The Effect Of Short- And Long-term CEO
Compensation On Cost Stickiness
Thesis MSc Accountancy & Control
Ilias Mesaoudi 10845747 20-06-2016
Amsterdam Business School Facult y Eco no mics and Business, Universit y o f Amsterdam MSc Accountanc y & Co ntrol
Abstract
Prior literature on the alternative cost behavior model has found that selling, general and administrative (SG&A) costs act asymmetrically. Anderson et al. (2003) find that SG&A costs increase on average 0.55% per 1% increase in sales but decrease only 0.35% per 1% decrease in sales. They focused on the decision of a manager to commit to recourses, which influences how costs behave. This paper examines the effect that short- and long-term CEO compensation has on cost stickiness. The expectation is that there is a positive relation
between long-term compensation schemes and cost stickiness and a negative relation between short-term compensation schemes and cost stickiness. The time period of the sample used to test the hypothesis covers the years 2007 – 2014 and consists of U.S. listed firms. No
evidence has been found regarding a positive relation between long–term CEO compensation schemes and the degree of cost stickiness. The expectation was that predicting future demand due to for example seasonality, would lead to managerial decisions to commit to resources because of a long-term view. Waiting too long with taking back committed resources, will lead to costs building up. However the assumption is that the firms in more recent samples are relatively more flexible and know how to cut committed resources when demand declines without it harming the long-term firm value. However this paper does find evidence regarding a negative relation between short–term CEO compensation and the degree of cost stickiness, which is in line with the expectation that came forth from prior literature. Managers are more inclined to manipulate the outcome of firm profit and cut committed resources when they are incentivized through short-term compensation schemes. This leads to a higher downward adjustment of SG&A costs when sales revenue decreases.
Statement of Originality
This document is written by student Ilias Mesaoudi who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its
references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents
Table of content
1. Introduction ...5
1.1 Motivation ...7
1.2 Structure ...7
2. Literature review and hypothesis ...8
2.1 Agency theory ...8 2.2 CEO compensation ...9 2.3 Cost stickiness ... 11 2.5 Hypothesis development ... 15 3. Research methodology ... 17 3.1 Sample selection ... 17 3.2 Cost stickiness ... 17 3.3 CEO compensation ... 18 3.4 Control variables ... 18 3.5 Empirical model ... 19 4. Results ... 22 4.1 Descriptive statistics ... 22
4.2 The effects of long-term CEO compensation on cost stickiness ... 26
4.3 The effects of short-term CEO compensation on cost stickiness ... 27
4.4 Additional analysis ... 30
4.4.1 Median split ... 30
4.4.2 Alternative measurement of short-term compensation ... 31
5. Conclusion ... 33
References ... 35
1. Introduction
Costs are one of the fundamental determinants of a firm’s earnings. Having a good understanding on cost behavior can lead to firms improving their earnings quality, their prediction of earnings, and tracking down earnings manipulation. Traditional costing models suggest that variable costs change corresponding to changes in the activity levels (Noreen, 1991). The model suggests that cost increases when revenue increases and decreases when revenue decreases. The total costs do not always change at the same rate as changes in sales. Prior literature on the alternative cost behavior model has found that cost act asymmetrically. These findings are not in line with the traditional cost models. Anderson et al. (2003) find evidence in their paper that cost act asymmetrically in relation to a firm’s activity. These results are tested by looking at the selling, general and administrative costs of a firm (SG&A). They found that SG&A costs increase on average 0.55% per 1% increase in sales but decrease only 0.35% per 1% decrease in sales. This type of cost behavior is referred to as “sticky costs”. The study by Roodzant (2012) finds that the same cost increase with 0.46% when the activity increases with 1% and decrease with 0.32% when the activity decreases with 1%. Anderson et al. (2003) define this asymmetrical cost behavior phenomenon as follows: “Costs are sticky if the magnitude of the increase in costs associated with an increase in volume is greater than the magnitude of the decrease in costs associated with an equivalent decrease in volume”.
There are many theories that explain the reasoning for costs to be sticky. The theory from Anderson et al. (2003) focuses on the alternative model of cost behavior, which is the decision of a manager to commit to recourses. When customer demand declines the manager has to decide whether to reduce the resources and pay the adjustment cost or to choose for the costs of not fully using the capacity of the firm. When it’s difficult to predict future demand and a firm has to make adjustment costs to reduce or take back committed recourses, managers may try to purposefully delay these reductions until it’s certain that the customer demand is stably declining. These decisions to commit to resources could be influenced by a manager’s compensation scheme or behavioral biases. The compensation scheme can lead to a CEO making this decision based upon personal benefits instead of making decisions in the firm’s interest, which in turn can lead to an agency conflict (Jensen and Meckling 1976).
Murphy and Jensen (2011) find that a CEO, who is incentivized through short-term compensation based on accounting profits, is more inclined to cut research and development expenses even if this has negative impact on firm value in the long run. The paper of
Dierynck et al. (2012) repeats the notion of the manager, who is for the most part incentivized through short-term incentives, is more focused on current earnings when sales decline and is therefore more likely to cut committed resources. Cutting committed resources by the manager at the expense of long-term earning would lead to a lower degree of cost stickiness. On the other hand managers that receive equity-based long-term incentives are more inclined to focus on the long-term performance of the firm. When future customer demand is
uncertain, the manager is inclined to avoid cutting committed resources because he is afraid that doing so is at the expense of long-term earnings. When sales increase the next period, the cost stickiness of the previous period would be reversed in the following period so that the stickiness may be less noticeable. However when sales keeps declining and the CEO waits too long with reducing committed resources, the costs of unutilized resources and of unused capacity of the firm will build up. This in turn will lead to a higher degree of cost stickiness for the firm. Chen et al. (2012) also finds that managers who are incentivized on a long-term basis to increase long-term firm value, can lead to an increase of sticky cost. To grow the firm, managers keep resources that are unutilized to increase personal status, power, compensation and prestige.
The study of Anderson et al. (2003) has found a positive relation between sticky cost and deliberate decision making by managers who weigh the economic consequences of their actions. The type of compensation the manager receives plays a role in the decision to commit to resources. Therefore I examine the following research question: What is the effect of short- and long-term CEO compensation on cost stickiness?
This paper examines the effect of CEO compensation on cost stickiness and focuses on the type of compensation granted to CEOs. The types of CEO compensation that will be researched are divided in short- and long-term promotion schemes.
1.1 Motivation
Anderson et al. (2003) mention that their paper provides a platform for further research on the causes and consequences of sticky cost behavior. This study aims to contribute to existing literature by researching the effects of managerial incentives on cost behavior. To be more specific, this paper aims to contribute to existing literature by gaining insight in to whether short –and long-term compensation schemes have different effects on managerial decisions with regards to the commitment of resources and in turn the cost stickiness of a firm.
Dierynck et al. (2012) focus their research on the influence of managerial incentives to meet or beat the zero earnings benchmark on labor cost behavior of private Belgian firms. Instead of researching the stickiness of labor cost, this paper aims to contribute to the research of Dierynck et al. (2012) by researching the effects that the chosen CEO compensation scheme has on the stickiness of selling, general and administrative costs (SG&A) of a firm.
Furthermore Dierynck et al. (2012) also note in their research that their focus on labor cost behavior in private Belgian firms makes it difficult to generalize the outcome of the research. They mention that further research may test and verify whether their findings are
generalizable in other institutional settings with other types of incentives and other costs. Anderson et al. (2004) note that there are different drivers of cost stickiness in different industries. Therefore a research that is conducted across a different time period, with a different set of research data could lead to different results when comparing to prior research. This difference could possibly result in an outcome where cost stickiness may be more or less prevalent. To my knowledge this topic has not been extensively researched, which is the reason for this study being conducted.
1.2 Structure
This paper proceeds as follows. The following is comprised of a literature review, where the agency theory, CEO compensation and cost stickiness will be discussed. Furthermore the hypothesis will be developed in this section. Section three will contain the research
methodology. In this section the chosen samples and the way the data research is conducted will be discussed. The fourth section will give an overview of the results. The final section, which is section five, will contain the conclusion and the limitations of this research paper.
2.
Literature review and hypothesis
2.1 Agency theory
The agency theory describes the conflict that arises between the principal and the agent. The research of Jensen and Meckling (1976) mentions that the agency theory focuses on an
agency relationship where the principal (shareholder) delegates some decision rights and work to an agent (CEO). The agent will not always act in the best interest of the principal which is due to the cooperating parties having different goals and division of labor. Furthermore Eisenhardt (1989) mentions in his research that agents and principals have a different preference when it comes to the amount of risk that should be taken. The agents and
principals might prefer a different approach in actions due to different goals. The agent has self-interest and might want to achieve his own goals at the expense of the firm’s goals. Cole et al. (2006) have found that long-term compensation such as stock options, provide
incentives to engage in riskier projects. This is due to the presence of upward potential in risky project and the downward risk these options are missing. The stock price might go up and in turn the bonus of the CEO goes up when a risky project succeeds. However when the stock price goes down the CEO will not incur any losses because of the pre-specified exercise price. The agency theory explains that managers want to have more wealth than less, but that this desire marginally decreases when more wealth is acquired. When a manager has more ownership in the firm, he will be less inclined to take very risky projects. Through the proper level of monitoring and incentive scheme the principal wants to limit these differences, to minimize the cost of monitoring, the preferences of the agent and the agent's opportunity cost of accepting employment with the firm (Eaton & White, 1983). Monitoring is the observing and measuring of the agent by the principal. Jensen and Meckling (1976) add to this by including efforts of the principal to control the behavior of the agent. One of those efforts is designing a control system that contests the manager’s tendency to act in his own interest at the expense of the firm’s interest. The need for a control system exists because of the presence of information asymmetry. Information asymmetry limits the extent to which a manager’s action can be controlled (the agent having more information than the principal). Since the principal cannot directly ensure that the agent always acts in the best interest of the principal, the need for a control system arises. To ensure that agents act in the best interest of the principal, firms establish incentive system which should maximize shareholder value by inducing optimal CEO effort. CEO compensation can be used as a means to contest the agency problem by using them in such a way that the interests of both the managers and
shareholders are aligned (Jensen & Meckling, 1976). One of the ways of doing this is by granting the CEO an ownership stake in the firm. These equity compensations create incentives for executives to act in the best interest of the shareholder. On the other hand Short-term incentives could prevent from moral hazard of occurring. Moral hazard occurs when the CEO does not put effort into reaching the goals of the agent (Eisenhardt, 1989). Information asymmetry makes it difficult to control whether the agent actually puts in enough effort to reach the firm’s goals. Short-term incentives motivate the CEO and could lead to an increase of the agent’s effort in the current period and therefore align the interest of the principal and the agent.
2.2 CEO compensation
Most CEO compensation schemes contain five basic components, namely: salary, annual bonus, payouts from long-term incentive plans, restricted option grants, and restricted stock grants (Frydman & Jenter, 2010). CEOs can also receive contributions to defined-benefit pension plans, and when they leave the firm they often receive severance payments.
The research of Frydman & Jenter (2010) finds that the major pay components for CEOs of large firms from 1936 to 1950 was mainly composed of salaries and annual bonuses. Long-term incentives plans became more popular in 1960. In these years long-term incentive plans were bonuses based on the performance of CEOs throughout multiple years and was paid in either stock or cash. From the year 1980 onwards the use of stock option to incentivize CEOs started to make its entrance. The use of stock options to incentivize CEOs was
introduced to tie remuneration to share prices and therefore give managers an incentive to increase the value of the principal. The use of stock options became more prevalent in the 1950s because of a tax reform that permitted payouts of certain options to be taxed at lower capital gains rate compared to the labor income tax rate. The use of stock options as an
incentive became more popular but didn’t really hit until the 1970s (Frydman & Jenter, 2010). During the 1980s and 1990s stock options became the main component of CEO
compensation. In 1992 the payments to CEOs consisted for 20% of stock options. In 2000 this amount more than doubled to 49%. However, this growth didn’t occur at the expense of other components of pay which means that the total compensation for CEOs also grew in this period (Frydman & Jenter, 2010).
Murphy (2012) has found annual bonus plans to be strong incentives, because it is easier for the manager to predict what the outcome of their action would mean for their
reward. This is more difficult to predict when it comes to the use of long-term equity incentives. When using short-term incentives, managers are for example able to figure out how their behavior affects the return on assets. This is more difficult to understand when it comes to the link between CEO behavior and stock price. The link is more difficult to be made because equity incentives are impacted by different factors that are beyond the control of the CEO. Accounting based performance measures on the other hand are known to be less noisy (Murphy 2012). In order for a CEO to receive a non-discretionary incentive he has to meet pre-specified requirements. Annual bonuses like return on assets, earnings per share and return on equity are accounting based measures and are linked to firm performance. Using these measures as incentives makes it possible to influence the level of cost stickiness. On the other hand bonuses can also be forward-looking. A discretionary bonus that is non-forward-looking is not linked to firm performance and is not pre-specified.
CEO compensation can be split into short- and long-term incentive schemes.
According to Hall & Liebman (1998), the right incentive scheme for executives could be the solution to the agency problem. They find that the optimal contract has to be a one-to-one correspondence between firm value and CEO pay. It is important that compensation and managerial interests are aligned with shareholders’ interests in order to solve agency problems (Murphy, 1999). Choosing for the wrong compensation scheme could lead to a weak
congruence which in turn could lead to an agency conflict (Holmstrom, 1979). The CEO might be inclined to focus on increasing personal wealth instead of increasing firm value by taking back committed resources instead of making adjustment costs when customer demand declines (Anderson et al., 2003). An example of an agency conflict is the CEO that receives short-term compensation based on accounting profits and decides to cut research and
development expenses even if this has negative impact on firm value in the long run (Murphy and Jensen 2011). These decisions of the CEO that are based on a short-term view would lead to cost stickiness being less noticeable.
The research of Gopalan et al. (2014) argues that firms who offer long-term
compensation contracts have lower accruals and less positive earnings enhancing accruals. This suggests that term compensation provides incentives for CEOs to focus on short-term earnings. They find that short-short-term CEO compensation contracts are linked with greater managerial incentive to manipulate short-term performance. They also find evidence of interaction between the duration of CEO compensation and characteristics of an industry or firm. They find that the duration is longer for “firms that are larger, firms with more growth opportunities, firms with a higher proportion of long-term assets, firms with higher R&D
intensity, less risky firms, and firms with better past stock performance” (Gopalan et al., 2014) CEOs that are incentivized through long-term contracts receive higher compensation, but on average a lower bonus.
Dikoli (2001) states that in order to motivate executive managers to increase long-term value, the agents should also be awarded based on long-term incentives. Long-term incentives are backwards looking performance measures that relate the reward to performance measures over periods greater than one year, and where the focus is on rewarding agents for creating long-term value (Murphy 1999). The performance measures that are used to reward CEOs, such as the return on assets (ROA), return on equity (ROE) or the earnings per share (EPS), are often accounting based measures. The use of long-term compensation scheme also has its disadvantages. Improving long-term profitability is not necessarily linked with maximizing stock wealth. The CEO could make the firm more profitable over few years, but this could be accomplished in such a way that this is not in the interest of the shareholders. The CEO could for example choose to sell preferred stock at a discount, include overproduction or repurchase debt (Kaplan and Atkinson, 1989).
2.3 Cost stickiness
The study from Noreen and Soderstorm (1997) suggests that certain costs increase faster with an activity increase than they decrease when an activity decrease takes place. Anderson et al. (2003) find evidence in their paper that cost act asymmetrically in relation to a firm’s activity. These results are tested by looking at the selling, general and administrative costs of a firm. The study uses a sample of 7,629 U.S. firms over a period of 20 years and suggests that selling, general and administrative costs increase on average 0.55% per 1% increase in sales but decrease only 0.35% per 1% decrease in sales. Another study by Roodzant (2012) uses a sample of 39,738 US-listed firms over a period of 14 years and finds that the same cost increase with 0.46% when the activity increases with 1% and decrease with 0.32% when the activity decreases with 1%. Sticky costs emerge out of the asymmetric frictions in adjusting resources. Anderson et al. (2003) explain that the reason for this is when it is difficult to predict future demand and a firm has to make adjustment costs to reduce or take back committed recourses. Taking back committed resources is costly because of the costs that come with it like hiring and firing costs for labor, or installation and disposal costs for machinery. Managers may try to delay these reductions on purpose until it’s certain that the
customer demand is stably declining. This would lead to the cost stickiness of one period to be reversed in the following period so that the stickiness may be less noticeable. When demand keeps declining however, the committed resources will be unutilized until the decision is made to cut them. Predicting and evaluating whether demand was temporary declining and will go up in the near future is difficult. When CEOs decide to keep unutilized resources rather than incur adjustment costs when volume declines, stickiness of SG&A costs occur (Anderson et al., 2003). Furthermore they find that costs are lumpy, because it is difficult to modify committed resources in an incremental fashion so that this matches the small change in customer demand. The cost lumpiness could lead to a surplus or insufficient capacity but it does not lead to sticky costs by itself. There are forces that lead to the process of downward adjustment of SG&A costs being slower than the upward process.
Calleja et al. (2006) find cost to be “sticky” on operating costs. They tested this by using samples consisting of firms from the United States, United Kingdom, France and Germany. In their research they find that operating costs of the firms to be stickier in France and Germany than the operating costs of the firms in the U.S. and U.K.
Subsequently the research of Banker and Chen (2006) is conducted by examining differences in the sticky behavior of operating expense across several countries. They used a sample of 12,666 firms from 19 countries that were members of the Organization for
Economic Co-operation and Development during 1996 until 2005. They find that the degree of cost stickiness varies strongly across the firms which were located in different countries. This study suggests that there is empirical support that the way the labor market is set up effects the degree of cost stickiness in the firms. Prior research on cost stickiness has also proven that the degree of cost stickiness varies across different industries (Dalla & Perego, 2014).
Determinants of cost stickiness
Cost stickiness is influenced by many factors. Prior academic literature has given us a number of factors that determine the cost stickiness behavior. One of the factors explained by
Anderson et al. (2003) is the characteristics of the firm like the employee intensity, asset intensity and debt intensity. Other literature’s explanatory factor is the different kind of characteristics of a firm across different countries. This could be growth, the labor market, and regulations of the specific firm in different countries (Banker and Chen 2006). The third type of literature focuses on the agency theory and the managerial incentives which cause the
cost stickiness behavior in a firm (Banker et al., 2011; Chen et al., 2012; Dierynck et al., 2012).
The key factor in these studies discussing the determinants of cost stickiness is the presence of adjustment costs. These adjustment costs are difficult to measure therefore most papers use observable proxies to measure adjustment costs. The research of Anderson et al. (2003) focuses on the correlation between asset –and employee intensity with the cost
stickiness of a firm. They find that it is less costly to make adjustment costs when it comes to adjusting purchased material and services which in turn leads to a lower degree of cost stickiness. However making adjustment costs because of the firm’s own assets and
employees, is more costly and leads to a higher degree of cost stickiness. Chen et al. (2012) find the opposite in their study, suggesting a lower degree of cost stickiness is present in firms that have relatively higher employee intensity. They assume that the reason for this is due to the different samples that have been used by both research papers which in turn lead to different results. The sample used in the research of Chen et al. (2012) is more recent and temporary labor is used relatively more by the firms. Temporary labor increases flexibility of labor costs which in turn would mean that it is easier to make adjustment costs when demand fluctuates and this decreases the degree of cost stickiness. The research of Weidenmier and Subramaniam (2003) confirms that firms with a high fixed asset intensity leads to a higher degree of cost stickiness.
Kama and Weiss (2010) use the model of Anderson et al. (2003) to assess whether the adopted technology of firms restricts them from the ability to respond accordingly to activity changes and therefore influence cost stickiness. The uncertainty in demand is used as a proxy that indicates the restriction that firms face when a higher demand uncertainty occurs. More flexible technology is preferred by these firms because this leads to lower adjustment costs and in turn also lower cost stickiness.
The expectation of the CEO with regards to future demand is an important part of the research conducted by Anderson et al. (2003). They find that cost stickiness is less prevalent when customer demand declines for two consecutive periods. The CEO is able to predict the duration of declining demand more effectively if this declination occurs in successive periods. This leads to the CEO being more prepared to cut committed resources which in turn leads to costs being less sticky (Anderson et al., 2003).
The research of Banker et al. (2011) has revealed that the degree of cost stickiness was high after a previous sales increase for 18 out of 19 countries. The opposite is the case when sales decreased for 6 out of 19 countries this led to cost stickiness being less prevalent.
Banker et al. (2011) also find that cost stickiness plays a bigger role in the manufacturing sector with growing industries. The paper of Anderson et al. (2003) also suggests that
managers take the influencing factors on a macro-economic level in consideration to evaluate the reduction of customer demand. The growth percentage in real GNP is used as a proxy to test whether this is the case. Their findings suggest that CEOs are less likely to cut the committed resources when they find themselves in a period of macro-economic growth. This leads to a higher degree of stickiness. In summary, I find that these papers suggest that optimistic CEO expectation who believe customer demand will rise after declination are not inclined to cut committed resources which in turn leads to a higher degree of cost stickiness. The opposite is the case when a prior period of decrease has occurred; CEOs are more likely to cut committed resources which lower the degree of cost stickiness.
Another determinant that Banker et al. (2011) find has influence on the cost stickiness is the type of compensation that is used for the managers of a firm. When managers’
incentives are used they can be linked to either firm profit or share prices on the stock market to reward the CEO when an increase in firm value is realized. Anderson et al. (2003) focus on the decision of the agent to commit to recourses. The compensation scheme can lead to a CEO making this decision based upon personal benefits instead of making the choice in the firm’s interest (Jensen and Meckling 1976). Therefore managerial incentives also play a big part in determining sticky cost behavior.
Dierynck et al. (2012) find that when managers are incentivized to meet or beat the zero earnings benchmark, this reduces the labor cost asymmetry. This research confirms that the manager, who is for the most part incentivized through a short term incentive, is more focused on current earnings when sales decline. Because the manager is not focused on the firm’s long-term earnings and is more concerned about the firm’s current earnings, he is more likely to cut committed resources to achieve higher current earnings. Cutting these committed recourses leads to a reduction of sticky costs. However the manager that is mostly
incentivized through equity-based long-term incentives will focus more on the long-term performance of the firm. The manager wants to avoid the adjustment cost when sales rise in the future and is therefore less likely to cut committed recourses. On the contrary to short-term incentives this leads to an increase of cost stickiness.
Kama and Weiss (2010) argue that depending on the underlying motivations of the agents, sticky cost might increase or decrease. They find two potential causes for cost
stickiness: managers who adjust recourses and are doing so because of self-interest instead of the interest of the firm and constraints which are imposed by past technological choices to
maximize firm value. They also find that CEOs that are incentivized to meet earnings targets causes managers to cut recourses in a faster rate when there is a decline in activity. The motivation to meet these targets leads to managers to act on self-interest which causes the decrease of sticky costs.
The research of Chen et al. (2012) argues that a manager’s empire building incentive is associated with the degree of SG&A cost asymmetry. Empire-building can lead to
overspending of managers which subsequently leads to cost stickiness. They find that managers who are incentivized to grow the firm at a faster rate than the firm is capable to, keep resources that are unutilized to increase personal status, power, compensation and prestige. Empire-building is more likely to occur when managers have the means to overinvest. The presence of good corporate governance can restrict the capability of the manager to overinvest.
2.5 Hypothesis development
Prior literature indicated that cost stickiness could be attributed to many factors. The theory of Anderson et al. (2003) focuses on the decision of the CEO to commit to recourses. Some firms have difficulty in predicting future demand due to for example seasonality. This makes it difficult for a CEO to make the decision whether to make adjustment costs or to take back committed resources when customer demand declines. Furthermore prior research has shown that the compensation scheme of the CEO influences managerial decisions with regards to making adjustment costs or taking back committed resources. The expectation is that a CEO that receives his bonus based on a long-term compensation scheme, will be less inclined to make adjustment cost until it is certain that the customer demand is stably declining. This would lead to the cost stickiness of one period to be reversed in the following period so that the stickiness may be less noticeable. However when the demand keeps declining and the CEO waits to long with taking back committed resources, he will have costs building up because of the firm’s unutilized resources and of unused capacity. This will lead to cost stickiness being more prevalent. To test whether long-term CEO compensation actually influences managerial decisions and subsequently has a positive relation with cost stickiness, the following hypothesis has been developed:
Hypothesis 1: there is a positive relation between long–term CEO compensation and the
The type of CEO compensation scheme used by a firm could also lead to an agency conflict and might influence decisions taken by the manager. The CEO could choose to act upon self-interest instead of the self-interest of the firm (Jensen and Meckling 1976). The expectation is that cost stickiness is less prevalent when CEOs are incentivized through short-term compensation schemes compared to long-term compensation schemes. This is expected because of the fact that the CEO of a firm can manipulate the outcome of firm profit and therefore also his annual bonus at the expense of increasing the long-term value of the firm (Gopalan et al., 2014). Manipulation of current earnings takes place by the CEO deciding to cut committed recourses to achieve higher current earnings at the expense of long-term earnings. To test this
expectation the following hypothesis has been developed:
Hypothesis 2: there is a negative relation between short–term CEO compensation and the
3.
Research methodology
3.1 Sample selection
The data for this research paper will be collected from the database the Wharton Research Data Service (WRDS). Compustat will be used to obtain the data regarding asymmetric cost behavior. To measure cost stickiness the sample will consist of U.S. listed firms over the period 2007-2014. This is a more recent period and differs from the periods used in prior literature. Execucomp will be used to gather data with regards to CEO compensation.
Financial Institutions have been excluded from the sample because they have a different cost structure and business approach (Kama and Weis 2010; Banker et al. 2011). This resulted in to 59871 observations. Next the sample has been filtered so that the cases that contained missing variables were removed. Furthermore the cases that contained SG&A costs that were higher than revenue were removed. Following the paper of Dalla & Perego (2014), the top and bottom 1% observations of the sample were winsorized. In order to eliminate the extreme skewness of SG&A costs and revenue, a log transformation has been performed. The reason for this transformation is to have a better normal distribution and to be able to give a better economic interpretation (Dalla & Perego, 2014). This ultimately results in having a final sample of 2429 observations to test the effects of long-term compensation schemes on the degree of cost stickiness. The final sample to test the effects of short-term compensation scheme on the degree of cost stickiness contains 3408 observations.
3.2 Cost stickiness
The primary variables that will be used to measure cost stickiness are SG&A costs (selling, general and administrative expenses) and net sales revenue. In line with the papers of Anderson et al. (2003) and Chen et al. (2012), net sales revenue are used as a proxy to measure the activity levels of a firm. The model that will be used to test asymmetric cost behavior is the model that is developed by Anderson et al. (2003). This model allows me to measure how SG&A costs react when changes in sales revenue occur and distinguishes periods where revenue increases and when revenue decreases. In the empirical model the interaction variable “Dummyrev” is used for the sales revenue.
3.3 CEO compensation
CEO compensation will be divided into short-term and long-term incentives. To be more specific the variables that are going to be used for long-term compensation are CEOs that are incentivized through stocks and/or options. Whilst managers that are incentivized through bonuses and/or non-equity based compensation are classified as short-term compensation. Dividing the firms into what extent the firms use one of the two types of compensation
schemes will show the effects on asymmetric cost behavior. The variables that are going to be used to measure the total CEO compensation is go to be TDC1 from Execucomp. This is comprised of: salary, bonus, non-equity incentive plan compensation, grant-date fair value of option awards, grant-date fair value of stock awards, deferred compensation earnings
reported as compensation, and other compensation. In the Execucomp database the short-term incentive are the variables bonus and non-equity incentive plan compensation, which gives the total value of the bonus earned by the CEO during the current year. The long-term incentives consist of the variables grant date fair value of options granted and grant date fair value of stock awarded under plan-based awards.
3.4 Control variables
The model of Anderson et al. (2003) will be extended to test whether the degree of cost stickiness is more prevalent when CEOs are incentivized through short-term compensation or more prevalent when they are incentivized through long-term compensation schemes. For the first hypothesis it is expected that firms that use mostly long-term incentives to compensate their CEOs experience a higher degree of cost stickiness compared to firms that don’t use long-term incentives. Consequently the expectation is that these firms have a larger increase of SG&A costs when sales increase and smaller decrease of these costs when sales decrease.
In order for the second hypothesis to be supported, SG&A costs of the firms in the sample that use mostly short-term compensation to incentivize their CEOs should show more symmetrical behavior compared to the firms that incentivize their CEOs with mostly long-term compensation. This means that the firms that use short-long-term compensation should have a lesser increase in SG&A costs when sales go up in comparison to firms that don’t use short-term compensation and greater decrease in SG&A costs when sales decline. Interaction variables will be added to the model to test the hypotheses. The interaction variables depict to what extent the firm uses short- and long-term incentives. In other words the variable will describe to what extent the firm uses annual bonus relative to total compensation as an
incentive (annual bonus/ total compensation). And to what extent the firm uses equity incentives (equity/ total compensation) will also be described by the interaction variable.
To control for the factors that influence cost stickiness, I have used three control variables, namely: asset intensity, employee intensity and firm size. The research paper of Anderson et al. (2003) suggests that cost stickiness is induced when firms have higher asset intensity. Asset intensity is calculated by dividing total assets of a firm by total sales revenue. Anderson et al. (2003) also find correlation with employee intensity and stickiness of costs. The employee intensity is the total number of employees to sales revenue. However Chen et al. (2012) suggests that employee intensity is negatively correlated with the stickiness of SG&A costs, which means that a lower degree of SG&A costs is present for firms that have relatively more employees. Chen et al. (2012) assumes that the reason for this is the different periods of the samples. The sample of Chen et al. (2012) covers the years 1996–2005, whilst the sample of Anderson et al. (2003) covers the years 1979–1998. The assumption is that the firms in the more recent samples make relatively more use of temporary labor, which leads to the firms being more flexible with their labor costs. The last control variable that is added to the model is the firm size, which is the natural logarithm of total assets of a firm per year. Watts and Zimmerman (1986) find that larger firms can become a political target, which effects manager’s choice and can lead to the manager reporting higher costs or lower profits.
3.5 Empirical model
The model that will be used to test the cost stickiness is the model of Anderson et al. (2003) Model 1 (Anderson et al., 2003):
In this model the SG&A costs are divided to represent the percentage change in costs and activity levels for firm i in year t, relative to year t-1. The dummy variable is equal to 1 if sales of firm i decrease in year t and zero if otherwise. The last component of the model
shows the error term. The log specification makes an economic interpretation of the estimated coefficients possible. When the Decrease Dummy is 0 in this model, the revenue increases and the coefficient β1 measures the percentage increase in SG&A costs with a 1% increase in sales revenue. Consequently when the value of Decrease Dummy is 1 this means that revenue decreases and the sum of the coefficients β1 +β2 measures the percentage increase in SG&A costs with a 1% decrease in sales revenue (Anderson et al., 2003). If the variation of SG&A costs with revenue increases are greater than the variation for revenue decreases, this shows us that SG&A costs are sticky. The limitation of the model is the use of sales revenue to reflect the more correct volume of sales.
To test whether there is a positive relation between long-term CEO compensation and the degree of cost stickiness (H1), the model of Anderson et al. (2003) will be extended.
Extended model 1:
Log (SG&Ai,t/ SG&A i,t-1)= β0 + β1 Log (Revenuei,t/Revenue i,t-1) + β2 Dummyrev i,t * Log
(Revenuei,t/Revenue i,t-1) + β3 Dummyrev i,t + β4 DummyLT i,t * Log
(Revenuei,t/Revenue i,t-1) + β5 Dummyrev i,t * DummyLT i,t * Log(Revenuei,t/Revenue i,t-1)
+ β6 DummyLT i,t * Controlsi,t
Specifically the model will be extended by adding interaction variables to test whether long-term incentives schemes lead to a higher degree of cost stickiness. The model will be extended with the interaction variable “DummyLT”. A median split will be made on the variable equity-compensation/total compensation, and classify all firm-year observations as 1 if they have a value above the median, and will be classified as 0 otherwise. Furthermore a two-way interaction term (β4) will be added to depict the different outcome in the increase of SG&A costs when sales increase for the firms that use long-term schemes in comparison to the other firms. And a three-way interaction term (β5) depicts the way SG&A costs react when sales decrease for firms that focus on providing mostly long-term schemes compared to
other firms. In the model “Controlsi,t” stands for the used control variables asset intensity,
employee intensity and firm size.
The expectation is that the use of this model shows us that the firms that use long-term schemes have smaller decrease in SG&A costs when sales decrease in comparison to firms
that use different compensation schemes. The expectation is that the use of long-term schemes leads to the CEO being less inclined to immediately adjust SG&A costs when sales decline and therefore leads to a higher degree of cost stickiness. In order for this to be the case, β5 needs to be significantly smaller than zero.
To test whether there is a negative relation between short–term CEO compensation and the degree of cost stickiness (H2), the same basic cost stickiness model of Anderson et al. (2003) will be extended.
Extended model 2:
Log (SG&Ai,t/ SG&A i,t-1)= β0 + β1 Log (Revenuei,t/Revenue i,t-1) + β2 Dummyrev i,t * Log
(Revenuei,t/Revenue i,t-1) + β3 Dummyrev i,t + β4 DummyST i,t * Log
(Revenuei,t/Revenue i,t-1) + β5 Dummyrev i,t * DummyST i,t * Log(Revenuei,t/Revenue i,t-1)
+ β6 DummyST i,t * Controlsi,t
This model is extended by a different interaction variable, namely “DummyST”. A median split will be made on the variable annual bonus/total compensation, and classify all firm-year observations as 1 if they have a value above the median, and will be classified as 0 otherwise. Furthermore, like the first model, a two-way interaction term (β4) will be added to depict the different outcome in the increase of SG&A costs when sales increase for the firms that use mostly short-term schemes in comparison to the other firms. And the three-way interaction (β5) term is going to show us how SG&A costs react when sales decrease for firms that focus
on providing mostly short-term schemes compared to other firms. In the model “Controlsi,t”
stands for the used control variables asset intensity, employee intensity and firm size. The expectation is that there will be a larger change in SG&A costs for firms that provide mostly short-term compensation when sales decrease compared to other firms. In order words for this to be the case, β5 needs to be significantly greater than zero. This would suggest that the use of short-term schemes make the CEO more inclined to immediately adjust SG&A costs when sales decline and therefore leads to a lower degree of cost stickiness.
4.
Results
4.1 Descriptive statistics
Table 1 contains the descriptive statistics of the variables that are needed to use the cost stickiness model developed by Anderson et al. (2003). The numbers reported are in millions. I find that the SG&A costs have a mean value of $946,77 million. The sales revenue per year has a mean value of $6.330,04 million. The ratio of SG&A costs to revenue in the sample is 14,96%. In comparison with the sample of Anderson et al. (2003), I find that SG&A costs as a percentage of revenue is lower. They report a percentage of 26,41%. They also report a lower mean value for SG&A costs and revenue of respectively $229,45 million and $1277,09
million. This difference could be due to a different time period of the samples and the fact that a smaller sample is used in this paper. The average ratio of SG&A costs, which represents the costs of firm i in year t, relative to year t-1, is 1,3. The average ratio of revenue which
represents the sales revenue of firm i in year t, relative to year t-1 is 1,9.
Table 1: descriptive statistics
Mean Std.
Deviatio n
10% 25% 75% 90%
SG&A Costs ($ mil) 946,77 3.776,05 8,65 28,27 430,5 1704,95
Revenue ($ mil) 6.330,0 4 27.795,9 1 31,45 128,09 2.540,0 8 10.364,5 3 SG&A Costs as a % of Revenue 14,96% 13,58% 27,50 % 22,07 % 16,95% 16,45%
Log(SG&Ai,t/ SG&A i,t-1) 1,3 2,8 0,54 0,73 1,42 2,09
The model of Anderson et al. (2003) is extended with three control variables. Table 2 reports the descriptive statistics of the control variables. I find that the ratio for asset intensity (total assets divided by total sales revenue) averages 1,49 with a median of 1,09. The ratio for employee intensity (total sales revenue divided by total amount of employees) averages 4,67 and has a median of 3,22. The firm size is measured by the natural logarithm of total assets of a firm per year. I find that the average value of total assets is $9.657,80 million and has a median of $2.136,77 million.
Table 2: descriptive statistics control variables
Mean Median Std. Deviation 10% 25% 75% 90% Firm size ($ mil) 9.657,80 2.136,77 23.659,18 25,9 113,9 2.611,30 9.847,66 Asset Intensity 1,49 1,09 1,53 0,41 0,64 1,95 3,68 Employee Intensity 4,67 3,22 5,79 0,55 1,65 5,92 10,33
Table 3 gives the statistics with regards to the compensation scheme used by the firms. The numbers reported are in thousands. Firms provide their CEO on average a BONUS of $155,32 thousand per year. The median of BONUS is $0,00. This median is so low due to the fact that a lot firms do not provide their CEOs with BONUS which in turn leads to a high amount of variables with 0 BONUS. The average amount of NON-EQUITY INCENTIVES received by CEOs is 951,10 thousand (median = $495,00 thousand). The total value of OPTIONS
awarded to CEO has a mean of $975,49 thousand. The total value of STOCK awarded to the CEO amounts to a mean of $1,6 million (median = $529,60 thousand). Firms have provided CEOs a total average amount of compensation of $4,44 million (median = $2,87 million). I also find that the firms award the CEO on average with a lower total amount of short-term compensation compared to long-term compensation. The average total value of SHORT-TERM as a percentage of TOTAL COMPENSATION is 25,10% whilst LONG-SHORT-TERM as a percentage of TOTAL COMPENSATION is 65,78%.
Table 3: descriptive statistics CEO compensation schemes
($ Thousands) Mean Median Std.
Deviation
10% 25% 75% 90%
Bonus 155,32 0.000 8.526,16 0 0 0 450
Non-Equity 951,1 495 1.616,92 0 0 1180,26 2249,33 Fair value Options 975,49 0.000 2.471,69 0 0 1113,98 2843,1 Fair value Stocks 1.671,02 529,6 3.123,27 0 0 2300 5007,79 Total-Comp (TDC1) 4.441,05 2.869,18 5.037,80 741,44 1472,31 6152,96 10732,02 Total LT-Comp 2.921,42 1.489,62 4.952,88 0 297,84 3698,01 7197,49 Total ST-Comp 1.114,70 620 1.839,76 0 168,66 1377,5 2587,04 % ST-Comp 25,10% 21,61% 36,52% 0,00% 11,46% 22,39% 24,11% % LT-Comp 65,78% 51,92% 98,31% 0,00% 20,23% 60,10% 67,07%
In table 4 the Pearson correlation is reported for the model used to determine the cost
stickiness induced by long-term compensation schemes. The table reports the Spearman and Pearson correlation coefficients between the SG&A costs, revenue and control variables. I find that there is high correlation (0.768) between the logarithm of the SG&A costs ratio (SG&A i,t / SG&A i,t-1) and the logarithm of the revenue ratio (Revenue i,t /Revenue i,t-1). This means that the yearly changes in SG&A costs and yearly changes in revenue maneuver in similar direction. Furthermore I find that the three control variables have low correlation with the changes in SG&A costs and changes in revenue. Moreover I find that the results for the control variables are all negative which would imply that they move in the opposite direction. For both the short- and long-term models used, I find that multicollinearity is not an issue. I find that the VIF of the independent variables are lower than 5 which means that the tolerance (1/VIF) is higher than 0.1.
Table 4: Pearson correlation long-term regression model
Log (SG&Ai,t/
SG&A i,t-1)
Log (Revenuei,t/ Revenuei,t-1)
Log (SG&Ai,t/ SG&A i,t-1) 1.000 .768
Firm size -.152 -.214
Asset intensity -.011 -.013
Employee intensity -.032 -.158
In table 5 the Pearson correlation is reported for the model used to determine the cost stickiness induced by short-term compensation schemes. I find that there is high correlation
(0.845) between the logarithm of the SG&A costs ratio (SG&A i,t / SG&A i,t-1) and the
logarithm of the revenue ratio (Revenue i,t /Revenue i,t-1). Furthermore I find that the three
control variables here also have very low correlation with the changes in SG&A costs and changes in revenue. Moreover I find that the results for the asset intensity and employee intensity are negative, except for the correlation between the logarithm of employee intensity and the logarithm of SG&A costs. The logarithm of firm size has low correlation with the logarithm of SG&A costs and revenue but is positive in contrast to the first model.
Table 5: Pearson correlation short-term regression model
Log (SG&Ai,t/
SG&A i,t-1)
Log (Revenuei,t/ Revenuei,t-1)
Log (SG&Ai,t/ SG&A i,t-1) 1.000 .845
Firm size .035 .015
Asset intensity -.007 -.045
4.2 The effects of long-term CEO compensation on cost stickiness
The first hypothesis of this research paper reflects the expectation that there is a positive relation between long–term CEO compensation and the degree of cost stickiness. This means that the firms that make relatively more use of long-term compensation schemes to incentivize the CEO, have a higher degree of cost stickiness. In the first model this hypothesis is tested.
I have added the control variables asset intensity, employee intensity and firm size because they could have an effect on the outcome of this research. In the regression model, I find that both firm size and employee intensity are statistically significant (p < 0.05).
However I find asset intensity does not have a significant effect with a p-value of 0.173 (p > 0.05). This would imply that asset intensity does not have an effect on the outcome of the research. This is not in line with the research paper of Anderson et al. (2003) which focuses on the decision made by the CEO to commit to resources, which in turn leads to asymmetric cost behavior. Despite the high p-value I decided to keep asset intensity in the model because they are used in most papers (e.g. Chen et al., 2012; Dierynck et al., 2012; Dalla & Perego, 2014) that test for cost stickiness and therefore I think it is necessary that this variable is kept in the model. Furthermore this model will also be used to test the effects of short-term CEO compensation on cost stickiness, where asset intensity could in fact have an effect on the outcome of the research.
The model summary shows that the R square is 0.601. This means that approximately 60% of the variance in the dependent variable is explained by the model. The regression analysis is conducted with 2425 observations. Table 6 reports the results of the conducted regression analysis. I find that the increase in sales revenue of 1 % leads to 0.81 % (β1) increase in SG&A costs. When sales revenue decline with 1 % I find that the SG&A costs decrease with 4.69 % (β1+ β2). When CEOs are mostly incentivized through long-term incentives, an increase of 1 % in sales revenue then leads to 0.86 % (β1+ β4) increase in SG&A costs. The use of long-term incentives in combination with a decrease of sales revenue of 1 % leads to a decrease of 1.66 % in SG&A costs (β1 + β2 + β5). The difference that occurs between sales revenue decrease without the use of long-term incentives and the occurrence of sales revenue decrease when CEOs are incentivized with long-term
compensation schemes can be found by looking at the β5 coefficient. In order for the first hypothesis to be supported, the β5 coefficient needs to be significantly smaller than zero. In line with the formulated expectation, the results of the regression show that the β5 coefficient is -3.03 % (β5: Dummyrev i,t * DummyLT i,t * Log (Revenue i,t /Revenue i,t-1). This
would suggest that the use of long-term compensation schemes leads to 3.03 % smaller decrease in SG&A costs for when 1% decrease in sales revenue occurs. In turn this would lead to the assumption that the provision of long-term incentives leads to stickiness of SG&A costs being more prevalent. However, I also find that the β5 coefficient is not significant with a p-value of 0.447 (p > 0.05), which means that this outcome is not statistically reliable. This is probably due to the high standard error of 3.98. It could also be due to the more recent samples used in this paper and firms nowadays are relatively more flexible and know how to cut committed resources when demand declines without harming the long-term firm value. Therefore I cannot interpret the outcome of this model and I do not find support for the first hypothesis, which states a positive relation between long–term CEO compensation and the degree of cost stickiness.
Model summary:
N R Square Adjusted
R Square P-value
2425 .601 .600 0.000
Table 6: effect of long-term compensation schemes on cost stickiness
B Std. Error t Sig. Firm size .009 .004 2.572 .010 Asset intensity .009 .007 1.362 .173 Employee intensity .039 .005 7.896 .000
β1: Log (Revenuei,t/Revenue i,t-1) .814 .019 42.504 .000 β2: Dummyrev i,t * Log (Revenuei,t/Revenue i,t-1) 3.876 3.620 1.071 .284
β3: Dummyrev i,t -.058 .035 -1.677 .094
β4: DummyLT i,t * Log (Revenue i,t /Revenue i,t-1 )
.043 .028 1.553 .121
β5: Dummyrev i,t * DummyLT i,t *
Log(Revenuei,t/Revenue i,t-1) -3.032 3.985 -.761 .447
4.3 The effects of short-term CEO compensation on cost stickiness
For the second hypothesis of this research paper it is expected that there is a negative relation between short–term CEO compensation and the degree of cost stickiness. This would mean that the firms that make relatively more use of short-term compensation schemes to
incentivize the CEO, results into SG&A costs being sticky. Table 7 reports the results of the regression analysis. I find that all three control variables are statistically significant (p < 0.05). The model summary shows that the R square is 0.722. This means that approximately 72% of the variance in the dependent variable is explained by the model. The regression analysis is conducted with 3408 observations. I find that the increase in sales revenue of 1 % leads to 0.89 % (β1) increase in SG&A costs. When sales revenue decline with 1% I find that the SG&A costs decrease with 0.75 % (β1+ β2). When CEOs are mostly incentivized through short-term incentives, an increase of 1% in sales revenue then leads to 0.84 % (β1+ β4) increase in SG&A costs. The use of short-term incentives in combination with a decrease of sales revenue of 1% leads to a decrease of 0.93 % in SG&A costs (β1 + β2 + β5). The
difference that occurs between sales revenue decrease without the use of short-term incentives and the occurrence of sales revenue decrease when CEOs are incentivized with short-term compensation schemes can be found by looking at the β5 coefficient. In order for the first hypothesis to be supported, the β5 coefficient needs to be significantly larger than zero. In line with the formulated expectation, the results of the regression show that the β5 coefficient is 0.18 % (β5: Dummyrev i,t * DummyST i,t * Log(Revenue i,t /Revenue i,t-1). This would mean that the use of short-term compensation schemes leads to 0.18 % larger decrease in SG&A costs for when 1% decrease in sales revenue occurs. In turn this would lead to the assumption that the provision of short-term incentives leads to stickiness of SG&A costs being less prevalent. I find that the β5 coefficient is significant with a p-value of 0.025 (p < 0.05), which means that this outcome is statistically reliable. Hence, I find support for the second hypothesis which states a negative relation between short–term CEO compensation and the degree of cost stickiness.
Model summary:
N R Square Adjusted
R Square P-value
3408 .722 .721 0.000
Table 7: effect of short-term compensation schemes on cost stickiness
B Std. Error t Sig. Firm size .008 .003 2.762 .006 Asset intensity .025 .006 4.408 .000 Employee intensity .033 .004 7.922 .000
β1: Log (Revenuei,t/Revenue i,t-1) .888 .036 24.981 .000 β2: Dummyrev i,t * Log (Revenuei,t/Revenue i,t-1) -.141 .075 -1.885 .059
β3: Dummyrev i,t -.006 .006 -.971 .332
β4: DummyST i,t * Log (Revenue i,t /Revenue i,t-1 )
-.046 .038 -1.216 .224
β5: Dummyrev i,t * DummyST i,t *
Log(Revenuei,t/Revenue i,t-1) .182 .081 2.248 .025
4.4 Additional analysis
4.4.1 Median split
An additional analysis has been conducted to determine the effects of short- and long-term compensation schemes on cost stickiness. The original model of Anderson et al. (2003) will be used, without the extension made prior in this paper. This model is tested through a regression analysis with the subsample of firms that mostly use long-term compensation schemes versus firms that mostly don’t use long-term compensation schemes (LT = low vs. high). A median split is made to look at firms that mostly use grant date fair value of options granted and grant date fair value of stock awarded under plan-based awards versus firms that don’t. The same model will be used with a subsample of firms that mostly use short-term compensation schemes versus firms that mostly don’t (ST = low vs. high). Another median split is made to find out how cost stickiness is affected by firms that mostly use bonus and non-equity incentive plan compensation versus firms that don’t. This makes it possible to compare the differences and gives a different insight on whether CEO compensation has an effect on cost stickiness. The β1 and β2 coefficient will be used from each model to analyze and compare the differences.
Table 8 in the appendix reports the results for the subsample of firms of LT = low. The β1 coefficient is 0.824 with a p-value of 0.000 (p < 0.05) and the β2 coefficient is 0.014 with a p-value of 0.779 (p > 0.05). I find that the increase in sales revenue of 1 % leads to 0.82 % (β1) increase in SG&A costs. When sales revenue decline with 1 %, I find that the SG&A costs decrease with 0.84 % (β1+ β2). The results suggest that there is no presence of cost stickiness. However, since the P-value of the β2 coefficient is not significant, the outcome is not statistically reliable. Therefore I cannot interpret the outcome.
Table 9 in the appendix reports the results for the subsample of firms of LT = high. The β1 coefficient is 0.874 with a p-value of 0.000 (p < 0.05) and the β2 coefficient is -0.012 with a p-value of 0.794 (p > 0.05). This seems to suggest that the increase in sales revenue of 1 % leads to 0.87 % (β1) increase in SG&A costs. When sales revenue decline with 1 % however, the SG&A costs decrease with 0.86 % (β1+ β2). The results suggest that there is a very miniscule presence of 0.01 % cost stickiness. Comparing this outcome with the outcome of Anderson et al. (2003), they found that SG&A costs increase on average 0.55% per 1% increase in sales but decrease only 0.35% per 1% decrease in sales, which means cost stickiness is more prevalent in their research paper. However since the P-value of the β2
coefficient is once again not significant, the outcome is not statistically reliable. Therefore I cannot interpret this outcome.
Ultimately the outcome of this additional analysis leads to me having to reject the hypothesis. This means I do not find a positive relation between long–term CEO
compensation and the degree of cost stickiness. Not only does LT = low suggest that there is no presence of cost stickiness and LT = high suggests a miniscule 0.01 % stickiness of SG&A cost, but the results are also not significant due to a p-value larger than 0.05. Hence, the test provided little or no evidence that the null hypothesis is false.
In table 10 the results can be found for the subsample of firms of ST = low. The β1 coefficient is 0.869 with a p-value of 0.000 (p < 0.05) and the β2 coefficient is -0.212 with a p-value of 0.015 (p < 0.05). I find that the increase in sales revenue of 1 % leads to 0.87 % (β1) increase in SG&A costs. When sales revenue decline with 1 %, I find that the SG&A costs decrease with 0.66 % (β1+ β2). The results suggest that there is presence of cost stickiness.
In Table 11 the results are reported for the subsample of firms of ST = high. The β1 coefficient is 0.845 with a value of 0.000 (p < 0.05) and the β2 coefficient is 0.063 with a p-value of 0.014 (p < 0.05). This means that the increase in sales revenue of 1 % leads to 0.85 % (β1) increase in SG&A costs. When sales revenue decline with 1 % however, I find that the SG&A costs decrease with 0.91 % (β1+ β2). The results suggest that cost stickiness is less prevalent when CEOs are mostly incentivized through short-term compensation schemes. To conclude I find that the second hypothesis, which states that there is a negative relation between short–term CEO compensation and the degree of cost stickiness, is supported. The hypothesis is supported even though ST = high reports a relatively small increase in the declination of SG&A costs when sales revenue decrease with 1%. Furthermore I find that ST = low does lead to cost stickiness whilst ST = high seems to lead to cost
stickiness being less prevalent which strengthens the support for the hypothesis.
4.4.2 Alternative measurement of short-term compensation
Finally an additional analysis has been performed by using the extended model of Anderson et al. (2003). A different approach is taken to find out whether this would also lead to the second hypothesis being supported. In this analysis the variable non-equity incentive plan compensation is left out. This means that for this test the short-term compensation scheme only exists of bonus. Bonus and non-equity incentive plan compensation are both considered
to be short-term compensation. However bonus is discretionary based while non-equity incentive plan compensation is considered to be non-discretionary based. This means that bonus is not a forward-looking measure like non-equity incentive plan compensation is. The CEO knows that good performance will get him a bonus at the end of the year. However, he does not know how good he has to perform to acquire that bonus. On the other hand, in order for the CEO to acquire the non-equity incentive plan compensation he has to meet the pre-specified requirements. Performing this additional analysis allows me to find out whether a negative relation exists between short-term compensation schemes that are discretionary based and the degree of cost stickiness.
In order for the test to be supported, the β5 coefficient needs to be significantly larger than zero. The results of the regression show that the β5 coefficient is 1.583 % with a p-value of 0.675 (p > 0.05). This would suggest that the use of bonus as CEO compensation leads to 1.58 % larger decrease in SG&A costs for when 1 % decrease in sales revenue occurs. In turn this would lead to the assumption that the use of bonus as CEO compensation leads to
stickiness of SG&A costs being less prevalent. However, since the P-value of the β5 coefficient is not significant, the outcome is not statistically reliable. Therefore I cannot interpret this outcome. The reason for this could be due to a high standard error of 3.78. It could also mean that the use of a non-forward-looking incentive does not give the CEO the incentive to manipulate to outcome of firm profit and also his annual bonus. This could be due to the fact that bonus is not pre-specified and does not have to be tied to firm profit. To
conclude, I do not find support for the notion that the use of bonus as CEO compensation leads to stickiness of SG&A costs being less prevalent.
5.
Conclusion
This paper aims to contribute to prior literature by researching whether CEO compensation leads to costs stickiness. Anderson et al. (2003) suggest that managers’ decision to reduce the resources and pay the adjustment cost or to choose for the costs of not fully using the capacity of the firm determines the asymmetrical cost behavior. Managers’ decision making behavior could be influenced by a manager’s compensation scheme or behavioral biases. Murphy and Jensen (2011) find that the CEO, who is incentivized through short-term compensation based on accounting profits, is more inclined to cut research and development expenses even if this has negative impact on firm value in the long run. This leads to a lower degree of cost stickiness. Dierynck et al. (2012) repeats this notion. On the other hand long-term
compensation schemes could lead to the manager being inclined to avoid cutting committed resources because he is afraid that doing so is at the expense of long-term earnings. Hence, this would lead to the cost stickiness degree of the firm to increase. To my knowledge this is one of the first papers that examines the effect of CEO compensation on the degree of SG&A cost stickiness by making a distinction between short- and long-term CEO compensation schemes.
The study is conducted using the cost stickiness model of Anderson et al. (2003). This model is extended to test whether CEO compensation affects the degree of cost stickiness. Three control variables have been added to control for the factors that influence cost stickiness, namely: asset intensity, employee intensity and firm size. The final sample for short-term CEO compensation schemes consists of 3408 observations. The final sample for the long-term CEO compensation schemes consists of 2429 observations. The research is conducted over the sample period of 2007 to 2014.
The outcome of this paper deviates from prior literature that examines the impact of managerial incentives on cost stickiness. No evidence has been found regarding a positive relation between long–term CEO compensation and the degree of cost stickiness. The expectation was that predicting future demand due to for example seasonality, would lead to managerial decisions to commit to resources even though demand is declining. When demand keeps declining and the CEO waits too long with taking back committed resources, he will have costs building up because of unutilized resources and of unused capacity of the firm. However the assumption is that the firms in the more recent samples are relatively more flexible and know how to cut committed resources when demand declines without it harming the long-term firm value.
