The moderating effect of CEO horizon on compensation and cost stickiness

The moderating effect of CEO horizon on compensation and cost

stickiness

Name: Sander Vermeulen Student number: 11130822

Thesis supervisor: M. Schabus Msc. Date: 16 June 2017

Word count: 10579

MSc Accountancy & Control, specialization Control

Statement of Originality

This document is written by student Sander Vermeulen 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.

Abstract

This paper investigated the relation between short-term and long-term compensation on cost stickiness. Short-term compensation incentivized CEOs to meet annual goals while long-term compensation motivates them to meet goals with a longer horizon than one year. Furthermore it is indicated that those two different terms of compensation make decisions differently. Therefore, I expect that those different terms of compensation effects cost stickiness differently. Besides looking for different effects of short-term and long-term compensation I looked at the moderating effect of CEO-horizon on these relationships. To test the effect of compensation on cost stickiness I conducted a sample of 7,622 observations. For testing the moderating effect of CEO-horizon I split the sample into two subsamples. One sample contains data of CEOs that works relatively long at the firm and a sample with data of CEOs that works relatively short at the firm. I find that short-term compensation leads to weaker cost stickiness. I did not find that long-term compensation makes the effect of cost stickiness stronger. The results did not support the moderating effect of CEO-horizon on both terms of compensation on cost stickiness.

Contents

1 Introduction ... 6

2 Literature review and hypotheses formulation ... 9

2.1 Cost stickiness ... 9

2.1.1 Cost stickiness ... 9

2.1.2 Organizational determinations of cost stickiness ... 9

2.1.3 Managerial factors of cost stickiness ... 10

2.2 Executive compensation plans ... 12

2.2.1 Agency theory ... 12

2.2.2 Executive compensation ... 13

2.2.3 Effects of long-term and short-term executive compensation on CEO behavior .... 14

2.2.4 Horizon problem and CEO tenure ... 16

2.3 Hypothesis formulation ... 17

3 Research method ... 20

3.1 Sample selection ... 20

3.2 Measurement of costs stickiness ... 21

3.3 Measurement of types of compensation ... 22

3.4 CEO horizon and control variables ... 22

3.5 Empirical model ... 23

4 Descriptive statistics and empirical results ... 25

4.1 Descriptive statistics ... 25

4.3 The moderating effect of CEO-horizon on the relation between short-term

compensation and cost stickiness ... 28

4.4 The moderating effect of CEO-horizon on the relation between long-term compensation and cost stickiness ... 31

5 Conclusion ... 34

6 Tables and figures ... 36

7 Referencing and literature list ... 37

1 Introduction

In cost accounting textbooks is assumed that costs have a fixed or variable characteristic regarding to activity. Recently academics found more evidence that these characteristic of either fixed or variable is not consistent with how costs are actually managed. Bad understanding of cost behavior can lead to inaccurate decision making and therefore more knowledge of cost stickiness is required. Noreen and Soderstrom (1997) were one of the first who found the inconsistency and were supported by a large scale data study Anderson, Banker, and Janakiraman (2003). They found that costs are reacting on whether activity is increasing or decreasing. Whether activity is increasing or decreasing the costs are changing in different magnitudes. Costs are sticky if they increase more with activity increase than they do decrease with activity decrease. More specifically they found that selling, general, and administrative (SG&A) costs increase 0.55% per 1% increase in revenue but decrease only 0.35% per 1% in revenue decrease and thus behave sticky. This can be explained through an alternative model of cost behavior in which managers on purpose adjust resources when there are changes in volume. Managers have to deal with uncertainty whether sales are going up or going down in the future but have occurred committed resources and need to make adjustment costs to reduce or restore this. Managers may be deliberately delay reductions to committed resources until they have more certainty about whether the decline in demand is for a longer period.

Cost stickiness is related to managerial behavior and thus to the decisions managers make to adjust resources to changes in activity. In firms where ownership and control are separated owners influence the behavior of the managers. The agency theory explains that when ownership and control are separated a conflict of interest occurs between managers and owners (Jensen and Meckling, 1976). To bring the interests of the manager in line with the interests of the owner, firms make use of executive compensation plans. Common components of these plans are base salary, annual bonus, long term incentive plans, restricted option grants, and restricted stock grands. Salary and annual bonus plans are short-term compensation which consists of payments based on performance of one year and shorter. The other components are based on performance of multiple years and are therefore part of long-term compensation. Both types of compensation plans motivate managers differently as is argued by Jensen and Murphy (2011). They explain that managers who are awarded

according annual accounting profits, they will be motivated to increase short-term profits. Compensate managers with equity-based incentives which are mostly seen as long-term incentives and they will make actions and decisions what increase the firm value for longer period. Thus owners can influence managers to make decision in a certain way and thus also the strength of costs stickiness. To my knowledge the relationship between short-compensation and cost stickiness and the relation between long-term short-compensation is not intensively studied because I was not able to find many results.

Managers that have compensation plans dominated by short-term incentives will cut costs more when sales are going down to meet annual targets. Hence, short-term compensation may lead to less cost stickiness. Managers who are compensated more with long-term compensation will wait longer until they are more certain about the persistence of a sales decrease. Hence, I expect long-term compensation to be positively related to cost stickiness. Although I could not find many results of papers about these relationships, the studies I found used the base model of Anderson, Banker, and Janakiraman (2003). In this paper the less widely direct measure of Weiss (2010) is used because this study will go one step further. This study will look at the moderating effect of CEO-horizon on the relation between both short-term and long-term compensation on cost stickiness. Jian and Lee (2011) argue that CEOs with higher level of reputation are choosing investments with higher net present value. A high level of CEO tenure might be a result of high level of reputation because firms are selecting those CEOs that maximize their firm value the most. CEOs with a high level of tenure at the firm are probably CEOs that creates high firm value otherwise another CEO would have been selected. Dechow and Sloan (1991) argue that CEOs that are almost retired are less incentivized by the level of compensation. In other words, CEOs with relatively long tenure are less sensitive to marginal changes in the compensation structure. Hence, I expect that for CEOs with high tenure, compensation (independent of whether it is short-term or long-term compensation) has less influence on cost stickiness. Therefore, I hypothesize that CEO-horizon will moderate the effect of short-term and long-term compensation on cost stickiness. In my knowledge there are no studies so far that are looking at a moderating effect of CEO-horizon on the relation between CEO compensation and cost stickiness. The moderating effect on CEO compensation and cost stickiness makes this study more interesting and more unique. To measure the effects of compensation on cost stickiness and the moderating effect of CEO-horizon this research will control for the change in GDP,

asset intensity, employee intensity, sales decreases in two successive years and operational leverage because different researcher found effects between these variables and cost stickiness.

As far as the author knows, a comparable study has not been executed, thus this research fulfilled the gap in the literature about the influence of compensation on the behavior of CEOs. More specifically the effect of short-term compensation and long-term compensation on cost stickiness and the moderating effect of CEO-horizon on these relationships. Studying the phenomenon cost stickiness will help us to better understand how costs actually behave and it allows us to improve making decisions.

The remainder of this paper is structured as followed. In the second section an in depth analysis of the literature is provided about cost stickiness, CEO-compensation and CEO-horizon. This literature review is the base of the formulation of the hypotheses which are conducted as final in this sector. The third sector explains the development of the sample, the different variables and information about the model. Section four shows and describes the descriptive statistics and the results. In the last section the conclusions and implications will be highlighted.

2 Literature review and hypotheses formulation

2.1 Cost stickiness

2.1.1 Cost stickiness

Traditional cost accounting literature assumes that costs react proportionally to changes in activity level. When the level of activity increases with one percent, the costs also rise with one percent and the same applies for decreases in activity level. The last two decades this assumption has become part of a debate. Although some studies already showed that costs don’t behave proportionally (Banker and Johnson, 1993; Noreen and Soderstrom, 1997), the first large scale data study to confirm cost stickiness was the study of Anderson, Banker, and Janakiraman (2003). Their article provide evidence, using a large sample of almost 65,000 observations, that selling, general, and administrative costs (SG&A) are sticky. SG&A costs increase by 0.55% per 1% increase in sales while they decrease by 0.35% per 1% decrease in sales. Dierynck, Landsman and Renders (2012) focus on labor costs which have a more variable characteristic in contrary with SG&A costs. They find that when revenues increase with 1%, the labor costs increase with 0.6% and when revenues decrease with 1% the labor costs decrease with 0.34%. Thus, whether the costs are fixed or variable, for both types costs stickiness is present.

The phenomenon of cost stickiness exists but it seems that it arises not at all level of activity changes. Subramaniam and Weidenmier (2003) examined if the magnitude of activity changes influence the level of cost stickiness. They found that SG&A and Cost of goods sold (CGS) do not show asymmetric cost behavior if revenue changes are small. When revenues change by more than ten percent, the cost behaves asymmetric. Both fixed, SG&A costs, and variable, CGS costs, behave sticky and this behavior arise when the level of activity changes by more than 10 percent.

2.1.2 Organizational determinations of cost stickiness

Subramaniam and Weidenmier (2003) found evidence about inter-industry differences in the determinations of stickiness. They examined five determinations of sticky cost behavior: fixed assets intensity, employee intensity, inventory intensity, interest ratio, and decrease in revenue for two consecutive years. They find that manufacturing industry is highly sticky because of the high level of fixed assets and inventory. Also, the service industry faces a certain level of stickiness because of employee- and inventory intensity while industries with high level of competition are less sticky. Calleja, Steliaros, and Thomas (2006) confirm that the level of cost stickiness is impacted by firm-specific and industry characteristics. In their study they also compared the level of cost stickiness among US, UK, French, and German firms and find differences. UK and US firms exhibit less cost stickiness behavior than German and French firms. They presume that this is caused by differences in corporate governance and managerial oversight systems. Another determination of cost stickiness is the usage of capacity (Balakrishnan, Petersen, and Soderstrom, 2003). Companies that have irrationally excess capacity will adapt their level of costs quicker when activity level is decreasing than companies that face strained capacity. The latter will first make the capacity less strain before they decrease the level of activity and thus show more cost stickiness behavior. Banker, Byzalov, and Chen (2013) studied the association between employment protection legislation (EPL) and cost stickiness. EPL determines the level of firing costs and therefor labor adjustment costs. If it is more expensive to fire people than it is more costly to adjust resources downwards. They find that stricter EPL is associated with a greater degree of stickiness of operating costs.

The level of cost stickiness is different among types of firms, industries, economic, legal environment and a lot of other factors. Besides these determinations, cost stickiness arises because managers make decisions how they change the allocation of resources when sales or activity levels are changing. Anderson and Lanen (2007) explain in their paper that the traditional model of cost behavior is not in line with how managers actually manage costs. Managers are actively engaged with cost management and therefore cost are not asymmetric. The next paragraph provides more insight in the role and behavior of managers by the phenomenon of cost stickiness.

2.1.3 Managerial factors of cost stickiness

The findings of Anderson, Banker, and Janakiraman (2003) on cost stickiness are consistent with an alternative model of cost behavior where managers purposefully adjust resources as reaction on changes in volume. This model distinguishes between costs that move mechanistically with changes in volume and costs that are determined by the resources committed by managers. Many firms face uncertainty about future demand and must incur adjustment costs for committed resources. CEOs delay deliberately reductions to committed resources until more certainty arise about the permanence of a decline in demand. Cost stickiness in subsequent periods may be reversed.

Chen, Lu, and Sougiannis (2012) find that cost stickiness is explained by empire building incentives which are a result of the agency problem. Empire-building theory explains that managers are increasing personal benefits through the tendency of growing firms beyond its ideal size. For example managers taking over other companies even if the net present value of the merge is negative to create certain status or prestige (Chen, Lu, and Sourgiannis, 2012). They find that empire building behavior is positively associated with the degree of cost stickiness. SG&A costs include most overhead costs and thus empire building managers are more likely to increase these costs when sales go up and are less willing to decrease costs when sales go down. They also find that good corporate governance mitigates this. Another example of managerial behavior is found by Dierynck, Landsman, and Renders (2012) by looking at the influence of managerial incentives to meet or beat zero earnings benchmark. They used Belgium firms in their sample because external auditors have to issue the going-concern opinion of these firms when a loss is reported in two consecutive years. This going-concern opinion can impair the creditworthiness of the firm. Belgium firms exercise a lower degree of cost stickiness because they have more incentives to cut labor costs when sales are decreasing and are less eager to increase labor costs when sales are increasing. They also conclude that cost stickiness arise form cost management or managerial decision making, although in this case the cost stickiness is lower than firms who not face the pressure form earnings management. Kama and Weiss (2010) find that managers who face incentives to meet targets or avoid losses accelerate downward adjustment of slack resources when sales are going down and thus decrease the level of costs stickiness. Like more papers they find that understanding the determinants of cost stickiness should be found with

managers’ motivation and agency-driven incentives. The previous papers in this subparagraph showed that cost stickiness is related to managerial behavior.

Wiersma (2010) studies the impact of the reward structure on stickiness. More precisely he looked at how the relative level of short term oriented bonuses in the CEO’s reward system effects cost stickiness. He claims that when managers have a reward structure that consists relatively more short term compensation, they have more incentives to decrease the resource level directly after sales decreases. When annual bonuses are a smaller part of the total compensation, managers will adjust resources later because they are not certain whether sales are going up again or keep on decreasing. Chen, Lu, and Sougiannis (2012) find support that the degree of costs stickiness decreases with the percentage of fixed pay in a CEO’s total compensation. More non-fixed payments in compensation package leads to greater empire building behavior which is effecting cost stickiness positively.

2.2 Executive compensation plans

2.2.1 Agency theory

Eisenhardt (1989) describes two problems that arise within organizations where ownership and control are separated. In these firms principals (owners) delegate decision-making rights to agents (controllers) to manage the organization. The first problem arises because agents possess information about actions and decisions they make that principals do not have (Jensen and Murphy, 1990). The principal faces uncertainty whether the agent makes decisions that is in best interest of the principal and the principal is not well able to verify what the agent is doing because that is too difficult or too expensive (Eisenhardt, 1989). The second problem is about differences within the attitude towards risks between the agent and the principal. The principals’ objective is to maximize the value of the firm while agents also care about assuring personal wealth, job security and image (Baysinger, Kosnik. And Turk, 1991).

The problem where the agents’ interests are not in line with the interests of the principals needs to be prevented which is leading to agency costs (Jensen and Meckling, 1976). Jensen and Meckling distinguish three types of agency costs: (a) monitoring

expenditures, (b) bonding expenditures, and (c) residual loss. Monitoring expenditures arise because the principal wants to monitor and control the decisions made by the agent. Bonding cost are made to create incentives to align agents’ interests with the interest of the principal. Residual lost are the remaining agency costs despite using controls of monitoring and bonding.

The agency theory suggests that the design of executive compensation structure is dealing with the agency problem because it brings the interests of both agent as principal in alignment. Executive compensation is a way to improve the alignment of interests between agent and principal. Meckling and Jensen (1976) and Fama and Jensen (1983) claim that if the agent owns equity of the firm the agency problem decline because the agent is than sensitive for changes in corporate wealth. For example, purchasing a luxurious office will also cost the agent personal wealth when he possessed number of shares of the firm.

2.2.2 Executive compensation

Frydman and Jenter (2010) recently investigated the evolution of executive compensation. There has been done an enormous amount of research to CEO payments and there is a large heterogeneity of executive compensation structure (Murphy, 2002; Frydman and Jenter, 2010; Murphy 2012.) Commonly, pay packages are divided into five different main components of CEO compensation plans: (a) salary, (b) annual bonus, (c) payouts from long-term incentive plans, (d) restricted option grants, and (e) restricted stock grants. The last decades there is an obvious shift from a short oriented compensation to a more long-term oriented. (Frydman and Jenter, 2010; Murphy, 1999 and 2012).

Figure 1 shows the evolution of CEO compensation within the last seventy years. Till the seventies and eighties CEO compensation consist mainly salary and annual bonus (Hall and Liebman, 1998). Since then the absolute level of salary and bonus have remained the same but relatively it becomes a smaller part of the total compensation. Nevertheless, the importance of base salary did not become any less because CEOs dedicate significant attention to the level of their base salary (Murphy, 1999). Base salary is the fixed component of the compensation plan and determines the minimum payment of the CEO but also the variable payments after meeting certain performance targets. The annual bonus provides the incremental cash compensation to base salary. The bonus the CEO receives depends on the

degree to which the performance targets are achieved. Mostly the CEO bonus is based on one or more performance measures and has an explicit minimum, target, and maximum payout (Murphy, 1999).

As is visible in figure 1 the total compensation has been increased exponentially since 1980. This is caused by the amount of stocks and options as part of the total compensation that has made their rapid rise. Stocks and options are mostly used as long-term incentives (Frydman and Jenter, 2010; Murphy, 2012; Jensen and Murphy, 1990). Matolcsy, Shan and Seethamraju (2012) believe that the change from cash bonus to equity based compensation has been driven by changes in firm characteristics and by the event of CEO turnover. Businesses have been become more complex recent years and are operating in much more uncertain environments which makes it harder for the principal to monitor the agent. They also find that when firms change their CEO they also change the compensation structure which is part of the learning process of optimal contracting (Matolcsy, Shan, and Seethamraju, 2012). Restricted stocks are shares that forfeits under certain conditions, for example the manager needs to stay at the company for a certain period. Options give the manager the right to buy the shares at a pre-specified exercise price for a per-specified term. These options become vested over time: for example, 20% become exercisable in each year for five years. Many companies also make use of long-term incentive plans which is similar to annual bonus plans but are based on rolling-average three-or five-year cumulative performance.

2.2.3 Effects of long-term and short-term executive compensation on CEO behavior

Jensen and Murphy (1990) conclude in their paper that it is not about how much you pay but how you pay CEOs. Short-term incentives lead to different behavior than long-term incentives do (Larcker, 1983; Tehranian and Waegelein, 1985) describe how long-term compensation plans differ from short-term. Long-term plans commonly measure performance over a time period longer than one year change and use change in share price as a performance measure in contrast to short-term that is using accounting measures. Salary and annual bonus are thus short-term compensation based on performance of one year or less. Long-term incentive plans, restricted options and restricted stocks are often long-term focusing. Murphy and Jensen (2011) believe that almost all executive compensation plans are

designed far from optimal which limits their benefits. They also find that CEOs manipulate earnings through the use of discretionary accruals. Discretionary accruals are adjustments to cash flows which can be influenced by the CEO (Healy, 1983). For example, the CEO can choose the method of depreciating long-lived assets or change the moment of delivery of inventory.

Murphy and Jensen (2011) explained that awarding CEOs based on annual accounting profits, they will be motivated to increase short-term profits. Short-term incentives plans consist short-term backwards-looking incentives and thus CEOs will accept projects that have positive net present value on short-term while they will decline projects that add value on the long-term because CEOs do not benefit from it on short-term. The effect of short-term incentives is well shown by Dechow and Sloan (1991). They found that CEOs spend less on research and development during their final years at the firm to improve short-term earnings performance. They also find that this spending reduction is mitigated with equity ownership. Davidson et al. (2007) argue that also the compensation structure contributes to the horizon problem when the compensation is mainly based on short-term performance incentives. They found that firms where CEOs retire and the compensation is mostly short-term based, a higher amount of discretionary accruals exist.

Murphy (1985) and Hall and Liebman (1998) find a strong positive relationship between executive compensation and firm performance when equity compensation is used to motivate managers. Ofek and Yermack (2000) state that compensation consisted with stocks and options do provide incentives to managers when they are not yet substantial owners of the company. When their percentage of ownership increases the incentives of stock and option compensation decreases because managers can sell their previously owned stocks. Coles, Daniel, and Naveen (2006) did look at the relation between managerial incentives and risk-taking. They show that when the CEO’s personal wealth depends more on the stock volatility they make more risky decisions. CEOs that adopts more risky policies having compensation that have more long-term incentives and less short-term incentives. Jensen and Meckling (1976) claim that CEOs who are also owners of the firm makes decision what is more in line with the interests of the firm holders than CEOs who are not owners. Although tying compensation to stock price will not always result in the expected alignment of interests. CEOs who receive equity-based rewards face a substantial wealth penalty because their holdings cannot be diversify. In this situation, risky investments will be avoided because

their personal wealth is vulnerable for negative outcome of projects. Stock options on the other hand will tackle this problem and encourage the agent to accept more risky projects (Ryan and Wiggins, 2001).

Executive compensation has been changed over the past seventy years and has been researched for unlimited times. Executive compensation helps the owners to ensure that the managers make decisions what maximizes the value of the firm. Depending on the emphasis of the compensation plan whether it is more focusing on short term or long term it influence the behavior of the manager. If the compensation plan is more focused on the short term the managers will be more focused on short term performance. Managers will make those decisions what influence the performance on short base rather than influence performance on longer term because that is out of their horizon. Projects that are costly in the beginning but creating positive net present value over longer term would be rejected because the performance on short term is be influenced negatively and thus his annual compensation.

2.2.4 Horizon problem and CEO tenure

Smith and Watts (1982) were among the first to publish about the horizon problems. Their paper showed that the effect of incentives of future salaries decrease as the CEO is near to his retirement. These CEOs will not be motivated by future salary adjustments. The horizon problem is later confirmed by Dechow and Sloan (1991). Dechow and Sloan investigated the behavior of CEOs in their final years at the firm. CEOs become more focused on the short term through cut R&D expenditures to meet short-term performance. This horizon problem can be solved with the provision of long-term incentives (Smith and Watts, 1982; Dechow and Sloan, 1991). Horizon problem is not only the case for retirement but also when CEO decides to leave to firm for another. Gibbons and Murphy (1992) explain that CEOs who change from one firm to another have besides contract incentives also incentives from the labor market.

Also, CEO tenure influences the behavior of the CEOs and thus the level of cost stickiness. Chen, Lu, and Sougiannis (2011) find that CEO tenure influences managers’ empire building incentives. They argue that those CEOs are more likely to have them entrenched in their position because of the accumulation of power and coalition building. They have more control over the board and other internal monitoring mechanisms in the firm

and therefor are more able to pursue their own interests instead of that of the firm. When CEOs are longer at the firm they have more empire building incentives which leads to stronger cost stickiness. Zhang (2010) looks at the different effects on earnings quality between CEO’s with long tenures and CEO’s with short tenures. He concludes that CEOs that are just at the start of their tenures, need to build reputation and therefor inflate earnings. Once CEOs have established their reputation they report less aggressively to maintain their reputation. However, in their final year at the firm they report more aggressively again. Brickley, Linck, and Coles (1999) find support that longer CEO tenure weakens a positive relationship between CEO pay and stock returns. They also find support for a stronger positive relationship between CEO pay and size if the CEO tenure is longer. Finally, they find support for a stronger positive relationship between CEO pay and risk if the CEO tenure is longer. They believe that the results are consistent with the suggestion that longer tenure gives CEO time to gain power and influence to tie compensation to their own preferences.

Jian and Lee (2011) looked at the relation between CEO reputation and corporate capital investments. What they see is that the stock market has a better response to capital investments’ announcements if those firms are directed by a CEO that has a better reputation. CEOs that have better reputation are more able to choose those investments that are have the highest net present value. You might also say that CEOs choose continuously investments with positive net present value are the CEOs that build the up high levels of reputation capital. Moreover, CEO with higher tenure have more knowledge and are better informed to make better investment decisions.

2.3 Hypothesis formulation

As literature has shown cost stickiness is influenced by managerial decision making. Chen, Lu, and Sougiannis (2012) find that cost stickiness is explained by empire building incentives which are a result of the agency problem. Empire-building theory explains that managers are increasing personal benefits through the tendency of growing firms beyond its ideal size. For example, managers taking over other companies even if the net present value of the merge is negative to create certain status or prestige (Chen, Lu, and Sourgiannis, 2012). They found that empire building behavior is positively associated with the degree of cost stickiness.

SG&A costs include most overhead costs and thus empire building managers are more likely to increase these costs when sales go up and are less willing to decrease costs when sales go down. They also found that good corporate governance mitigates this. Kama and Weiss (2010) found that managers who face incentives to meet targets or avoid losses accelerate downward adjustment of slack resources when sales are going down and thus decrease the level of costs stickiness. Like more papers, they find that understanding the determinants of cost stickiness should be found with managers’ motivation and agency-driven incentives.

In public companies where ownership and control are separated arises the problem that CEO have interests that are not in line with the interest of the company. Nowadays companies are operating in a complex environment and it is more difficult and more costly to monitor CEO’s decisions and actions. As shown in the literature, executive compensation structure is used to bring the interest of the CEO in line with the interest of the shareholders. If the CEO is in his last year of contract and is dominated with short-term incentives, than the CEO will take actions that benefits him directly while there is no focus for the value of the firm on the long run. Long-term incentives will make CEOs take actions and decisions that creates value in the long run. If CEOs only have incentives for the current year than they will take actions that benefits them in short run and they are thus willing to adjust resource levels dramatically in case of level decrease to eliminate unutilized resources. CEOs adapt resources more in line with changes in sales and thus cost stickiness will be weaker. Therefore, the hypothesis 1a is formulate as followed:

H1a: Short term compensation leads to less cost stickiness

As is showed by Jian and Lee (2011) a higher level of CEO tenure leads to better investment decisions. When managers make continuously better decisions during their career they build up a higher level of reputation. Higher CEO tenure arise from higher level of CEO reputation because the market will eliminate CEOs that are making bad decisions. Investors simply do not accept those CEO losing their value and compensation of investment. Dechow and Sloan (1991) argue that when managers are reaching their retirement age they are less motivated by compensation. Thus, when CEOs are working relatively long at a firm they determine their decisions more based on experience than based on their structure of compensation. In other

words the CEO horizon problem moderates the effect of short term compensation on cost stickiness. Therefore, hypothesis 1b is formulate as followed:

H1b: CEO horizon moderates the relationship between short-term compensation and cost stickiness

On the contrary long-term compensation plans incentivize CEOs to meet targets over a longer time period. CEOs are more willing to accept costs now if it is more profitable for the future. If sales are going down, they are willing to wait for a period to see if the decrease is structural or not. CEOs are not too worried about their compensation because it is more based on long term performance targets. In a scenario where sales decreases, costs may not change for a certain time before the CEO adapts costs to the sales decrease. Therefore, I hypothesize that long-term compensation leads to more cost stickiness.

H2a: Long term compensation leads to more cost stickiness

CEO horizon will also moderate the effect of long-term compensation on cost stickiness. CEOs that have high tenure and are almost reaching their age of retirement will make their decisions more based on their experiences and less based on their structure of compensation. Therefore, hypothesis H2b is formulated as followed:

H2b: CEO horizon moderates the relationship between long-term compensation and cost stickiness

3 Research method

3.1 Sample selection

Data for this study was collected from the databases Compustat and Execucomp of Wharton Research Data Service (WRDS). The data regarding cost stickiness and control variables are contained from Compustat. The Execucomp database provides the data of CEO compensation and the moderator CEO horizon. Data of US gross domestic product is gathered from the bureau of economic analysis. The sample to measure cost stickiness consists of 475,032 observations of S&P1500 firms for the years 1992 to 2017. The data of cost stickiness is merged with the data from CEO compensation and the merged sample consists of 299,772 observations. Data where sale and cost of goods were below zero are dropped. Cases where sale were lower than the annual income before extraordinary items (ib) are also eliminated. Subsequently extreme changes in sale and ib which is the basis of calculating cost stickiness are dropped. Furthermore the data where sale and the difference between sale and ib are going in opposite direction are removed. Finally the sample is corrected for negative values of CEO tenure which results in a total drop of 52,536 observations and a remaining sample of 247,236 observations. Within the remaining sample there are still 239,614 missing variables which are also dismissed. At last the top and bottom 1% of the observations is winsorized in order to deal with outliers. The final sample used to test the hypotheses consists 7,622 observations. In table 1 is an overview of the used variables with a description and the database source.

3.2 Measurement of costs stickiness

There are two different approaches to measure costs stickiness according the literature. The first and the most widely used approach to measure costs stickiness is from Anderson, Banker, and Janakirama (2003). This approach is a cross-sectional regression model and determines costs stickiness through SG&A costs response to simultaneous changes in sales revenue and distinguishes between periods when revenue increases and revenue decreases are presented. SG&A costs are sticky if the variation of SG&A costs with revenue increases is greater than the variation for revenue decreases. The second approach is a direct measure of cost stickiness and is more accessible and applicable for this study. Weiss (2010) introduced a new measurement of costs stickiness because of a major restriction of the approach of Anderson, Banker, and Janakirama. The new approach allows for a large-scale study without constraining the analysis to firms with at least ten valid observations and at least three sales reductions during the sample period (Weiss, 2010). I use the approach of Weiss for this study to determine cost stickiness. The main variables for measuring cost stickiness are sales and earnings. For practical reasons I use periods of years instead of quarters which is normally used. Both variables are used to determine the change in cost and together with change in sale, cost stickiness can be determined:

STICKYi ,t=log

(

)

(

)

This model estimates for recent years the difference between the slope of cost decrease with decreasing sales and the corresponding slope of cost increase with increasing sales. Cost stickiness (STICKY) is defined as the difference in the cost function slope between the two most recent of the last four years from year t-3 through year t, such that sales decrease in one year and increase in the other. Costs are sticky if the measure results in a negative value.

3.3 Measurement of types of compensation

In this study I try to answer the question whether the structure of compensation is effecting cost stickiness. To measure the effect I discriminate between short-term compensation and long-term compensation. From Execucomp I retrieve data of bonus, non-equity incentive plan compensation, grant-date fair value of option awards, and grant-date fair value of stock awards. Short-term compensation (Stcomp) is the total bonus a CEO received in the current year and is the sum of bonus and non-equity incentive plan compensation long-term compensation (LTcomp) is the sum of grant-date fair value of option awards and grant-date fair value of stock.

3.4 CEO horizon and control variables

I expand this research of the effect of CEO compensation on cost stickiness through adding the moderating variable CEO horizon. If CEOs work relatively long at the firm they are more likely to determine their decisions based on their experience instead of the structure of their compensation. So whether the CEO is compensated on the short-term or on the long term does not matter when the CEO has been employed for a longer period of time at the firm. To measure the CEO horizon I build an interaction variable which is 1 for the highest 25 percent of the distribution of CEO tenure and which is 0 for the lowest 75 percent. Thus the variable is classified as 1 when the CEO is much less incentivized by compensation structures.

Anderson, Banker, and Janakiraman (2003) use three control variables to test their hypothesis in their research of cost stickiness. First, they find that the degree of cost stickiness is effected by growth in real Gross National Product (GDP). In periods where the economy is rising the degree of cost stickiness is greater because managers have a stronger believe that sales declines are just a one-off event and will increase the next period and thus they do not cut costs directly. I control for growth in GDP and I measure this by change in GDP per year. Secondly, costs are stickier in firms that require relatively more assets to support their sales. Thirdly, costs is also more sticky in firms that have a high level of employees. Both asset intensity and employee intensity increase cost stickiness because of adjustment costs that incurs to reduce committed resources. Asset intensity is measured through total assets divided by sales and employee intensity is measured by number of

employees divided by sales. Chen, Lu, and Sougiannis (2012) control for successive sales decreases because managers become more pessimistic when sales are dropped multiple periods in a row which influence the decision making whether to adapt capacity or not. They argue that sales decreases in two successive years leads to lower cost stickiness. I measure declines in sales in two successive years through creating an indicator variable which is 1 if the sales of last year is lower than the year before last year and 0 if otherwise. The last variable that I control for is operating leverage which is also used by Weiss (2010). The operating leverage is determined by sales minus cost of goods sold divided by sales. The higher the result of this variable the more the cost structure of the firm is relatively fixed. Fixed assets are harder to dismiss than non-fixed assets.

3.5 Empirical model

In this study two relational hypotheses and two moderating hypotheses are tested. To make testing the moderating hypotheses convenient I make use of a direct measure of cost stickiness at the firm level consistent with Weiss (2010):

Coststickinessi ,t=log

(

)

(

)

To test hypothesis 1a whether short-term compensation leads to less cost stickiness the following model is used:

Coststickiness❑=β 0+β 1 STcomp❑+β 2GDPchange❑+β 3 Assetintensity❑

+β 4 Employeeintensity❑+β 5 Successive❑+β 6Opleverage❑+ε

This model captures the relationship between short-term compensation and cost stickiness given that the control variables also relate to cost stickiness. The last component of the model shows the error term. To test hypothesis 1b whether CEO horizon is moderating the relationship between cost stickiness and short-term compensation the same model is used but only the sample is divided into two subsamples. One subsample contains the data with CEOs

with longer CEO-horizon which is the top 25% of CEO tenure. The other subsample contains the data with CEO-horizon equals 0 which is the bottom 75% of CEO tenure.

To test hypothesis 2a whether long-term compensation leads to more cost stickiness the model below is used:

Coststickiness❑=β 0+β 1 LTcomp❑+β 2GDPchange❑+β 3 Assetintensity❑

+β 4 Employeeintensity❑+β 5 Successive❑+β 6Opleverage❑+ε

To test the last hypothesis whether CEO horizon moderates the relationship between long-term compensation and cost stickiness the same sample split is used as hypothesis 1b.

4 Descriptive statistics and empirical results

4.1 Descriptive statistics

Table 2 contains the descriptive statistics of all the relevant variables used in this study. The dependent variable cost stickiness shows more negative values than positive values. When the value of costs stickiness is negative than the costs are sticky and thus will increase more when activity rises than it will decrease when activity falls. These dominating negative results means that the costs show a more sticky behavior. Although I looked back for four years instead of four quarters to determine cost stickiness it does not seem to influence the results. The numbers of long-term compensation (LTcomp) and short-term compensation (STcomp) are in thousands. Firms provide their CEO a long-term compensation of 2,336,480 dollars on average and a short-term compensation of 974,830 US dollars on average. The median of long-term compensation is 1,623,880 and the median of short-term compensation is 751,940. The maximum long-term compensation is 6,785,340 US dollars while the short-term compensation is 2,646,000 US dollars.

CEO horizon is an interaction variable which is either a 1 or a 0. The mean is 0.22 and the median is 0 which is the result of splitting the sample into two subsamples based on the 75th _{percentile. All observation that belongs to the bottom 75 percent results in a 0 and the} observations that belongs to the top 25 percent results in a 1. These two subsamples are used to test the moderating effect of CEO horizon. The first control variable change in GDP is measured in percentages. The average GDP change is 2.93 percent and the median is 3.70 percent. There has been a decline in GDP growth once during the financial crisis from 2008 to 2009. The mean and median of asset intensity and employee intensity are different than the results of Chen, Lu, and Sougiannis (2012). They only have used a period from 1995 to 2005 while this study covers the period from 1992 to 2017. This study has a lower average

employee intensity (3.50 against 6.21 of Chen et al.) but a higher average asset intensity (2.07 versus 1.13). The ratio between sale minus cost of goods sold and sale has a mean (median) of 0.37 (0.34). Where 1 means an extremely high level of fixed cost structure and 0 extremely low. The last control variable, sales decreases in two successive years, is an interaction

variable which is either a 0 or a 1. Sale decreases in two successive years is less common because the median is 0.

4.2 Pearson correlation regression models

Table 3 reports the Pearson correlation for the model used to determine cost stickiness induced by long-term compensation, short-term compensation and the control variables. I find that there is weak positive correlation between cost stickiness and long-term

compensation (0.0445) and this is significant. I also find that there is a weak positive correlation between cost stickiness and short-term compensation (0.1571) and this is

significant as well. Although both correlations are weak, short-term compensation is slightly stronger than long-term compensation. All control variables are positively correlated with cost stickiness except employee intensity but this correlation is not significant. Also the correlation between cost stickiness and asset intensity is not significant. All control variables are weakly correlated with cost stickiness.

Furthermore, I split the sample into two subsamples where on one hand the CEO horizon is short and on the other hand the CEO-horizon is long. Table 4 consists of the sample of short CEO-horizon. The correlation between cost stickiness and long-term

compensation became slightly stronger (0.0473) while the correlation between cost stickiness and short-term compensation became slightly weaker (0.1481). Change in GDP, operational leverage and successive decrease are still weakly correlated while asset intensity and

employee intensity remain not significant. Table 5 reports the results of the sample of long CEO-horizon which contains all CEOs with a tenure of 10 years or more. The correlation between cost stickiness and long-term compensation became in this subsample not significant and weaker (0.0233). The correlation between cost stickiness and short-term compensation on the other hand has become stronger (0.1800). The correlations between cost stickiness and the control variables do not change much except for asset intensity. The correlation between cost stickiness and asset intensity became significant.

4.3 The moderating effect of CEO-horizon on the relation between short-term compensation and cost stickiness

First I looked at the effect of short-term compensation on cost stickiness. Firms that incentivize CEOs with short-term compensation will have less cost stickiness because the CEO will adapt the resources quickly to meet short-term performance goals and to receive short-term rewards. This model uses a direct measure for cost stickiness to test hypothesis 1a: short-term compensation leads to less cost stickiness. If costs behave sticky the value of the direct measure is negative and if not the value is positive. Firms that use higher short-term compensation will thus have higher positive values of cost stickiness. Besides the variables cost stickiness and short-term compensation this model also consists of five control variables because this could have influenced the results of this test. The results of this test are reported in table 6. The regression analysis is conducted with 7,622 observations and at least least one independent variable influences the dependent variable (F = 47.94 and p=0.000). The model explains 3.64% of the variance in the dependent variable.

I find that short-term compensation has a positive relation to cost stickiness and this relation is significant (0.00 < 0.05). Firms that use higher short-term compensation will face weaker cost stickiness and therefore hypothesis 1a is supported. GDP change has a coefficient of 0.0328 which means that if the GDP increases with 1% cost stickiness increases with 0.0328. Operational leverage has the strongest significant relation with cost stickiness (0.2131) and successive decreases has a coefficient of 0.0402. Both asset intensity and employee intensity are not significant which is also the result of Wiersma (2010) only I have a positive relation between asset intensity in contrast with Wiersma. Wiersma also had a surprisingly result according GDP change where the level of cost stickiness is significant lower in years of economic growth. I find a significant positive effect of GDP change on cost stickiness. Wiersma (2010) also looked at the remaining two control variables successive decreases and operational leverage. Both I and Wiersma found a negative effect of Successive decrease on cost stickiness but my result was significant. Wiersma did not find a significant effect of operational leverage on cost stickiness but I did find a positive effect that is also significant.

Next I looked at the moderated effect of CEO-horizon on the effect of short-term compensation on cost stickiness. A longer CEO-horizon should moderate the effect of short-compensation on cost stickiness and thus the relationship between short-short-compensation and

cost stickiness should become less positive or more negative. I compared the results of the relationship between short-term compensation and cost stickiness for the group with the lowest 75% of CEO horizon (table 7) and the group with the highest 25% of CEO-horizon (table 8). The regression analysis for short CEO-horizon (long CEO-horizon) is executed with 5,661 (1,961) observations and at least one independent variable affects the dependent variable in both subsamples. The model for short CEO-horizon (long CEO-horizon) explains 3.6% (4.3%) of the variance in the dependent variable.

If the CEO works for a long amount of time at the firm costs are slight less sticky (0.00014) than if they work for a short amount of time at the firm (0.00012). Thus CEO-horizon does not moderate the relationship between short-term compensation and cost stickiness but makes it even less sticky and therefore hypothesis 1b is not supported. The effects of control variables on the other hand changed through longer CEO-horizon. Longer CEO-horizon increased the effect of successive decreases on cost stickiness (0.057). The effect of GDP change on cost stickiness decreased while the effect of the remaining control variables on cost stickiness is still not significant.

4.4 The moderating effect of CEO-horizon on the relation between long-term compensation and cost stickiness

Secondly, I looked at the effect of long-term compensation on cost stickiness. Firms that make use of long-term compensation will face more cost stickiness because these CEOs do not need to adapt their resources quickly because they are more focused on the long term. To test hypothesis 2a: long-term compensation leads to more cost stickiness, I used the same model that is used for the hypothesis 1a only with the variable long-term compensation instead of short-term compensation. In this case I am looking for lower stickiness and thus for negative values. The results of this test are reported in table 9. The regression analysis is conducted with 7.622 observations and at least least one independent variable influences the dependent variable (F = 20.05 and p=0.000). This model explains 1.56% of the variance in the dependent variable.

I find that long-term compensation has a positive relation to cost stickiness (0.0000) and this effect is significant(0.00 < 0.05) and thus hypothesis 2a is not supported. Higher

long-term compensation will lead to less cost stickiness. GDP change, operational leverage, and successive decreases are significant related to cost stickiness while asset intensity and employee intensity are not significant. The control variables are positive related to cost stickiness except for asset intensity and employee intensity.

To test the last hypothesis I used the same model that tests hypothesis 1b only the variable short-term compensation is replaced by long-term compensation. Long-term compensation leads to stronger cost stickiness and a longer CEO-horizon should moderate this effect. To test this hypothesis I split the sample into two subsamples and compared the results of both samples. Table 10 show the results of the relationship between long-term compensation and cost stickiness if CEOs work relatively short at the firm. Table 11 on the other hand provides results of the relationship between long-term compensation and cost stickiness if CEOs work relatively long at the firm. The regression analysis for short CEO-horizon (long CEO-CEO-horizon) is executed with 5,661 (1,961) observations and at least one independent variable affects the dependent variable in both subsamples. The model for short CEO-horizon (long CEO-horizon) explains 1.78% (1.43%) of the variance in the dependent variable.

In the sample with CEOs with longer CEO-horizon the effect of long-term compensation on cost stickiness is almost 0 in contrast with shorter CEO-horizon (0.00001). Although the difference is extremely small hypothesis 2b is not supported. Long CEO-horizon makes the effect of long-term compensation on cost stickiness not less strong.

5 Conclusion

This study contributes to prior literature by investigating whether CEO compensation influences the level of cost stickiness. Anderson, Banker, and Janakiraman (2003) were the first researchers that used a large set of data to support that managers’ decisions determine sticky behavior of costs. Managers need to decide whether to dismiss unused capacity or to accept losses of not using the full capacity when the level of activity of the firm has been declined. Executive compensation is a way to affect the behavior of CEOs (Meckling and Jensen (1976). Wiersma (2010) is one of the few authors that investigated the relationship between short-term compensation and cost stickiness. His study claims that highly short-term incentivized CEOs adjust resources earlier to meet annual targets and thus the level of stickiness will be lower. Murphy and Jensen (2011) confirm that annual targets motivate CEO’s to increase short-term profits and thus increase the CEO’s willingness to adjust the resources if the activity level goes down. If CEOs are incentivized with long-term compensation they make decisions which are best on the long run. The future is uncertain and after a decrease in activity change the CEO should not be willing to instantly make costs to adjust their resources and therefore those costs will slack more. The different effects of the different types of compensation on cost stickiness should be eliminated by a long CEO-horizon. Jian and Lee (2011) show that a higher level tenure is a result of CEOs making continuously good decisions over their whole career. Dechow and Sloan (1991) argue that managers who are reaching their retirement age, make decisions that are not influenced by compensation plans. CEOs with longer tenure should moderate the effect of compensation on cost stickiness.

To my knowledge there has not been done a lot of research that aims to understand the relation between compensation plans and cost stickiness. Studies that look at the moderating effect of CEO-horizon on the relationship between compensation and cost stickiness seem to be even more rare. Besides the lack of studies to these relationships, I also used a unique setting by aiming to understand the relation between CEO-compensation and cost stickiness by using a direct measure of cost stickiness. Most studies used the model of Anderson, Banker, and Janakiraman (2003) while I used the direct measure from the model of Weiss (2010) that helped me to better study the moderating effect of CEO-horizon on the

relationship between compensation and cost stickiness. With all these different angles I contributed to the available knowledge and came to new insights.

The final sample consists of 7,622 observations from 1992 to 2017. The first model looked at the relation between short-term compensation and cost stickiness and contains 5 control variables. To test the effect of CEO-horizon on that relationship, the sample is split in two subsamples. One subsample contains data of CEOs who worked for a relatively long period at the firm and another subsample of CEOs who worked for a relatively short period at the firm. Subsequently I looked at the different results between the two subsamples. I did the same for the effect of long-term compensation on cost stickiness and the moderating effect of CEO-horizon on this relationship.

According to prior literature, short-term compensation leads to weaker cost stickiness. I did find support regarding this relationship. CEOs that are incentivized with higher short term compensation make decisions that affect annual results to meet their short-term bonus targets. After splitting up the sample into a subsample containing CEOs with the highest 25% tenure and a subsample with the lowest 75% tenure, I did not find that CEO-horizon moderates the relationship between short-term compensation and cost stickiness. This is not as expected and therefore hypothesis 1b is not supported. Subsequently I looked at the effect of long-term compensation on cost stickiness. This study predicted that long-term compensation leads to stronger cost stickiness but did not find support for it. Finally this study looked at the moderating effect of CEO-horizon on the relationship between long-term compensation and cost stickiness. CEO-horizon did not moderate the effect and therefore hypothesis 2b is not supported.

There are several limitations on the results of this study. First, I used a period of four years instead of four quarters to determine cost stickiness which makes the changes in cost stickiness slightly larger. I looked at compensation plans that influence CEO behavior while there are more factors that influence the behavior of a CEO. These include base salary, contribution to pension plans or making promotion. I used five different control variables within my models and decided to exclude other variables. Wiersma (2010) controls for the economies of scale of a firm while Chen, Lu, and Sougiannis (2012) used different corporate governance variables that affects cost stickiness that I could have controlled for. I used data

from 1992 to 2017 while effects might have been changed over time. The use of shorter periods of data will lead to different results and insights.

6 Referencing and literature list

Anderson, M. C., Banker, R. D., & Janakirama, S. N. (2003). Are selling, general, and administrative costs "sticky"? Journal of Accounting Research, 41 (1), 47-63.

Anderson, S. W., and Lanen, W. (2007). Understanding cost management: What can we learn from the evidence on ‘sticky costs’? Working paper, University of California, Davis and University of Michigan.

Balakrishnan, R., Peterson, M. J., & Soderstrom, N. S. (2004). Does capacity utilization affect the "stickiness" of costs? Journal of Accounting Auditing and Finance, 19 (3), 283-299. Banker, R., Byzalov, D., and Chen, L. (2013). Employment protection legislation, adjustment costs and cross-country differences in cost behavior. Journal of Accounting and Economics, 55, 111-127.

Banker, R., and Johnson, H. (1993). ‘An empirical study of cost drivers in the U.S. airline industry.’ Accounting Review, 68, 576-601.

Baysinger, B., Kosnik, R., and Turk, T. (1991). Effects of board and ownership structure on corporate R&D strategy. Academy of Management Journal, 34, 205-214.

Brickley, J. A., Linck, J. S., and Coles, J. L. (1999). What happens to CEOs after they retire? New evidence on career concerns, horizon problems, and CEO incentives. Elsevier, 52(3), 341-377.

Calleja, K., Steliaros, M., & Thomas, D. (2006). A note on cost stickiness: Some international comparisons. Management Accounting Research,17, 127–140.

Chen, C. X., Lu, H., & Sougiannis, T. (2012). Managerial empire building, corporate governance, and the asymmetrical behavior of selling, general, and administrative costs.

Contemporary Accounting Research, 29 (1), 252-285.

Coles, J. L., Daniel, N. D., & Naveen, L. (2006) Managerial incentives and risk-taking.

Elsevier, 79, 431-468.

Davidson, W. N., Xie, B., Xu, W., and Ning, Y. (2007). The influences of executive age, career horizon and incentives on pre-turnover earnings management. Journal of Management

Dechow, P.M. & Sloan, R.G. (1991). Executives incentives and the horizon problem: An empirical investigation. Journal of Accounting and Economics, 14, 51-89.

Dierynck, B., Landsman, W. R., & Renders, A. (2012). Do managerial incentives drive cost behavior? Evidence about the role of the zero earnings benchmark for labor costs behavior in private Belgium firms. The Accounting Review, 87 (4), 1219-1246

Eisenhardt, K. M. (1989). Agency theory: An assessment and review. The academy of

management review, 14, 57-74.1

Fama, E., & Jensen, M. (1983). Separation of ownership and control. Journal of Law and

Economics, 26, 301-325.

Frydman, C., & Jenter, D. (2010). CEO compensation. Annual Review of Financial

Economics, 2, 75- 102.

Gibbons, R. & Murphy, K. J. (1992). Optimal incentive contracts in the presence of career concerns: theory and evidence. Journal of political economy, 100(3), 468-505.

Healy, P., (1983). The impact of bonus schemes on accounting choices, Dissertation (Univer- sity of Rochester, Rochester, NY).

Hall, B. J., & Liebman, J. B. (1998). Are CEOS Really Paid Like Bureaucrats? Quarterly

Journal of Economics, 113 (3), 653–691. http://dx.doi.org/10.1162/003355398555702

Jensen, M. C., & Meckling, W. H. (1976). Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure. Journal of Financial Economics, 3 (4), 305-360.

Jian, M., K.W. Lee (2011). Does CEO reputation matter for capital investments? Journal of

Corporate Finance. 17. 929–946.

Kama, I., & Weiss, D. (2010). Do managers' deliberate decisions induce sticky costs? Working paper.

Larcker, D. F. (1983). The association between performance plan adoption and corporate capital investment. Journal of Accounting and Economics, 5, 3-30.

Matolcsy, Z., Shan, S., and Seethamraju, V. (2012). The timing of changes in CEO

compensation from cash bonus to equity-based compensation: determinants and performance consequences. Journal of Contemporary Accounting & Economics, 8 (2), 78-91.

Murphy, K. J. (1999). Chapter 38, Executive compensation. In C. A. Orley & C. David (Eds.), Handbook of Labor Economics (Volume 3, Part 2, pp. 2485–2563).

http://dx.doi.org/10.1038/11777

Murphy, K. J. (1985). Corporate performance and managerial remuneration: An empirical analysis. Journal of accounting and economics, 7(1), 11-42.

Murphy, K. J. (2002). Explaining executive compensation: Managerial power versus the perceived cost of stock options. The university of Chicago law review, 69 (3), 847-869. Murphy, K.J. (2012). Executive Compensation: Where We Are, and How We Got There. Unpublished manuscript, University of Southern California, Marshall School of

Business.

Murphy, K. J., & Jensen, M. C. (1990). ‘CEO incentives – It’s not how much you pay, but how.’ Harvard Business Review, 138-153.

Murphy, K. J., & Jensen, M. C. (2011). CEO bonus plans: and how to fix them. Working paper.

Noreen, E., & Soderstrom, N. (1997). The accuracy of proportional cost models: Evidence from hospital service departments. Review of Accounting Studies, 2, 89–114.

Ofek, E., & Yermack, D. (2000). Taking stock: Equity‐based compensation and the evolution of managerial ownership. The Journal of Finance, 55(3), 1367-1384.

Ryan, H. and Wiggins, R. (2001). The influence of firm- and manager-specific characteristics on the structure of executive compensation. Journal of corporate Finance, 2, 101-123.

Smith Jr, C. W. & Watts, R. L. (1982). Incentive and tax effects of executive compensation plans. Australian Journal of Management, 7(2), p139.

Subramaniam, C., & Weidenmier, M. L. (2003). Additional Evidence on the Sticky Behavior of Costs. Retrieved from Working: http://papers.ssrn.com/sol3/papers.cfm?

abstract_id=369941

Tehranian, H. and Waegelein, J. F. (1985). Market reaction to hort-term executive compensation plan adoption. Journal of Accounting and Economics, 7 (1-3), 131-144.

Weiss, D. (2010). Cost behavior and analysts’ earnings forecasts. The accounting review, 4, 1441-1471. DOI: 10.2308/accr.2010.85.4.1441

Wiersma, E. (2010). The impact of the reward structure on stickiness. Working paper available at SSRN: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1668758.

Zhang, W. (2009). CEO tenure and earnings quality. Working paper university of Texas at Dallas.

Appendices

Figure 1: Evolution of CEO compensation the last seventy years

Note. From “CEO Compensation’’ by C. Frydman and D. Jenter, 2010, Annual Review of Financial Economics,

Tables

Table 1: variables, description and database source

VARIABLE DATA FROM

Cost Stickiness Compustat

Cost = Sale -/- Annual income before extraordinary items

Compustat

Short-term compensation Bonus + Non-equity incentive plan

Execucomp

Long-term compensation Grant-date fair value of option

awards + Grant-date fair value of stock awards

Execucomp

CEO horizon Fyear -/- CEO start Execucomp

Annual change in GDP GDPt=0 -/- GDPt=-1 Bureau of economic analysis

Asset intensity Assets / sales Compustat

Employee intensity Employees / sales * 1000 Compustat

Successive decreases Salest=-1 < salest=-2 Compustat

Operational leverage (sales -/- cost of goods sold) /

sales

Compustat

Table 3: Pearson correlation regression model for complete sample

Table 6: effect of short-term compensation on cost stickiness

Table 8: effect of short-term compensation on cost stickiness if CEO-horizon = 1

Table 10: effect of long-term compensation on cost stickiness if CEO-horizon = 0