CAPÍTULO II. LA TAREA DE SÍNTESIS DISCURSIVA: PROCESOS Y
5. El papel de los conocimientos previos
Education and child labor are measured using three different sources of information that provide complementary evidence of the effects of gold mining. School administrative records follow enrollment, grade promotion, repetition, and dropout rates at different education levels. Exit exam databases have individual information on test scores and working situa- tion of senior high school students. DHS surveys, focus on households and allow estimating the effect of mining on school attendance and child labor.
School administrative records are collected annually by the National Statistic Depart- ment (DANE). School principals are required to complete the C600 form, which includes questions on previous year enrollment, grade promotion and repetition, dropouts, and transfers since 2005. Using the 2005-2013 datasets, I measure the initial enrollment and
at least eight years. These schools are for the most public (94.6%) and rural (78.8%). Primary schools are not always separate from middle schools and high schools. 95% of the schools in the sample offer primary education (grades 1-6), of which 20% also have middle school (7-9) and high school (10-11). Descriptive statistics of initial enrollment and progress throughout the year are presented in Panel A of Table 3.1. Primary schools enroll on average 73 students at the beginning of each school year, of which 82.7% get promoted, 6.7% repeat, 6.9% dropout, and the remaining 3.7% transfer to other schools. There are much less middle and high schools, but each one of them enrolls more students. Dropout rates are higher in middle school (7.4%), than in high school (4.7%), which reflects that students who enroll in high school are a relatively self-selected group. Recall that schooling is mandatory until age 15 or grade 9.
National exit exams -SABER 11 - are administered by the Colombian Institute for the Promotion of Higher Education (ICFES). They are required for college application and in some cases for graduation as well. Most senior students take them. Anonymized individual test scores are available since 2000 but I restrict the sample to the 2004-2012 period to have comparable results. While registering for the exam, students complete a survey that provides basic information such as date of birth and gender.8 Since 2008, students are also asked about their labor situation. Excluding metropolitan areas, there are 6,245 schools, of which 2,703 have at students taking the exam in at least eight years. The share of public and rural schools is smaller at this level, with 83.8% and 40%, respectively. As can be seen
8 Other characteristics, such as parents’ education and family income are not included in some of the years of analysis, and therefore cannot be used as controls.
in Table 3.1, schools have on average 52.2 students taking the exam every year. The test scores are slightly bellow the national mean 0.013 standard deviations) and 11% of the students report working.
The Ministry of Education, in coordination with the Secretaries of Education of each department, has geocoded over 24,000 schools in the country. There are nine depart- ments in which more than 85% of the schools (excluding metropolitan municipalities) are geocoded: four in the central region -Antioquia, Caldas, Quind´ıo and Risaralda- and five in the Orinoco-Amazon region -Meta, Putumayo, Guan´ıa, Guaviare and vaup´es. I focus on the four departments from the central region for two main reasons. First, they are historically connected and have comparable living standards. Caldas, Quind´ıo and Risar- alda were in fact colonized by Antioquian settlers during the nineteenth century, and they all benefited from the Coffee booms throughout the twentieths century. According to the 2005 Census, the poverty rates of the four departments oscillate between 16.2 and 22.9%, which is considerably bellow the Orinoco-Amazon region (39.9%) and the national average (27,7%). Second, the region has some of the biggest mining clusters in the country, but also enough municipalities without this activity. This is not the case of the Orinoco-Amazon region where there are relatively few mines. This point is further developed in the next
working (13%). (See Table 3.1). Approximately half of the schools without coordinates are classified as urban in the C600 records. In these cases, I impute the coordinates of the corresponding municipal towns. With this correction, over 97% of the schools in the region of study are geocoded. While most of the schools are located in the the central region of Antioquia, and the Coffee Region, there are relatively few schools in the Northeast and West of Antioquia, which reflects the low population density of these areas (Panel (a) of Figure 3.3).
The 2005 and 2010 waves of the DHS provide information on school attendance and labor situation of a sample of children aged 6 to 17. While the 2010 wave has GPS information, in 2005 it is only possible to accurately geocode the clusters classified as urban. I do so by imputing the coordinates of the corresponding municipal towns. Given the data limitations, the main estimates consider only households living in the urban area of municipalities under 200.000. There are 4,261 children aged 6 to 17 in non-metropolitan municipalities of Antioquia and the Coffee Region, representing 11% of the total sample. On average, 84% of the children go to school. There is some late entrance, only 66% of children under 8 attend school, and dropouts increase after 15. The percentage of children working increases with age; less than 3% work under 11, 8% in the group 12-14, and 19% over 15. The location of all the geocoded DHS clusters are presented in Panel (a) of Figure 3.3. Although there are less clusters, their spatial distribution is similar to that of the schools (Panel (a) of Figure 3.3).