This paper analyzes how teachers sort themselves across schools in Colombia. In particular, I use longitudinal data to study to what extent high quality teachers, measured by teachers’ education, experience and tests scores, are placed in high quality schools, measured by the sociodemographic characteristics of the students and their results in standardized examinations.

Perhaps more than in most education systems, Colombian laws provide clear channels for high quality teachers to be placed in high quality schools. First, the structure of the recruitment process can cause teachers’ sorting. Since 2005 teachers are recruited through a process that maps education, experience and tests scores into a single score. This single score is important because teachers can choose which school they want to go in strict descending order of the score, so high quality teachers have more options to choose from, than low quality teachers.

Moreover, sorting does not take place only at the beginning of the teachers’ career, but also throughout it. In effect, among incumbent teachers, seignority plays a major role on how teachers’ transfers work. When a position is opened in a given school, and more than one incumbent teacher is interested in it, it is explicitly stated that more experienced teachers will be more likely to be transferred than less experienced teachers.

Considering that wages are fixed at the national level, and assuming that more affluent and abler students are easier to teach, it can be expected that high quality teachers will choose to go to schools with more affluent and abler children.

To test this hypothesis, in the first part of the paper I study whether novice teachers that do well in the recruitment process tend to go to high quality schools. Preliminary results indicate that high performing teachers tend to go to schools with high quality students, where quality of the students is measured by past students’ test scores.

In the second part of the paper I study sorting among incumbent teachers. Using administrative records of all public school teachers in the country for years 2009 and 2012, I construct a panel that allows me to observe transitions such as moving to another school, moving to another municipality, or leaving the public education system. Roughly 20 percent of teachers experienced any of these transitions between 2009 and 2012. I analyze whether teachers’ quality, measured by their education, experience and test scores, are correlated with the quality of the students in schools they leave and are transferred to.

This is one of the first papers that analyze teachers’ sorting in Latin America using longitudinal data. The use of longitudinal data allows me to disentangle how students’ characteristics affect teachers’ quality, which in general is not identified in the context of cross sectional data. The results of this study, will contribute to the debate on the extent to which teachers value non-pecuniary factors, such as students’ characteristics.