Chapter 10This is quite a high cost. So the policy maker might ask
a different question: Is there any other measure by which
I could raise the educational performance of a poor child?
As online Table A10.1 shows, there is a high correlation
between intellectual performance and the quality of sec-
ondary schooling. Anticipating Chapter 14, we find that
the difference in performance between the best and worst
performing three secondary schools in Avon (holding child
and parent characteristics constant) is 0.46 standard devi-
ations of GCSE points. Suppose it took £2,000 a year per
pupil to lift a school that far— or £10,000 over five years.
That would be an increase of 0.46 standard deviations per
£10,000— much more than could be achieved by a direct
income transfer to parents costing the same amount.
If you are wondering whether these results are too
negative about the quantitative impact of income on aca-
demic performance, they are in fact consistent with earlier
research.^9 For example Jo Blanden and Paul Gregg used
three earlier British datasets (BCS; BHPS; and the National
Child Development Study, NCDS) to examine how family
income at age 16 affected GCSE performance.^10 They con-
cluded that when family income falls by 33% the propor-
tion of children who obtain any GCSE A*- C grades falls by
3– 4 percentage points.^11 This corresponds to a β- coefficient
of around 0.1— similar to our estimate of 0.15.
For the United States, Daron Acemoglu and Jörn- Steffen
Pischke have used three longitudinal studies of school leav-
ers sponsored by the National Center for Education Sta-
tistics. These show that a 10% rise in family income leads
to a 1.4 percentage point rise in the probability of college
attendance, which implies a β- coefficient of around 0.14—
again similar.^12 Likewise at the earlier ages of three and five,