Data and Program for "Loan Spreads and Credit Cycles: The Role of Lenders’ Personal Economic Experiences"

Published: 8 February 2023| Version 1 | DOI: 10.17632/k8z6xpjmrb.1
Janet Gao,
Pengfei Ma


This folder contains the code and datasets to recreate the tables in "Loan Spreads and Credit Cycles: The Role of Lenders’ Personal Economic Experiences."


Steps to reproduce

Sample and variable construction procedures are described in Section 2.1 and 2.2 of the paper: “Loan Spreads and Credit Cycles: The Role of Lenders’ Personal Economic Experiences.” Our data come from several sources. The sample period is from 2000 to 2018. We start with 24,459 syndicated loan contracts in LPC Dealscan with available information on loan contract terms (e.g., spreads, loan amount, and maturity) that are issued to U.S. public firms outside of financial and utility industries (SIC codes in 6000-6999 or 4900-4999) with available firm characteristics. We then identify loan officers responsible for originating syndicated corporate loans following the procedure outlined in Bushman et al. (2020). For each loan, we search for the borrower’s SEC filings (8-K’s, 10-Q’s, and 10-K’s) around the loan issuance date and identify credit agreements from exhibits attached to these filings. We then scrape the signature panel at the end of the credit agreements to identify the names of bankers underwriting the loan. Bankers’ employment affiliations are mapped to Dealscan data to ensure that those institutions are also reported by Dealscan. We focus on loan officers from lead arranger banks (i.e., “lead bankers”), who are primarily responsible for setting loan terms. Next, we identify the property ownership records for these loan officers, which allow us to pin down the location and ownership dates of their real estate properties. We examine how shocks to individual officers change their lending terms over time, and this requires more than one loan per officer. We thus focus on officers who lead-arranged at least two loans (in separate deals taking place over different years). We search for bankers’ property ownership records in LexisNexis Public Records database and follow closely the procedures described in Cheng, Raina, and Xiong (2014). After finding a loan officer in the LexisNexis database, we gather all the addresses related to the loan officer and then collect all the deed transfer records, mortgage records, and tax assessment records of those addresses. Using this information, we determine the dates when the officer gains and releases control of each property as well as the locations of these properties. With the above-listed data, we link each loan contract to its lead officer(s). The unit of observation is a loan contract-lead officer. This database allows us to contrast decisions by the same officer before and after shocks to these local experiences. Additionally, we construct a loan-level sample, where there is a single observation for each loan. For loans with more than one lead officer in our main sample, we select the officer that issues the largest number of loans.


Indiana University Bloomington, Georgetown University


Banking, Corporate Finance, Financial Institution, Supply of Credit, Behavioral Finance, Credit Market