Heterogeneity in the Effect of Managerial Equity Incentives on Firm Value

• DOI: http://doi.org/10.1111/fire.12185
• Published date: 01 August 2019
• Author: Jung Chul Park,Hui L. James,Bradley W. Benson
• Date: 01 August 2019

The Financial Review 54 (2019) 583–638

Heterogeneity in the Effect of Managerial

Equity Incentives on Firm Value

Bradley W. Benson

Ball State University

Hui L. James∗

University of Texasat Tyler

Jung Chul Park

University of South Florida

Abstract

We document significant heterogeneity in the relation between chief executive officer

(CEO) equity incentives and firm value using quantile regression. We show that CEO delta

is more effective in the presence of ample investment opportunities, while CEO vega is more

beneficial for firms lacking investment opportunities. Further, Tobin’s Qincreases in CEO

delta for more risk-tolerant firms but increases in CEO vega for more risk-averse firms. We

also observe that higher monitoring intensity after the Sarbanes-Oxley Act reduces CEO delta’s

role in compensation. Risk aversion alters the optimal incentive-value relation, and the nature

of this relation also depends on the level of Tobin’s Q.

Keywords: executive compensation, managerial equity incentives, firm value, SOX

JEL Classifications: G30, G34, J33, M52

∗Corresponding author: Department of Accounting, Finance, & Business Law, University of Texas at

Tyler,Tyler, TX 75799; Phone: (903) 566-7360; Fax: (903) 566-7372; E-mail: hjames@uttyler.edu.

The authors thank an anonymous referee and the editors for many helpful and insightful comments. They

also thank Garrett Smith (discussant) and participants at the Financial Management Association 2017

Annual Meeting. All errors remain the responsibility of the authors.

C2019 The Eastern Finance Association 583

584 B. W. Benson et al./The Financial Review 54 (2019) 583–638

1. Introduction

An expansive body of research examines the relation between managerial equity

incentives and firm value with mixed findings.1Previous studies, however, use tradi-

tional linear regression models (e.g., ordinary least squares [OLS] regression, fixed

effect [FE] regression) to investigate whether the conditional mean of the dependent

variable (i.e., firm value or performance) is associated with a change in covariates

(e.g., chief executive officer[CEO] delta and vega). Linear regression models assume

the estimates of the relation between firm value and CEO equity incentives are the

same regardless of where the researcher examines the conditional distribution. But

the estimated effects on the conditional mean of firm value are not necessarily in-

dicative of the nature of the relation between CEO equity incentives and firm value

when firms are at noncentral locations. Moreover, the coefficient estimates generated

from this estimation methodology are unbiased and efficient only if the residuals

are normally distributed, the variance of errors is constant, and the sample has no

influential outliers (Greene, 2008; Li, 2015).

We employ several statistical methods to test whether the aforementioned con-

ditions are met in our sample. First, a Shapiro-Wilk test rejects the normality of the

residuals at the 1% significance level, indicating nonnormality in the errors. Second,

both a Breusch-Pagan/Cook-Weisberg test and White’s test for heterogeneity rejects

the null hypothesis of constant variance at the 1% level. Finally, the absolute value

of the studentized residuals is larger than or equal to 2 in 5% of our sample firms,

suggesting the existence of outliers.2This suggests that traditional conditional mean

estimation techniques may be problematic when examining the effect of manage-

rial equity incentives on firm value. Figure 1 describes these issues graphically by

presenting the highly skewed distribution of Tobin’s Q.

We employ a quantile regression (QR) analysis to examine the effect of CEO

delta and vega on firm valueto mitigate these issues. This analysis produces estimates

by finding the values that minimize the sum of absolute residuals while OLS models

minimize the sum of squared errors. This estimation methodology overcomes many

problems associated with methods using a conditional mean framework (Koenker

1Studies find a positive correlation between managerial ownership and financial performance (Mehran,

1995; Core and Larker, 2002), a positive but decreasing relation (Morck, Schleifer and Vishney, 1988;

McConnell and Servaes 1990, 1995; Hermalin and Weisbach,1991; Hubbard and Palia, 1995; Holderness,

Krosner and Sheehan, 1999; Anderson and Reeb, 2003; Tian, 2004; Davies, Hillier and McColgan, 2005;

Adams and Santos, 2006; Pukthuanthong, Roll and Walker, 2007; McConnell, Servaes and Lins, 2008;

Tong, 2008; Benson and Davidson, 2009) or no relation (Demsetz, 1983; Demsetz and Lehn, 1985;

Agrawal and Knoeber, 1996; Loderer and Martin, 1997; Cho, 1998; Himmelberg, Hubbard and Palia,

1999; Demsetz and Villalonga, 2001; Palia, 2001; Coles, Lemmon and Meschke, 2012; Brick, Palia and

Wang, 2005; Cheung and Wei, 2006). Brick, Palmon and Wald (2012) find a negative relation between

vega and firm future stock returns. Chen and Ma (2011) find that CEO stock options increase both long-

term and short-term stock returns. Shen and Zhang (2012) show that higher vega encourages greater and

suboptimal research and development investments,leading to a lower stock and operating performance.

2The test results are not tabulated but are available from the authors on request.

B. W. Benson et al./The Financial Review 54 (2019) 583–638 585

Figure 1

Distribution of Tobin’sQ

Graph plots the distribution of Tobin’sQ. The sample is from COMPUSTATand Execucomp and includes

21,308 firm-year observations from fiscal year 1997–2013.

and Hallock, 2001; Li, 2015). First, QR makes no assumption on the distribution of

the error term. Second, QR allows different estimates to be calculated at different

points on the conditional distribution of Tobin’s Q. This is preferable to segmenting

the sample based on the distribution of Tobin’s Q(e.g., large Tobin’s Qvs small

Tobin’s Q), which truncates the sample and may lead to sample selection bias.

Finally, QR can be used to demonstrate that the conditional distribution may not

be homogeneous. Specifically, the conditional distribution is not homogenous when

estimates of different quantiles are found to be significantly differentfrom each other.

The application of the QR method is common in the economics literature and

gaining popularity in the finance literature. The QR method has been applied recently

in studies on the effect of diversification on firm value (Lee and Li, 2012), pay-

performance sensitivity (Hallock, Madalozzo and Reck, 2010), bankruptcy prediction

(Li and Miu, 2010), corporate governance and tax avoidance (Armstrong, Blouin,

Jagolinzer and Larcker, 2015), and pecking order (Chay, Park, Kim and Suh, 2015).

This study represents the first attempt to apply the QR method to the relation between

CEO equity incentives and firm value.

Weexamine the relation between CEO delta (the sensitivity of managerial wealth

to stock price) and CEO vega (the sensitivity of managerial wealth to stock return

volatility) and Tobin’s Qfor a sample of 21,308 firm-year observations from 1997 to

2013 using QR. Tobin’s Qis used in the literature as a proxy for both performance