¿Son las escuelas particulares subvencionadas mejores que las municipales? Estimación de la ecuación de logro escolar para Chile
This work considers the relative performance of municipal schools (MUN) and individuals subsidized (PS) with different strategies from identification, ordinates from most basic - supposing the absence of selection slant and characteristics multilevels in the observations until which it relaxes such assumptions. The used source of intelligence is test SIMCE of mathematics rendered by the students of 4º basic year in the 2002, the characteristics of the students and their relatives whom the SIMCE in the survey gathers of parents, and the information on the educational establishments that it gives the Ministry of Education. Following the used methodology, the estimations revealed that a student who attends a subsidized particular school, obtains a gain in the test SIMCE of mathematics that fluctuates between 11.6 points (considered by Mínimos Cuadrados Ordinarios) and 17.3 points (considered by Mínimos Cuadrados in two stages with robust errors) versus if the student had attended a municipal school. On the other hand, the estimations that use instrumental variables show that, following the used method, a student that attend a subsidized particular school obtains between 14.1 points and 17.3 additional points in test SIMCE of mathematics that if she had attended a municipal school. In relation to the variability of the estimations, the more basic strategies of identification show a smaller difference in the scholastic performance between subsidized and municipal particular schools. On the other hand, between the techniques that correct by the characteristic multilevels and slant of selection in the observations, the method of Heckman - assuming different returns from the characteristics of schools MUN and PS, as well as different nested specifications from instrumental variables it reported a smaller variability in the estimation of the relative performance of both types of schools. In as much, for a certain specification of instrumental variables, all the techniques revealed noticeable differences in their results, and those that they controlled by the fact that the students are nested in establishments (clusters) exhibited major variance in the estimation. Finally, although the present thesis is focused in the puntaje differences enters types of schools both, the coefficients considered for the rest variable (characteristic of the students and establishments) agree with the awaited thing and the found thing in other studies. However, it is possible to mention that a limitation of the present study is that does not control by effect even (to peer effects).