publications

2023

1. 

   Using artificial intelligence in the request of pretrial measures: approaching the uncertainty in recidivism risk

   Andrés Camacho Baquero, Daniel Mejia, and Lina Navas

   2023

   Working Paper

   Abs PDF

   This paper presents an innovative approach to enhance decision-making processes related to pretrial detentions in the Colombian judicial system. By incorporating a machine learning model for predicting recidivism risk, the study aims to provide an objective, impartial, and uniform methodology to assess the benefits of using decision-aid tools. The prosecutors in Colombia face significant challenges in deciding whether a defendant should go to prison based on their recidivism risk. Current practices are scrutinized for their reliance on incomplete and non-uniform information, leading to two critical types of decision-making errors: Type I, involving the unwarranted detention of low-risk individuals, and Type II, the failure to detain high-risk individuals with potentially detrimental effects on public safety. The proposed analytical tool is designed to minimize these errors and biases, ensuring judicial decisions align with principles of equality and impartiality. The anticipated outcome is a more cautious use of detention measures, effectively balancing the rights of the accused with community safety and reducing the number of crimes by recidivists without proportionally increasing detention rates. There is evidence of welfare gains by using this approach when comparing with actual decisions through different levels of leniency in prosecutors

2. Educación Superior en Colombia: Logros en Calidad, Eficiencia y Equidad

   H. Bayona-Rodriguez, J. Bedoya, A. Camacho, and 5 more authors

   2023

   Forthcoming
3. Is Being a Woman a Hurdle for Having a Housing Credit?

   Andrés Camacho, and Nora De Libertun

   2023

   Sunmitted to Housing Studies

2022

1. 

   Reducing Alcohol-Related Violence with Bartenders: A Behavioral Field Experiment

   Andrés Ham, Darío Maldonado, Michael Weintraub, and 2 more authors

   Journal of Policy Analysis and Management, 2022

   Abs Bib HTML

   Abstract This paper evaluates the randomized Good Drinks program in four localities of Bogotá, Colombia. The intervention encourages bartenders to adopt standardized practices that promote responsible behavior in terms of alcohol consumption with the goal of reducing alcohol-related violence and was implemented in cooperation with Colombia’s largest brewery and the city’s Secretariat of Security, Coexistence, and Justice. Tracing out the relationship between alcohol consumption and violence is useful because alcohol-related incidents often lead to more serious crimes. Our experimental design allows estimating direct and spillover effects on reported incidents within and around bars. Results show that bartenders in treatment locations sell more water and food, thus contributing to more responsible behavior by patrons. However, we find no direct or spillover effects of these changes in consumption on brawls five months after the program, but some improvement on other alcohol-related incidents. The experience of the Good Drinks program provides a better understanding of three aspects related to alcohol regulation and policy: (i) the role bartenders can play to curb excessive alcohol consumption and promote good behavior among customers, (ii) a practical experience of using less restrictive interventions for alcohol regulation, and (iii) the value of public-private partnerships.

   @article{https://doi.org/10.1002/pam.22365,
     author = {Ham, Andrés and Maldonado, Darío and Weintraub, Michael and Camacho, Andrés Felipe and Gualtero, Daniela},
     title = {Reducing Alcohol-Related Violence with Bartenders: A Behavioral Field Experiment},
     journal = {Journal of Policy Analysis and Management},
     volume = {41},
     number = {3},
     pages = {731-761},
     keywords = {alcohol, bartenders, brawls, alcohol-related violence, regulation},
     doi = {https://doi.org/10.1002/pam.22365},
     url = {https://onlinelibrary.wiley.com/doi/abs/10.1002/pam.22365},
     eprint = {https://onlinelibrary.wiley.com/doi/pdf/10.1002/pam.22365},
     year = {2022},
     dimensions = {true},
   }

2018

1. 

   Using Machine Learning to Predict Recidivism Risk in the Colombian Prison System

   Andrés Camacho

   2018

   Abs Bib PDF

   Recidivism represents an expensive risk for society, not only because of crimes committed but also due to the costs of incarceration in a country with high levels of overcrowding. Every day, judges face a decision of whether the defendants must be deprived of freedom based on a prediction of what a defendant would do if released. The importance of that decision, combined with recent and rich data on Colombian inmates in the Prison System, makes this problem an ideal application for machine learning. In this paper, I expand the efforts of recent literature on how computer algorithms can be used to improve judges decisions about the incarceration of defendants by using predictions about the risk of recidivism amongst released ex-convicted people. The main contribution of this paper suggests that there is a potential mis-prediction of the recidivism risk when deciding who to incarcerate and who not to. Potential welfare gains could be achieved, measured as reductions in crime and jailing rates. Moreover, these predictions can be used to develop public policies that address the growing prison overcrowding problem by suggesting an early release of less risky inmates based on the algorithm prediction.

   @unpublished{TesisIEng,
     author = {Camacho, Andrés},
     title = {Using Machine Learning to Predict Recidivism Risk in the Colombian Prison System},
     pages = {1-24},
     keywords = {Recidivism, machine learning, prison system, overcrowding, Colombia},
     year = {2018},
   }

2017

1. 

   Forced Migration Effects on Trust and Attitudes Toward Immigrants in Destination Countries: Evidence from the Refugee Crisis in Europe

   Andrés Felipe Camacho Baquero

   Aug 2017

   Abs HTML PDF

   This article estimates the effect of refugee influx on widespread trust and attitudes towards immigrants in European destination countries between 2002 and 2015. Based on recently published data from the European Social Survey (ESS) and following an instrumental variable approach, I find that a greater influx of refugees negatively affects interpersonal trust, and that people’s attitudes toward immigrants improve between 2014 and 2015. I find evidence of two mechanisms that may be shaping the results: labor uncertainty and educational level of respondents, which reflects the perception of the residents in the destination countries about the economic impact resulting from the arrival of refugees. The results found in this paper are robust to different classifications of interest variable, to the level of economic activity and to the inclusion of terrorist events that could undermine trust.