2005
Results: Staying On – BCS
Males Females
Raw 22.0% 19.8%
+ Ability 13.1% 14.0%
+ Fam. background, region 7.2% 7.1%
Only stat. significant gaps 7.0% 6.2%
© Institute for Fiscal Studies, 2005
Results: Staying On – NCDS
Males
Raw 15.0%
© Institute for Fiscal Studies, 2005
Results: Staying On – NCDS
Males
Raw 15.0%
+ Ability 8.9%
© Institute for Fiscal Studies, 2005
Stay On Rates: Males – NCDS
Males
0.1
.2.3
.4.5
.6.7
.8.9
1
Bottom Middle Top
Low 2nd 3rd Top Low 2nd 3rd Top Low 2nd 3rd Top
Unadjusted
0.1
.2.3
.4.5
.6.7
.8.9
1
Bottom Middle Top
Low 2nd 3rd Top Low 2nd 3rd Top Low 2nd 3rd Top
Adjusted
© Institute for Fiscal Studies, 2005
Results: Staying On – NCDS
Males
Raw 15.0%
+ Ability 8.9%
+ Fam. background, region 1.3%
Only stat. significant gaps 0.0%
© Institute for Fiscal Studies, 2005
Results: Staying On – NCDS
Males Females
Raw 15.0% 14.2%
© Institute for Fiscal Studies, 2005
Results: Staying On – NCDS
Males Females
Raw 15.0% 14.2%
+ Ability 8.9% 9.3%
© Institute for Fiscal Studies, 2005
Stay On Rates: Females – NCDS
Females
0.1
.2.3
.4.5
.6.7
.8.9
1
Bottom Middle Top
Low 2nd 3rd Top Low 2nd 3rd Top Low 2nd 3rd Top
Unadjusted
0.1
.2.3
.4.5
.6.7
.8.9
1
Bottom Middle Top
Low 2nd 3rd Top Low 2nd 3rd Top Low 2nd 3rd Top
Adjusted
© Institute for Fiscal Studies, 2005
Results: Staying On – NCDS
Males Females
Raw 15.0% 14.2%
+ Ability 8.9% 9.3%
+ Fam. background, region 1.3% 2.3%
Only stat. significant gaps 0.0% 0.0%
© Institute for Fiscal Studies, 2005
HE Completion
• Target groups:– HE vs Anything Less– HE vs at least Level 2
• key marginal group who could access / would benefit from HE
• Attainment– Credit constraints might affect dropping out
© Institute for Fiscal Studies, 2005
Summary (stat. significant gaps - pp)
1958 1970
Stay-on – Males 0 7
Stay-on – Females 0 6
Males
HE vs Less 0 <3
HE vs ≥Level 2 0 <2
Females
HE vs Less <3 <2
HE vs ≥Level 2 6 3
© Institute for Fiscal Studies, 2005
Conclusions
• Short term ‘credit constraints’ have more impact on staying-on decisions for our younger cohort– Is 6-7% a large fraction?
• Upper bound• Conceptual policy experiment
– But: emerged between the two cohorts
• Less evidence of effect on HE completion– ‘Convergence’ for males and females to 2-3%
• Reduced for females• Emerged for males
• Policies earlier on may be well placed.