ALEJANDRO BELTRAN
BIENVENIDO, WELCOME!

MY BIO
I am a Postdoctoral Research Associate in the Public Policy Programme at the Alan Turing Institute.
I received my PhD from the School of Government and Public Policy at UArizona. My research uses computational social science approaches and causal inference techniques to study corruption and tax avoidance in various contexts.
Currently, I am studying trends in firm ownership across various countries and using machine learning to identify clusters of suspicious companies. In my dissertation I explore the institutional and political constraints of state audit agencies in Mexico. It evaluates which mayors are most likely to be sanctioned by state auditors based on political linkages. In my broader research agenda I use natural language processing and machine learning to extract information from text. I also frequently collaborate with specialists at the Inter-American Development Bank, helping build readily accesible subnational fiscal data across many countries in the region.
EDUCATION

August 2016 - May 2021
PHD IN GOVERNMENT AND PUBLIC POLICY, UNIVERSITY OF ARIZONA
Dissertation: Accounting for Corruption:Evaluating State Audit Agencies in Mexico
Committee: Edella Schlager (Chair), Javier Osorio (Co-Chair), Brint Milward

August 2016 - May 2018
MA IN GOVERNMENT AND PUBLIC POLICY, UNIVERSITY OF ARIZONA
Advisor: Edella Schlager

August 2011 - June 2015
LICENCIADO EN POLÍTICAS PÚBLICAS, UNIVERSIDAD AUTÓNOMA DE SINALOA
Graduated top of class
International Exchange Student: The University of New Mexico, 2013-2014
International Exchange Student: The Washington Center, 2014
PUBLICATIONS
FISCAL DATA IN TEXT: INFORMATION EXTRACTION FROM AUDIT REPORTS USING NATURAL LANGUAGE PROCESSING
February 2023
This paper details how I use audits from the Auditoria Superior del Estado de Sinaloa, to build a pipeline for extracting the monetary value of discrepancies detected in municipal budgets.
ACCOUNTING FOR CORRUPTION: EVALUATING STATE AUDIT AGENCIES IN MEXICO
May 2021
The dissertation tests these theoretical expectations by focusing on subnational SAIs in Mexico. The Auditoria Superior del Estado exists in all 32 states and these vary along the dimensions of autonomy and politicization. It compiles data on municipal audits for every municipality in Mexico. It then identifies the partisan affiliations of each auditor and mayor during this period as well as creating a composite score of organizational autonomy.
ENHANCING THE DETECTION OF CRIMINAL ORGANIZATIONS IN MEXICO USING ML AND NLP
July 2020
With Javier Osorio. In this paper we generate a geo-referenced database on the violent presence of Mexican Criminal Organizations between 2000-2018. We use NLP and Machine learning to classify over 150,000 Spanish articles. Published in the proceedings of IJCNN at the World Conference on Computational Intelligence 2020.
SUPERVISED EVENT CODING FROM TEXT WRITTEN IN ARABIC: INTRODUCING HADATH
May 2020
With Javier Osorio, Alejandro Reyes, & Atal Ahmadzai. In this article we introduces Hadath, a supervised protocol for coding event data from text written in Arabic. Using Machine Learning and NLP we classify over 100,000 articles in Arabic and extract information on conflict in Afghanistan from 2008 through 2018.
CIUDADES, DESARROLLO URBANO, Y AUTONOMÍA FINANCIERA: DILEMAS PARA LA GOBERNANZA LOCAL EN MEXICO
December 2017
With Mercedes Verdugo López. In this book we analyze the relationship between fiscal decentralization and good governance for municipalities in Mexico.