Association for Public Policy Analysis & Management

Education programs targeted towards children at early stages of life have been proven to have important causal consequences on the development of cognitive and non-cognitive skill and therefore, on long-term outcomes (e.g. Cunha and Heckman 2008 and Heckman et al. 2015). However, the success of these programs can be limited in cases where a “curricula mismatch” exists between the language of instruction and the students’ mother tongue, as evidenced in several contexts. There are two typical circumstances where this mismatch takes place. First, among migrant populations that do not speak the mainstream language, and second, across ethnic groups from developing countries where the mainstream tongue is that of the former colonial power (Glewwe et al. 2009). This study focuses on the latter. Specifically, I analyze a bilingual education (BE) reform in Peru, for which the main purpose is to correct the instructional mismatch in schools where the majority of students belong to an indigenous community and speak a Peruvian vernacular language. The reform has three components: (i) production of workbooks in vernacular language, (ii) on-the-job training on bilingual education for teachers, and (iii) monetary incentives for teachers in bilingual schools (they represent between 5% and 14% of a teacher's monthly earnings.) In this paper, I focus on the short-run effects of the reform on elementary school students.

Using a difference-in-difference strategy where I compare the evolution of treated schools (schools where the majority of students speak a vernacular language) versus non-treated schools (schools where the majority of students speak Spanish), I find that the probability for a school principal to report that BE is offered increase by 50%. Further, I also find that in schools targeted by the program the percentage of trained teachers in BE and the probability of receiving textbooks in mother tongue increased by 70% and 500%, respectively. Regarding the effects on achievement, preliminary results suggest that repetition rates for students in grades 2-4 decreased by 10%

The results of the study may have important pro-poor implications if we consider the fact that 15% of the world population under poverty conditions corresponds to indigenous communities (United Nations 2010) and recent cross-country evidence suggesting that former colonies that adopted colonial language as language of instruction (as opposed to local language) is correlated with low levels of human capital accumulation (Laitin and Ramachandran 2016).