Paul E. Soto

Stressed Banks? Evidence from the Largest-Ever Supervisory Review

with Puriya Abbassi, Rajkamal Iyer, and José-Luis Peydró

Management Science, 2025

We use the ECB's Asset Quality Review as a natural experiment, showing that banks temporarily reduce risk by shifting portfolios toward safer assets for supervisory compliance, but revert afterward, causing persistent negative real effects, particularly harming high-risk firms.

View Paper

Capital Controls, Domestic Macroprudential Policy and the Bank Lending Channel of Monetary Policy

with Andrea Fabiani, Martha López Piñeros, and José-Luis Peydró

Journal of International Economics, 2022

We examine how the simultaneous implementation of capital controls and domestic reserve requirements in Colombia during a credit boom effectively curtailed banks' risky lending practices, and show that these policies complementarily strengthened the transmission of monetary policy to credit supply by limiting foreign currency arbitrage and domestic funding, ultimately reducing financial vulnerability.

View Paper

Capital Controls, Corporate Debt and Real Effects

with Andrea Fabiani, Martha López Piñeros, and José-Luis Peydró

R&R at the Review of Financial Studies

We examine Colombia's foreign-debt capital controls using firm-level data, finding that these controls limit risky foreign borrowing during booms, thus safeguarding exports and employment during crises, especially benefiting financially constrained firms.

View Paper

Trademarks in Banking

with Ryuichiro Izumi and Antonis Kotidis

Finance and Economics Discussion Series, 2024

We study how trademark similarity among banks significantly increases deposit withdrawals following bank failures, particularly when media coverage is extensive, highlighting the underestimated role of trademarks in financial contagion and banking stability.

View Paper

Breaking the Word Bank: Measurement and Effects of Bank Level Uncertainty

Journal of Financial Services Research, 2021

I use NLP methods on bank earnings calls to measure idiosyncratic uncertainty to show that higher uncertainty significantly reduces lending and increases liquidity, particularly during volatile economic conditions, directly impacting broader credit availability.

View Paper

Manufacturing Sentiment: Forecasting Industrial Production with Text Analysis

with Tomaz Cajner, Leland D. Crane, Christopher Kurz, Norman Morin, and Betsy Vrankovich

R&R at Journal of Applied Econometrics

We show that NLP techniques, including transformer-based deep learning models, applied to purchasing managers' textual survey responses significantly enhance forecasting accuracy of U.S. industrial production.

View Paper

Research in Commotion: Measuring AI Research and Development through Conference Call Transcripts

Finance and Economics Discussion Series, 2025

I develop the AIR Index—a novel NLP-based measure from earnings calls—to quantify AI R&D at the firm level, and show that AI R&D leads to immediate market valuation gains and accelerated capital expenditures without immediate productivity increases.

View Paper

Inside the Boardroom: Evidence from the Board Structure and Meeting Minutes of Community Banks

with Rosalind L. Bennett and Manju Puri

Finance and Economics Discussion Series, 2024

We analyze unique board-meeting records from community banks, showing that during financial distress boards shift discussions from routine activities toward capital management and regulatory oversight, and experience increased turnover and sharply negative sentiment.

View Paper

Tracking Real Time Layoffs with SEC Filings: A Preliminary Investigation

with Leland D. Crane, Emily Green, Molly Harnish, Will McClennan, Betsy Vrankovich, and Jacob Williams

Finance and Economics Discussion Series, 2024

We develop a real-time indicator of layoffs using SEC 8-K filings analyzed through NLP and LLMs, providing timely signals that correlate strongly with labor market conditions.

View Paper

Measuring AI Uptake in the Workplace

with Leland D. Crane and Michael Green

FEDS Notes, 2024

We review multiple firm and employee surveys, revealing that AI adoption varies widely but consistently falls within a 20-40% range.

Featured in speech by Vice Chair for Supervision Michael S. Barr, Feb 18, 2025

Featured in Speech by Governor Adriana D. Kugler, Feb 7, 2025

View Paper

Measurement and Effects of Supply Chain Bottlenecks Using Natural Language Processing

FEDS Notes, 2023

I develop the Supply Chain Bottleneck Sentiment (SCB Sentiment) index, a novel NLP-based measure derived from Federal Reserve Beige Books, capturing both frequency and sentiment of supply chain disruptions.

Featured in Financial Times, March 11, 2025

Featured in Remarks from Vice Chair Philip N. Jefferson, Feb 21, 2025

View Paper