Three essays on human capital and labor markets for college graduates in Colombia Dissertation Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University By Norma Gomez, M. Graduate Program in Agricultural, Environmental and Development Economics The Ohio State University 2015 Dissertation Committee: Joyce Chen, Advisor Alessandra Faggian, Co-advisor Tim Haab c Copyright by Norma Gomez 2015 Abstract This research analyzes the relationship among human capital accumulation, return to education, and migration, for university graduates in Colombia. Taking advantage of newly available data on the graduate tests scores administered at the end of high school and at the end of college, information on labor market outcomes, and location; I use different models to study how much human capital is accumulated during college, how this capital is rewarded in the labor market, and how these incentives lead to reallocation of human capital across regions in Colombia by analyzing migration decisions. The first chapter is based on the uniqueness of the national tests scores in Colombia where the same students take standardized test both at the end of high school and at the end of college.
I use these scores to estimate the value added of college education by university and then aggregate it by region using an Average Residual model. I define 10 regions in the country and rank universities according to value added, presenting rankings by region. I find that public universities outperform private ones in terms of value added, for the regions on the top 5. These results suggest that public universities generate higher value added compared to private universities, also that universities in the regions where the main cities are located outperform universities in second-tier cities.
This regional disparity is relevant, since the development process ii in these smaller cities might benefit from generating higher value for their graduates, which could stay in the region, innovate, and attract physical and human capital. The second chapter estimates returns to college education in Colombia. Using data for wages of college graduates, as well as information on standardized test scores, and university where they graduated from, I estimate a Mincer equation type model to obtain the return of education, and I look at the hierarchical structure of this return by using a multilevel model. I obtain heterogeneous estimates of returns for college education in Colombia, and prove that accounting for this heterogeneity matters.
I find that the wage premium for graduating from top-ranked universities is around 12%, while the penalty for graduating from bottom-ranked universities is close to 10%, higher than the estimated wage gap (8%). The magnitude of the coefficients suggests that the importance of that heterogeneity is comparable to the gender issue in wage gap, although it has received significantly less attention in the literature. The third chapter uses data from different ministries in Colombia to follow college graduates since they are in high school, when they go to college and once they join the labor market. I focus on the differences between the probability of migration of repeat and return migrants, and a combination of both groups.
I estimate separate logistic regression models for the repeat migrants, the return migrants, and both groups combined, and I present results that suggest that there are significant differences between the repeat and return migrants, therefore, taking into account the destination of the worker leads to different conclusions. This suggests that previous studies on internal migration in Colombia might be biased. Furthermore, having an additional iii wave of migration -from high school to college- adds valuable information to the migration analysis. iv Acknowledgments During the last four years I have received support and encouragement from a great number of individuals.
Alessandra Faggian has been a mentor, and excellent advisor, and over the years has become a dear friend. Her guidance has made this a fulfilling journey. Joyce Chen provided her support not only as advisor, but as instructor. I have learned a great deal from her classes, and from watching her teach.
I would like to thank Dr. Tim Haab for his support over the past two years as an idea became a completed manuscript. In addition, thanks to Jenny Gnagey, who patiently went over previous drafts of this manuscript and provided valuable comments. I also acknowledge the support from the Colombian Institute for the Evaluation of Higher Education ICFES, which provided funding for one of these chapters, but also considerable technical support.
I would like to express my deepest gratitude to my fellow classmates and friends, from whom I learned so much, but also, who have made being away from my country, my family, and my husband, much more bearable. Finally, thanks to my mom and family who cared for me at every step, despite the distance. Thanks to Norman, my loving husband who stood by me in the darkest hours, and to my baby, Gabriela, who was my best companion while finishing this dissertation. Economics, Universidad Nacional de Colombia 2011.
Master in Public Administration M., Columbia University Awarded a Scholarship from Bank of Japan/World Bank Graduate Scholarship Program 2014. in Agricultural, Environmental and Development Economics, The Ohio State University Publications Books Economı́a Matemática en Matlab (2008) (Mathematical Economics in Matlab). With Norman Maldonado, Eduardo Sánchez and Lida Quintero. ISBN 978-958-719-110-3 Book Chapter Strategic Asset Allocation: Balancing Short Term Liquidity Needs and Real Capital Preservation for Central Banks.
With Javier Bonza and Reinaldo Pabón. Edited by: Arjan B. Berkelaar, Joachim Coche and Ken Nyholm, in: Central Bank Reserves and Sovereign Wealth Management. ISBN-13 978-0230580893 Fields of Study Major Field: Agricultural, Environmental and Development Economics Specialization: Labor Economics, Economics of Education, Regional Economics vi Contents Page Abstract.
vi List of Tables. ix List of Figures .6 Analysis of an extended dataset. 79 viii List of Tables Table Page 1.1 Descriptive Statistics for Application 2011-2 .2 Universities by Geographical Location .3 Universities by Type .4 Average Residuals all universities .5 Average Residuals: Public universities .6 Average Residuals: Private universities .1 Fields of Study .2 Estimates of multilevel model .3 Estimates for returns of college education .1 Definition of Variables .2 Migration type of Colombian graduates .3 Migration type by gender .4 Descriptive Statistics of the individual and university .5 Descriptive Statistics of the location .6 Results from the logistic regression .7 Results from the logistic regression - Controls .8 AME from the logistic regression. 70 x List of Figures Figure Page 1.1 Regions defined for the analysis .2 Value added: All universities .3 Value added in 2008: All universities .4 Value added in 2009-1: All universities .5 Value added in 2009-2: All universities .6 Value added in 2011-1: All universities .7 Value added in 2011-2: All universities .1 Cumulative distribution of βj (individuals) .2 Cumulative distribution of βj (universities) .1 Location of universities in the sample .2 Number of graduates per 1,000 inhabitants, per departamento .3 Graduates per 1,000 inhabitants, by departamento in 2011 .4 Number of graduates in the sample by location.
65 xi Chapter 1: Value Added in College Education: A Regional Approach 1.1 Introduction The development of indexes to measure the quality of higher education institu- tions (HEI from now on) is essential to improve a national education system and ultimately a national economy. Standardized student tests have been widely used as a tool to assess education quality. However, in most countries these tests are usually administered only before the students enter college and not later on. Colombia is an exception in this respect, and as far as I am aware, it is the only country in the world that requires students to take two standardized national tests: one, SABER 11, at the end of high school, and another one, SABER PRO, at the end of their tertiary education.
The Instituto Colombiano para la Evaluación de la Educación Superior (ICFES) collects data on students’ performance in both exams. Taking advantage of this peculiarity of the Colombian education system, in this chapter I use an Average Residual model to measure university and region value added based on the SABER 11 and SABER PRO tests scores. I also look at how, and if, a university location affects its value added. 1 There might be differences in value added in different regions or in different ge- ographical contexts (e.
The geographical variation in the HEIs value added offers new insight into the regional differences in educational outcomes in Colombia and sheds some light on why these differences exist. Although there is large heterogeneity in educational outcomes among the different regions in Colombia, this issue still remains rather under-explored. I produce a regional value added ranking which could be used to design education policy, but could also be linked in the future to other relevant variables, such as regional innovation or growth. Test scores represent a useful measure of the human capital individuals accumulate during college, and they are a practical method to assess the contribution of different universities in the economy of a country.
The performance measure developed in Colombia is unique in the world and has been implemented only recently, which means that the information collected from these tests can be a tool to design policies to increase the quality of HEI and to enhance the matching between local labor supply and demand, since SABER PRO could be used as a signal of student quality for employers in the country. The chapter is organized as follows.2 introduces the theoretical frame- work.3 describes the data, while section 1.4 presents the methodology used. The results are discussed in section 1.5 and the analysis of an extended dataset are presented in section 1.2 Theoretical framework Government spending in education is central to increasing the economic perfor- mance of a country. While the costs of education are clear and tangible even in the 2 short run (in terms of direct costs, but also opportunity costs as foregone income from labor), the returns to education are more difficult to quantify because they often take a longer time to become visible and because of the positive externalities, at the indi- vidual and local level, associated with more years of education, such as reduction in crime.
There is also an increasing interest in designing clear mechanisms to assess the return to the investment in education measured as performance of schools and HEI. Among the models to measure school quality and economic returns to education, the production function or input-output approach has been widely used. The model assumes that the student receives multiple inputs from her home and school which produce cognitive achievement (Todd and Wolpin 2003). Value Added Models (VAM) are a set of statistical models based on the production function approach, they use test score data to measure student achievement, and estimate the effect of school or teachers on student performance (McCaffrey et al.
It starts from the production function approach where the value that college education adds to students depends also on the skills they have previously acquired, in particular VAM assume that a lagged test score provides information for historical inputs that the student previously received, innate endowment or other unobservables (Todd and Wolpin 2003). The interest on the input-output approach to generate policy results in schools in the United States was initially motivated by the Equality and Educational Op- portunity (the “Coleman Report”) in 1996, where school effects were examined in a large sample of students from the US Department of Education. The findings of the report challenged the effect of teachers and schools on performance and highlighted 3 the correlation between family inputs and achievement (Rivkin, Hanushek, and Kain 2005, p. VAM have proven to be useful for providing an answer to whether the perfor- mance of a teacher affects cognitive achievement of a student or a class; since VAM incorporate student inputs that teachers and schools cannot influence, they control for these unobserved factors and offer a more precise measure of the impact of teachers or schools on student achievement.
