Teacher Value Added in a Low-Income Country

Abstract

Using data from Pakistan, we show that existing methods produce unbiased and reliable estimates of teacher value added (TVA) despite significant differences in context. Although effective teachers increase learning substantially, observed teacher characteristics account for less than 5 percent of the variation in TVA. The first two years of tenure and content knowledge correlate with TVA in our sample. Wages for public sector teachers do not correlate with TVA, although they do in the private sector. Finally, teachers newly entering on temporary contracts with 35 percent lower wages have similar distributions of TVA to the permanent teaching workforce. (JEL I21, J31, J41, J45, O15)

New evidence on learning trajectories in a low-income setting

“…Indeed, Andrabi et al (2011) showed that the value-added specification precisely replicated the gains among children switching to a private school, even though the gains specification yields an approximately 0 coefficient on private schooling. This result foreshadowed a large and growing literature using value-added models to estimate the productivity of teachers ( Chetty et al, 2014 ; Bau and Das, 2020 ) and schools ( Angrist et al, 2017 ; Andrabi et al, 2020 ). 26 Using as a control to address omitted variable bias is therefore well-established in the production function literature and in RCTs, where it serves to increase precision, given that ) = 0 ( Bruhn and McKenzie, 2009 ).…”

Engaging Teachers with Technology Increased Achievement, Bypassing Teachers Did Not

“…Further, the skills necessary to earn a university degree might not be applicable to the middle school curriculum. In Punjab, Bau and Das (2020) found that a teacher having a college degree is associated with only a 0.2 SD increase in a teacher's test score on a upper primary school exam. They estimate that including all available teacher characteristics (including training and experience) explains at most 9 percent of the variation in teacher content knowledge.…”

“…When we combine the project and standardized test scores in a single measure of student achievement, the Classrooms intervention improved test scores 0.26SD. This improvement is approximately equivalent to increasing the value-added of a teacher by 1.8SD in Punjab, Pakistan (Bau and Das 2020), 0.7SD in private secondary schools in Uttar Pradesh, India (Azam and Kingdon 2015), or 1.3 to 1.8SD in the US (Hanushek and Rivkin 2010;Jackson, Rockoff, and Staiger 2014). Further, the Classrooms intervention increased the likelihood that students passed the standardized grade 8 test by 5 percentage points.…”

Long‐run effects of teachers in developing countries

“…Such data are more common in high‐income countries that have extensive official digitalized testing infrastructure. The few papers focused on developing countries have instead made use of survey data (Araujo et al, 2016; Bau & Das, 2020; Buhl‐Wiggers et al, 2017) and administrative data from a private school network in India (Azam & Kingdon, 2015).…”