Is Fraud Contagious? Coworker Influence on Misconduct by Financial Advisors

• Author: NATHANIEL P. GRAHAM,WILLIAM C. GERKEN,STEPHEN G. DIMMOCK
• Published date: 01 June 2018
• DOI: http://doi.org/10.1111/jofi.12613
• Date: 01 June 2018

THE JOURNAL OF FINANCE •VOL. LXXIII, NO. 3 •JUNE 2018

Is Fraud Contagious? Coworker Influence

on Misconduct by Financial Advisors

STEPHEN G. DIMMOCK, WILLIAM C. GERKEN, and NATHANIEL P. GRAHAM∗

ABSTRACT

Using a novel data set of U.S. financial advisors that includes individuals’ employment

histories and misconduct records, we show that coworkers influence an individual’s

propensity to commit financial misconduct. We identify coworkers’ effect on miscon-

duct using changes in coworkers caused by mergers of financial advisory firms. The

tests include merger-firm fixed effects to exploit the variation in changes to coworkers

across branches of the same firm. The probability of an advisor committing miscon-

duct increases if his new coworkers, encountered in the merger, have a history of

misconduct. This effect is stronger between demographically similar coworkers.

IS FINANCIAL MISCONDUCT CONTAGIOUS? The media and a large academic liter-

ature document widespread financial misconduct in many different contexts,1

and both academics and policy makers have stressed the importance of under-

standing why financial misconduct occurs and how it spreads. In this paper, we

show that financial misconduct is contagious, that is, that the ethical conduct

∗Stephen G. Dimmock is in the Division of Banking & Finance, Nanyang Technological Univer-

sity. William C. Gerken is in the Department of Finance & Quantitative Methods, University of

Kentucky. Nathaniel P. Graham is in the Division of International Banking and Finance Studies,

Texas A&M International University. We are grateful to Leonce Bergeron, John Chalmers, Gjergji

Cici, Thomas Dudley,Joe Farizo, Allaudeen Hameed, Kristine Hankins, Michael Hertzel, Pat Hud-

dleston, Zsuzsa Husz´

ar, Russell Jame, Bill Johnson, Simi Kedia, Jussi Keppo, Ross Levine, Adair

Morse, Andy Puckett, Wenlan Qian, David Reeb, Tyler Shumway, Kenneth Singleton (the Editor),

Anand Srinivasan, Johan Sulaeman, Tracy Wang,Chishen Wei, Scott Weisbenner,Bernard Yeung;

an Associate Editor; two anonymous referees; seminar participants at Hong Kong Polytechnic Uni-

versity, National University of Singapore, Texas Christian University, University of New South

Wales, and WestVirginia University; and participants at the 2016 American Finance Association,

2014 American Law and Economics, 2015 European Finance Association, Federal Reserve Bank of

New YorkEconomics of Culture, Federal Reserve Bank of New York Conference on Culture and Fi-

nancial Stability,2014 Financial Management Association, Jim and Jack, NTU Finance, 2015 SFS

Cavalcade, Singapore Scholars Symposium, and 2015 Western Finance Association conferences.

We also thank the Institute for Fraud Prevention for financial support and the Arkansas Securities

Department (and Ann McDougal, in particular), the Florida Office of Financial Regulation, and

the New YorkState Office of the Attorney General for providing assistance with the data. We have

no conflicts of interest with interested parties (see the detailed disclosure statement in the online

version of this article).

1Dyck, Morse, and Zingales (2014) summarize the extent and cost of corporate fraud, Griffin

and Maturana (2016) and Piskorski, Seru, and Witkin (2015) document misconduct for residential

mortgages, Duffie and Stein (2015) review LIBOR manipulation, and Dimmock and Gerken (2012)

document fraud by asset managers.

DOI: 10.1111/jofi.12613

1417

1418 The Journal of Finance R

of many individuals is malleable, with their misconduct behavior influenced

by the misconduct behavior of their coworkers. Specifically, we show that fol-

lowing a shock to their coworker group, the misconduct behavior of financial

advisors2becomes more like that of their new coworkers.

Coworkers—advisors who work in the same branch of a firm—can influence

each other’s misconduct behavior in several ways. First, coworker influence

may occur through social learning (e.g., Banerjee (1992), Ellison and Fuden-

berg (1995)). By interacting with their coworkers, advisors can learn about the

profitability of misconduct as well as techniques for committing misconduct

or reducing the risk of detection (e.g., Sah (1991)). Second, coworker influence

may occur through social utility, where an advisor’s utility from an action de-

pends on the actions of his peers. By interacting with their coworkers, advisors

can absorb the branch’s culture and social norms, and conform their behavior

accordingly (e.g., Bernheim (1994)). Such social norms may be directly related

to misconduct, but may also include indirectly related social norms (e.g., a

branch culture of intense competition or a lack of respect for clients). Advisors

may also mimic their coworkers’ behavior due to concerns about their relative

position (e.g., DeMarzo, Kaniel, and Kremer (2004), Bursztyn et al. (2014)).

Empirically identifying peer effects such as coworker influence on misconduct

is challenging. First, coworker groups are endogenously formed, as individuals

select to work with others who share similar characteristics. Egan, Matvos,

and Seru (2018), for example, show that dishonest financial advisors choose

to work together. Second, within a firm, coworkers face similar economic in-

centives and monitoring procedures. Distinguishing coworker influence from

selection, common incentives, shared supervision, or the external environment

is challenging and requires very specific data features.

We test whether coworkers influence the propensity to commit misconduct

using data on financial advisors for several reasons. First, data on the financial

advisory industry have unique features that allow us to overcome the identifi-

cation challenges above. Second, the financial advisory industry is economically

important. In 2013 for instance, there were more than 635,837 financial ad-

visors in the United States,3and a sizable majority of U.S. investors used an

advisor for investment decisions (see Hung et al. (2008)). Third, the compen-

sation structures used in this industry create incentives for misconduct, as the

models of Inderst and Ottaviani (2009) and Stoughton, Wu, and Zechner (2011)

show. Consistent with the implications of these models, misconduct occurs fre-

quently in this industry: during the 1999 to 2011 period, our sample includes

more than 26,000 substantiated cases of misconduct.4

We define an advisor’s coworkers as the other advisors employed at the

same branch of the firm at the same time. To avoid the problem of endogenous

2Throughout the paper, “financial advisor” and “advisor” refer to individuals who register with

the Financial Industry Regulatory Authority (FINRA) as “registered representatives.” This in-

cludes individuals commonly referred to as brokers or financial planners.

3See http://www.finra.org/Newsroom/Statistics/.

4The cases of misconduct in our study are substantiated customer complaints related to financial

misconduct by advisors, most of which fall under the concept of civil fraud.

Is Fraud Contagious? 1419

coworker selection, we use changes to coworker groups caused by mergers

of financial advisory firms. Of course, while merger decisions are themselves

endogenously determined at the firm level, our data allow us to overcome

this issue by observing misconduct at the individual advisor rather than firm

level. In particular, because we observe information about advisory firms, the

geographically distinct branches of each firm, and the individual advisors of

each branch, our empirical tests focus on the within-firm variation by including

merger-firm fixed effects. The key to our identification is the fact that mergers

occur at the firm level, but changes to coworker groups occur within a firm

at the branch level. Our empirical tests thus exploit across-branch variation

and the impact of combining branches during a merger, while removing

all effects at the firm level, which addresses the most obvious endogeneity

concerns.

To illustrate, consider a merger between Target Firm, with branches in At-

lanta and Boston, and Acquirer Firm, with branches in Atlanta, Boston, and

Chicago. Suppose that the advisors at the Atlanta branch of Target Firm have

a history of misconduct, while the advisors at all other branches of both firms

have clean histories. When the firms merge, the branches in Atlanta and Boston

are combined but the branch in Chicago remains unchanged. Thus, following

the merger, there are changes to the coworker groups at the Atlanta and Boston

branches (of both Target Firm and Acquirer Firm), but only the coworker group

of advisors from the Atlanta branch of Acquirer Firm has changed to now in-

clude individuals with a history of misconduct (the advisors from the Atlanta

branch of Target Firm). The empirical question, then, is whether the advi-

sors from the Atlanta branch of Acquirer Firm are now more likely to commit

misconduct relative to advisors from the Boston branch of Acquirer Firm. The

comparison between the two branches removes any variation common to all

branches of the premerger firm and any common time series changes in the

propensity to commit misconduct.

Our measure of coworker behavior avoids the reflection problem described

in Manski (1993) by observing coworker and individual behavior in separate

time periods. We define Introduced Branch as the individuals that an advisor

encounters due to the merger (i.e., individuals who worked for the other firm

before the merger but work in the combined branch after). Our key measure of

coworker behavior, Introduced Branch High Misconduct, captures misconduct

committed by Introduced Branch coworkers before the merger, and thus

before the advisor meets the coworker group. We use this measure to explain

individual misconduct committed after the advisor meets these new coworkers

in the merger.

We find evidence of coworker influence on misconduct committed by financial

advisors. Controlling for merger-firm fixed effects and using changes to an

advisor’s coworkers due to a merger, we show that an advisor is 37% more

likely to commit misconduct if his Introduced Branch coworkers have a history

of misconduct. We further show that the results are robust to using observations

at the branch rather than individual level, and that the magnitude of the effect

is similar for advisors from the target firm and from the acquirer firm.