the dark side of promotion tournaments: the effect on ......kachelmeier, chan li, jeremy lill,...

57
The Dark Side of Promotion Tournaments: The Effect on Losing Competitors and Their Firms Eric W. Chan The University of Texas at Austin John H. Evans III* University of Pittsburgh Duanping Hong University of Pittsburgh Version as of November 2016 *Corresponding author Email: [email protected] Phone: +1 (412) 648-1714 238A Mervis Hall, Katz Graduate School of Business, University of Pittsburgh, Pittsburgh, PA 15260 We thank Sean Cao, Shuping Chen, Matthew DeAngelis, Mei Feng, Nick Hallman, Steve Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants at Georgia State University, the University of Pittsburgh, the University of Texas at Austin, participants at the 2016 Southeast Region Meeting, and reviewers from the 2017 Management Accounting Section Midyear Meeting for their helpful comments.

Upload: others

Post on 08-Jul-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

The Dark Side of Promotion Tournaments: The Effect on Losing Competitors

and Their Firms

Eric W. Chan

The University of Texas at Austin

John H. Evans III*

University of Pittsburgh

Duanping Hong

University of Pittsburgh

Version as of November 2016

*Corresponding author

Email: [email protected]

Phone: +1 (412) 648-1714

238A Mervis Hall,

Katz Graduate School of Business,

University of Pittsburgh,

Pittsburgh, PA 15260

We thank Sean Cao, Shuping Chen, Matthew DeAngelis, Mei Feng, Nick Hallman, Steve

Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady

Williams, Flora Zhou, workshop participants at Georgia State University, the University of

Pittsburgh, the University of Texas at Austin, participants at the 2016 Southeast Region Meeting,

and reviewers from the 2017 Management Accounting Section Midyear Meeting for their helpful

comments.

Page 2: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

The Dark Side of Promotion Tournaments: The Effect on Losing Competitors

and Their Firms

Abstract

Firms commonly use promotion tournaments to motivate employee performance. While prior

research has examined the incentive effects of such tournaments on the tournament winners and

their firms, we focus on certain additional consequences that arise following the end of

promotion tournaments. More specifically, we investigate tournaments for promotion to Chief

Operating Officer (COO) to examine whether and how the subsequent actions of non-promoted

VPs, whom we refer to as non-promoted executives (NPEs), affect subsequent firm performance.

When NPEs are passed over for promotion, their implicit, promotion-based incentives decline

because their future prospects for advancement within the firm fall significantly. In turn, we

predict and find that firms generally do not adjust the NPEs’ explicit compensation to replace the

diminished promotion-based incentives, and we attribute the absence of such adjustments to

associated adjustment costs. As a result, NPEs face reduced incentives after the COO tournament

ends, and we predict and find an increase in turnover among NPEs. Further, we document that

the change in firm performance following the end of the COO tournament is more negative for

more competitive (strong) tournaments than for less competitive (weak) tournaments, which we

attribute to a combination of the decreased incentives for NPEs who choose to stay and the loss

in human capital for NPEs who choose to leave the firm. Further, we show that these negative

effects on firm performance are only temporary and are remediated by the firm over time.

Overall, our results identify potentially negative consequences for firms immediately after a

promotion tournament ends resulting from the effects on non-promoted employees.

JEL classifications: J01, M12, M41, M51

Keywords: Promotion, Tournament, Executive Compensation, Executive Turnover, Firm

Performance

Page 3: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

1

1. Introduction

Promotion is an important, implicit ingredient of most organizations’ incentives systems

(Gibbs, 1996). Prior research has found that promotion tournaments motivate employee

performance (Main et al., 1993; Baker et al., 1994; Bognanno, 2001; DeVaro, 2006; Kale et al.,

2009).1 While many prior studies examine the incentive effects of promotion tournaments on

employees during the tournament, we are not aware of previous studies that have investigated

the incentive consequences at the end of the tournament, particularly for the employees who are

passed over for promotion.

Promotion tournaments are typically intermittent because firms have a fixed number of

positions to which employees can be promoted within their hierarchy (DeVaro, 2006). Therefore,

when a promotion tournament ends and a position is filled, the future advancement prospects for

those who are passed over for promotion fall significantly. Further, there is increased uncertainty

regarding the timing of the next tournament for the same position. This uncertainty increases

with the hierarchical level because there are fewer opportunities for promotion at the top of the

organization and job tenure is often longer. As a consequence, we expect promotion-based

incentives of non-promoted executives (NPEs hereafter) to decrease significantly after the end of

a tournament. The purpose of our study is to examine the actions that firms and NPEs take after a

promotion tournament ends, and to analyze how those actions affect firm performance.

Prior research posits that firms should optimize the combination of employees’ explicit

incentives from their compensation contracts and their implicit incentives from career concerns,

which include internal promotion-based incentives from the firm and external incentives from

1 Consistent with the extant literature, we assume all promotions involve tournaments in which multiple employees

compete for promotion. However, as discussed in detail later, we differentiate between strong and weak promotion

tournaments based on the number of competitors and their characteristics.

Page 4: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

2

the labor market (Gibbons and Murphy, 1992; Gibbs, 1995). To the extent that explicit and

implicit incentives are substitutes, when NPEs face a significant drop in promotion-based

incentives after losing a promotion tournament, firms should strengthen their explicit incentives

to offset the reduction in implicit incentives. However, prior archival studies provide mixed

evidence on the extent to which firms actively adjust employees’ explicit incentives when their

implicit promotion-based incentives change (Gibbs, 1995; Ederhof, 2011; Campbell, 2008).

Limited adjustment of explicit incentives is consistent with such adjustments being costly

for the firm. Prior research has found that various adjustment costs severely constrain firms’

ability to restructure employees’ compensation contracts when their incentives become

misaligned (Core and Guay, 1999; Core et al., 2003; Bushman et al, 2015). Further, increasing

non-promoted employees’ explicit incentives would effectively reduce the ex ante pay gap

between tournament winners and losers, thereby further reducing the strength of promotion-

based incentives. As such, we predict that firms will not fully adjust NPEs’ explicit incentives

despite the reduction in their implicit incentives following a promotion tournament.

Consequently, we expect NPEs to face lower incentives immediately following the end of an

executive promotion tournament.2

We expect NPEs to respond to a decrease in total incentives by choosing either to leave

the firm for better outside opportunities or to stay with the firm but to exert less effort in

response to reduced rewards for high performance. Importantly, whether the NPE remains with

the firm or not, we expect the NPE’s actions to adversely affect firm performance. Specifically,

NPEs who leave the firm voluntarily reduce the firm’s valuable, firm-specific human capital

2 To retain particularly talented NPEs, firms can also create new executive positions with expanded responsibilities

and titles. Such changes are nevertheless likely to require corresponding increases in explicit incentives to maintain

the NPE’s total incentives.

Page 5: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

3

(Hayes and Schaefer, 1999; Ittner et al., 2003; Fee and Hadlock, 2004). As a result, we expect

firm performance to be negatively affected by NPE turnover. For NPEs who choose to stay in the

same position with the firm, we predict that the firm’s inability to fully adjust their incentives

will result in their exerting lower effort. Because such NPEs are important decision-makers

within the firm, their reduced effort will also have a negative effect on firm performance.

To test our hypotheses, we examine internal promotions to COO and/or President that

occurred in firms covered by ExecuComp from 1993 to 2013. We include both COOs and

Presidents in our sample and refer to both COOs and Presidents as “COOs” in the remainder of

the paper because prior research finds that executives who hold either title generally perform the

same function (Hambrick and Cannella 2004). While prior tournament research focuses mainly

on CEO tournaments (e.g. Bognanno, 2001; Kale et al., 2009), we examine COO promotions

because they represent a setting in which a larger pool of executives typically compete for the

promotion, generating a larger sample. In addition, many firms use succession plans whereby the

COO is expected to become the next CEO. In such firms, the outcome of the subsequent CEO

tournament is largely determined by the results of the preceding COO tournaments (Cannella and

Shen, 2001; Shen and Cannella, 2003; Naveen, 2006), making such COO tournaments very

important to the firm’s overall performance. Therefore, we focus our sample on promotions in

which the promoted COO is likely to become the next CEO based on characteristics such as the

COO’s age and pay rank within the firm.

We document several patterns in responses by firms and NPEs following the outcomes of

COO tournaments. First, we find little evidence that firms significantly increase the explicit

incentives of NPEs subsequent to a COO promotion. This result holds even for those who were

apparently strong but unsuccessful competitors for the COO position, and hence are likely to

Page 6: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

4

continue to serve in important executive positions if they remain with the firm. Second, more

than 30% of our sample NPEs leave the firm following a COO promotion. An additional cross-

sectional test shows that the incremental turnover rate, when compared to a control sample of

non-CEO executives (VPs hereafter) in other firms that did not experience a COO promotion, is

significantly higher among NPEs who appear to be stronger competitors for the COO position.

Finally, we find that the change in firm performance, measured as industry-adjusted return on

assets (ROA) from one year prior to COO promotion to the year of promotion and one year after,

is worse for firms with a strong COO tournament than for firms with either no COO promotion

or a weak COO tournament.3 This worse firm performance following the end of a strong COO

tournament occurs whether NPEs leave the firm or stay with the firm after being passed over for

promotion. However, this negative performance effect is generally temporary, and is remediated

by the firm within two years after the COO promotion. Additional analysis shows that departed

NPEs on average find superior positions at other firms, suggesting that the NPEs have high

ability and are leaving voluntarily. We provide evidence that weak pre-promotion firm

performance and earnings management are not viable alternative explanations for the decline in

performance after the COO tournament ends.

Our study contributes to the literature on incentives and compensation in several ways.

First, we provide new evidence on how firms design and adapt employees’ incentives over time.

Prior studies that have examined this issue provide mixed evidence. Studies based on data from

individual firms (Ederhof, 2011, Gibbs, 1995) offer the advantage of holding constant other firm

institutional effects but are limited by this focus and may not generalize to other firms. We

3As discussed later in the background and hypotheses section, we define a COO tournament as “strong” (“weak”) when there are

(are not) multiple, strong competitors who appear to be suitable candidates for the COO position.

Page 7: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

5

analyze data for a large sample of firms over an extended period of time to test whether firms

dynamically adjust employees’ explicit incentives when their promotion-based incentives change

as predicted by the career concerns model (Gibbons and Murphy, 1992). Our results suggest that

firms on average do not significantly adjust NPEs’ explicit incentives when their promotion-

based incentives decrease. This result is consistent with prior research that finds that various

adjustment costs inhibit firms’ restructuring of compensation contracts when employee

incentives are misaligned (Core and Guay, 1999; Core et al., 2003; Bushman et al., 2015).

Second, our study offers new insight on the costs and benefits associated with promotion

tournaments. Prior research concludes that tournaments can induce higher employee effort and

lead to improved performance (e.g., Kale et al., 2009, Main et al., 1993; Baker et al., 1994;

Bognanno, 2001). However, prior research also analyzes certain costs associated with

tournaments. For example, employees competing in tournaments are less likely to cooperate and

more likely to engage in sabotage activities (Chen, 2003; Harbring and Irlenbusch, 2011). We

identify a new cost associated with promotion tournaments by providing evidence of a decline in

firm performance after the end of a COO tournament, which we associate with firms’ inability to

retain key executives and to adequately adjust the incentives of the executives who are retained.

Third, our results add to the prior literature that examines the value that executives bring

to firms. Prior studies, such as Hayes and Schaefer (1999), have examined the market reactions

to VP turnover. We document that firm performance is sensitive to how VPs respond to changes

in their promotion-based incentives. Our results suggest that it is important for firms, particularly

those that rely on promotion tournaments to provide incentives, to consider the reactions of non-

promoted employees when designing incentive schemes.

Next, Section 2 provides background information and develops our hypotheses; Section 3

Page 8: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

6

describes our research setting and sample selection methods; Section 4 reports the results;

Section 5 provides additional analysis and robustness tests; and Section 6 discusses the

implications of our results and concludes.

2. Background and Hypotheses Development

2.1 Promotion tournaments

Prior literature has extensively examined the role of promotion tournaments in firms

(e.g., O’Reilly et al., 1988; Gomez-Mejia and Balkin 1992; McConnell and Brue, 1992; Main et

al., 1993; Gibbs, 1995). In a typical tournament, employees compete against their peers based on

relative rather than absolute performance for rewards that accompany promotions. Economic

theory establishes conditions under which tournaments can provide optimal effort incentives

when employees face common uncertainty and/or when output is difficult or costly to measure

(Lazear and Rosen, 1981; Nalebuff and Stiglitz, 1983).

Empirical evidence generally supports the predictions of tournament theory (Gibbs, 1994;

Cichello et al., 2009). For example, prior studies have documented a positive relation between

the size of the tournament incentive, as measured by the pay gap between hierarchical levels, and

the number of competitors (Main et al., 1993; Bognanno, 2001). Kale et al. (2009) also provide

evidence that larger pay gaps between the CEO and VPs are associated with better firm

performance. Because a tournament structure requires rewards to increase at an increasing rate at

each higher job level, pay grows at an increasing rate with the hierarchical level (Gibbs, 1994).

The optimal convex pay pattern predicted by tournament theory is often used to explain the high

levels of executive pay observed in practice. Thus, the prospect of potentially being promoted to

COO, and then possibly to CEO, provides a very powerful implicit incentive for competitors in a

Page 9: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

7

COO tournament.4

In contrast to annual bonuses which provide recurring, relatively predictable and

continuous incentives, promotion tournaments occur irregularly. When a promotion tournament

ends and a position is filled, there is typically considerable uncertainty with respect to the timing

of the next tournament for the same position. Therefore, when employees are passed over for

promotion, their promotion-based incentives decline significantly because they are both less

likely to obtain future promotions in the firm and they also face increased uncertainty regarding

the timing of the firm’s next promotion tournament.

2.2 Adjustments to explicit incentives

As described earlier, we expect the promotion-based incentives of non-promoted

employees to decrease immediately following the end of a promotion tournament. In this case,

prior research argues that firms should adjust their explicit incentives accordingly.5 Specifically,

models of career concerns predict that firms will consider employees’ implicit incentives when

designing their explicit incentive contracts (Gibbons and Murphy, 1992; Gibbs, 1995). Implicit

incentives from career concerns can include the employees’ internal promotion opportunities as

4 As a rough indication of the magnitude of annual incentives associated with potential promotions to COO, consider

a VP with a 1/3 probability of being chosen COO, and if the VP wins the COO tournament, a 1/3 probability of later

being chosen CEO. Based on the mean pay values in our sample, the annual increase in total pay from being

promoted to COO would be $2.74M - $1.98M = $0.76M (Panel E of Table 2), and the annual increase in total pay

from being promoted to CEO would be $5.22M - $2.74M = $2.48M (untabulated). The expected value of the

increased annual compensation would be 1/3*$0.76M + (1/3)2*$2.48M = $0.53M, which would be a

$0.53M/$1.98M = 27% increase in the VP’s current total annual pay. Stated differently, learning that a VP had lost

the COO tournament could be interpreted as reducing the VP’s expected future annual pay by approximately 27%. 5 Prior research has established alternative conditions under which implicit and explicit incentives are complements

rather than substitutes (Baker et al., 1994; Holmstrom and Milgrom, 1994; Kaarboe and Olsen, 2006). For example,

Baik, Evans, Kim, and Yanadori (2016) examine the compensation schemes of mid-level managers at large,

hierarchical firms and find that both implicit and explicit incentives becomes stronger at higher job levels. Their

results suggest that the two types of incentives can be complementary when firms do not face constraints on

employees’ implicit incentives. However, in our COO promotion context, firms face significant constraints on

adjusting NPEs’ implicit incentives when a tournament ends because the number of job openings is fixed and there

is uncertainty regarding the timing of the next promotion (DeVaro, 2006). In this case, firms can only adjust NPEs’

explicit incentives to substitute for the loss in implicit incentives.

Page 10: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

8

well as their outside options in the external labor market. Career concern models predict that

when employees experience a decrease in implicit incentives from their career concerns, the firm

should realign their incentives by increasing their explicit, performance-based bonuses and

equity incentives to preserve their total incentives.

Several field studies have tested career concern predictions with mixed results. Ederhof

(2011) examines mid-level managers’ pay at a multinational firm and finds that bonus-based

incentives are stronger for managers who face weaker promotion-based incentives, such as

executives with fewer remaining levels to climb in the hierarchy. However, her study relies on

data for one firm in a single year and therefore is constrained in addressing the issue of the extent

to which firms in general actively adjust employees’ explicit incentives in response to changes in

their implicit incentives. Further, she acknowledges that an alternative explanation for her

finding that higher-level managers have greater bonus-based pay is that they have more decision-

making authority and therefore the available performance measures could more accurately reflect

their actions than would be the case for lower-level managers (Ederhof 2001, 148).

Gibbs (1995) provides a more direct test of whether employees’ explicit incentives are

adjusted when their implicit incentives decrease. Gibbs uses longitudinal data for a single firm to

examine whether incentives change over time for employees with longer tenure at the same

position who have been repeatedly passed for over promotion. He finds that the firm does not

adjust the terms of non-promoted employees’ incentive plans. Further, the performance ratings of

non-promoted managers decline over time, which he attributes to their facing lower incentives.

However, Gibbs indicates that the lack of evidence on an adjustment of managers’ explicit

incentives could also be due to the firm’s use of a centrally administered incentive plan, which

limits its flexibility to make individual adjustments to employees’ incentives (p.274).

Page 11: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

9

Taken together, prior research provides some evidence that firms account for employees’

implicit incentives when designing their explicit incentives across job levels. However, firms do

not appear to routinely and fully adjust employees’ explicit incentives on an individual basis.

One explanation for this result is that there are substantial adjustments costs associated with

restructuring employees’ incentive contracts (Core and Guay, 1999; Core et al., 2003). For

example, the firm’s ability to realign executives’ incentives could be constrained by the public’s

scrutiny of executive compensation (Murphy, 2013; Bushman et al., 2015).6 Further, equity

concerns could arise among employees when firms selectively adjust only certain employees’

incentives. This would compel firms to either accept potential employee discontent or

concurrently restructure the contracts of other unaffected employees, increasing the firm’s total

adjustment costs.

Researchers argue that such adjustment costs can explain why firms apparently do not

restructure employees’ compensation contracts even when their incentives become misaligned

(e.g., Morck et al., 1988; Core and Guay, 1999). For example, Bushman et al. (2015) conclude

that high adjustment costs inhibit firms from issuing equity grants to restore the pay performance

sensitivity in executives’ compensation contracts to the optimal level after certain shocks reduce

their equity holdings, such as when executives exercise their stock options or rebalance their

investment portfolios. Further, they argue that such adjustment costs contribute to firms’

persistent use of apparently suboptimal incentive contracts. Overall, these prior studies suggest

that in many cases the adjustment costs associated with restructuring employees’ compensation

contracts outweigh the related benefits of the adjustments, and therefore firms choose not to

adjust employees’ incentives.

6 The Say-On-Pay regulation in the Dodd-Frank Act of 2011 requires firms to justify any upward adjustment of

executive pay (Ertimur et al., 2013).

Page 12: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

10

In the context of a COO promotion tournament, we expect the costs associated with

adjusting NPEs’ explicit incentives at the end of a tournament to be even greater than those

documented in many prior studies. Specifically, the strength of executives’ promotion-based

incentives is partially determined by the prize of winning the promotion tournament as measured

by the pay gap between the promoted employee and the non-promoted employees. To the extent

that firms increase NPEs’ explicit incentives at the end of a promotion tournament, the pay gap

between the promoted executives and the NPEs is reduced, effectively reducing the strength of

the executives’ promotion-based incentives. In other words, if the executives competing in a

COO tournament expect their explicit incentives to be adjusted upwards if they lose the

tournament, they have less incentive to exert high effort to win the tournament. Thus, to preserve

the executives’ ex ante promotion-based incentives, firms may prefer to follow a recognized

policy of making only limited increases in NPEs’ explicit incentives following a tournament.

This reasoning implies that many firms will make only limited adjustments to NPEs’ explicit

incentives at the end of a promotion tournament, leading to our first hypothesis:

H1: NPEs’ explicit incentives do not change following the end of a promotion

tournament.

2.3 Non-promoted executives’ impact on firm performance

When firms fail to adjust NPEs’ explicit incentives after a tournament to offset the

reduced implicit promotion-based incentives, NPEs face a reduction in total incentives relative to

their incentives prior to the tournament outcome.7 This reduction in total incentives can create

problems in both retaining key executives and also in providing sufficiently strong incentives to

7 By an executive’s “incentives”, we refer to the executive’s expected compensation given the firm’s compensation

design and the executive’s action choices. As such, an executive’s “incentives” reflects both the level of expected

compensation and the design of the compensation system.

Page 13: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

11

those executives who are retained.8 Without sufficient incentive adjustments following a

tournament, NPEs are likely to either leave the firm for outside opportunities or to stay with the

firm but provide less effort than shareholders would prefer. Individual NPEs’ decisions in this

regard are likely to depend on the available outside options and also such idiosyncratic factors as

the NPE’s relationship with the newly promoted executive, their loyalty to the firm, etc.

However, we argue below that the firm’s performance is likely to be affected regardless of

whether the NPE stays with the firm or leaves.

Firm performance following a tournament will depend on the actions and decisions of the

newly promoted executive who wins the tournament, and the NPEs, the tournament losers. The

incentives of an executive promoted to COO will typically increase after the promotion both

because of increased explicit performance-based incentives and stronger implicit incentives

resulting from an increased probability of subsequent promotion to CEO, which would provide

an even greater increase in pay. Thus, we expect the newly promoted COO to respond to the

increase in incentives by exerting greater effort, which will result in a positive effect on firm

performance. Consistent with this, prior studies on CEO succession find some evidence that firm

performance increases following CEO turnover and they attribute this positive effect on firm

performance to the actions taken by the new CEO (Denis and Denis, 1995; Huson et al., 2004).

However, we expect the actions of the NPEs to have an offsetting negative effect on firm

performance whether the NPE leaves the firm or not.

When NPEs leave the firm to pursue outside opportunities, firm performance will often

suffer because of the firm’s loss in human capital. Prior research provides evidence that investors

and the external labor market value the firm-specific human capital possessed by VPs. For

8 Our empirical analysis examines whether firms adjust NPEs’ short-term pay, which includes primarily salary and

bonus, as well as long-term pay, which includes primarily equity compensation.

Page 14: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

12

example, Hayes and Schaefer (1999) document a negative stock market reaction when firms lose

VPs to a competitor firm. Fee and Hadlock (2004) find that VPs who leave voluntarily tend to

obtain superior employment elsewhere, which suggests that the external labor market recognizes

their value. Finally, in a small sample study, Shen and Cannella (2002) find that VP turnover

after a CEO has been replaced by an outsider is associated with subsequent poor firm

performance measured as ROA. Based on these prior studies, we expect NPE turnover at the end

of a promotion tournament to have a negative effect on firm performance.9

Alternatively, NPEs can choose to stay with the current firm at the end of a promotion

tournament. As noted earlier, we expect NPEs who remain with the firm to face weakened total

incentives to expend high levels of effort and to make decisions that maximize firm value to the

extent that the firm does not adjust their explicit incentives to replace the rewards for hard work

in the form of an increased probability of promotion that the tournament generated.10 In response

to reduced incentives, we expect NPEs to exert less effort after the promotion tournament.

Because NPEs will continue to make or influence important decisions within the firm, we predict

that their reduced effort will have a detrimental effect on firm performance.

Importantly, we expect the competitiveness of the promotion tournament to moderate

both the turnover rate of NPEs and firm performance following a promotion tournament.

Specifically, we expect NPE turnover to be higher and the change in firm performance to be

more negative at the end of a COO tournament when the tournament is more competitive; i.e.,

when it features more strong competitors. As discussed in detail later, we define a “strong

9 The magnitude of the negative effect will depend on various factors, such as the labor market supply of executive

talent. For example, if the firm operates in a specialized industry, it will generally be more difficult and costly for

firms to replace the departed NPE, increasing the negative impact. 10 From a multitask perspective, an NPE’s “reduced effort” is also likely to take the form of performing a mix of

tasks that deviate more significantly from the mix that shareholders prefer. That is, for a fixed level of total effort, an

NPE facing weaker incentives is likely to put more weight on tasks that they personally prefer versus tasks that

maximize firm value.

Page 15: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

13

competitor” for the COO position as a VP who is relatively highly paid compared to other VPs,

which is a characteristic frequently associated with newly appointed COOs. In turn, we define a

COO tournament as more competitive or “strong” when multiple strong competitors compete for

the COO position. In contrast, we define a COO tournament as “weak” when there is only one

executive who appears highly suitable for the COO position and is ex ante the most likely winner

of the tournament.

Because strong competitor NPEs competing in a COO tournament are more likely to be

promoted than weak competitor NPEs, we expect strong (weak) competitor NPEs’ promotion-

based incentives to decrease to a greater (lesser) extent after a tournament ends.11 Based on the

H1 prediction that firms will not subsequently adjust the explicit incentives of NPEs despite the

decrease in implicit incentives, we expect the turnover of strong competitor NPEs to be higher

than weak competitor NPEs following the end of a promotion tournament. Further, we expect

strong competitor NPEs competing in strong COO tournaments to reduce their subsequent effort,

resulting in a more negative effect on firm performance than in weak COO tournaments.

In summary, regardless of whether NPEs leave or stay, we expect their response to have a

negative impact on firm performance after the promotion tournament. Because we expect firm

performance to also be positively affected by the actions of the promoted executive, we make no

prediction regarding the net directional change in firm performance following the end of a

promotion tournament. However, we expect the change in firm performance to be more negative

for strong tournaments than for weak tournaments. Our above reasoning generates our second

11 We do not control for the total number of competitors in the COO tournament, but the greater the number of

competitors, the smaller will be the reduction in implicit incentives that an NPE experiences because their ex ante

likelihood of winning the tournament is smaller. The effect of not having this control is limited in our context

because in our sample there are typically relatively few high-level executives competing for the COO position in

most firms.

Page 16: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

14

and third hypotheses:

H2: Executive turnover increases following the end of a promotion tournament.

H2a: The increase in executive turnover is greater for strong competitors for the

COO position than for weak competitors.

H3 (null form): Firm performance does not change following the end of a promotion

tournament.

H3a: The change in firm performance following the end of a promotion tournament

is more negative in strong COO tournaments than in weak COO tournaments.

3. Research Setting and Sample Selection

3.1 Research setting

We test our hypotheses concerning consequences of executive tournaments using a

sample of promotions to COO. Analyzing COO tournaments offers two advantages. First,

promotion to COO is economically important because in many firms the COO becomes the heir

apparent to the CEO position and typically soon replaces the current CEO (Cannella and Shen,

2001; Shen and Cannella, 2003; Naveen, 2006). Further, NPEs who are passed over for

promotion in a COO tournament frequently remain in influential executive roles in the firm.

Therefore, to the extent that losing the COO tournament significantly reduces their promotion-

based incentives in the absence of a corresponding increase in explicit incentives, the result is

likely to have a negative effect on firm performance. Hence, the consequences of the COO

promotion process are likely to be economically significant.

Second, COO tournaments provide a relatively clean setting in terms of accurately

identifying the tournament competitors. This is because the tournament competition for the COO

position is largely among internal VPs, who are easier to identify than outside candidates, who

often play a more important role in tournaments for promotion to CEO. Further, after any

executive promotion tournament ends, firm performance will depend on the actions of the

Page 17: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

15

tournament winner and losers. Given our focus on tournament losers, we prefer a setting in

which the tournament winner is less likely to exert a dominant influence on post-promotion firm

performance. We expect a newly promoted COO to be less likely to exert a dominant impact on

subsequent firm performance than a newly appointed CEO would. Therefore, when compared to

a CEO tournament, we expect the COO tournament setting to provide us with a better

opportunity to identify potential negative effects on firm performance after the tournament ends

as a results of actions by the tournament losers, the NPEs.

As general background, we note that the COO position is a relatively new innovation in

U.S. corporations. In fact, as late as 1975 many prominent analyses of executive teamwork and

succession in major U.S. firms do not discuss the role of the COO position (Hambrick and

Cannella 2004, 960). The emergence of the COO position has been attributed to the increasing

complexity of large corporations, which has left many CEOs unable to attend adequately to both

external constituents and internal operations (Bass and Stogdill, 1990; Bennett and Miles, 2006).

Vancil (1987) provided one of the first systematic characterizations of COO positions in

U.S. firms. He described the external CEO/internal COO model in which the CEO retains overall

authority and focuses on external relationships while the COO has primary internal operating

responsibilities. Vancil developed the heir apparent or “relay” model in which the firm uses the

COO position to designate an heir apparent to the current CEO. The “relay” model provides a

more formalized CEO succession process in which the newly appointed COO initially holds that

position for a few years. During this time the board evaluates the COO’s fitness to become the

CEO and grooms the COO for the CEO position, with transition taking place when the outgoing

CEO retires and “hands off the baton” to the COO as the new CEO.

Page 18: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

16

Hambrick and Cannella (2004) provide an overview of how often firms employ a COO

position and how often the COO is the internal partner to the external CEO versus being the heir

apparent to the CEO position. They document that in 404 major U.S. firms from 1987 to 1996,

45% of the firms had a COO position. The authors estimate that this 45% of the sample is

comprised of 20% operating with an external CEO/internal COO team in which the COO is not

the heir apparent, and 25% using the COO position to designate a CEO heir apparent.12 The

remaining 55% of their sample firms had no COO position. As described below, we present

descriptive statistics on COO promotions for a larger and more recent sample over a longer time

span than those in prior studies.

3.2 Sample selection

To test our hypotheses, we first construct a sample of COO promotions, as shown in

Panel A of Table 1. First, following prior literature, we use job title and age to identify an initial

sample of COO promotions.13 We track the ages and changes in the titles of all executives in

firms covered by ExecuComp during our sample period from 1993 to 2013. We define an

executive as promoted to COO in year t if (1) among all VPs in the firm in year t, only this

executive holds the title of COO and/or President, and (2) in year t-1 the executive’s title in

ExecuComp in the same firm is non-empty and the executive is neither a COO nor a President of

the firm in that year.14 This process identifies 1,601 COO promotions over our sample period.

12 Hambrick and Cannella (2004) use the age of the COO relative to the CEO to distinguish the role of the COO.

Specifically, if the COO is no more than four years younger or is older than the CEO, then s/he is considered an

internal COO dedicated to performing operational activities; if the COO is more than four years younger than the

CEO, s/he is considered an heir apparent. 13 We follow prior literature in relying on an executive’s job title and age to proxy for the likelihood of the executive

being a serious candidate to succeed the current CEO; i.e., to participate in the firm’s next CEO tournament. For

example, Naveen (2006) defines an heir apparent as a VP who is not older than the CEO and who is the firm’s COO

or President. Cannella and Shen (2001) define a VP as the heir apparent if the VP is the only person holding the title

of COO or President and if he/she is at least five years younger than the CEO. 14 To identify the COOs in our sample firms, we search the executive’s title for keywords, including ‘COO’, ‘Chief

Operating Officer’, and ‘President’. COOs of a firm’s subdivision or subsidiary are classified as a non-COO VP.

Page 19: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

17

[Table 1]

Next, we further narrow our sample of newly promoted COOs to focus on those whose

characteristics suggest that they are likely to be serious candidates to win the firm’s next CEO

tournament. We focus on this subsample for whom the ex ante implicit incentives associated

with potential promotion to COO and then to CEO are particularly strong. These strong

incentives mean that the corresponding changes in NPEs’ implicit incentives from losing the

COO tournament are large, providing a more powerful test of our hypotheses. Following prior

research, we use two additional requirements to eliminate COOs who are less likely to be serious

candidates for subsequent promotion to CEO. To be retained in our sample, we also require (3)

the new COO is not older than the CEO, and (4) the new COO ranks first or second in salary

and/or total pay in the year of promotion.15,16 We also include in our sample all new COOs who

subsequently become the firm’s CEO even if they do not meet the age and pay-rank

requirements on the basis that their selection confirms that they were serious candidates. Our

sample size is reduced to 1,186 after we drop 354 COOs who do not meet the age requirement

and 61 COOs who do not meet the pay-rank requirement.17

To test our hypotheses, it is important to distinguish the effects of the COO promotion

15 CEO age and executive age are not always available in ExecuComp, particularly in the 1990s. We require both the

CEO’s age and the COO’s age to be non-missing, unless the new COO is subsequently promoted to CEO. This

requirement eliminates 22% of our observations. 16 To validate our pay-rank requirement, we examine the likelihood of the new COO becoming the firm’s next CEO

among firms that subsequently change CEOs. Within our sample, if the new COO’s salary or total pay ranks at least

second highest among all VPs in the firm in the year of becoming COO, then he/she subsequently becomes the

firm’s next CEO 63% of the time. In contrast, if the new COO ranks below second in both salary and total pay

among all VPs, then he/she becomes the firm’s next CEO only 31% of the time. Naveen (2006) uses a similar pay-

rank requirement in which the pay of the heir apparent must be the highest among all VPs and at least 10% higher

than the next highest paid VP. 17 Of the 1,186 new COOs in our sample, we can identify the firm’s next CEO within our sample period in 881

cases (74%). Panel B of Table 1 shows that the promoted COO becomes the firm’s next CEO in 570 (64.7%) of

these 881 cases. In addition, 56.2% (23.7% + 32.5%) of these 570 COOs are promoted to CEO within two years of

their promotion to COO. Consistent with prior research, these descriptive statistics indicate that the promoted COO

often becomes the next CEO, highlighting the important role that COO tournaments play in many firms (Cannella

and Shen, 2001; Shen and Cannella, 2003).

Page 20: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

18

from other concurrent changes in top management. Therefore we exclude from our COO

promotion sample the 318 cases in which a new CEO was appointed either in the year of the

COO promotion or the following year. Finally, we also drop 23 new COOs who leave the firm in

the year after promotion. Following these procedures, our final sample of COO promotions

consists of 845 observations.

Table 2 provides descriptive statistics on our final COO promotion sample. As described

later, we use these statistics to further break down the NPEs in these 845 COO promotion

tournaments into the strongest competitors for promotion to COO versus those who are less

strong competitors. Panel A of Table 2 reports the number of COO promotions in each sample

year and shows that the sample is evenly distributed over our sample period. Panel B summarizes

the promoted COO’s title in the year prior to promotion and shows that 75.6% of new COOs

held operations titles prior to being promoted to COO. Examples of operations titles include Vice

President, Executive Vice President, Vice Chairman, or CEO of a subsidiary. The next most

common title is CFO, which accounts for 19.4% of our sample of COOs. Panel C indicates that

77.9% of new COOs are promoted between the ages of 41 and 55 with a median age of 48. Panel

D shows that 77.8% (76.7%) of new COOs are among the top two highest paid VPs in salary

(total pay) in their firm prior to being promoted.

[Table 2]

Panel E of Table 2 compares the mean and median salary and total pay of the newly

promoted COO in the year prior to, and the year of promotion. Any change in pay represents a

part of the prize of winning the COO tournament. All pay variables are measured in constant

year 2000 dollars using the Consumer Price Index (CPI). As expected, the mean (median) salary

and total pay of the promoted COO increases significantly (p < 0.01) by $45k and $756k ($33k

Page 21: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

19

and $238k), respectively, after being promoted. In addition to this immediate increase in explicit

incentives, the promoted COO’s implicit incentives also increase based on the increased

likelihood of becoming the next CEO and a further significant increase in compensation.

We next identify the NPEs who were passed over for promotion in the COO promotion

tournaments associated with our sample of 845 COO promotions. For all firms included in the

845 COO promotions, the NPEs are all non-CEO and non-COO executives who are among the

top five executives in total compensation as identified by ExecuComp in the year prior to COO

promotion, who are also not promoted to COO. This yields a final sample of 2,389 NPEs.

As a benchmark against which to compare the NPEs, we construct a control sample of

VPs outside of a COO promotion event window. We first construct a control sample of firm-

years in which no COO promotion occurs. Observations in the control sample are firm-years

from ExecuComp in which: (1) the firm has no COO or CEO promotion in the current year and

the next year, and (2) the firm continues to exist in ExecuComp in the next year.18 Thus, the

control firm-years (control years, hereafter) are similar to the firm-years in our COO promotion

sample except there are no COO promotions in the control years. We then identify all non-COO

VPs in the control sample firms (control VPs hereafter) in the year before the control year. These

control VPs are similar to the NPEs in our COO promotion sample except there is no COO

promotion in the control year. Our final control sample consists of 2,031 firm-years and 7,328

control VPs from 1994 to 2012.

4. Results

4.1 Tests of H1 - Firm adjustments to NPE pay

H1 predicts that firms will not increase explicit compensation for NPEs following the end

18 If multiple firm-years meet this requirement for a firm, we include in our sample only one randomly selected year.

Page 22: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

20

of a COO tournament. To test H1, we collect information about each NPE’s compensation in the

year prior to, the year of, and the year following COO promotion. We examine changes in an

NPE’s short-term and long-term compensation following the COO tournament. We conduct this

analysis at the individual VP level, using three compensation variables: ST_pay, which includes

salary, bonus, and other annual payments; LT_pay, which includes restricted stock grants, option

grants, long-term incentive payouts, and total other annual payments; and Total_pay, the sum of

the VP’s short-term and long-term compensation.19 We measure all compensation variables in

constant year 2000 dollars using the annual CPI index, and winsorize all pay variables at the top

and bottom one percentile to reduce the effect of extreme values.

We estimate an NPE’s post-COO promotion pay as the average of the NPE’s pay in the

year of the COO promotion and the following year to account for any lag in pay adjustments.20 If

the NPE leaves the firm in the year following promotion, post-promotion pay is the NPE’s pay in

the year of promotion. If the NPE leaves the firm in the year of promotion, we code the NPE’s

post-COO promotion pay as missing. The high turnover of sample NPEs immediately following

a COO promotion, as described more fully in section 4.2, censors our data in the sense that we

only observe the post-promotion pay of those who stay with the firm. We expect this effect to

work against our prediction that firms do not adjust employees’ explicit incentives because

NPEs are more likely to stay if the firm increases their explicit incentives than if the firm makes

no such adjustment.

19 We follow Kale et al. (2009) in defining short-term and long-term compensations for years prior to 2006 as

described above. Data for years after 2006 reflect the SEC’s expanded disclosure requirements for executive

compensation, and we define ST_pay as the sum of salary, bonus, and non-equity incentive plan, and LT_pay as the

sum of grant date fair value of stock awarded, grant date fair value of options granted, total portion of deferred

earnings reported as compensation, and all other compensation as defined in ExecuComp. 20 As a robustness check, we redefine post-promotion pay as the pay in the year following COO promotion or the

control year and only keep VPs who stay through this year. This reduces our NPE and control VP samples by 21%

and 13%, respectively. Our results in Tables 3 and 4 do not change using this modified sample.

Page 23: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

21

Table 3 reports the percentage changes in short-term pay, long-term pay, and total pay for

NPEs of firms with a COO promotion versus the corresponding changes in compensation for

VPs in the control sample of firms in which no COO promotion occurs. Compensation before the

COO promotion year (control year) is measured in the year prior to the COO promotion year

(control year). Because the distribution of all percentage changes is right-skewed as evidenced

by the means exceeding the medians, it is typically more meaningful to compare the median

percentage changes across groups.21

[Table 3]

As shown in Panel A of Table 3, NPEs receive increases in short-term, long-term and

total pay that are not statistically different from the corresponding increases that control sample

VPs receive. Consistent with H1, these results suggest that firms generally do not significantly

adjust NPEs’ explicit incentives following the end of a promotion tournament to offset the

reduction in the implicit incentives.22

We next compare pay changes after categorizing VPs as strong or weak competitors in

the COO tournament. This distinction is important because we use it to generate cross-sectional

predictions between pay changes for strong versus weak competitors that provide further

assurance that observed changes in NPE turnover and firm performance are associated with

changes in NPE incentives, as we hypothesize. In particular, hypothesis H2a predicts a larger

increase in NPE turnover for strong competitors for COO promotion than for weak competitors

21 The median is a more meaningful statistic in this case because the calculation of the percentage change in VP pay

involves using VP pay before the COO promotion year (control year) as the denominator, for which some pay

components are very small, resulting in very large percentage increases . This problem is particularly severe for

long-term pay because firms typically do not grant equity to VPs every year, resulting in very small denominators

and very large percentage changes that distort the mean results but not the median results. 22 Untabulated results show that mean percentage changes in NPEs’ long-term pay and total pay are significantly

greater than the corresponding amounts for control VPs’ long-term pay and total pay (p=0.029 and 0.062

respectively, two-tailed tests). However, as discussed in footnote 21, these results are heavily influenced by extreme

values even though we winsorize pay variables at the top and bottom one percentile.

Page 24: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

22

and H3a predicts a larger decline in firm performance when NPEs in COO promotion

tournaments are strong rather than weak competitors. We define “strong competitors” for

promotion to COO as VPs whose salary ranks first or second among all VPs competing for the

COO promotion in the year prior to COO promotion and “weak competitors” as VPs whose

salary is ranked third or lower. The rationale for this classification is that higher paid VPs have a

higher likelihood of being promoted to COO as documented in Panel D of Table 2 above because

they are viewed by the firm as being more talented, valuable, and difficult to replace than lower

paid VPs. As such, we expect strong competitor NPEs to experience a greater reduction in

implicit incentives after being passed over for promotion to COO and thus to require a larger

adjustment to their explicit incentives to maintain the same level of initial total incentives. To

ensure a meaningful comparison between NPEs and control VPs, we use the same classification

of strong and weak competitors for both groups.

Consistent with H1, the results in Table 3 indicate that NPEs do not receive statistically

significantly larger increases in short-term pay, long-term pay, or total pay than control VPs,

regardless of whether they are strong competitors (Panel B) or weak competitors (Panel C).

We next conduct multivariate regressions to examine firms’ adjustments of NPE pay after

controlling for other factors that could influence VPs’ pay using the following model:

Log(pay)post = β0 + β1 COO promotion + β2 Strong Competitor + β3 COO promotion*Strong

Competitor + β4 Log(pay)pre+ β5 AROA + β6 Size + β7 Leverage + β8 Market-to-book + β9 Stock

volatility + Year + Industry + e (1)

We again separately analyze changes in short-term pay, long-term pay, and total pay for

NPEs versus control VPs from before to after the COO promotion year (control year for control

VPs). COO promotion is a dummy variable that takes the value of one for NPEs and zero for

Page 25: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

23

control VPs, which distinguishes our treatment sample from our control sample. We code Strong

Competitor as one if the NPE or control VP salary rank is first or second among all non-COO

VPs in the year prior to COO promotion or in the year prior to the control year for the control

sample, and zero otherwise. We also control for firm performance, measured as ROA adjusted

by Fama-French 48 industry median ROA, firm size, measured as the natural log of the firm’s

total assets, leverage, measured as the ratio of long-term liabilities over total assets, market-to-

book, and stock volatility. All control variables are measured in the year of COO promotion for

NPEs or in the control year for control VPs. Finally, we control for year and industry fixed

effects and cluster standard errors at the industry level.

As shown in Table 4, we test H1 using two model specifications for each pay measure.

Columns 1-3 report the results of our first model to test for changes in short-term pay, long-term

pay, and total pay, respectively. Our main test is based on β1, the coefficient of COO promotion,

which represents the change in pay subsequent to COO promotions. Consistent with H1, the

coefficient of β1 is not significantly different from zero for each of the three pay variables.

[Table 4]

Because the failure to reject the null hypothesis of no change is a relatively weak test,

columns 4-6 of Table 4 report the results of our second model in which we add the variables

Strong Competitor and the interaction COO promotion*Strong Competitor. The second model

provides additional cross-sectional evidence concerning whether strong competitor NPEs receive

a larger positive pay adjustment than weak competitor NPEs. Our primary test is based on β3, the

coefficient of COO promotion*Strong Competitor, which represents the change in post-COO

promotion pay for strong competitor NPEs.

The results indicate that the coefficients for β3 are negative for all three pay measures and

Page 26: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

24

significantly negative for long-term-pay and total pay (β3=-0.117 and -0.040, p=0.005 and 0.040,

respectively). These results suggest that not only do strong competitor NPEs not receive a larger

increase in pay after being passed over for promotion, their change in pay is actually less than the

estimated change in pay in the absence of a COO promotion.

Overall, these results provide strong support for the H1 prediction that firms do not adjust

NPEs’ explicit incentives to replace the loss in implicit, promotion-based incentives following

the end of a COO tournament. The results hold even for strong competitor NPEs, despite their

experiencing a greater reduction in implicit incentives at the end of the COO tournament. We

attribute this result to the presence of adjustment costs that inhibit a firm’s ability to adjust VP

explicit pay components. Importantly, our results suggest that NPEs who choose to stay with

their firm likely face a reduction in total incentives after being passed over for promotion. We

next examine the implications of these patterns in NPE pay for the retention of NPEs.

4.2 Turnover of NPEs: Tests of H2 and H2a

H2 predicts that NPE turnover, i.e. turnover of VPs who are passed over for promotion to

COO, will increase following the end of a COO tournament. We test this prediction by

comparing the turnover rate of NPEs to that of control VPs at the individual VP level. We code

NPE (control VP) turnover as one if the NPE (control VP) leaves the firm during the COO

promotion year (control year) or the following year, and zero otherwise. We further differentiate

turnovers in which the VP leaves at pre-retirement age versus retirement age. Turnovers at pre-

retirement age are more likely to be associated with being passed over for promotion to COO,

whereas turnovers at retirement age are more likely to be independent of the COO tournament

outcome.23 We classify turnover as “pre-retirement” if the departed VP’s age in the departing

23 Our tests on pre-retirement turnover and retirement exclude VPs with missing age data in ExecuComp. This

reduces our NPE sample size from 2,389 to 1,711 and our control VP sample size from 7,328 to 5,866.

Page 27: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

25

year is less than 63; and “retirement” otherwise.24

Table 5, Panel A reports that the overall mean turnover rate of 30.3% for NPEs is

significantly greater (p < 0.01) than the 22.5% overall mean turnover rate for control VPs.

Further, the mean pre-retirement turnover rate and the retirement turnover rates of NPEs of

23.4% and 6.9%, respectively, are significantly greater (p < 0.01) than the corresponding rates of

18.0% and 4.6%, respectively, for control VPs. Consistent with H2, these results show a

significant increase in the turnover of NPEs following the end of a COO promotion tournament.

[Table 5]

H2a predicts that NPE turnover will be greater for strong competitors for the COO

position than for weak competitors. We test this prediction using the same “strong competitor”

and “weak competitor” criteria as in our earlier tests of H1. Table 5, Panel B shows that the

overall turnover, pre-retirement turnover, and retirement turnover rates of 34.3%, 25.2%, and

9.1% for NPEs are significantly greater (p <0.01) than the corresponding turnover rates of

20.0%, 15.0%, and 4.9% for control VPs, respectively. Given the magnitudes of baseline

turnover rates for the control VPs, the differences of 14.3%, 10.2%, and 4.2% in overall

turnover, pre-retirement turnover, and retirement turnover rates suggest potentially large

economic effects. In contrast, as shown in Panel C of Table 5, we find no statistically significant

differences in overall turnover, pre-retirement turnover, and retirement turnover rates between

weak competitor NPEs and control VPs. Overall, these results provide initial support for the H2a

prediction that turnover rates are higher for strong competitors than for weak competitors

following the end of a promotion tournament.

24 Prior literature generally uses age cutoffs ranging from 60 to 65 in defining retirement. We use the midpoint of 63

as the age cutoff in defining retirement in our analyses above. As a robustness test, we perform the same analyses

using 60 and 65 as two alternative age cutoffs and find qualitatively similar results.

Page 28: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

26

We next use multivariate probit regressions to further examine the incremental turnover

of NPEs after controlling for factors that could influence turnover using the following model:

Pr (Turnover=1) = β0 + β1 COO promotion + β2 Strong Competitor + β3 COO promotion*

Strong Competitor + β4 Retire_age + β5 AROA + β6 Size + β7 Leverage + β8 Market-to-book +

β9 Stock volatility + e (2)

The dependent variables are overall turnover and pre-retirement turnover as defined

earlier. Similar to our earlier multivariate test of H1, we use two model specifications and we

cluster standard errors at the individual firm level. The first model tests H2 based on β1, the

coefficient of COO promotion, which represents the incremental difference in turnover of NPEs

as compared to control VPs. We expect β1 to be positive because we expect a higher rate of

turnover for NPEs than for the control VPs.

As shown in Columns 1 and 3 of Table 6, both coefficients of COO promotion are

significantly positive (β1 = 0.070 and 0.053, respectively; p < 0.01) after controlling for firm-

specific factors such as firm performance and size. Consistent with H2, these results indicate that

NPEs are 31.1% (7.0%/22.5%) more likely than control VPs to leave the firm after being passed

over for promotion to COO, and NPEs are 29.4% (5.3%/18.0%) more likely than control VPs to

leave the firm prior to retirement age, where the base turnover rates of 22.5% and 18.0% are

from Table 5, Panel A.

[Table 6]

Our second model includes the variables Strong Competitor and COO promotion*Strong

Competitor, to analyze the incremental differences in turnover of strong competitor NPEs versus

weak competitor NPEs, after controlling for the baseline differences in turnover across similar

VPs in the population. We expect β3 to be positive due to the greater loss in implicit incentives

Page 29: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

27

for strong competitor NPEs.

As shown in Columns 2 and 4, the coefficients of COO promotion*Strong Competitor are

significantly positive in the overall turnover and pre-retirement turnover regressions (β3 = 0.130

and 0.109, respectively; p < 0.01). Consistent with H2a, these results indicate that after

controlling for the baseline turnover rates of control VPs, strong competitor NPEs are 13.0%

(10.9%) more likely to leave the firm (leave the firm before retirement age) than weak

competitor NPEs after being passed over for promotion.

We note that the coefficients of COO promotion (β1) in regressions (2) and (4) of Table 6

are not significantly different from zero, indicating that NPEs who are weak competitors are not

more likely to leave the firm than similar control VPs, providing additional support for H2a.25

Overall, our results indicate that NPEs are more likely to leave the firm following the end

of a COO tournament than they would be in the absence of a COO promotion tournament.

Further, this increased turnover rate is greater for strong competitor NPEs, whom we expect to

suffer the greatest reduction in incentives as a result of being passed over for promotion, as

compared to weak competitor NPEs. Collectively, these results support H2 and H2a.

4.3 Firm performance: Tests of H3 and H3a

The null form H3 hypothesis predicts that firm performance will not change after a COO

tournament ends, reflecting our expectation that the positive influence of the newly promoted

COO will be offset by the expected negative influence of the NPEs, either because the NPEs

leave the firm or they stay but exert less effort. To test H3, we calculate changes in accounting

25 In addition, we find that the coefficient of Strong Competitor (β2) is significantly negative for both the overall

turnover and pre-retirement turnover models. This result indicates that in the control sample, high salary-ranked VPs

are less likely to leave the firm than low salary-ranked VPs. One factor that could contribute to this result is that

executives higher in the management hierarchy are typically older, have longer tenure, and thus are more likely to

stay in the firm until retirement.

Page 30: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

28

performance for sample firms, measured as the return on assets (ROA) adjusted by the Fama-

French 48-industry median ROA. We measure pre-promotion firm performance as the firm’s

industry adjusted ROA in the year prior to COO promotion and post-promotion firm

performance as the firm’s industry-adjusted ROA in each of the three years starting from the

year of COO promotion. We examine performance changes within a relatively short three-year

window after the promotion event to limit performance effects caused by events other than the

promotion in the subsequent years, such as replacement of the CEO. In conducting this test, we

drop 105 treatment firms and 560 control firms that already had a COO in the year prior to COO

promotion. We drop the 105 treatment firms because they promoted a VP to COO to replace an

existing COO and the change in performance in these firms could be unduly influenced by the

previous actions of the replaced COO rather than by the actions of the new COO or NPEs.26 We

drop the 560 control firms with an existing COO because the COO promotion tournament has

ended in these firms and we might capture end-of-tournament performance effects if they are

included in our analysis. After applying these procedures, our final sample consists of 739

treatment firms and 1,471 control firms.

To test H3 and H3a, we first classify our treatment firms as having either a strong or

weak COO tournament based on the number of “strong” competitors in the COO tournament.

We adapt procedures from prior literature to identify a subsample of strong competitors based on

their pay rank, as described earlier, as well as their age.27 We must apply more stringent criteria

26 Similar to CEO turnover, the decision by the firm to replace an existing COO might be caused by poor COO

performance. In our sample, most COOs who are replaced subsequently leave the firm. It is unclear how the

departure of a replaced COO affects firm performance. 27 We must use multiple criteria to define strong competitors in order to generate a sample of treatment firms with

strong and weak tournaments. Because we constructed our sample earlier to include at least two strong competitors

for promotion to COO, by construction all treatment firms must have at least one strong competitor who is passed

over for promotion. Thus, we use age to further differentiate between strong and weak competitors to define firms

with strong versus weak tournaments. Prior studies use similar characteristics to identify an heir apparent to the

CEO position. For example, Naveen (2006) defines an heir apparent as a VP who is younger than the CEO and

Page 31: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

29

in classifying “strong competitors” to test H3a than in our earlier test of H2a because our H2a

test is at the individual VP level while our H3a test is at the firm level. Because we expect a

strong competitor for the COO and eventually the CEO position to be young enough to be able to

serve in both roles for a number of years prior to retirement, we require the strong competitor to

be relatively young. Given the median of two years between COO promotion and CEO

succession in our sample firms as reported earlier and the third quartile of CEO appointment age

of 55 for all ExecuComp firms over our sample period, we consider VPs younger than 53 to be

strong competitors for the COO position. For the 19% of cases in which VP age data are missing,

we assume that such VPs are sufficiently young to be considered as strong competitors.28

Therefore, we classify a VP as a “strong competitor” in a COO tournament if in the year

prior to promotion, the VP’s salary ranks first or second among all VPs and the VP is younger

than 53. NPEs and control VPs who do not meet both criteria are classified as weak competitors.

In addition, we classify any VP who is promoted to COO as a strong competitor for COO

promotion even if they fail to meet the pay rank and age criteria.29 We then define a COO

tournament as a “weak tournament” if we identify only one strong competitor in the year prior to

the year of COO promotion or control year, and a “strong tournament” when there are at least

two strong competitors for the COO promotion.

Among the 2,185 sample NPEs competing in 739 COO promotion tournaments, we

receives at least 10% more pay than the next highest-paid VP. Kale et al. (2009) add the criterion that the heir

apparent holds the title of either President or COO but is not the Chair of the Board. We use a slightly different set

of criteria because we seek to define a COO competitor rather than a CEO successor. 28 Among observations in our sample with available age data, the median VP age is 53. If VPs without age data have

a similar distribution in age, then about 50% of them are misclassified as being young enough for the COO

promotion when they are actually not. However, we note that any misclassification from our coding of the missing

data should work against our finding significant results. Our results are qualitatively the same when we exclude

firms with missing age data. 29 Excluding the 158 treatment firms that promoted a VP with low pay-rank to COO yields qualitatively similar

results.

Page 32: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

30

classify 544 NPEs as strong competitors and the remaining 1,641 NPEs as weak competitors.

Among the 739 COO promotions, we classify 485 as strong tournaments and 254 as weak

tournaments. To test our hypotheses, we apply the same criteria to create subsamples within our

control sample for comparative purposes. Among the 5,710 control VPs competing in 1,471

control tournaments, we classify 1,789 VPs as strong competitors and 3,921 VPs as weak

competitors. Among the 1,471 control firms, we classify 892 as strong tournaments and 579 as

weak tournaments.

To test H3 and H3a, we conduct multivariate regressions to test the change in firm

performance after controlling for other factors that could influence firm performance using the

following model:

ΔAROA = β0 + β1 COO_promotion + β2 Strong_Tournament + β3 COO_promotion*

Strong_Tournament + β4 Size + β5 Leverage + β6 Market-to-book + β7 Stock volatility + Year +

Industry + e (3)

The dependent variable is the firm’s change in performance, measured as the change in

industry-adjusted ROA from the year before COO promotion to each of the three years starting

from the year of COO promotion for the treatment sample or the control year for the control

sample. Strong_Tournament is coded as one for COO promotion firms with a strong tournament

as defined earlier, and zero otherwise. We control for firm size, leverage, market to book, stock

volatility, year fixed effect and industry fixed effect. We cluster standard errors at industry level.

Results for the first regression model in Columns 1, 3, and 5 of Table 7 show that the

coefficients of COO_promotion are not significantly different from zero (β1 = 0.0036, 0.0010,

and -0.0004; p=0.301, 0.804, and 0.934, respectively). These results suggest that, on average, the

performance changes for firms before versus after COO promotion are not different from the

Page 33: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

31

performance changes for control firms. This result is consistent with our null H3 which predicts

that the subsequent positive influence of the promoted COO will offset any subsequent decline in

performance resulting from the NPEs’ actions. However, this model does not account for

differences in tournament strength.

[Table 7]

Results in Columns 2, 4, and 6 of Table 7 use the full model from equation (3) to test

H3a. We expect a negative coefficient of COO_promotion*Strong_Tournament, which

represents the incremental change in firm performance before and after COO promotion for

firms with a strong tournament versus those with a weak tournament.

We first describe the results in Columns 2 and 4 and later discuss the results in Column 6.

Consistent with H3a, we find that the coefficient of COO_promotion* Strong_Tournament is

negative and significant in Columns 2 and 4 (β3 = -0.014 and -0.020; p=0.033 and 0.007,

respectively) when the dependent variable is the performance change from one year prior to

COO promotion to the year of promotion and to one year after promotion, respectively. This

result supports the H3a prediction that the change in firm performance at the end of a promotion

tournament is more negative in strong COO tournaments than in weak COO tournaments.

In addition, we find that the coefficient for COO_promotion is positive and significant in

both Columns 2 and 4 (β1 = 0.0128 and 0.0137), suggesting a net positive change in firm

performance following the end of a weak COO promotion tournament. Again, this improvement

in performance is consistent with positive results of actions taken by the promoted COO who has

stronger incentives following the promotion. We also find the coefficient for Strong_Tournament

is positive and significant in Columns 2 and 4 (β2 = 0.0061 and 0.0098), which is consistent with

stronger tournaments having a positive effect on firm performance, consistent with the results in

Page 34: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

32

Kale et al. (2009).30

In Column 6, where the dependent variable is the performance change from one year

prior to COO promotion to two years after the promotion, none of the corresponding coefficients

(β1 = 0.0026, β2 = -0.0008, β3 = -0.004) are significantly different from zero. There are several

possible explanations for this result. First, any negative changes in firm performance after the

COO promotion as a result of NPEs leaving the firm or staying in the firm but exerting less effort

are likely to be temporary. We expect firms to remediate the negative actions of the NPEs over

time. For example, firms will take actions to train or to hire personnel to replace the NPEs who

leave the firm as a result of losing the COO tournament. Second, other events that occur

subsequent to the COO promotion may offset the negative performance changes. For example,

our descriptive statistics indicate that one-third of new COOs are promoted to CEO within two

years of the COO promotion. Such CEO succession events could have a positive effect on firm

performance (Denis and Denis, 1995; Huson et al., 2004).

We next perform additional analyses to shed more light on whether the more negative

performance changes in firms with strong tournaments as compared to those with weak

tournaments is driven by the loss in human capital when NPEs leave the firm, or reduced effort

by NPEs who stay with the firm and face lower incentives, or both. We divide our subsample of

firms with strong promotion tournaments into two groups according to whether at least one

strong competitor leaves the firm by the year following COO promotion, which we code as

Strong_Tour_Leave equal to 1 versus all strong competitors remain with the firm, which we code

as Strong_Tour_Stay equal to 1.We include these two subgroup variables and their interactions

30 When we net the positive performance effect from having a strong tournament (β2) with the negative end-of-

tournament performance effect from strong tournament (β3), the total effect has a negative sign in both Columns (2)

and (4) and is significantly smaller than zero at p=0.057 in Column (4), where the change in AROA from the year

prior to promotion to the year after promotion is the dependent variable.

Page 35: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

33

with COO_promotion in the same multivariate regression model from Equation (3). We again

analyze the changes in firm performance from the year before COO promotion to each of the

three years after.

[Table 8]

Table 8 shows that when we first consider the results in Columns 1-2, the coefficients for

both COO_promotion * Strong_Tour_Stay (β4) and COO_promotion * Strong_Tour_Leave (β5)

are negative and at least marginally significant (β4 = -0.016 and -0.012; p=0.092 and 0.069; β5 =

-0.013 and -0.032, p=0.08 and 0.01). These results are consistent with our argument for H3a that

regardless of whether NPEs competing in strong tournaments choose to leave the firm or stay

and face reduced incentives, their actions will have an adverse effect on firm performance.

Similar to our earlier results in Table 7, the results in column (3) of Table 8 reveal no

significantly negative performance effects in strong tournament firms two years after the end of

the COO promotion tournament.

5. Additional Analysis and Robustness Tests

This section analyzes three alternative explanations for key components of our results.

Specifically, we first consider whether the increased NPE turnover following tournaments could

be explained by poor NPE performance rather than by the reduction in implicit incentives when

the tournament ends, as we contend. Our second and third analyses address alternative

explanations for the decline in firm performance after the tournament ends. One alternative

explanation is that the appointment of the COO and the subsequent decline in firm performance

both stem from the same initial unfavorable circumstances facing the firm. The final alternative

explanation for the decline in firm performance is that executives competing in the tournament

engaged in earnings management to inflate earnings, with the result that the firm subsequently

Page 36: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

34

experienced reduced reported performance when the earnings management effects reversed.

5.1 Future career of departed NPEs

We first provide further evidence concerning whether the higher turnover of strong

competitor NPEs following the end of the COO tournament as compared to similar control VPs

is the result of a reduction in their promotion-based implicit incentives rather than the NPEs’

poor performance. If the NPEs are leaving voluntarily because of a reduction in promotion-based

incentives as we propose, we expect the NPEs to find positions at other firms that are

comparable to the positions that control VPs find when they exit their firms. Alternatively, if the

NPEs are leaving involuntarily because of their poor performance, we expect the NPEs to find

inferior positions after leaving the firm.

We begin with a sample of 393 NPEs and 1,053 control VPs who leave their firms by the

second year following the year of COO promotion or control year (i.e., year t +2) and are

younger than 63 when they leave to isolate pre-retirement turnover. For these VPs, we search the

Execucomp database to determine whether they subsequently work for other S&P 1,500 firms as

a top-five executive, and if so, whether they become a CEO or COO. We consider the

opportunity to work for another S&P 1,500 firm at the executive level and become the firm’s

CEO/COO to be good future career outcomes. Based on our reasoning above, we expect a

similar or higher percentage of departed NPEs to have these career outcomes as compared to

departed control VPs.

Table 9 shows that among the 393 NPEs who leave their firm after the COO promotion,

18.6% (10.7%) subsequently work for another S&P 1,500 firm as a top-five executive (CEO or

COO). For the 1,053 control VPs, the corresponding percentage is 12.4% (6.0%), which is

Page 37: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

35

significantly smaller (p < 0.01).31 Table 9 next divides the full sample into two subsamples

according to whether the VP is classified as a strong or weak competitor based on their salary

rank as defined earlier. The results show that 21.7% (13.9%) of departed strong competitor NPEs

subsequently work for another S&P 1,500 firm as a top-five executive (a CEO or COO), and

these percentages are significantly greater (p < 0.05) than the corresponding percentages of

15.1% (7.6%) for strong competitor control VPs with the same career outcomes. Likewise, the

percentage of departed weak competitor NPEs who subsequently work for another S&P 1,500

firm as a top-five manager (15.6%) is also greater than the corresponding percentages of similar

control VPs (10.0%), although the percentage who become the firm’s CEO or COO (7.5% versus

4.5%) is not significantly different at conventional levels.

[Table 9]

Overall, these results provide further support for our theory that NPEs, especially those

likely to be strong competitors for the COO position, leave the firm voluntarily after being

passed over for promotion due to a reduction in their promotion-based incentives and

subsequently obtain superior job positions as compared to similar VPs who leave but do not

experience a reduction in their promotion-based incentives. In addition, our results suggest that

while the NPEs are deemed to be less capable than the promoted COO, these tournament losers

are highly valued by the external labor market and thus are unlikely to have been viewed as

“corporate deadwood” whom the firms would prefer to exit the firm.

5.2 Endogenous COO promotion decision

31 These percentages reflect the fact that departing NPEs move next to a wide variety of positions. Other than

working as a top-five executive in another S&P 1,500 firm, a departed VP could retire, or work in a private firm, a

smaller public firm, or another S&P 1,500 firm at a rank lower than top-five. Fee and Hadlock (2004) performed a

comprehensive search of news articles to track the subsequent careers of departed executives at public firms covered

by Execucomp and could identify the future careers of 16.3% (26.8%) of all executives in their sample of departed

executives (a subsample of departed executives younger than 60).

Page 38: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

36

We next address the alternative explanation that the documented differences in VP

turnover and changes in firm performance between our treatment and control samples are due to

endogenous factors related to sample firms’ decision to establish a COO position and initiate the

CEO succession process. For example, a firm that is facing a decline in performance might

decide to promote a VP to COO and eventually to CEO in order to improve firm performance

and appease its shareholders. This alternative explanation suggests that the firm’s poor

performance before the COO promotion drives the higher VP turnover and lower firm

performance after the COO promotion.

We argue that our results are unlikely to be driven by such factors related to the firm's

COO promotion decision because they cannot explain the results from our cross-sectional tests

presented earlier. Recall that we find higher NPE turnover for strong competitors but not for

weak competitors. If our results are driven by poor firm performance, turnover should be high

for all NPEs, including the weak competitor, which is not the case. In addition, we find evidence

of negative firm performance for strong COO tournaments but not for weak COO tournaments.

Again, a firm’s decision to promote a VP to COO is unlikely to be simultaneously related to poor

pre-promotion firm performance and the strength of the COO tournament (i.e., number of likely

competitors in the tournament).32

We next formally test whether there is a difference in pre-promotion firm performance

between our treatment and control samples. First, Columns 1 and 2 of Table 10 compare the level

of industry-adjusted ROA between our treatment and control firms in the year prior to COO

promotion (t-1), after controlling for various firm characteristics. We find no significant

32 Another reason that firm performance could be weaker for firms with strong versus weak tournaments is the

greater uncertainty about who will become the new COO in the case of a strong tournament, which could delay the

new COO’s ability to exert influence to improve performance. In contrast, in the case of a weak tournament in

which the new COO’s identity is generally recognized sooner, the new COO can exert influence sooner.

Page 39: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

37

differences in the pre-promotion AROA between our treatment sample and control sample

overall based on β1 in Column 1. Results are similar in Column 2 for firms with weak

tournaments, based on β1 and for firms with strong tournaments, based on the results for β1+β3

reported at the bottom of Column 2. We also find no significant difference in pre-promotion

AROA within our treatment sample for those with weak tournaments versus those with strong

tournaments (results for β2+β3, bottom of column 2).

[Table 10]

Next, in Columns 3 and 4 we perform the same regressions using the change in each

firm's performance from two years prior to COO promotion (t-2) to the year prior to COO

promotion (t-1) as the dependent variable. Again, we find no significant difference in the pre-

promotion change in AROA between our treatment and control firms for both strong and weak

tournaments. Overall, these results indicate that weaker or declining pre-promotion firm

performance cannot explain our findings.

5.3 Pre-promotion earnings management

A third alternative explanation for our finding that firm performance declines at the end

of a COO tournament is that competitors for the COO position, anticipating the upcoming COO

promotion, manage earnings up during the pre-promotion period to increase their likelihood of

winning the tournament. When the tournament ends, this temporary increase in firm performance

reverses, resulting in a decline in performance.

To address this alternative explanation, we compare the level of earnings management

between our treatment and control samples in the year prior to COO promotion. Consistent with

prior research, we use discretionary accruals as our proxy for earnings management. Specifically,

we estimate non-discretionary accruals using the modified Jones model and include ROA in the

Page 40: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

38

prior year as a regressor to control for the effects of performance on discretionary accruals

(Kothari et al., 2005). We then regress signed discretionary accruals on our sample classification

variables, controlling for various firm characteristics and CEO equity incentives. The regression

results in Table 11 provide little support for greater accrual-based earnings management in

treatment firms with strong tournaments than in those with weak tournaments or in control firms.

In addition, in untabulated results, we rerun the same firm performance regression analyses to

test H3 described earlier in Table 7 after controlling for the firm’s discretionary accruals in the

year prior to COO promotion and find qualitatively the same results. Overall, these results

suggest that pre-promotion earnings management is unlikely to explain our findings.

[Table 11]

6. Conclusion and Discussion

This study examines how the actions of both promoted and non-promoted executives

(NPEs) can affect firms following a COO promotion tournament. We focus on the influence of

NPEs and find little evidence that firms increase the explicit incentives of NPEs to replace the

significant decline in their implicit promotion-based incentives, even for those who are likely to

have previously played key executive roles and to have been strong competitors for the COO

position. In response, almost 35% of NPEs who are strong competitors for the COO position

choose to leave the firm after they are passed over for promotion. This turnover rate is

significantly greater than that for a control sample of firms in which no COO promotion took

place during our sample period. Further, we find evidence that the change in firm performance

following the tournament is more negative for firms with a strong tournament characterized by

having multiple strong competitors for the COO promotion, than for firms with a weak

tournament. This worse firm performance following the end of a strong COO tournament occurs

Page 41: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

39

whether NPEs leave the firm or stay with the firm after being passed over for promotion. This

result suggests that while a strong tournament with multiple strong competitors for the COO

promotion offers the advantage of strengthened incentives for competitors prior to the

tournament, such strong tournaments also appear to generate more negative consequences when

the tournament ends.

Overall, our results advance our understanding of how dynamic changes in employees’

implicit and explicit incentive affect their firms. Specifically, we provide evidence of higher

executive turnover and lower firm performance following the end of a COO tournament event.

We attribute this result to the firms’ inability to adjust NPEs’ incentives to retain key executives

with appropriate incentives. Our results highlight the need to motivate NPEs who add value to

their firms despite not being selected for promotion. Although it is costly for firms to adjust

NPEs’ incentives, it could be more costly to allow NPEs to leave the firm or to allow them

remain in the firm with weaker incentives.

Our results are subject to the important limitation that we cannot observe board of

director proceedings that would provide more direct evidence concerning whether a firm uses a

COO tournament and whether specific NPEs are competitors for the COO position. Therefore,

our results could be influenced by misclassifications of the data. We attempt to mitigate this

concern in several ways. First, we use actual descriptive data of promoted COO characteristics to

identify the VPs who are most likely to have been primary competitors for the COO position.

Second, we rely on prior research to guide our set of criteria of competitor NPEs. Third, we

perform various robustness checks and cross-sectional analyses to provide corroborative

evidence for our hypotheses tests. Finally, as far as we know, misclassification should operate

against our finding significant results.

Page 42: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

40

References

Baik, B., Evans, J.H., Kim, K., Yanadori, Y., 2016. White collar incentives. Accounting,

Organizations and Society 53, 34-49.

Baker, G., Gibbs, M., Holmstrom, B., 1994. The internal economics of the firm: evidence from

personnel data. The Quarterly Journal of Economics 109, 881-919.

Bass, B.M., Stogdill, R.M., 1990. Handbook of leadership (Vol. 11). New York: Free Press.

Bognanno, M.L., 2001. Corporate tournaments. Journal of Labor Economics 19, 290-315.

Bushman, R.M., Dai, Z., Zhang, W., 2015. Management Team Incentive: Dispersion and Firm

Performance. The Accounting Review 91, 21-45.

Campbell, D., 2008. Nonfinancial Performance Measures and Promotion‐Based

Incentives. Journal of Accounting Research 46, 297-332.

Cannella, A.A., Shen, W., 2001. So close and yet so far: Promotion versus exit for CEO heirs

apparent. Academy of Management Journal 44, 252-270.

Chen, K.P., 2003. Sabotage in promotion tournaments. Journal of Law, Economics, and

Organization 19, 119-140.

Cichello, M.S., Fee, C.E., Hadlock, C.J., Sonti, R., 2009. Promotions, turnover, and performance

evaluation: Evidence from the careers of division managers. The Accounting Review 84,

1119-1143.

Core, J., Guay, W., 1999. The use of equity grants to manage optimal equity incentive

levels. Journal of Accounting and Economics 28, 151-184.

Core, J., Guay, W., 2002. Estimating the value of employee stock option portfolios and their

sensitivities to price and volatility. Journal of Accounting Research 40, 613-630.

Core, J.E., Guay, W.R., Larcker, D.F., 2003. Executive equity compensation and incentives: A

survey. Economic Policy Review 9, 27-50.

Denis, D.J., Denis, D.K., 1995. Performance changes following top management dismissals. The

Journal of Finance 50, 1029-1057.

De Varo, J., 2006. Internal promotion competitions in firms. RAND Journal of Economics 37,

521-542.

Ederhof, M., 2011. Incentive compensation and promotion-based incentives of mid-level

managers: Evidence from a multinational corporation. The Accounting Review 86, 131-

153.

Ertimur, Y., Ferri, F., Oesch, D., 2013. Shareholder votes and proxy advisors: Evidence from say

Page 43: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

41

on pay. Journal of Accounting Research 51, 951-996.

Fee, C.E., Hadlock, C.J., 2004. Management turnover across the corporate hierarchy. Journal of

Accounting and Economics 37, 3-38.

Gibbons, R., Murphy, K.J., 1992. Optimal Incentive Contracts in the Presence of Career

Concerns: Theory and Evidence. Journal of Political Economy 100, 468-505.

Gibbs, M., 1994. Testing tournaments? An appraisal of the theory and evidence. Labor Law

Journal 45, 493-500.

Gibbs, M., 1995. Incentive compensation in a corporate hierarchy. Journal of Accounting and

Economics 19, 247-277.

Gibbs, M., 1996. Promotions and incentives. Unpublished working paper, University of Chicago.

Gomez-Mejia, L.R., Balkin, D.B., 1992. Compensation, organizational strategy, and firm

performance. South-Western Pub.

Hambrick, D.C., Cannella, A.A., 2004. CEOs who have COOs: Contingency analysis of an

unexplored structural form. Strategic Management Journal 25, 959-979.

Harbring, C., Irlenbusch, B., 2011. Sabotage in tournaments: Evidence from a laboratory

experiment. Management Science 57, 611-627.

Hayes, R.M., Schaefer, S., 1999. How much are differences in managerial ability worth?. Journal

of Accounting and Economics 27, 125-148.

Holmstrom, B., Milgrom, P., 1994. The firm as an incentive system. The American Economic

Review 84, 972-991.

Huson, M.R., Malatesta, P.H., Parrino, R., 2004. Managerial succession and firm

performance. Journal of Financial Economics 74, 237-275.

Ittner, C.D., Lambert, R.A., Larcker, D.F., 2003. The structure and performance consequences of

equity grants to employees of new economy firms. Journal of Accounting and

Economics 34, 89-127.

Kale, J.R., Reis, E., Venkateswaran, A., 2009. Rank‐order tournaments and incentive alignment:

The effect on firm performance. The Journal of Finance 64, 1479-1512.

Kaarbøe, O.M., Olsen, T.E., 2006. Career concerns, monetary incentives and job design. The

Scandinavian Journal of Economics 108, 299-316.

Kothari, S.P., Leone, A.J., Wasley, C.E., 2005. Performance matched discretionary accrual

measures. Journal of Accounting and Economics 39, 163-197.

Lazear, E.P., Rosen, S., 1981. Rank-Order Tournaments as Optimum Labor Contracts. Journal of

Political Economy 89, 841-64.

Page 44: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

42

Main, B.G., O'Reilly III, C.A., Wade, J., 1993. Top executive pay: Tournament or

teamwork?. Journal of Labor Economics 11, 606-628.

McConnell, C., Brue, S., 1992. Critiques of Orthodox Wage Theory. Contemporary Labor

Economics, 480-493.

Morck, R., Shleifer, A., Vishny, R.W., 1988. Management ownership and market valuation: An

empirical analysis. Journal of Financial Economics 20, 293-315.

Murphy, K.J., 2013. Executive Compensation: Where We Are, and How We Got

There. Handbook of the Economics of Finance 2, 211-356.

Nalebuff, B.J., Stiglitz, J.E., 1983. Prices and Incentives: Towards a General Theory of

Compensation and Competition. Bell Journal of Economics 14, 21-43.

Naveen, L., 2006. Organizational Complexity and Succession Planning. Journal of Financial and

Quantitative Analysis 41, 661-683.

O'Reilly III, C.A., Main, B.G., Crystal, G.S., 1988. CEO compensation as tournament and social

comparison: A tale of two theories. Administrative Science Quarterly 33, 257-274.

Shen, W., Cannella, A.A., 2002. Revisiting the performance consequences of CEO succession:

The impacts of successor type, postsuccession senior executive turnover, and departing

CEO tenure. Academy of Management Journal 45, 717-733.

Shen, W., Cannella, A.A., 2003. Will succession planning increase shareholder wealth?

Evidence from investor reactions to relay CEO successions. Strategic Management

Journal 24, 191-198.

Vancil, R.F., 1987. Passing the baton: Managing the process of CEO succession. Harvard

Business School Press.

Page 45: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

43

Appendix. Variable Definitions

Variables related to the classification of NPEs, control VPs, and firms NPE = Non-promoted Executive is a non-COO VP who is a competitor in a tournament for

promotion to COO but does not receive the promotion.

Competitor = 1 if the NPE or VP’s salary in the year prior to COO promotion year or control year is at

least second among all non-COO VPs, = 0 otherwise.

COO promotion = 1 for all NPEs; = 0 for all control VPs.

Strong_Tournament For a COO promotion firm, =1 if the firm has at least one NPE who is classified as a

competitor and whose age is younger than 53, = 0 otherwise. For a control firm, = 0 if

the firm has only one NPE who is classified as a competitor and whose age is younger

than 53, = 1 otherwise.

Dependent variables Turnover = 1 if the NPE (control VP) is not listed as the firm’s executive by ExecuComp in the year

of or the year following COO promotion year (control year); = 0 if the NPE (control

VP) continues to be listed as the firm’s executive by ExecuComp in the year of and the

year following COO promotion year (control year).

Retirement = 1 if Turnover = 1 and the age of the NPE (control VP) in the year prior to COO

promotion (control year) is non-missing and at least 63; Missing if Turnover =1 and the

age of the NPE (control VP) in the year prior to COO promotion (control year) is

missing; = 0 otherwise.

Pre-retirement

turnover

= 1 if Turnover = 1 and the age of the NPE (control VP) in the year prior to COO

promotion (control year) is non-missing and smaller than 63; Missing if Turnover =1

and the age of the NPE (control VP) in the year prior to COO promotion (control year)

is missing; = 0 otherwise.

ST_pay = Salary + bonus + other annual payments if the firm reports executive compensation in

the pre-2006 formula; = salary + bonus + non-equity incentive plan if the firm reports

executive compensation in the post-2006 formula.

LT_pay = Restricted stock grants + option grants + long-term incentive payouts + other annual

payments if the firm reports executive compensation in the pre-2006 formula; = stock

awards + option awards + total portion of deferred earnings reported as compensation +

other compensation if the firm reports executive compensation in the post-2006

formula.

Total_pay ST_Pay + LT_pay.

ROA Net income before special items divided by the average of the firm’s total assets at the

beginning and the end of the year.

AROA ROA – median ROA of all firms in the same fiscal year in the same industry based on

Fama-French 48 industry classification.

Control variables used in regressions Retire_age = 1 if the age of the NPE (control VP) in the year prior to COO promotion year (control

year) is non-missing and at least 63; = 0 otherwise.

Size

Leverage

Natural log of the firm’s total assets.

= Long-term Liability/Total Assets.

Market-to-book Market value of the firm’s common stock divided by book value of equity.

Stock volatility Standard deviation of the company’s monthly stock returns over the past 60 months, then

converted to annual volatility.

Page 46: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

44

Table 1. Selection of COO promotion sample Panel A reports our sample selection process. Starting from 1,601 new COOs over the period from 1993 to 2013, we screen and identify 1,186 COO promotions

where the COO is a strong candidate to be the current CEO’s successor. After placing additional requirements to remove COO promotions with

contemporaneous events, our final COO promotion sample includes 845 new COOs. Panel B provides descriptive statistics on the future career of the 1,186 new

COOs in the same firm subsequent to their promotion to COO.

Panel A. Sample selection process

Promotions from a non-COO and non-President position to COO and/or President position 1,601

Less: new COOs older than the CEO or whose age is missing, unless subsequently promoted to CEO (354)

Less: new COOs whose ranks in salary and total pay are both below second among all VPs,

unless subsequently promoted to CEO (61)

COO promotions in which the new COO is a strong candidate to be the next CEO 1,186

Less: cases in which the firm changes its CEO in same year or the year following COO promotion (318)

Less: new COOs who leave the firm in the year following promotion (23)

Final COO promotion sample 845

Panel B. The new COO’s subsequent career path in the firm

Number

Percent of

trackable observations

Percent of

subsample

All new COOs who are designated CEO successors 1,186

Less: COOs in firms where no subsequent CEO appointment is observed (305)

COOs whose future CEO career in the firm can be tracked 881 100%

Part 1. the new COO becomes the next CEO 570 64.7% 100%

CEO promotion occurs at t+1 135 23.7%

CEO promotion occurs at t+2 185 32.5%

CEO promotion occurs at t+3 103 18.1%

CEO promotion occurs later than t+3 147 25.8%

Part 2. the new COO does not become the next CEO 311 35.3%

Page 47: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

45

Table 2. Description of COO promotion sample Our final COO-promotion sample includes 845 COO promotions from 1993 to 2013. Panels A to D report the

distribution of this sample by year (Panel A), the promoted COO’s title (Panel B), age (Panel C), and pay rank

among all VPs (Panel D) in the year prior to promotion. Panel E reports the mean and median pay levels for the VP

promoted to COO in the year prior to and the year of COO promotions. For Panel E, tests on changes in mean are

based on t-test of change=0. Tests on changes in median are based on Wilcoxon two-sample (before versus after

promotion) tests. *, ** and *** denote statistical significance for the change in pay at the 10%, 5%, and 1% levels,

respectively, for two-tailed tests.

Panel A. Sample distribution across years

Year Number Percentage

1993 15 1.78%

1994 36 4.26%

1995 38 4.50%

1996 58 6.86%

1997 44 5.21%

1998 45 5.33%

1999 48 5.68%

2000 38 4.50%

2001 39 4.62%

2002 46 5.44%

2003 34 4.02%

2004 35 4.14%

2005 52 6.15%

2006 46 5.44%

2007 51 6.04%

2008 50 5.92%

2009 48 5.68%

2010 34 4.02%

2011 36 4.26%

2012 39 4.62%

2013 13 1.54%

Total 845 100%

Panel B. COO’s title prior to promotion to COO

Title Number Percentage

Operations function 639 75.6%

CFO 164 19.4%

General Counsel 13 1.5%

CTO 18 2.1%

CAO 21 2.5%

Total 845 100%

Page 48: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

46

Table 2 (continued)

Panel C. COO's age prior to promotion to COO

Age Number Percent of all obs.

40 or younger 87 10.3%

41 – 45 170 20.1%

46 – 50 256 30.3%

51 – 55 232 27.5%

56 – 60 86 10.2%

60 or older 13 1.5%

Missing 1 0.1%

Panel D. COO's pay rank among all VPs in year prior to promotion to COO Rank Salary Total pay

1 432 51.1% 430 50.9%

2 226 26.7% 218 25.8%

3 109 12.9% 131 15.5%

4 or lower 78 9.2% 56 6.6%

Panel E. COO’s pay levels prior to and after promotion to COO

The year prior

to COO promotion

Year of COO

promotion Change

New COO ($ in thousands)

Mean salary (845 obs.) 367 412 45 ***

Median salary (845 obs.) 342 382 33 ***

Mean total pay (835 obs.) 1,983 2,740 756 ***

Median total pay (835 obs.) 1,267 1,576 238 ***

Page 49: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

47

Table 3. Changes in VP pay, comparisons across groups This table reports the mean and median changes in annual pay for NPEs (control VPs) before and after the COO

promotion year. Definitions of short-term pay, long-term pay, and total pay are in the appendix. The sample consists

of 1,665 NPEs and 5,481 control VPs over the 1993-2013 period with pay data available before and after the COO

promotion year or control year. Tests on differences in median % changes are based on Wilcoxon two-sample (NPEs

versus Control VPs) tests. *, ** and *** denote statistical significance at the 10%, 5%, and 1% levels, respectively,

for two-tailed tests.

Panel A. All NPEs versus Control VPs

A. NPEs

(1,665 obs.)

B. Control VPs

(5,481 obs.)

Difference

(A-B)

Mean % change short-term pay 10.1% 10.0% Median % change short-term pay 5.7% 4.9% 0.8%

Mean % change long-term pay 91.2% 66.0%

Median % change long-term pay 12.7% 12.0% 0.7%

Mean % change total pay 24.3% 20.8%

Median % change total pay 8.3% 8.5% -0.2%

Panel B. Strong Competitor NPEs versus Control VPs

A. NPEs

(758 obs.)

B. Control VPs

(3,222 obs.)

Difference

(A-B)

Mean % change short-term pay 8.4% 8.2% Median % change short-term pay 4.6% 3.5% 1.1%

Mean % change long-term pay 88.9% 68.1%

Median % change long-term pay 9.8% 10.1% -0.3%

Mean % change total pay 20.1% 19.2%

Median % change total pay 5.7% 6.8% -1.1%

Panel C. Weak Competitor NPEs versus Control VPs

A. NPEs

(867 obs.)

B. Control VPs

(2,079 obs.)

Difference

(A-B)

Mean % change short-term pay 11.5% 12.5% Median % change short-term pay 6.6% 6.7% 0.0%

Mean % change long-term pay 93.1% 64.0%

Median % change long-term pay 14.5% 14.8% -0.3%

Mean % change total pay 27.8% 22.8%

Median % change total pay 10.7% 10.7% 0.1%

Page 50: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

48

Table 4. Firms’ adjustments to VP pay, multivariate regressions This table reports the results of regressions where the dependent variable is the natural log of the VP’s post-

promotion short-term pay (Columns 1 and 4), long-term pay (Columns 2 and 5) and total pay (Columns 3 and 6).

Definitions for pay variables and all independent variables are in the appendix. The sample consists of all NPEs and

control VPs whose firms’ control variables are non-missing in CompuStat and CRSP and who stay with the firm at

least through the year of COO promotion or the control year. Standard errors are clustered at the industry level. P-

values are shown in parentheses. *, ** and *** denote statistical significance at the 10%, 5%, and 1% levels,

respectively, for two-tailed tests.

Log (ST

pay) post

Log (LT

pay) post

Log (Total

pay) post

Log (ST

pay) post

Log (LT

pay) post

Log (Total

pay) post

(1) (2) (3) (4) (5) (6)

COO promotion (β1) -0.013 0.012 0.001 -0.005 0.081 0.027

(0.347) (0.754) (0.966) (0.749) (0.103) (0.239)

Strong_Competitor (β2) 0.025** 0.128*** 0.062***

(0.015) (<0.001) (<0.001)

COO promotion *

Strong_Competitor (β3) -0.011 -0.117*** -0.040**

(0.613) (0.005) (0.040)

Log (ST pay) pre 0.694*** 0.686***

(<0.001) (<0.001) Log (LT pay) pre

0.426*** 0.420***

(<0.001) (<0.001) Log (Total pay) pre

0.596*** 0.586***

(<0.001) (<0.001)

AROA 0.070 0.302*** 0.216*** 0.0702 0.296** 0.215***

(0.302) (0.009) (0.003) (0.303) (0.011) (0.003)

Size 0.089*** 0.320*** 0.174*** 0.091*** 0.325*** 0.179*** (<0.001) (<0.001) (<0.001) (<0.001) (<0.001) (<0.001)

Leverage -0.076* -0.468*** -0.222*** -0.077* -0.472*** -0.224***

(0.099) (0.001) (0.006) (0.097) (0.001) (0.006)

Market-to-book 0.010*** 0.049*** 0.025*** 0.010*** 0.049*** 0.025*** (<0.001) (<0.001) (<0.001) (<0.001) (<0.001) (<0.001)

Volatility 0.037 0.365*** 0.136** 0.037 0.369*** 0.140**

(0.223) (0.003) (0.018) (0.225) (0.003) (0.015)

Constant 3.328*** 5.008*** 4.155*** 3.392*** 4.984*** 4.227*** (<0.001) (<0.001) (<0.001) (<0.001) (<0.001) (<0.001)

Year fixed effect Yes Yes Yes Yes Yes Yes

Industry fixed effect Yes Yes Yes Yes Yes Yes

Observations 6,688 6,292 6,688 6,688 6,292 6,688

R-squared 0.719 0.530 0.704 0.719 0.532 0.705

Page 51: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

49

Table 5. VP turnover, comparisons across groups Panel A reports the comparison of mean turnover, pre-retirement turnover, and retirement rates between NPEs and

control VPs. The sample consists of 1,711 NPEs and 5,866 control VPs over the 1993-2013 period whose age is

non-missing in ExecuComp. Definitions of overall turnover, retirement and pre-retirement turnover are in the

appendix. In Panels B (C) we repeat the comparison in Panel A, but only include NPEs and Control VPs who are

classified as Strong (Weak) Competitors. *, ** and *** denote statistical significance at the 10%, 5%, and 1%

levels, respectively, for one-tailed tests.

Panel A. All Control NPEs and Control VPs

A. NPEs

(1,711 obs.)

B. Control VPs

(5,866 obs.)

Difference

(A-B)

Overall Turnover 30.3% 22.5% 7.8%***

Pre-retirement Turnover 23.4% 18.0% 5.4%***

Retirement 6.9% 4.6% 2.3%***

Panel B. Strong Competitor NPEs and Control VPs

A. NPEs

(789 obs.) B. Control VPs

(3,321 obs.) Difference

(A-B)

Overall Turnover 34.3% 20.0% 14.3%***

Pre-retirement Turnover 25.2% 15.0% 10.2%***

Retirement 9.1% 4.9% 4.2%***

Panel C. Weak Competitor NPEs and Control VPs

A. NPEs

(922 obs.) B. Control VPs

(2,545 obs.) Difference

(A-B)

Overall Turnover 26.9% 25.8% 1.1%

Pre-retirement Turnover 21.9% 21.8% 0.1%

Retirement 5.0% 4.0% 1.0%

Page 52: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

50

Table 6. VP turnover, multivariate regressions This table reports the results of probit regressions where the dependent variable is Turnover (Columns 1 and 2) or

Pre-retirement turnover (Columns 3 and 4). Definitions for Turnover, Pre-retirement turnover and all independent

variables are in the appendix. The sample consists of all NPEs and control VPs whose firms’ control variables are

non-missing in CompuStat and CRSP and whose age is non-missing in ExecuComp. Marginal effects of the

independent variables are reported. Standard errors are cluster at the individual firm level. P-values are shown in

parentheses. *, ** and *** denote statistical significance at the 10%, 5%, and 1% levels, respectively, for one-tailed

tests when there is a signed prediction for the independent variable and two-tailed tests when no signed prediction is

given for the independent variable.

Predicted

sign

Pr (Turnover=1) Pr (Pre-retirement

Turnover=1)

(1) (2) (3) (4)

COO promotion (β1) ? 0.070*** 0.006 0.053*** 0.0004

(<0.001) (0.702) (<0.001) (0.978)

Strong Competitor (β2) ? -0.075*** -0.074***

(<0.001) (<0.001)

COO promotion * Strong

Competitor (β3)

+ 0.130*** 0.109***

(<0.001) (<0.001)

Retire_Age + 0.221*** 0.229***

(<0.001) (<0.001)

AROA - -0.042 -0.043 -0.028 -0.028

(0.338) (0.332) (0.492) (0.489)

Size ? 0.005 0.005 0.002 0.002

(0.158) (0.178) (0.499) (0.563)

Leverage ? -0.017 -0.017 -0.012 -0.012

(0.425) (0.426) (0.537) (0.552)

Market-to-book ? 0.0006 0.0006 -0.0002 -0.0003

(0.732) (0.747) (0.880) (0.864)

Volatility + 0.048** 0.048** 0.048** 0.048**

(0.047) (0.045) (0.027) (0.028)

Observations 7,142 7,142 7,142 7,142

Pseudo R2 0.027 0.033 0.0065 0.0136

Page 53: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

51

Table 7. Change in firm performance, multivariate regressions This table reports the results of regression where the dependent variable is the firm’s change in AROA from the year prior to COO promotion year or control year

to each of the subsequent three years. We include variable Strong_Tournament to denote firms with strong COO promotion tournament. Definitions for AROA

and all independent variables are in the appendix. The sample consists of all COO promotion firms and control firms for which the change in AROA is

measurable and control variables are non-missing in CompuStat and CRSP. Standard errors are cluster at the industry level. P-values are shown in parentheses. *,

** and *** denote statistical significance at the 10%, 5%, and 1% levels, respectively, for one-tailed tests when there is a signed prediction for the independent

variable and two-tailed tests when no signed prediction is given for the independent variable.

Predicted

sign

ΔAROA

t-1 to t

ΔAROA

t-1 to t+1

ΔAROA

t-1 to t+2

(1) (2) (3) (4) (5) (6)

COO_promotion (β1) ? / + 0.0036 0.0128** 0.0010 0.0137** -0.0004 0.0026 (0.301) (0.026) (0.804) (0.021) (0.934) (0.354)

Strong_Tournament (β2) + 0.0061** 0.0098** -0.0008 (0.047) (0.013) (0.479)

COO_promotion *

Strong_Tournament (β3)

- -0.014** -0.020*** -0.004 (0.033) (0.007) (0.319)

Size ? -0.004*** -0.004*** -0.002 -0.002 -0.002 -0.002 (0.006) (0.005) (0.306) (0.269) (0.270) (0.262)

Leverage ? 0.035*** 0.035*** 0.051*** 0.051*** 0.077*** 0.077*** (<0.001) (<0.001) (<0.001) (<0.001) (<0.001) (<0.001)

Market-to-book ? 0.0004 0.0004 -0.0003 -0.0003 -0.0024** -0.0024** (0.553) (0.547) (0.737) (0.746) (0.0176) (0.0174)

Volatility ? -0.011 -0.011 -0.006 -0.006 -0.023 -0.023 (0.440) (0.442) (0.767) (0.768) (0.287) (0.292)

Constant 0.022 0.020 -0.005 -0.009 -0.001 0.001 (0.184) (0.235) (0.745) (0.576) (0.981) (0.981)

Year fixed effect Yes Yes Yes Yes Yes Yes

Industry fixed effect Yes Yes Yes Yes Yes Yes

Observations 1,973 1,973 1,956 1,956 1,844 1,844

R-squared 0.035 0.037 0.037 0.040 0.051 0.051

F-test p value: β1 + β3 < 0 0.477 0.138 0.455

F-test p value: β2 + β3 < 0 0.185 0.057* 0.294

Page 54: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

52

Table 8. Change in firm performance, the effects of staying and leaving strong competitors This table reports the results of regression where the dependent variable is the firm’s change in AROA from the year

prior to COO promotion year or control year to each of the subsequent three years. We divide firms with strong

COO promotion tournaments into two groups, Strong_Tour_Stay and Strong_Tour_Leave, according to whether at

least one strong competitor subsequently leaves the firm. Definitions for AROA and all independent variables are in

the appendix. The sample consists of all COO promotion firms and control firms for which the change in AROA is

measurable and control variables are non-missing in CompuStat and CRSP. Standard errors are clustered at the

industry level. P-values are shown in parentheses. *, ** and *** denote statistical significance at the 10%, 5%, and

1% levels, respectively, for one-tailed tests when there is a signed prediction for the independent variable and two-

tailed tests when no signed prediction is given for the independent variable.

Predicted

sign

ΔAROA

t-1 to t

ΔAROA

t-1 to t+1

ΔAROA

t-1 to t+2

(1) (2) (3)

COO_promotion (β1) + 0.0128* 0.0138** 0.0027

(0.0515) (0.0408) (0.696)

Strong_Tour_Stay (β2) + 0.0051 0.0081* -0.0053

(0.118) (0.067) (0.225)

Strong_Tour_Leave (β3) ? 0.0086 0.0140** 0.0097

(0.163) (0.036) (0.436)

COO_promotion * Strong_Tour_Stay (β4) - -0.016* -0.012* 0.0021

(0.092) (0.069) (0.425)

COO_promotion * Strong_Tour_Leave (β5) - -0.013* -0.032** -0.0170

(0.08) (0.01) (0.143)

Size ? -0.004*** -0.002 -0.002

(0.004) (0.267) (0.256)

Leverage ? 0.036*** 0.050*** 0.076***

(<0.001) (<0.001) (<0.001)

Market-to-book ? 0.0004 -0.0003 -0.0024**

(0.536) (0.715) (0.018)

Volatility ? -0.011 -0.007 -0.025

(0.430) (0.745) (0.279)

Constant 0.022 -0.012 0.0009

(0.196) (0.493) (0.970)

Year fixed effect Yes Yes Yes

Industry fixed effect Yes Yes Yes

Observations 1,973 1,956 1,844

R-squared 0.038 0.041 0.053

Page 55: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

53

Table 9. Future career of leaving NPEs and control VPs

This table compares the frequency of subsequent employment in S&P 1,500 firms for NPEs and control VPs who leave their firms. The sample consists of 393

NPEs and 1,053 control VPs over the 1993-2012 period who leave their firms by the second year following the year of COO promotion or control year, at an age

younger than 63. *, ** and *** denote statistical significance at the 10%, 5%, and 1% levels, respectively, against the treatment sample, for two-tailed tests.

Treatment Control

Total # Obs.

# Obs. subsequently employed in

another S&P 1,500 firm as

Total # Obs.

# Obs. subsequently employed in

another S&P 1,500 firm as

Top executive CEO or COO Top executive CEO or COO

Full Sample 393 73 (18.6%) 21 (10.7%) 1,053 85 (12.4%***) 31 (6.0%***)

Strong Competitors 194 42 (21.7%) 27 (13.9%) 498 75 (15.1%**) 38 (7.6%**)

Weak Competitors 199 31 (15.6%) 15 (7.5%) 555 56 (10.0%**) 25 (4.5%)

Page 56: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

54

Table 10. Firm performance prior to COO promotion This table reports the results of regression where the dependent variable is the firm’s change in AROA from two years prior to one year prior to the COO

promotion year or control year. We include variable Strong_Tournament to denote firms with strong COO promotion tournament. Definitions for AROA and all

independent variables are in the appendix. The sample consists of all COO promotion firms and control firms for which the change in AROA is measurable and

control variables are non-missing in CompuStat and CRSP. Standard errors are clustered at the industry level. P-values are shown in parentheses. *, ** and ***

denote statistical significance at the 10%, 5%, and 1% levels, respectively, respectively, for two-tailed tests.

AROA, t-1 ΔAROA, t-2 to t-1

(1) (2) (3) (4)

COO_promotion (β1) -0.0000 -0.0086 0.0014 -0.0050

(0.990) (0.163) (0.724) (0.387)

Strong_Tournament (β2) -0.0082** -0.0027

(0.034) (0.643)

COO_promotion * Strong_Tournament (β3) 0.0132 0.0099

(0.135) (0.162)

Size 0.0070** 0.0071** 0.0033 0.0001

(0.011) (0.010) (0.973) (0.904)

Leverage -0.165*** -0.165*** 0.0056 0.0056

(<0.001) (<0.001) (0.623) (0.623)

Market-to-book 0.0092*** 0.0092*** 0.0022*** 0.0022***

(<0.001) (<0.001) (<0.001) (<0.001)

Volatility -0.0670*** -0.0671*** 0.0412*** 0.0409***

(0.003) (0.003) (0.003) (0.003)

Constant 0.0136 0.0175 -0.0026 -0.0021

(0.772) (0.711) (0.913) (0.938)

Year fixed effect Yes Yes Yes Yes

Industry fixed effect Yes Yes Yes Yes

Observations 1,962 1,962 1,953 1,953

R-squared 0.168 0.169 0.029 0.030

F-test p value: β1 + β3 = 0 0.562 0.318

F-test p value: β2 + β3 = 0 0.495 0.092

Page 57: The Dark Side of Promotion Tournaments: The Effect on ......Kachelmeier, Chan Li, Jeremy Lill, Michael Majerczyk, Paul Newman, Jaime Schmidt, Brady Williams, Flora Zhou, workshop participants

55

Table 11. Accrual-based earnings management prior to COO promotion This table reports the results of regression where the dependent variable is the firm’s discretionary accruals in the

year prior to the COO promotion year or control year . We include Strong_Tournament to denote firms with strong

COO promotion tournament. Discretionary accruals is the difference between total accruals and non-discretionary

accruals, where non-discretionary accruals is estimated using the modified Jones model while controlling for prior

year ROA. Std(Sale) and Std(CashFlow) are the standard deviations of sale and operating cash flow over the past

five years, scaled by the firm’s total assets. CEO equity incentive is the logarithm of the CEO total pay-performance

sensitivity, measured following the methodology in Core and Guay (2002). Definitions for all other variables are in

the appendix. The sample consists of all COO promotion firms and control firms for which discretionary accruals

are measurable and control variables are non-missing in CompuStat and CRSP. Standard errors are clustered at the

industry level. P-values are shown in parentheses. *, ** and *** denote statistical significance at the 10%, 5%, and

1% levels, respectively, for two-tailed tests.

Discretionary accruals

(1) (2)

COO_promotion (β1) 0.00589 0.00158

(0.276) (0.813)

Strong_Tournament (β2) -0.000719

(0.926)

COO_promotion * Strong_Tournament (β3) 0.00643

(0.448)

Size (0.630) (0.615)

-0.00322 -0.00311

Loss -0.00322 -0.00311

(0.835) (0.840)

Market-to-book 0.00203*** 0.00202***

(0.00556) (0.00557)

Std(Sale) -0.0116 -0.0116

(0.521) (0.522)

Std(CashFlow) -0.245** -0.246**

(0.0226) (0.0225)

CEO equity incentive -0.00941** -0.00943**

(0.0201) (0.0208)

Constant 0.0136 0.0175

(0.772) (0.711)

Year fixed effect Yes Yes

Industry fixed effect Yes Yes

Observations 1,592 1,592

R-squared 0.036 0.036

F-test p value: β1 + β3 = 0 0.231

F-test p value: β2 + β3 = 0 0.535