SUBASTAS PUBLICAS

I will devote only relative short space to what most might regard as the single most important element of improving education: identifying and expanding the particular actions that lead to increased learning achievement. That is, while most will admit that budget expansion will not necessarily improve performance, nearly everyone has a “favorite” input, action, or activity that if the incremental resources were devoted to that activity then budget increases would lead to greater attainment and learning achievement. There are three classes of inputs that are frequently identified as promising in the truly voluminous empirical literature relating

measures of student learning to inputs (see reviews of the ‘education production function’ literature by Fuller and Clark, Hanushek, Villegas and a slightly different tack on the literature in Pritchett and Filmer 1997).

• Instructional materials. Many studies find that some measure of instructional materials at either the class room (presence of absence of textbooks, for instance) or school (having a library) is associated with higher test scores. Many education projects and programs build on those findings and focus on increasing those identified inputs.

• Key Infrastructure. Many studies find that some measure of infrastructure (e.g. having desks, having latrines, having running water, having windows) is an important

determinant of progress. Again, many education projects and programs are structured around not just school construction but school upgrading of one type or another. • Teacher training. Many studies find that some measure of teacher knowledge of the

content and/or pedagogical practices is associated with higher test scores. Nearly every educational project or program has some component for pre-service or in-service “teacher training” that attempts to improve either teacher content knowledge.

Conclusion on expanding known quality improvements. I will not even attempt to review this literature, rather I will go straight to four points.

First, almost certainly there are interventions in each of these dimensions, which, if successful in their intermediate objectives (e.g. raising textbook availability, improving teaching) will have substantial, cost-effective, impacts on attainment and learning achievement.

Second, what the particular interventions are will vary across countries, regions, schools, etc. Since the “learning achievement” function is so complex and depends on the presence of a variety of inputs the impacts will vary.

Third, while “associations” are easy to establish using the data at hand the question of causal impact has been underplayed and the findings from studies using randomization find that the usual methods (OLS or HLM) would produce estimates that were quite wrong as predictors of program impact (see below on Kenya). This goes back to the issue that, since most of the observed variation in the data is the result of choices made by parents or producers, without a positive model and “identification” the associations are consistent with a very wide range of underlying causal models. For instance, lack of key inputs may just be a proxy for a

dysfunctional school situation. Take the example of textbooks. Suppose in the data there is a strong correlation of textbook availability and scores in a cross section of schools. This could be because the schools that did not have textbooks were dysfunctional in a variety of other ways. If this is the case then a program that gave textbooks to all schools might have no impact on

performance—even if it saturated the school with books--in spite of the strong positive

association of textbooks and performance--because the incremental textbooks would all be used in a still dysfunctional school and hence might have little or no impact.

Fourth, the key pragmatic objection is whether programs designed at eliminating

“bottlenecks” to learning by devoting incremental resources to high

productivity activities can be successfully

implemented. For instance, the world is littered with projects and programs for teacher training that had no impact on subsequent teacher classroom practices (see Box 1 for a

particularly depressing example from Pakistan).

In examining the claims that incremental spending on this or that activity would increase scores enormously, one needs to ask the question “then why is this not happening already?” Convincing evidence that some inputs have

incredibly high marginal

products relatively to other inputs is at the same time convincing evidence that existing producers are not efficient (since the implication of efficiency is equalization of marginal products per dollar). But if the system is such that there are no built-in tendencies towards

Box 1: The dismal state of teacher training in Pakistan, early 1990s.

Teacher training in this province is a mockery. We should close down the teacher training institutes and stop this nonsense. I have been teaching in a B.A./B.Ed. program for many years and see no signs that I have any impact on the students I teach.

—A university education instructor quoted in Warwick and Reimers (1995).

Most inmates of this system [two teacher training institutes] have no respect for themselves, hence they have no respect for others. The teachers think the students are cheats, the students think the teachers have shattered their ideals. Most of them are disillusioned. They have no hopes, no aims, no ambitions. They are living from day to day, watching impersonally as the system crumbles around them.

—Nauman (1990).

A national survey of Pakistan's primary schools suggests that these anecdotal accounts are only too true. Survey data on teaching practices "provide no basis for statements that . . . teacher training makes a substantial difference to how teachers teach." A 1998 study of teacher training suggests that "staff and faculty are professionally untrained, political interference is common, resources and facilities are poor and badly utilized, motivation and expectations are low and there is no system of accreditation to enforce standards." Embedded in an education system that was fundamentally unaccountable and lacked any outcome orientation, teacher training reflected worst practice.

Sources: Adapted from World Bank 2004, Warwick and Reimers

productive efficiency then why will a particular intervention to raise efficiency with no changes in systemic incentives succeed? I am not arguing that such a case can never be made—there are numerous successful piecemeal initiatives to act as counter-examples--but it cannot simply be assumed that piecemeal policy action can be effective as the many failures attest.

Opportunity 2.C) Quality improvements, supply side: Experimentation with rigorous