Personal Lending Relationships

• DOI: http://doi.org/10.1111/jofi.12589
• Author: STEPHEN ADAM KAROLYI
• Published date: 01 February 2018
• Date: 01 February 2018

THE JOURNAL OF FINANCE •VOL. LXXIII, NO. 1 •FEBRUARY 2018

Personal Lending Relationships

STEPHEN ADAM KAROLYI∗

ABSTRACT

I identify the effects of personal relationships on loan contracting using executive

deaths and retirements at other firms as a source of exogenous variation in executive

turnover. After plausibly exogenous turnover, borrowers choose lenders with which

their new executives have personal relationships 4.1 times as frequently, and loans

from these lenders have 20 basis points lower spreads and 12.5% larger amounts. Per-

sonal relationships benefit firms across loan terms, especially during macroeconomic

downturns. Increased financial flexibility from personal relationships insulated firms

from financial shocks during the recent financial crisis: they exhibited less constrained

investment and were less likely to layoff employees.

COASE (1937,1988)PROPOSES THAT VALUABLE relationships form in long-term

contracting environments subject to uncertainty and information asymmetry.

Because the private loan market is subject to these conditions and long-term

relationships between firms and banks are an empirical regularity, several

theories and empirical tests have been developed to understand the nature of

relationships between firms and banks. Theoretically, borrowers and lenders

stand to gain from relational lending (Berger and Udell (1995)) if they can

split the gains from monitoring cost savings, but either party might also suffer

from moral hazard problems that can prevent relationships from forming (e.g.,

Sharpe (1990), Rajan (1992), von Thadden (2004)).

∗Stephen Adam Karolyi is at Tepper School of Business, Carnegie Mellon University. I thank

my dissertation committee members Gary Gorton (chair), Andrew Metrick, Justin Murfin, and

Matt Spiegel for helpful advice and comments during the course of this project. This paper has

also benefitted from my discussions with Viral Acharya, Turan Bali, Nicholas Barberis, Allen

Berger, Andrew Bird, Shane Corwin, David Denis, Diane Denis, Jean Helwege, Ryan Israelsen,

Andrew Karolyi, Camelia Kuhnen, Ken Lehn, Stefan Lewellen, Tyler Muir,Marina Neissner, Alan

Moreira, Manju Puri, Adriano Rampini, David Robinson, Tom Ruchti, Ren´

e Stulz, Shyam Sunder,

Geoff Tate (Discussant, Society for Financial Studies 2014 Cavalcade), Greg Udell, Mike Weisbach,

Scott Yonker, and my fellow students at Yale University. I would also like to thank seminar

participants at Boston College, Carnegie Mellon University, Dartmouth College, Duke University,

the Federal Reserve Board, Georgetown University, Imperial College London, Indiana University,

London Business School, The Ohio State University, Pennsylvania State University, University

of Pittsburgh, Southern Methodist University, University of North Carolina, and University of

Notre Dame, as well as participants at the 10th Annual Washington University Corporate Finance

Conference, 2013 Financial Management Association Meetings, and Society for Financial Studies

Cavalcade 2014 for questions and comments. All remaining errors are my own. I have read the

Journal of Finance’s disclosure policy and have no conflicts of interest to disclose.

DOI: 10.1111/jofi.12589

5

6The Journal of Finance R

Empirical evidence on these theories about institutional lending relation-

ships is mixed (e.g., Petersen and Rajan (1994), Berger and Udell (1995),

Schenone (2010), Bharath et al. (2011)). One possible explanation for the mixed

evidence is that these relationships are not between firms and banks at all,

but rather between firms’ executives and banks. Importantly, institutions are

subject to several layers of organizational constraints that may prevent them

from committing to an action despite the potential consequences. Individuals,

however, are subject to simpler constraints, which, under some conditions,

could allow them to credibly commit to avoid moral hazard actions (Tirole

(2006)). This insight suggests that, while firms may not be able to overcome

moral hazard problems in lending markets, executives might.

In this paper, I analyze the extent to which endogenous personal business

connections between executives and lenders—a previously unobserved factor—

can explain the formation and dissolution of institutional lending relationships

and determine lending relationships’ theoretical sources of value. In theory,

relationships between corporate executives and loan officers may differ from

firm-bank relationships if they subject corporate executives to behavioral con-

straints, which would arise if these relationships promote trust, mutual ca-

reer dependence, or any personally borne breakup cost, such as reputation

loss (Duarte, Siegel, and Young (2012)). Empirically, firm-bank and executive-

loan officer relationships are correlated, which suggests that the presence of

personal lending relationships may confound previous studies of institutional

lending relationships.

These results contribute three new findings about lending relationships.

First, I find causal evidence that endogenous personal relationships between

corporate executives and lenders are a key determinant of lender choice and

loan contracting outcomes. My identification strategy allows me to study the

dynamics of personal relationship effects over time rather than across firms,

which mitigates the effects of firm-bank and executive-firm matching. Second,

I find larger economic effects of personal lending relationships during NBER

recessions and the 2008 to 2009 financial crisis. Third, I find that firms with

intense personal lending relationships exhibit less constrained investment and

a lower probability of employee layoffs when the effect of personal lending re-

lationships on loan contracting outcomes is largest. My estimates suggest that,

among borrowers in my sample, those with intense personal lending relation-

ships would have invested $30.8 billion less during the recent financial crisis

if they instead had weak personal lending relationships. Together, the paper’s

results suggest that personal lending relationships are an important source of

financial flexibility during macroeconomic downturns.

Personal lending relationships pose an identification challenge because they

arise endogenously. An important contribution of this paper is to identify the

value of endogenous personal lending relationships, rather than focus on a

subset that form exogenously.I use two new empirical strategies to identify the

effect of personal lending relationships on loan contracting outcomes. Under

the first strategy, I exploit the fact that executive turnover provides a setting

in which a firm’s bank relationships remain constant, while the firm’s personal

Personal Lending Relationships 7

lending relationships change. This setting reduces the endogeneity problem

to finding a source of variation in executive turnover that is exogenous with

respect to lender choice and loan contracting outcomes and implicitly controls

for potential biases from firm-executive matching.

Specifically, I use executive deaths and quasi-exogenous retirements at other

firms to analyze sample firms’ relationships. These deaths and retirements

serve as a demand shock for the executives of interest or as a supply shock

affecting their willingness to depart. This variation is relevant to executive

turnover because the market for executives is geographically constrained (Bro-

man, Nandy, and Tian (2014), Yonker (2017)) and executives have industry-

specific expertise (Parrino (1997)). Deaths and quasi-exogenous retirements

should therefore increase the likelihood of executive turnover at other firms

with strong candidates or with executives that have lower switching costs.

In contrast with own-firm executive deaths and quasi-exogenous retirements,

which could impact firm value and the cost of loan financing (Bennedsen, Perez-

Gonzalez, and Wolfenzon (2006), Salas (2010), Israelsen and Yonker (2012)), ex-

ecutive deaths at other firms are also more likely to satisfy the exclusion restric-

tion because they do not directly affect differences in loan contract terms that

firms receive across lenders with which they have relationships and lenders

with which they do not.1

I use this plausibly exogenous variation in executive turnover to explore

the effects of personal and institutional lending relationships on lender choice

and loan terms. To analyze personal lending relationships in this executive

turnover context, I estimate the effect of personal relationships between re-

placement executives and potential lenders on lender choice and loan terms.

Linear probability model (LPM) estimates with this endogeneity correction in-

dicate that firms are not only more likely to initiate loans with new lenders after

executive turnover, but also 16.3% more likely to choose a lender with whom

their new executive has an intense personal relationship rather than a weak

relationship. These effects are not limited to the extensive margin of lender

choice. After a turnover event, loans from lenders with whom the replacement

executive has an intense personal relationship have 22 bps lower spreads and

are 12.5% larger, on average. Importantly, these effects on lender choice and

loan terms do not exist until after turnover occurs, which excludes executive-

firm matching as an alternative explanation. All else equal, the causal effect

of personal lending relationships on lower spreads is worth $1.9 mm per out-

standing loan annually, which is approximately the same magnitude as the

total compensation of the average CEO in my sample. This comparison sug-

gests that the personal relationship savings, while measured in basis points,

are economically large and potentially contribute to corporate executives’ hu-

man capital.

1Note that this is a weaker assumption than would be required for an analysis of loan terms

across firms. In such a study,the exclusion restriction would require that executive deaths at other

firms do not directly affect the level of loan contract terms.