Watch Your Basket ‐ to Determine CEO Compensation

• Date: 01 September 2019
• Published date: 01 September 2019
• Author: Neal Galpin,Hae Won (Henny) Jung,Lyndon Moore,Ekaterina Volkova
• DOI: http://doi.org/10.1111/fima.12243

Watch Your Basket - to Determine CEO

Compensation

Neal Galpin, Hae Won (Henny) Jung, Lyndon Moore,

and Ekaterina Volkova∗

CEOs (chief executive officers) are paid moreif they outperform other firms in their blockholders’

portfolios. For every percentage point by which their own firm’s return exceeds the return of

the largest blockholder’s basket of investments in a year, their compensation increases by over

$9,800. Once we benchmark to this portfolio, industry returns and own firm returns are of

little importance. When the firm is a larger portion of the blockholder’s portfolio and when the

blockholder is experienced, the reward for outperforming the blockholder’s portfolio is greater.

Our results are robustto alternate industry classif ications and definitions of blockholders.

John T. Schuessler, the chief executive officer (CEO) of Wendy’s International, was no doubt

pleased to receive a near 50% increase in his compensation package in 2002. Wendy’s stock price

had declined by almost 6.5% during the year, whichappeared weak compared to the 1.7% return

to the fast food sector. Wendy’s largest shareholder was Barrow, Hanley, Mewhinney, and Strauss,

who owned 9.9% of the firm at the close of 2001. Although Wendy’s performance may have

looked bad compared to its industry peers, it had done nicely compared to the other firms owned

by Barrow, Hanley, Mewhinney, and Strauss, whose portfolio had lost 19% of its value during

2002. By those standards, Schuessler appeared quite deserving of his $1.9 million bonus and

$4.3 million in option grants. We suggest one rationale for a principal to benchmark an agent’s

performance to her portfolio, rather than the industry: owners with information about other firms

in their portfolio will optimally choose to overweight co-owned firms relative to non-co-owned

firms as benchmarks in the compensation process.

To highlight our main implication, we modify the standard principal-agent framework of

H¨

olmstrom (1979) and H¨

olmstrom and Milgrom (1987). These models are the building blocks

from which much of the empirical CEO compensation literature has evolved. These models

predict benchmarking an agent’s compensation based upon performance relative to other firms.

Irregardless, evidence of relative performance evaluation (RPE), using typically a market index

such as the S&P 500 index and two-, three-, or four-digit Standard Industrial Classification (SIC)

industry portfolios as the benchmarks, is mixed (Edmans, Gabaix, and Jenter, 2017). We find

strong evidence of RPE, when we use the owner’s portfolio as a benchmark. A firm’s own stock

return performance by itself, and outperformance relative to industry peers, appear to be of little

incremental importance.

The authors thank an anonymous reviewer, Bing Han (Editor), Dan Zhang, Alex Edmans, and seminar participants at

the 2018 Executive Compensation Conference at Erasmus University Rotterdam, theDeakin Business School, Monash

University, and the KAIST College of Business fortheir helpful comments.

∗Neal Galpin is an Associate Professor in the Department of Banking and Finance at Monash Universityin Melbourne,

Australia. Hae Won (Henny) Jung is a Senior Lecturer in the Department of Finance at the University of Melbourne in

Melbourne, Australia.Lyndon Moore is a Senior Lecturer in the Department of Finance at the University of Melbourne in

Melbourne, Australia.Ekaterina Volkovais a Senior Lecturer in the Department of Finance at the University of Melbourne

in Melbourne, Australia.

Financial Management •Fall 2019 •pages 945 – 971

946 Financial Management rFall 2019

Weadd an assumption, that a principal’s information allows her to make more precise forecasts

for firms in her portfolio than f irms outside her portfolio, into the standard agency model.

When a principal has more information about firms in her por tfolio than other firms, she faces

less uncertainty about the future performance of fir ms in her portfolio than firms outside her

portfolio. Thus, the principal can offer the manager a contract less sensitive to forces outside the

(risk-averse) manager’s control. Such a contract will place a larger weight on the more certain

co-owned firms’ profits rather than any other f irm’s profit to compensate the agent.

There may be other explanations that are consistent with blockholdersusing their own portfolios

as a benchmark in CEO compensation. Models of limited attention, like Tversky and Kahnemann

(1973)’s “availability bias,” would imply that blockholders could use information more easily

available, such as data on the other firms they own, in order to set CEO compensation. Therefore,

the positive relation we observe between CEO pay and own-f irm performance relative to the

blockholders’ portfolio may be driven by an availability bias, instead of the optimal contracting

effect we emphasize.

We empirically test our prediction by estimating the pay-performance sensitivities for CEOs

and relate those sensitivities to the portfolio of the largest institutional blockholder (an institution

who holds at least 5% of the shares outstanding for the given company). The economic effects

of considering firm performance relative to the blockholder’s portfolio are large. When a CEO’s

firm outperforms the largest blockholder’s portfolio by an additional one percentage point, the

CEO’s pay is $20,979 higher, on average, or $9,861 if we use a natural logarithm specification.

In an alternative regression, we estimate that the CEO receives $105 of ever y $1,000 of excess

performance that she delivers. A CEO’s pay does not increase for returns that simply keep

pace with the largest blockholder’s portfolio because of, for example, general market upturns

or downturns. We find the largest sensitivities to the portfolio benchmark in option grants and,

to a lesser extent, in bonuses. Sensitivities are greater when the firm is a larger fraction of a

blockholder’s portfolio, and when the blockholder is more experienced in the sense that it has

been the largest blockholder or a blockholder in any firm for the past consecutive years.

Standard agency models, starting with H¨

olmstrom (1979), imply CEO compensation should

be benchmarked to other firms to reduce the risk of the contract for the CEO. However, an

extensive literature shows mixedempirical evidence on RPE (Edmans et al., 2017). Recent work

explores potential explanations for the weak evidence. Aggarwal and Samwick (1999) suggest

that product market competition reduces RPE. Gopalan, Milbourn, and Song (2010) suggest that a

CEO can partly influence in which industry the firm operates, which also reduces RPE. DeMarzo

and Kaniel (2017) show that “keeping up with the Joneses” preferences can also diminish the

importance of RPE. Dittmann, Maug, and Spalt (2013) and Chaigneau, Edmans, and Gottlieb

(2018) argue that RPE erodes incentives for managers and therefore may not be part of an

efficient contract. Other research suggests that the typical benchmark groups chosen by prior

studies (market or industry performance) are misspecified. Albuquerque (2009) suggests the use

of similar industry-size firms, whereas Drake and Martin (2018) propose including peer f irms

at the same point in their life cycle as performance benchmarks. Both papers find support of

RPE. De Angelis and Grinstein (2018) examine the stated structure of compensation contracts

and show that a majority of firms benchmark to a peer group that is not simply the industry or

market index. We contribute by showing that a standard agency model, augmented with informed

owners, predicts benchmarking not to the market or industry peers, but to owners’ portfolios. We

show strong empirical evidence for this version of RPE.

Kempf, Manconi, and Spalt (2016) argue that institutions cannot monitor all parts of their

portfolio at all times, and that when an institution’s por tfolio is hit with a large shock, positive

or negative, the institution is distracted. They find that a more distracted institution is more