Slide1 Slide2
C LASS S IZE:M AIMONIDES R ULE
? Strong debate about impact school quality,in particular class
size.
? I.e Kremer (2002) argues that class size in Kenya is too small,
could increase education by reallocating inputs (textbooks) or
making school free (uniform,fee).
? In the US: very contrasted results
? Angrist-Lavy: Impact of class size on test scores in Israel
D ETERMINATION OF C LASS SIZE
? Small class size in poorer areas in Israel:
? →class size negatively correlated with test score.
? Rule inspired from Maimonides (talmudic scholar) determines
class size
? up to 39, one teacher, 40-79,2 teachers, etc....
? Generates seesaw in predicted student teacher ratio.
? Some variation around that, but generally quite tight.
I DENTIFICATION STRATEGY AND RESULTS
? Idea: school with 39 students in a divisions should not be very
different from school with 40 student in a division, but
student-teacher ratio drops suddenly.
? Do we see the same for tests score? See Figure: test score as a
Slide3
function of student per division.
? Generalization: use predicted class size as an instrument for
actual class size, after controlling for smooth function of number
of pupil per divisions (square, cubics etc...).
? Find signi?cant improvements in tests scores: a decrease of 10 in
the pupil/teacher ratio increases reading scores by 2.5 points
(from an average of 72 points)
I DENTIFICATION STRATEGY AND RESULTS 1 I DENTIFICATION STRATEGY AND RESULTS 2