[Editor’s Note: This is the sixth in a series of excerpts from chapter 4 (education policy) of Abhijit Banerjee and Esther Duflo’s Poor Economics.]

Parents are not alone in focusing their expectations on success at the graduation exam: The whole education system colludes with them. The curriculum and organization of schools often date back to a colonial past, when schools were meant to train a local elite to be the effective allies of the colonial state, and the goal was to maximize the distance between them and the rest of the populace. Despite the influx of new learners, teachers still start from the premise that their mandate remains to prepare the best students for the difficult exams that, in most developing countries, act as a gateway either to the last years of school or to college. Associated with this is a relentless pressure to ‘modernize’ the curriculum, toward making it more scientific and science oriented, towards fatter (and no doubt weightier) textbooks – to the point where the Indian government now sets a limit of 6.6 pounds on the total weight of the book bag that first- and second-graders can be asked to carry.

We once followed some Pratham staff to a school in the city of Vadodara, in western India. Their visit was preannounced and the teacher clearly wanted to make a good impression: His idea was to draw an enormously complex figure on the board, representing one of the fiendishly clever proofs that Euclidian geometry is famous for, accompanied by a long lecture about the diagram. All the children (students in third grade) were neatly arranged in rows on the floor, and sat very quietly. Some might have been trying to draw a simulacrum of the figure on their tiny plates, but the quality of the chalk was so low that it was impossible to tell. It was clear that none of them had a clue what was going on.

This teacher was not an exception. We have seen countless examples of this kind of elite bias among teachers in developing countries. In collaboration with Pascaline Dupas and Michael Kremer, Esther helped design a reorganization of Kenyan classrooms, taking advantage of an extra teacher to divide the class in two. Each class was separated by prior achievement, to help children learn what they did not know yet. Teachers were then randomly assigned to the ‘top’ or ‘bottom’ track by a public lottery. Teachers who ‘lost’ the lottery and were assigned to the bottom track were upset, explaining that they wouldn’t get anything out of teaching and would be blamed for their students’ low scores. And they adjusted their behavior accordingly: During random visits, teachers assigned to the bottom track were less likely to teach, and instead more likely to be having tea in the teachers’ room, than those assigned to the top track.

The problem is not the high ambition per se; what makes it really damaging is that it is combined with low expectations of what the students can accomplish. References to a certain old-fashioned sociological determinism, whether based on caste, class, or ethnicity, are rife in conversations involving the poor. In the late 1990s, a team led by Jean Dreze prepared a report on the state of education in India, the Public Report on Basic Education in India (PROBE). One of the findings was:

Many teachers are anxious to avoid being posted in remote or ‘backward’ villages. One practical reason is the inconvenience of commuting, or of living in a remote village with poor facilities. Another common reason is alienation from the local residents, who are sometimes said to be squandering their money on liquor, to have no potential for education, or simply to ‘behave like monkeys’ . Remote or backward areas are also seen as infertile ground for a teacher’s efforts.

Narendra Modi’s government is evaluating an intriguing model of creating large schools (KG through 12) for rural Gujarat with free transportation and free food in the middle of a cluster of 10 villages to bring economy of scale. Larger schools with more students means more subject-specialist teachers and less likelihood of alienation. We wrote about this in Rurbanization in Gujarat – early signs of success.

A young teacher simply told the team that it was impossible to communicate with ‘children of uncouth parents.’

In a study designed to find out whether this prejudice influenced teachers’ behavior with students, teachers were asked to grade a set of exams. The teachers did not know the students, but half the teachers, randomly chosen, were told the child’s full name (which includes the caste name). The rest were fully anonymous. They found that, on average, teachers gave significantly lower grades to lower-caste students when they could see their caste than when they could not. But interestingly, it was not the higher-caste teachers who were doing this. The lower-caste teachers were actually more likely to assign worse grades to lower-caste students. They must have been convinced these children could not do well.

Children themselves use this logic when assessing their own abilities. The social psychologist Claude Steele demonstrated the power of what he calls ‘stereotype threat’ in the US context: Women do better on math tests when they are explicitly told that the stereotype that women are worse in math does not apply to this particular test; African Americans do worse on tests if they have to start by indicating their race on the cover sheet. Following Steele’s work, two researchers from the World Bank had lower-caste children in the Indian state of Uttar Pradesh compete against higher-caste children in solving mazes. They found that the low-caste children compete well against the higher-caste children as long as caste is not salient, but once low-caste children are reminded that they are low castes competing with high-caste children (by the simple contrivance of asking them their full names before the game starts), they do much worse. The authors argue that this may be driven in part by a fear of not being evaluated fairly by the obviously elite organizers of the game, but it could just as well be the internalization of the stereotype.