African countries the expansion of education has exceeded the growth of wage
employment leading to declining returns to education. Third, that the quality of
schooling may be so low that it fails to raise cognitive skills or productivity.
There is ample evidence that the quality of education in developing coun-
tries is often very poor. A proxy measure for education quality is teacher absen-
teeism. The Public Report on Basic Education (PROBE, 1999) provided the
first serious evidence-based study of the quality of primary schooling in India.
It was based on a survey of schooling facilities in 242 villages in five North
Indian states (Bihar, Madhya Pradesh, Rajasthan, Uttar Pradesh and Himachal
Pradesh) in 1996. PROBE found poor school infrastructure: 26 per cent of
schools did not have a blackboard in every classroom, 52 per cent had no play-
ground, 59 per cent no drinking water, 89 per cent no toilet, 59 per cent no
maps/charts, and 77 per cent no library. In half of the schools surveyed there
was no teaching activity at the time of the observations, though often teachers
were physically present. Another survey of teacher absence in rural India in
2003 was based on three unannounced visits to each of 3,700 schools in 20
major states. The study found that on average 25 per cent of teachers in govern-
ment primary schools were absent from school on a given day and of those
teachers present only half were found to be teaching (Kremer et al.,2005). The
results from surveys when enumerators made two unannounced visits to
primary schools (and health clinics) in Bangladesh, Ecuador, India, Indonesia,
Peru and Uganda showed that averaging across countries, 19 per cent of teach-
ers (and 35 per cent of health workers) were absent (Chaudhury et al.,
2006:92). Together these results imply that around two-thirds of employed
teachers are not teaching at any particular moment during the school day.
The best type of data for ‘education quality’ would be comparable direct
measures of the quality of education. There are only a few such examples. One
measure of cognitive skills is for children enrolled in the ninth grade in 288
schools from the Indian states of Rajasthan and Orissa. The results show that
42 per cent of enrolled children in Rajasthan and 50 per cent in Orissa are
unable to ‘show some basic mathematical knowledge’. Since secondary enrol-
ment in India is only 53 per cent, the performance of the median child (which
includes those not enrolled in school) is certainly considerably worse. Results
of the top 5 per cent of performers in these two states are comparable to those
in high-income countries, though the average child in both states studied is
being badly failed (Das and Zajonc 2010).
Using this new research the link between student achievement tests and
economic growth has been re-examined internationally. One study uses data
between 1960 and 2000 with a sample of 50 countries that have comparable
quality measures and data on GDP growth. The measure of the quality of
education, a simple average of the mathematics and science scores in interna-
tional tests of school children, has a statistically significant impact on the
growth of real GDP per capita. Adding educational quality (to a model that
only uses initial income and years of schooling) increases the share of varia-
tions in economic growth explained from 25 per cent to 73 per cent. Sharp
124 Sources of Growth in the Modern World Economy since 1950