Humans vs Machines: Soft and Hard Information in Corporate Loan Pricing

50 Pages Posted: 15 May 2020

See all articles by Manuel Adelino

Manuel Adelino

Duke University; Duke Innovation & Entrepreneurship Initiative; National Bureau of Economic Research (NBER); Centre for Economic Policy Research (CEPR)

Ivan Ivanov

Board of Governors of the Federal Reserve System

Michael Smolyansky

Board of Governors of the Federal Reserve System

Date Written: November 18, 2019

Abstract

This paper uses novel regulatory data on internal loan-level risk metrics of US banks to show that corporate loan interest rates line up closely with measures of hard information. We show that the variation in interest rates in excess of what internal models suggest provides limited information for predicting loan default. These results hold similarly for large and small banks, and are stronger for large firms, where hard information is more readily available. These results show that virtually all credit-relevant information contained in corporate loan pricing is hard information.

Keywords: Loan pricing, hard information, soft information, corporate loans

JEL Classification: G21, G17, G32

Suggested Citation

Adelino, Manuel and Ivanov, Ivan and Smolyansky, Michael, Humans vs Machines: Soft and Hard Information in Corporate Loan Pricing (November 18, 2019). Available at SSRN: https://ssrn.com/abstract=3596010

Duke Innovation & Entrepreneurship Initiative ( email )

215 Morris St., Suite 300
Durham, NC 27701
United States

National Bureau of Economic Research (NBER) ( email )

1050 Massachusetts Avenue
Cambridge, MA 02138
United States

Centre for Economic Policy Research (CEPR) ( email )

London
United Kingdom

Ivan Ivanov

Board of Governors of the Federal Reserve System ( email )

20th Street and Constitution Avenue NW
Washington, DC 20551
United States

Michael Smolyansky

Board of Governors of the Federal Reserve System ( email )

20th Street and Constitution Avenue NW
Washington, DC 20551
United States

Here is the Coronavirus
related research on SSRN

Paper statistics

Downloads
85
Abstract Views
201
rank
317,404
PlumX Metrics