Research

Working papers

The effects of two-way lending between financial conglomerates in bilateral repo markets (with Jorge Flórez, Banco de la República y Carlos Cañon, Bank of England)
Journal of Empirical Finance (2025): R&R
We examine how market structure, market power, and systemic risk respond to close and intense lending relationships between financial conglomerates (FCs) in non-centrally cleared bilateral repo markets. Using transaction-level data from Mexico, we document persistent and stable funding relationships between FC-affiliated banks and funds, characterized by two distinctive features: First, the funding transactions are two-way, meaning that a pair of rival FCs provide lending to each other on the same day. Second, two-way transactions are executed at lower average rates than one-way transactions. We show that two-way lending (TWL) between FCs contributes to increased market concentration and market power for FC-affiliated funds, while worsening the lending terms for independent banks and funds. Furthermore, we find that the bank-level contribution to systemic risk rises with increased TWL.

Attention-Driven and Fundamental Information Acquisition: Asymmetric Effects on Consensus Analyst Forecast Bias (with José Gabriel Astaíza, EAFIT)

Journal of Economics and Business (Under review)

This study examines how investor information acquisition affects bias in consensus analysts’ target price forecasts. We distinguish between two channels: fundamental information acquisition, proxied by EDGAR file downloads, and attention-driven information acquisition, captured by Bloomberg news search and readership activity. Using quarterly data on 2,449 U.S. firms, we show that these channels have asymmetric and context-dependent effects. Greater acquisition of fundamental information is associated with significantly lower one-year-ahead consensus forecast bias, consistent with a monitoring mechanism through which investors detect and respond to overly optimistic forecasts. In contrast, attention-driven information acquisition does not systematically improve forecast accuracy. We further find that media attention is positively associated with forecast bias and attenuates the bias-reducing effect of fundamental information acquisition. This effect is weaker for hard-to-value firms and during periods of elevated economic uncertainty. Overall, our results highlight that the impact of investor information acquisition depends critically on both its underlying source and the information environment in which it operates.

FinTech and Digital Fraud: Evidence from Self-Reported Complaints on Social Media

(with Hernando Hernández)

Journal of Digital Economics (Under review)

Digital Financial Inclusion has revolutionized financial access; however, its rapid expansion has triggered a surge in financial fraud that official statistics fail to capture in real time. This research proposes a novel framework to quantify this phenomenon using Big Data from social media. We developed the Fraud Risk of Social Knowledge index, a high-frequency indicator derived from more than 300,000 user complaints on X (formerly Twitter) using transformer-based language models and clustering techniques. We validate the index against traditional metrics and estimate its impact on DFI in Mexico, Colombia, and Chile through panel regressions. Our findings demonstrate that a 1% increase in social media-reported fraud leads to a 1% reduction in DFI penetration, confirming that distrust is a critical barrier to digital adoption. These results underscore the vital role of AI-driven monitoring in digital finance and the need to understand fraud mechanisms to protect consumers and ensure the long-term sustainability of the digital financial ecosystem.

Work in progress

Digital Wallet Adoption and Financial Fraud: A Staggered Difference-in-Difference analysis (with Hernando Hernández, Roy Mersland and Kjetil Andersson)

This paper investigates the causal impact of digital wallet adoption on financial fraud. Using a unique monthly panel dataset from Colombia covering all regulated financial institutions between 2009 and 2022, we exploit the staggered introduction of digital wallets across banks within a difference-in-differences framework. Results show that digital wallet adoption significantly increases the number of reported fraud cases. However, the surge in digital transactions is substantially larger, indicating that the expansion of transactions far outpaces the rise in fraud incidents. As a result, the relative frequency of fraud per 100 digital transactions remains statistically unchanged. Heterogeneity analysis reveals asymmetric effects across fraud typologies: social engineering fraud rises persistently after adoption, while authorized push payment fraud declines. Identity theft and account takeover display moderate, platform-specific vulnerabilities. These results suggest that digital financial innovation reshapes the opportunity structure for crime, highlighting the need for regulatory frameworks that balance innovation and security through typology-sensitive approaches.

Higher Rates, More Credit? A Difference-in-Differences Analysis of an Increase in the Lending Rate Cap (con Rubén Salas, Felipe Ramos y Yudi Pereira)

This paper analyzes how an increase in the legal interest rate ceiling affects the supply of productive popular credit in Colombia. Exploiting a regulatory change in the methodology for calculating the Current Banking Interest Rate, we estimate a difference-in-differences model using weekly data at the credit establishment level. The results show that the increase in the regulatory ceiling did not translate into a credit expansion. On the contrary, the adjustment manifested primarily as an increase in the spread between the lending rate and the regulatory ceiling, with no statistically significant evidence of changes in lending rates following the reform. These results suggest that the regulatory ceiling did not constitute the main constraint determining lending rates. Rather, the evidence indicates that the market response is conditioned by the competitive structure and market power of financial institutions. In particular, we find that the reaction of institutions depends on their prior competitive position, suggesting that market power plays a central role in the transmission of regulatory changes in this credit market.

A Continuous Time AR(2) with Random Coefficients: Exact Discretization, Estimation, and Forecasting (with Milena Hoyos and Daniela Cardenas)

This paper studies a second-order continuous-time autoregressive model with random coefficients driven by Brownian increments. We derive the exact discrete-time analogue and show that it follows an ARMA(2,1) process. The moving average component arises endogenously from the second-order structure. It induces serial correlation in the discrete time errors even when the underlying continuous time disturbances are serially uncorrelated — a feature absent in existing first-order continuous time random coefficient models. Estimation is based on a first-order approximation that yields a tractable representation with explicitly characterized conditional heteroscedasticity, and parameters are recovered by maximum likelihood. A Monte Carlo experiment shows that correctly modelling the MA(1) component reduces finite sample bias and delivers substantially better forecasting performance relative to a misspecified AR(2) model. An empirical application to the OECD Composite Leading Indicator for the United States illustrates the practical relevance of the framework.

Financial capability and Financial well-being: A study among rural population in Colombia? (with Sebastian Cardenas)

This paper explores the link between financial capability and the financial well-being of rural populations with limited credit and insurance services are limited and weak state support. We use data collected from 404 individuals living in rural areas of Tolima-Colombia to compute an indicator of financial well-being. Results from a regression analysis showed significant differences in the financial well-being level by various factors, including financial behaviors and attitudes, when controlling for demographic and socioeconomic information and financial characteristics.

How optimal are fines in cartel cases? An evaluation of the case of Colombian fining policy.

In Colombia, the available sanctions for hard-core cartels include monetary fines against both business entities and individuals. For business entities, the maximum fine per offense charged is the greater of (i) 100,000 current legal minimum monthly wages, presently equal to 98,065 billion pesos or about USD 28.8 million or (ii) 150% of the profits derived from the anticompetitive conduct (Article 25 of Law 1340). The SIC sanctioned between 2009 and 2019 around eight hardcore cases imposing fines using the first criteria exclusively, arguing that the option permitting fines of up to 150% of the illicit profits is unavailable as a practical matter because the profits cannot effectively be calculated. In such circumstances, is the maximum sanction of 28.8 million sufficient to produce a deterrent effect? What happens when very large firms are involved? The purpose of this study is to analyze the deterrence property of this peculiar rule. In doing so, we expect to provide evidence in favor of applying the second criteria available in the law and propose a practical formula to compute it.

Liquidity premium and return predictability in U.S. Inflation-linked Bonds Market

This paper discusses the predictive role of alternative measures of the liquidity premium of TIPS relative to Treasury bonds for government excess bond returns. The results show that the liquidity premium predicts positive (negative) TIPS (nominal Treasury) excess returns. The explanatory power of the TIPS liquidity premium is statistically significant and economically meaningful for short-term excess TIPS maturities and for long-term nominal Treasury bonds. I also find that the out-of-sample forecasting power of liquidity for nominal Treasury excess returns appears to have been addressed by the events during the recent financial crisis. By contrast, I have evidence of out-of-sample forecasting ability during both normal and bad times for TIPS’ excess returns. Available here.

A triple difference estimator with placebos to deal with systematic unobserved factors (with Milena Hoyos and Jorge Muñoz)

Gobernanza institucional y de datos en finanzas abiertas. En: Inclusión financiera y Open Data: El caso colombiano. Editorial FCE-UNAL. (Capítulo de libro. Con Diana Forero)

Oportunidades y retos para la creación de valor a través de una política de datos abiertos. En: Inclusión financiera y Open Data: El caso colombiano. Editorial FCE-UNAL. (Capítulo de libro)

Page updated

Report abuse