Eventos pasados

Seminario Pepe (políticas públicas) - Ahorro sin barreras en Colombia

Con el objetivo de conseguir que la población de América Latina y el Caribe ahorre para su vejez, el BID puso en marcha el Laboratorio de Ahorro para el Retiro en Colombia, Chile, México y Perú, en 2016. Esta iniciativa busca soluciones innovadoras para fomentar el ahorro mediante intervenciones basadas en la economía del comportamiento y uso de nuevas tecnologías, con el fin de generar evidencia confiable a través de evaluaciones de impacto rigurosas.    Este evento, co-organizado por la Universidad de los Andes y el BID, se enfocará en los aprendizajes del Laboratorio y sus implicaciones de política pública en el caso de Colombia, donde se llevó a cabo cuatro intervenciones. Se aprovechará la conversación para repensar el alcance de estos aprendizajes, aplicables a la promoción de autoaseguramiento de trabajadores de plataformas digitales y a la contribución a la seguridad social.   El evento contará con presentaciones de Oliver Azuara, Economista sénior de la División de Mercados Laborales del BID, y Gustavo Caballero, Consultor de la División de Mercados Laborales del BID, seguido de un panel de políticas públicas integrado por Alice Gutiérrez Stanley, Directora de Negocio Mandatorio de SURA Asset Management, Juan Miguel Villa, Presidente de Colpensiones, Óscar Becerra, Facultad de Economía, Universidad de los Andes, y Luz Magdalena Salas, Departamento de Economía de Pontificia Universidad Javeriana. El evento será en español y contará con interpretación simultánea al inglés.

Seminario Pepe (políticas públicas) - Luis Omar Herrera

The new millennium brought with it a new challenge: Students reaching higher education were not only greater in number, they were also less prepared than their predecessors. This caused a sharp deterioration in the quality of the entire higher education system, which in turn affected enrollment and graduation rates. Colombia faced this challenge with severely scarce economic resources, but the Ministry of Education designed a plan to keep students in higher education and eventually increased the overall enrollment rate from 20% in 2002 to over 42% in 2010. One of the tools used to implement this plan was the System for the Prevention and Analysis of School Dropouts in Higher Education (SPADIES). This software dashboard facilitates the collection, analysis, and visualization of student data. Spadies was instrumental in allowing information to flow easily among all the agents.  This paper provides evidence that SPADIES was vital to achieving higher educational attainment outcomes across the country. Using a differences-in-differences approach, I find that SPADIES reduced the probability of dropping out by 1.2% (impact of 4.35%), increased the probability of graduating by 0.6% (impact of 1.3%), and increased the probably of graduating on-time by 0.6% (impact of 1.9%). I also find that the number of transitions (i.e. student status changes from absent or dropout to enrolled) decreased 3.4% (impact of 48.4%), and the average duration of a student's gap in enrollment decreased 0.2 semesters (impact of 14.4%).

Seminario PePe (políticas públicas) - Luis Fernando Mejía

Seminario PePe (políticas públicas) - José Ernesto López-Córdova

Seminario PePe (políticas públicas) - Fabio Sánchez

Seminario PePe (políticas públicas) - Carlos Arteta

Seminario de políticas públicas - Luis Omar Herrera

Using administrative data, I track the path of 5.4 million secondary school graduates to higher education and the formal labor market in Colombia. I compare graduates from the same secondary school cohort to estimate the premium of higher education and workers that earned a bachelors against their college classmates that finished 90% or more of the college course-work but did not graduate to estimate the sheepskin effect. Using a modified Mincer equation, I find that the Colombian labor market values a college graduate at the time of graduation the same as a secondary school graduate with five years of formal labor market experience. I also find high positive correlations between the quality of higher education institutions and students’ skills and earnings, and between on-time graduation and earnings. High-quality tertiary institutions boost the entry-level salary for its graduates, but this boost fades over time as others gain experience and graduates’ skills as workers are revealed. I found evidence of how higher education is slowly reducing the gender gap and improving income distribution. Finally, the sheepskin effect is about 12.6% on average and the returns for bachelors, diplomas, and masters are 15.1%, 33.6%, and 53.2%, respectively.

Seminario de Políticas Públicas - Paulina Restrepo

Seminario de Políticas Públicas - Alvaro Riascos

We propose a novel conditional GANs architecture for crime (robberies) prediction in downtown Bogota, capital city of Colombia. The model uses several layers of ConvLSTM neural nets in both the generative and the discriminatory networks.  We further condition on past crime intensity maps,  weekdays,  and holidays.The trained network is able to capture spatiotemporal patterns and outperforms state-of-the-art predictive models such as spatiotemporal Poisson point process, as well as other models trained with the same dataset.The model accuracy reaches an area under the Hit Rate – Percentage Area Covered by Hotspots curve of 0.86.  However, our model suggests that there are potential biases with heterogeneous effects on vulnerable populations.  We address the fairness consequence of this model in low income vs.  high income residents by estimating a calibration test conditional to protected variables.  Finally, we introduce a fairness – accuracy balancing technique that quantifies the tradeoffs between accuracy and fairness in this type of models.