Long-term behavior and cost stickiness.

(1)

Amsterdam Business School

Long-term behavior and cost

stickiness

Ilan Fernhout (10297952) 10-6-2014

Faculty of Economics and Business, University of Amsterdam MSc Accountancy & Control, specialization Control

Thesis supervisors: dr. ir. S. van Triest, prof. dr. F.H.M. Verbeeten MBA

Abstract: This research investigates the relation between long-term behavior and cost stickiness.

The proxy for long-term behavior is CEO tenure. Results imply that cost stickiness is positively related to long-term behavior. Subsequently I investigated whether this long-term behavior benefits future performance. I find that the 20 percent of the companies with the highest revenue growth (in two years’ time) currently display higher cost stickiness than other companies. Prior research argues that cost stickiness (for a part) can be interpreted as agency costs. I argue that cost stickiness can be caused by exactly the opposite of agency costs; adding value by long-term behavior.

Key words: Cost stickiness, asymmetrical cost behavior, discretionary resource adjustment,

(2)

Table of contents

1 Introduction ... 4

1.1 Background ... 4

1.2 Research questions ... 5

1.3 Motivation and contribution ... 5

1.4 Structure ... 6

2 Literature review ... 7

2.1 Cost stickiness ... 7

2.1.1 Traditional cost model ... 7

2.1.2 Asymmetrical cost model ... 7

2.2 Short-termism ... 9

2.3 CEO tenure ... 11

2.3.1 Low CEO tenure and CEO attitudes ... 11

2.3.2 High CEO tenure and CEO attitudes ... 12

2.3.3 CEO tenure and firm performance ... 13

2.4 Hypothesis development ... 18

3 Research methodology ... 19

3.1 Sample selection ... 19

3.2 Model specification and variable definitions ... 20

3.2.1 Basic model of cost stickiness ... 20

3.2.3 Robustness check ... 21

3.2.4 Future performance ... 22

3.2.5 Contingencies to industry dynamism ... 22

4 Results... 23

4.1 Descriptive results ... 23

4.2 Results of hypothesis tests ... 25

(3)

4.2.2 Cost stickiness and CEO tenure ... 25

5 Conclusion ... 33

3.2 Model specification and variable definitions ... 19

3.2.1 Basic model of cost stickiness... 19

4 Results... 22_Toc383958198 4.1 Descriptive results ... 22

4.2 Results of hypothesis tests ... 24

4.2.1 Cost stickiness – basic model ... 24

5 Conclusion ... 32

(4)

1 Introduction

1.1 Background

Costs are an important part of earnings. Cost accounting plays an important role in the analysis of costs and profitability. The better managers understand the behavior of costs, the better they can make informed decisions towards improved profitability. Traditional models of cost accounting assume variable costs to change linearly with activity. However, contemporary research indicates that costs do not change linearly with activity and react differently to an increase in activity then they react to a decrease in activity. Anderson et al. (2003) show that costs behave “sticky”. This means in the case of an upward change in activity, costs increase to a greater extent than they decrease in the case of a downward change in activity. This phenomenon is referred to them as “cost stickiness”.

There has been extensive prior research which examines the determinants of cost stickiness. This research shows that cost stickiness differs per country and per industry. Prior research also shows that cost stickiness is less when CEO incentives have a short-term character. Dierynck et al. (2012) show that firms “meeting or beating” the zero earnings benchmark display less cost stickiness than firms reporting healthy profits. Their paper shows that, when the right incentives are in place, the extent of cost stickiness can be decreased.

UvA students van der Weijden (2013) and Dekker (2012) investigated if there is a relation between CEO incentives and costs stickiness. Both papers concluded that cost stickiness was lower when short-term incentives were in place for the CEO. The results of these two researches are in line with the results of Dierynck et al. (2012); short-term incentives cause short-term behavior and are negatively associated with cost stickiness.

However, taking the literature of criticism against short-term incentives into consideration, I got interested in the question if cost stickiness of companies with a long-term vision differs from other companies. When a company faces a decrease in activity, cutting resources could be an appropriate response. But I am interested in the question if a company with a long-term vision reduces resources less drastically than other companies. A company with a long-term vision can decide to cut fewer resources given the expectation that activity will increase in the future and subsequently resources of the company are (partly) kept in place to be able to be prepared for this future increase in activity. I want to investigate if there is a positive relation between a long-term behavior and cost stickiness.

(5)

This research argues that cost stickiness can be acceptable if companies, out of long-term behavior, do not cut resources drastically as a response to a decrease in activity. A CEO with long-term behavior will not cut resources if this only results in short-term results at the cost of adding value. The retaining of unused resources can be a deliberate decision, explained by strategic persistence. If companies with long-term behavior accept cost stickiness to a certain extent, it is interesting to add a second part to this research; do these companies display a better future performance than other companies? So subsequently to my first investigation, I also want to investigate how future performance relates to current cost stickiness levels.

1.2 Research questions

Recent literature suggests short-term incentives are negatively associated with cost stickiness. I want to investigate if a long-term vision is positively associated with cost stickiness. Therefore my basic research question is:

“To what extent does long-term behavior influence the asymmetrical behavior of costs?”

This research investigates if companies with long-term behavior may accept cost stickiness because of strategic persistence. Therefore it is interesting to investigate if there is a relation between future activity growth and current cost stickiness levels. This results in the following research question:

“To what extent is future activity growth related to asymmetrical behavior of costs?”

1.3 Motivation and contribution

This research contributes to the current literature investigating cost stickiness and its determinants. Secondly, this research contributes to the literature investigating the impact of short-term behavior. More specifically it contributes to the literature criticizing short-term incentives. Thirdly this research contributes to the research of contingencies of the relation between CEO tenure and performance. Finally this research contributes to the literature investigating the contingencies of cost stickiness.

(6)

1.4 Structure

In section two I will discuss prior literature and the development of the hypotheses. In section three I will discuss the research methodology; sample selection, empirical models and variable measurement. In section four I will present the results of the research. Section five contains the conclusion.

(7)

2 Literature review

2.1 Cost stickiness 2.1.1 Traditional cost model

In the traditional model of cost behavior costs are assumed to be fixed or to be variable with respect to changes in the activity volume. In this classical model, the variable costs change proportionately with the changes in the activity drivers. (Noreen 1991). This model implies costs to change proportionately regardless the direction of the driver of the change. If activity increases, costs increase proportionately and if activity decreases costs decrease proportionately as well.

2.1.2 Asymmetrical cost model

Based on cross-sectional data from hospitals, Noreen and Soderstrom (1994) were the first to test the proportionality of costs to activities. In this research they reject the proportionality hypothesis for most overhead accounts. The average cost per unit is 40% to 100% higher than marginal costs. Therefore they conclude that in the field of cost accounting, managerial decisions regarding average cost per activity should be approached cautiously. In another research, Noreen and Soderstrom (1997) do not examine cross-sectional data but this time they examine time series of overhead costs from hospitals. Using the data of 108 hospitals for the years 1977 to 1992 their study is the first to identify that costs change more readily in response to increase in activity than they do to decrease in activity.

Anderson et al. (2003) investigate this issue more in depth. Their research is aimed at finding evidence about the behavior of costs in relation to changes in levels of activity. They are first to label asymmetrical cost behavior as “sticky”. In their research they observed 7.629 firms over 20 years. The focus of the research was on the behavior of SG&A costs in relation to revenue. By estimating an empirical model that relates the change in SG&A costs to the change in revenue they test for sticky behavior of costs. To distinguish between revenue-decreasing and revenue-increasing they include an interaction dummy variable. The outcome of their investigation was that SG&A costs increase at a rate of 0.55% per 1% increase in revenue but decrease only 0.35% per 1% decrease in revenue. They argue that cost stickiness is explained by the delay in adjustment of committed resources. If activity decreases and there is uncertainty about future developments, managers may be inclined to delay adjustments until there is

(8)

certainty about future demand. They find support for this by showing that cost stickiness is reduced with the aggregation of measurement periods. Using an expanded version of their model, they also find support that cost stickiness increases with asset intensity, employee intensity and with macroeconomic growth. This contradicts the traditional model of cost accounting and the authors argue that this research supports an alternative view of cost behavior. This alternative view of cost behavior emphasizes managerial discretion and resource adjustment costs. When revenue decrease managers opt to retain unused resources in order to avoid adjustment costs. They have a choice to retain or not retain unused resources. A variety of motives can lead to a decision to retain unused resources. On the contrary, when revenue increases, managers do not have much of an option but to increase resources. This discretion leads to asymmetrical cost behavior.

Expanding the research on cost stickiness, Subramanian and Weidenmier (2003) investigate whether the magnitude of the change in activity is related to cost stickiness. Their research is based on data of 9,592 firms and 22 years of annual data for the period of 1979 - 2000. Dividing the level of activity changes into six ranges, they examine the level of cost stickiness as a function of the magnitude of activity change. They conclude that for small changes in activity SG&A costs and COGS do not show sticky cost behavior. If activity changes by more than 10%, costs show sticky behavior. They argue that this is caused by manager’s asymmetrical response to large changes of market demand. A large increase in activity causes an immediate increase of costs while large decreases in activity may not result in an immediate decrease of costs. This happens because firms cannot reduce employees, assets and/or other costs in the short-term. In this research they study four different industries; manufacturing, merchandising, service and financial services. As a result of this inter-industry research they find that fixed assets intensity, employee intensity, inventory intensity and interest expense are determinants of stickiness. Empirical results from their research document that manufacturing is the industry which displays the highest level of cost stickiness and merchandising displays the lowest level of cost stickiness.

Calleja et al. (2006) compare cost stickiness in different countries. They use data of listed firms from the UK, US, French and German stock markets. In their research they confirm the results of previous literature; costs exhibit sticky behavior but the level of cost stickiness decreases when time intervals are aggregated or when firms sustain large drops in revenue. They also find that firm-specific characteristics and industry-characteristics influence the level of cost stickiness. The overall results show that operating costs, on average, increase 0.97% per 1% increase in revenue but decrease only 0.91% per 1% decrease in revenue. Using separate models

(9)

they find that French and German firms exhibit more cost stickiness than UK and US firms. They argue that this difference can be explained by the differences in systems of corporate governance and managerial oversight. The research of Banker et al. (2013) has a similar conclusion stating countries with a more efficient judicial system and a higher level of development show higher cost stickiness because these factors force the company into long-term commitments.

UvA student Dekker (2012) investigates the potential influence of corporate governance and the type of CEO incentive compensation on the degree of cost stickiness. He makes a separation between board governance and institutional governance and between short-term incentives and long-term incentives. The impact of each of these four variables on cost stickiness is investigated. The results of board governance display that only a higher proportion of independent directors on the board mitigates cost stickiness. A higher proportion of institutional holdings does not mitigate cost stickiness. Besides the results in the field of governance, the research shows that companies with short-term incentive schemes display less cost stickiness. In line with this, the research shows that companies with long-term incentives schemes display more cost stickiness. The author states: “the provision of relative more long-term incentives facilitates sticky cost behavior.” (p. 2)

UvA-student van der Weijden (2013) investigates the association of the magnitude of cost stickiness with the type of incentive schemes awarded to managers. He finds that firms that have compensation schemes that are dominated by short-term incentives exhibit a lesser degree of cost stickiness compared to firms that have compensation schemes that are dominated by long-term incentives. The author states: ”long-term incentives have a facilitating effect on asymmetrical costs.” (p. 1)

2.2 Short-termism

Myopia theory criticizes short-term management behavior. In the past two decades average CEO tenure has decreased from eight to less than four years (Le Breton-Miller and Miller, 2006; Weisman, 2008). Therefore the pressure on CEOs to deliver quick results has increased subsequently. Long-term investments that will not become profitable in the short-term are avoided (James, 1999).

The theoretical overview of section 2.1 has the underlying assumption that cost stickiness is per definition negative. Cost stickiness is (for a part) interpreted as agency costs, which should be kept at a minimum. I argue that, to some extent, cost stickiness could be allowed from a long-term perspective. Immediately cutting resources as a response to a decrease in revenue may not

(10)

be adding value in the long run. There is a fine line between doing what needs to be done and opportunistic resource adjustment. Short-term incentives promote short-term behavior. But is short-term behavior always adding value? Cost stickiness basically comes down to discretionary decisions on retaining resources. I argue that these decisions be different from a short-term perspective than from a long-term perspective. Decisions regarding retaining of resources are not any different than decisions regarding acceptance of a project based on its NPV. A short-term view may lead to decisions which do not add most value to the company and its’ shareholders.

Jensen (2004) investigated managerial goals. He described how short-term managerial goals, often supported by earnings management, can serve as potential sources of overvalued equity and agency costs that end up in a decreasing firm value. A CEO with a short-term focus could inflate short-term profits by cost cutting, which is not a sustainable source of profit growth, instead of preferring to add value by investing in positive NPV projects which only are not profitable immediately.

Brauer (2013) has investigated the effects of short-term managerial behavior compared to long-term managerial behavior. He summarizes the core of current criticism on short-term behavior. This criticism focuses on CEOs maximizing quarterly results at the expense of long-term performance;

“In the face of continuously decreasing chief executive officer (CEO) tenure, CEOs, however, seem to have few incentives to embrace long-term oriented behavior. Instead, the question of foremost importance to self-interested CEOs is whether short-term orientation already harms financial performance in the three to four years of their own tenure, and whether CEOs stand a chance of benefiting from long-term orientation while still in office. CEOs thus face an intriguing ethical dilemma between optimizing their financial pay-off within their own tenure and securing the longer-term well-being of the corporation, its employees, and other major stakeholders.” (p. 386)

Brauer’s study focuses on the medium-term performance implications of short-term and long-term orientation in large publicly listed European companies. He finds that short-long-term behavior is negatively associated with medium-term performance and that long-term behavior positively impacts medium-term performance. These findings give a new perspective on myopic behavior of CEOs and stress the ethical dilemmas they currently face.

In a survey of Graham et al. (2005) the researchers reported that 80% of the respondents would decrease discretionary spending on R&D, maintenance, advertising and personnel costs to

(11)

meet company targets. Over 50% would postpone a project even if this would lead to giving up value. Delaying value-adding projects or cutting resources, even if this means giving up value, seems to be evidence of the fact that companies attach too much weight on meeting short-term expectations.

2.3 CEO tenure

In this section I will discuss prior literature on CEO tenure. Firstly I will discuss literature which argues that CEOs with low tenure have incentives to display short-term behavior. Secondly I will discuss literature on CEO attitudes which are developed over his tenure. Lastly I will discuss literature on the impact of these attitudes on firm performance.

2.3.1 Low CEO tenure and CEO attitudes

Zhang (2010) investigates the relation between CEO tenure and earnings quality. The author takes CEO tenure as a measure of CEO ability. The author states this measure has been used in prior research by Milbourn (2003) because long tenured CEOs would have survived several retention/dismissal decisions by their board of directors. Zhang’s research shows that CEOs with long tenures report earnings less aggressively than CEOs with short tenures. The outcome is in line with the notion that at the beginning of their tenures CEOs have incentives to inflate earnings, in order to build a reputation of ability. After the CEOs have established their reputations, results show they report less aggressively in order to protect their reputation. The paper mentions that a reputation is built up by current results and results in the past. As CEOs with low tenure cannot show results from the past. The only way to be perceived as a capable CEO is to report positive results; one way or the other. Once CEOs can fall back on positive results from the past, the dynamics change. Instead of building a reputation the focus shifts to protecting a reputation and therefore the CEO moves away from short-term behavior aimed at short-term results. Furthermore the research mentions that reputation is associated with long-term results. This paper confirms that short tenured CEOs face external pressures to display short-term behavior.

A similar research has been done by Ali and Zhang (2013). They investigate how the increase of tenure impacts the incentive to manage earnings. They document that CEOs with a low tenure are more likely to overstate earnings than CEOs with higher tenure because the market is likely to be more uncertain about their ability in the early years and earnings reported at that time would have a greater effect on the market’s assessment of their ability (Hermalin and

(12)

Weisbach, 2012). The authors state that these results are consistent with the incentive of a CEO to influence the market perception of his ability in the early years of his service. CEOs with high ability are more likely to survive retention/dismissal decisions. Once a reputation of high ability has been established CEOs will have the incentive to avoid overstatement of earnings in order to protect their reputation and their incentive to engage in opportunistic behavior would decrease. Chen and Zheng (2013) investigate the effect of CEO tenure on risk-taking. The research documents a positive relation between CEO tenure and risk-taking. The author points to the fact that the results are inconsistent with other research documenting tenure to be only an indicator of human capital investment. The author states that there are also other theories which explain the risk taking by power effects of tenure and by experience effects of tenure. The outcome of this research, that career concerns are related with risk taking, is consistent with recent research that interprets tenure as the career concerns of a manager.

The literature discussed in this paragraph confirms that CEOs with low tenure have incentives to focus on short-term results and therefore may be inclined to quickly adjust resources when activity decreases. CEOs with high tenure have no incentive to create short-term results. They can make a more balanced decision on retaining or releasing unused resources.

2.3.2 High CEO tenure and CEO attitudes

Upper echelon theory states that top managers are the strategists who set the direction of the company and subsequently set the pace of competition in the industry. Strategic choices made are therefore the reflection of the values, cognitions, knowledge and skills of top management through intervening information processing steps such as selective perception of information, field of vision, and interpretation of information (Hambrick and Mason 1984).

Prior research suggests there is a positive relationship between CEO tenure and

reluctance to change (Musteen et al 2006), lack of adaptability (Miller, 1991) and commitment to the status quo (Hambrick et al. 1993). This phenomenon is called the “fixed paradigm problem”. CEOs with high tenure operate with a paradigm of how the environment exists, which strategies are available and how the organization should operate (Hambrick and Fukutomi, 1991). These attitudes underlie why previous studies have consistently found that increasing tenure of top management leads to strategic persistence (Finkelstein and Hambrick, 1990; McClelland et al 2010). Westphal (2001) looks at this matter from a social psychological angel. In his paper he summarizes the field of social psychological research on tenure. He summarizes that his research suggests that CEO tenure in an organization is associated with psychological commitment to the status quo, including commitment to the current strategy of the company. As tenure increases,

(13)

through participation in implementing a firm’s strategy, a manager’s internal sense of career accomplishment and external reputation becomes more strongly tied to the strategy, increasing their psychological commitment to it (Fox and Staw, 1979; Hambrick et al., 1993; Staw, 1981). Therefore, long-tenured CEOs come to believe more strongly in the strategy.

Above mentioned literature allows me to take CEO tenure as a proxy for long-term behavior as the attitudes which a CEO develops over his tenure will lead to strategic persistence. De Angelis and Grinstein (2010) have taken the same approach in their research; they have taken log tenure as a proxy for CEO experience and the stability of the firm’s strategy.

2.3.3 CEO tenure and firm performance

The literature of the previous paragraph displays the attitudes a CEO develops over his tenure. This paragraph will look at the effect of CEO tenure on firm performance. Extent research has been performed on this subject and the outcomes show mixed results. Literature shows two different views on tenure. The first view emphasizes that paradigms over a CEOs tenure become obsolete and finally negatively influence firm performance. The second view emphasizes the benefits of the accrued knowledge and experience which positively influences firm performance. (Resource-based view). These two views imply contradictory effects of CEO tenure on firm performance (Bergh 2001). The following is an overview of literature that finds CEO tenure has a negative relation with firm performance. This relation is contingent to the environment of the company. In a dynamic environment the relation is negative. In more stable industries the relation is not negative.

The research of Henderson et al (2006) finds that the relation between tenure and performance improvement is contingent on the dynamism of the external environment. They display that CEOs' paradigms will become increasingly obsolete as their tenure increases and this process negatively impacts future performance in dynamic industries. However, such performance declines may not occur in more stable industries. Among firms in the stable food industry firm performance increased for at least ten years of a CEO's tenure before declining. Yet in the dynamic IT industry, they could find no evidence of performance improvements over tenure. In this industry, CEOs were at their best during their first year and displayed decreasing performance as tenure increased.

These results were confirmed by McLelland (2012). In his research he found some evidence that CEO tenure is negatively associated with future performance. Especially the fixed paradigm problem becomes prominent as a CEO gains tenure. If CEOs do not effectively

(14)

respond to environmental changes by updating their paradigms, companies are likely to experience a decreasing performance due to environment-organization misalignment. However, also these results are nuanced by contingencies; in a dynamic industry CEO tenure negatively impacts future performance but in a stable industry there is no negative relation. This research was more detailed than the research of Henderson and could generalize Henderson’s results for “dynamic industries” (instead of only the IT industry) versus “stable industries” (instead of only the food industry).

Richard et al. (2009) investigate how the relation between EO and firm performance is impacted by CEO position tenure and CEO industry tenure within a sample of 579 US banks. Entrepreneurial Orientation (EO) is a firm-level construct in strategic management and entrepreneurship studies. EO captures specific entrepreneurial aspects of decision-making styles, methods and practices which lead to new entry (Lumpkin and Dess 1996). An EO firm has characteristics which are similar to the prospectors of Miles and Snow (1978) as they both are focusing to pursue opportunities outside their current scope. In his research he summarizes prior literature stating that CEOs with relatively long position,- and industry tenure would possess unique and nontransferable firm- and industry specific knowledge which is unavailable to CEOs with shorter tenures (Govindarajan 1989; Haspeslagh and Jemison 1991; Cannella and Hambrick 1993). CEOs who are early in their tenures work at learning and implementing strategy (Hambrick and Fukutomi 1991; Miller 1991; Miller and Shamsie 2001). As the tenure of a CEO increases, he becomes more confident and the learning declines (Miller and Shamsie 2001). Therefore, CEOs with short position tenure are more likely to gain real time knowledge from the market and from operations. Meanwhile, he can make emergent strategies to take advantage of this acquired knowledge (Covin et al. 2006). Prior literature displays that emergent strategies would increase the chances of successful innovative, risky and proactive actions (Covin et al. 2006). Focusing on emergent strategies also could cause a firm with an EO to postpone commitments to a particular course of action and subsequently retain the strategic flexibility which is needed for a good performance in an uncertain environment (Covin et al. 2006). On the other hand, this also means that it is difficult for CEOs with high tenure to implement emergent strategies. CEOs with high tenure formulate and implement strategies not based on their real-time knowledge of the firm and its environment, but based on their prior organizational experience (Hambrick et al. 1993). As a result, strategies executed by CEOs with high tenure may not be appropriate for emerging environmental opportunities. This is in line with Miller (1991) who examined the match between CEO tenure and the environment and strategy. The results of Richard’s research strongly support that CEO industry tenure positively moderates the EO to

(15)

performance relationship and CEO position tenure negatively moderates the EO to performance relationship. This means that in a dynamic, outward focused, prospector-like company long CEO tenure is not contributing to the performance. As the CEO is not able to respond effectively to a firm’s EO and implement strategies that will enhance the benefits of EO on performance. This might be caused by the reluctance to change and the commitment to the status quo which appears not to work in an outward focused company. However no strong evidence has been found that OE is associated with better performance. A positive relation was found with ROE but not with ROA. If one can say OE and a dynamic environment are related; than similar to the previous paragraph this means that position tenure does not positively influence performance in a dynamic environment. As prior research states, there may be little or no difference in the performance of entrepreneurial and conservative firms (Zahra et al. 1999). This would suggest that the banks in the investigation of Richard are capable of competing each other along different strategic dimensions and still achieve similar results. This is indeed also reflected in the results of the research, suggesting that long industry tenure and short position tenure probably compensate for each other’s weaknesses and allow CEOs to have an unbiased assessment of opportunities and interpret these from their own strategic perspective.

The following is an overview of literature that finds a positive influence of CEO tenure on firm performance. Optimal contracting theory and resource-based view shed a different perspective on the impact of CEO tenure on risk appetite and performance. This theory argues that companies benefit by the accumulated knowledge and skills a CEO accumulates over his tenure. Dikolli et al. (2013) investigate the relation between CEO tenure and performance-related dismissal. The outcome of their research is that the likelihood of a performance-related dismissal is negatively associated with tenure. Among other findings, their main finding is that CEO survival is associated with superior performance. Another research in the same field, but from a different angle is the research of Pan et al. (2013) in which they find that uncertainty on CEO quality leads to stock return volatility. This volatility decreases over tenure.

Simsek (2007) investigated the impact of tenure on risk-taking behavior. His results contradict the view of the upper echelon theory claiming that risk appetite decreases over tenure. He argues that tenure shapes the risk-taking behavior that the CEO brings to the task of evaluating, rewarding, and motivating the members of the top management team (TMT). He argues that CEO tenure leads to TMT risk-taking propensity and TMT risk-taking propensity leads to better firm performance. Results indeed show that CEO tenure had a positive net effect on TMT risk taking, rather than the inertial and complacent effects suggested by upper echelon theory. So, the image of a risk-taking CEO that emerges from this study is not that of young and

(16)

energetic CEO as commonly displayed in literature. The risk taking CEO is an experienced individual with idiosyncratic and tacit knowledge of the company and its environment. The advantages of long tenure, firm-specific human and social capital, knowledge and power seem to outweigh the disadvantages of inflexibility and maintaining the status quo. If strategic risk taking indeed relies on idiosyncratic and tacit knowledge of the TMT, the company might necessitate TMT experience accumulated via long tenure. Short-tenured CEOs may not have sufficient awareness to effectively notice and assess the strategic risks the company faces.

Antia et al. (2010) investigated the relation between expected CEO tenure and agency costs, information risk and market valuation. These results too contradict upper echelon theory. Antia tests whether expected tenure is significantly related with market valuation. Based on the notion that a CEO's expected tenure is a good proxy of his decision horizon, he hypothesizes that firms should benefit when CEOs are expected to have a long tenure with their company. The benefits from longer expected tenure should be lower agency costs, lower information risk and higher market valuation. He finds strong evidence that shorter CEO decision horizons are associated with significant agency costs. The research also confirms the hypothesis that decision horizon has a positive influence on firm performance and market valuation, as measured by Tobin's q and alternative measures of valuation. These findings are consistent with the view that high CEO tenure leads to an alignment of the interests of managers and shareholders, and, consequently to higher valuation. Subsequently they show that CEO myopia diminishes company performance. Antia argues that CEOs with high tenure could be set in their ways and might ignore competitive pressures that are a risk to a business model that has proven effective for a long time. But he argues that high tenure should be a requisite for implementing a long-term strategy that would respond adequately to a dynamic environment. A CEO with short-long-term horizon may be tempted to make decisions that are not in the best (long-term) interest of the company. The results from this research also imply that companies should design incentive structures that discourage myopic behavior and conversely encourage CEOs to embrace long-term behavior.

Above overview on the literature gives mixed results regarding the impact of CEO tenure on performance and risk appetite. Upper echelon theory and optimal contracting theory /resource-based view have a different interpretation of the effect of CEO tenure on firm performance. The results of the investigations which focused at the impact of CEO tenure on risk and performance have been summarized in table 1.

(17)

Br au er (201 3) Antia_et a l. (201 0) Sim sek (2007) D iko lli e t a l. (201 3) Ri ch ard et al. ( 2009) M cL el lan d (201 2) He nde rso n et a l. ( 2006 ) Res ea rch er The im pa ct of sh or t-te rm / long -te rm or ient atio n on me diu m -te rm re sul ts. The re lati on shi p be twe en CE O d eci sion hor izo n a nd ag enc y c ost s, f irm v alu ation and inf or m ation r isk . Th e ef fect o f C E O ten ure, via T M T a ttit ud e, o n f irm pe rfor m anc e. The re lati on shi p be twe en CE O te nu re a nd per fo rm an ce r el ated d H ow the re lati on be twe en E O a nd fir m pe rfor m anc e i s m od era ted by C E O posi tio n ten ure an d CE O in dus try ten ure . The re lati on shi p be twe en E nt repr ene uria l O rie nt atio n (E O ) a nd fir m pe rfor m anc e. Im pa ct o f C E O c are er hor izo n a nd C E O te nu re on fut ure p erf orm an ce. Im pa ct C E O te nu re on f irm pe rfor m anc e. In ves tig atio n Shor t-ter m / long -te rm or ient ation M ed iu m ter m re sul ts CEO de cisi on hor izo n CE O ten

ure tenure CEO CEO

pos

itio

n

ten

ure EO tenure CEO Ca_ree

r hor izo n CE O ten ure Va_ria ble V aria bilit y in wor kfor ce, T M T si ze, acq uis itio n a ctiv ity an d Ca pe x r atio Th ree yea r a ve rag e ind us try ad jus ted ROA E xp. C E O te nu re = (Ind ust ry m ed ian of CE O T en ure - C E O ten ure ) + (In du str y m ed ian o f CE O a ge - CE O a ge) Y ear s i n o ffic e Y ear s i n o ffic e Y ears in in dus try Nin e-it em E O sc ale (C ov an a nd S lev in 1989) Y ear s i n o ffic e M ea su red b y a ge Y ear s i n o ffic e M ea su rem en t Shor t-te rm orie nta tio n n eg ativ ely in flu en ces me diu m -te rm pe rfor m anc e. L ong -te rm o rie nta tio n p osit iv ely in flu en ces me diu m -ter m p er fo rm an ce. Sh ort er ex pect ed ten ure i s a sso ci ated w ith a gen cy co sts . E xp ect ed ten ure h as a po sit iv e in flu enc e on f irm pe rfor m anc e. Fir m s t ha t ha ve C E O s wi th l on g e xpe cte d t enu re ha ve hi gh m ark et v alu atio n a s m ea su red b y T ob in's q . E xp ect ed ten ure mit ig ate s t he in fo rma tio n r isk in ve sto rs f ace . CE O ten ure p osi tiv el y i nflu en ces th e r isk -ap pe tite of the top m ana gem ent te am a nd the fir m 's pu rsu it of e nt repr ene uria l in itia tiv es. T his in tu rn ha s a p osit iv e in flu en ce o n f irm pe rfor m anc e. CE O ten ure n eg ativ el y i nflu en ces th e l ikel iho od o f a pe rfo rma nc e r ela ted d ismis sal. CE O posi tio n t enu re ne gat ive ly m od era tes th e E O to pe rfor m anc e r ela tions hi p. Som e su ppor t f or pos itiv e E O -pe rfor m anc e r ela tions hi p. W ill le ad to fixe d p ara dig ms w hic h in tu rn w ill n eg ativ ely inf luen ce f utu re p er fo rm an ce i n d yn am ic i nd ust ries . T his m ay not be the c ase in m or e st abl e in du str ies . W ill le ad to ris k-a ve rse str ate gie s, w hic h in tu rn w ill ne gat iv ely inf lue nc e f ut ure firm p erf orm an ce. N ega tiv e f or c om pu ter in du str y. Po sit iv e f or fir st te n y ear s i n food ind ust ry. Res ult s No No No No Y_es Y_es No Y_es Co nti ng en t?

(18)

2.4 Hypothesis development

As mentioned in the theory section there has been a lot of research which documents the empirical prevalence of cost stickiness and that the degree of cost stickiness is different across firms, countries and industries. However there has been no prior research which investigates the relation of long-term behavior and cost stickiness.

To investigate this I take CEO tenure as a proxy for long-term behavior. I argue that CEOs with high tenure display higher levels of cost stickiness as they will respond less opportunistic to a decrease in revenue. To test this, I will hypothesize the opposite:

Hypothesis 1: CEO tenure has no relation with the level of cost stickiness.

Results that may come from investigating hypothesis 1 look at how long-term behavior can result in accepting cost stickiness. Deliberately accepting cost stickiness from a long-term behavior perspective only can be justified if this is compensated by future performance improvement. Therefor I argue that current levels of cost stickiness are higher for firms which display the best revenue growth in two years’ time. To test this I will hypothesize the opposite:

(19)

3 Research methodology

3.1 Sample selection

I test my hypothesis using a sample from the ExecuComp database combined with a sample from the CompuStat database. For both samples data has been retrieved from 1998 until 2007. I want to test for the relation between CEO tenure and cost stickiness for the period of 2000 – 2005. The 1998 and 1999 observations only serve to determine t-1 and t-2 values for subsequent years. The 2006-2007 values will be used in the second part of this research which investigates the relations between revenue growth in two years’ time (t + 2) and cost stickiness.

From the CompuStat database a sample is obtained with financial data for the period 1998-2007. Variables downloaded are: company ID, fiscal year, profit before extraordinary income, revenue (>0), SG&A (>0) and total assets. This resulted in an initial sample of 69.833 observations.

From ExecuComp the following variables are retrieved: Company ID number, Fiscal year, Company name, Executive name, date became CEO and annual CEO flag. This resulted in an initial sample of 114.530 observations. As this research only focuses on CEO’s we delete all observations which do not have a mark at the field “CEO flag”. This decreases the sample to 17.895 observations. Merging the two samples using a unique combined “Company ID & fiscal year” variable leads to a sample of 14.720 observations. As “date became CEO” is a crucial variable for this research, all observations without “start date” are dropped. This decreases the sample to 14.278 observations. Based on the field “date became CEO” a new field is computed: “year became CEO”. The start date of this CEO enables me to determine the start year of the CEO with the company. The fiscal year minus the start year plus 1 is a new variable; “years of service”. Observations where “year became CEO” are larger than the value of “fiscal year” (= negative years of service) are dropped. This decreases the sample with 443 observations to a sample of 13.835 observations.

With the data from this database I calculate the Industry adjusted 3-years average return on assets, SGG&Ait/SG&Ait-1, Revenueit/Revenueit-1, a decrease dummy for decrease in

revenue, log (SGG&Ait/SG&Ait-1) and log (Revenueit/Revenueit-1). The result for the year 1998

and 1999 now can be dropped as these only served to determine t-1 and t-2 values. This decreases the sample from 13.835 observations to 11.263 observations.

If a company had its first observation within the period of the final sample (2000-2005), the first available observations will be used to calculate t-1 of the following year’s observation

(20)

and after this, the first observation is dropped. This decreases the sample with 913 observations to 10.349 observations.

Subsequently the revenue for t+2 is added to each observation. After this the observations of 2006 and 2007 are removed. This leads to a final database of 7.614 observations. This database will be used for testing on the relation between CEO tenure and cost stickiness. Subsequently it will be used to test the relation of future revenue growth and current levels of cost stickiness.

3.2 Model specification and variable definitions 3.2.1 Basic model of cost stickiness

The basic model developed by Anderson et al. (2003) will be used to test the sample on cost stickiness. Basic cost stickiness is conceptually shown in model 1.

Model 1

ln(SG&Ait/SG&Ait-1)= β0+β1ln(Revenueit / Revenueit-1)+β2DecreaseDummyit *ln(Revenue/Revenueit-1)

+ β3DecreaseDummyit

ln(SG&Ait/SG&Ait-1) is the log-change in SG&A costs for firm i in year t and

ln(Revenueit/Revenueit-1) is the log-change in revenue. DecreaseDummy_it is a dummy variable

equal to 1 if revenue decreased in year t and 0 otherwise. In general; there is asymmetrical cost

behavior within the sample when β1 > (β1 + β2). So when β2 < 0. In line with Anderson (2003)

changes in Selling, General and Administrative expenses are used as a proxy for changes in costs and changes in annual revenue are used as a proxy for changes in activity levels. According to Anderson (2003) the log specifications of these changes improve comparability and reduce hetroskedasticy.

3.2.2 Cost stickiness and CEO tenure

To test whether CEO tenure is related with the magnitude in which costs decrease in reaction to a decrease of activity, the basic model is extended. Within the 2000-2005 sample I added a dummy variable of CEO tenure; with a value of 1 for the firms with the highest 20% of CEO

(21)

years of service and a value of 0 for the lowest 80%. This variable is added as an interaction variables to the basic cost stickiness model (model 2).

Model 2

ln(SG&Ait/SG&Ait-1) =

β0+ β1ln(Revenueit/Revenueit-1) + β2DecreaseDummyit * ln(Revenueit/Revenueit-1)

+ β3TenureDummyit * ln(Revenueit/Revenueit-1) + β4TenureDummyit * DecreaseDummyit

* ln(Revenueit/Revenueit-1)+ β5TenureDummyit

3.2.3 Robustness check

Managerial skills and knowledge accumulate over a CEOs tenure. This accumulated knowledge and skills can contribute to a firm’s ability to effectively manage and convert resources into profitable capabilities and earn superior returns (Castanias and Helfat 1991; Mahoney and Pandian 1992; Mahoney 1995; Kor and Mahoney 2000). Therefore I take 3 years industry adjusted ROA as a robustness check for CEO tenure. For each observation I calculate ROA as income before extraordinary items (CompuStat data 18) divided by total assets. Subsequently I

calculate the three-year rolling average ROA for each CEO firm-year. If the CEO has not been in office for the last three years, I use an average ROA computed over the CEO’s tenure. I abstract the three-year average industry ROA from the three-year average ROA of each CEO-firm. If the CEO has not been in office for the last three years, I take the average ROA over his tenure and the average industry ROA over that same period. A positive score on this ratio suggest that the CEO outperformed the industry. I added a dummy variable for 3 years industry adjusted ROA; “ROADummy”. This variable has a value of 1 for the highest 20% of 3 years industry adjusted ROA and 0 for the lowest 80% (model 3).

Model 3

ln(SG&Ait/SG&Ait-1) =

β0+ β1ln(Revenueit/Revenueit-1) + β2DecreaseDummyit * ln(Revenueit/Revenueit-1)

+ β3ROADummyit * ln(Revenue/Revenueit-1) + β4DROADummyit *DecreaseDummyit

(22)

3.2.4 Future performance

As the above regressions try to display that CEO’s with high tenure display more cost stickiness, the question remains whether this behavior will pay off in the future. To test this I want to focus on the growth of the companies. The 7.614 observations in my sample cover the period 2000-2005. For each observation I have added the revenue of two years later (t+2). Subsequently I calculated the future growth (Revenueit+2/Revenueit). I want to test if there is a relation between

future activity growth and the current level of cost stickiness. In order to do so, I added a new dummy variable to the sample; “GrowthDummy”. This variable has a value of 1 if the realized growth in revenue for an observation belongs to the top 20% of growth numbers for that year and 0 if otherwise. I included this variable as an interaction variable to the basic cost stickiness regression (model 4). The results of this regression will show the cost stickiness levels of companies who achieved the best growth results compared to the other companies.

Model 4

ln(SG&Ait/SG&Ait-1) = β0+ β1ln(Revenueit/Revenueit-1) +β2DecreaseDummyit

* ln(Revenueit/Revenueit-1) + β3GrowthDummyit* ln(Revenueit/Revenueit-1) + β4GrowthDummyit

* DecreaseDummyit* ln(Revenueit/Revenueit-1) + β5GrowthDummyit

3.2.5 Contingencies to industry dynamism

As discussed in the literature review; the environment in which an industry operates may be a moderating variable in the CEO tenure – performance relationship. I want to investigate if the outcomes of my regressions are contingent to the dynamism of the industry. Therefore I have divided my sample in two subsamples based on the dynamism of the environment. Per industry I calculated the standard deviation of annual revenue for the period of 2000-2005. Subsequently I divided this standard deviation of revenue by the mean revenue. This leads to a “dynamism” ratio for each industry. Subsequently I calculated the average “dynamism ratio of the whole sample. Industries with a ratio below the average sample ratio are labelled “stable”. Industries with a ratio above the average sample ratio are labelled “dynamic”. Annex 1 displays an overview of these calculations. Of each subsample I have calculated and compared the basis cost stickiness model, the basic model extended with the interaction variable “CEO tenure dummy” and the basic model extended with the interaction variable “future growth dummy”. This has been done in the same way as in previous paragraphs.

(23)

4 Results

4.1 Descriptive results

Table 2 displays the descriptive statistics of the main variables used; revenue, SG&A costs, CEO tenure, industry adjusted ROA and revenue growth in two years’ time. The revenue of the firms within the sample has a mean value of 4,835 million and a median of 1,150 million. The SG&A costs of the firms within the sample have a mean value of 912 thousand and a median of 225 thousand. On average, SG&A costs are 29% of the revenue. These results are in line with previous studies on cost stickiness (Anderson et al., 2003; Calleja et al., 2006). CEO tenure has a mean of 8 years and a standard deviation of 7,42 years. These statistics are in line with the research of Henderson (2006), McClelland (2012) and Dikolli et al. (2013). Industry adjusted ROA has a mean of -0,02 and a median of 0,00. Its standard deviation is 0,23. On average firms within the sample have realized a 10 percent revenue growth in two years’ time (mean = 1,10) with a standard deviation of 0,63.

Table 2 Descriptive statistics

Mean Median Std. Dev 10% 20% 50% 80% 90%

Revenue ($1000) 4.834,82 1.150,24 14.804,65 190,11 352,46 1.150,24 4.621,16 11.049,70

SG&A costs ($1000) 912,31 225,60 2.504,46 42,70 72,92 225,60 897,80 1.903,48

Revenueit / Revenueit-1 1,13 1,08 0,52 0,87 0,96 1,08 1,22 1,37

SG&Ait / SG&Ait-1 1,13 1,07 0,77 0,88 0,97 1,07 1,21 1,33

SG&Ait / Revenueit-1 0,29 0,23 1,01 0,07 0,11 0,23 0,38 0,50

CEO tenure - years in office 8,00 6,00 7,42 2,00 3,00 6,00 12,00 18,00

Industry adjusted ROA -0,02 0,00 0,23 -0,10 -0,05 0,00 0,05 0,08

Revenue growth

(24)

Table 3 shows the Pearson correlation coefficient between the main variables. There is a strong and significant positive correlation (0,629) between the natural logarithm of the change in SG&A costs and the natural logarithm of the change in revenue. This correlation is in line with cost stickiness literature. All remaining correlations are significant but not very strong.

Table 3 Correlation matrix

Ln SG&Ait/SG&Ait-1 Ln Revenueit/Revenueit-1 CEO

tenure Adjusted ROA Revenue growth (Revenueit+2/Revenueit)

Ln SG&Ait/SG&Ait-1 1,00

Ln Revenue/Revenueit -1 0,629 1,00

CEO tenure 0,057 0,052 1,00

Adjusted ROA 0,095 0,066 0,050 1,00

Revenue growth

(Revenueit+2/Revenueit) 0,093 -0,039 0,033 0,048 1,00

(25)

4.2 Results of hypothesis tests 4.2.1 Cost stickiness - basic model

Table 4 shows the regression summary statistics for the basic model of cost stickiness. Based on my sample of 7.614 observations for the period of 2000 until 2005 the results show presence of cost stickiness within this sample. The significant values of the coefficients of β0 and β1 indicate

that 1 percent increase in revenue will lead to a 0,014 +0,595= 0,609 percent increase of SG&A costs. However, a decrease of 1 percent in revenue will only lead to a decrease of SG&A costs of 0,424% (0,609-0,185). This is consistent with the findings of Anderson et al (2003).

Table 4

Summarized regression results

Variables Predicted sign Coeff. t-stat p-value

β0 : Intercept 0,014 9,609 0,000

β1 : ln(Revenueit / Revenue_it-1) + 0,595 47,234 0,000

β2 : DecreaseDummyit

* ln(Revenueit/Revenue_it-1) -

-0,185 -10,558 0,000 β₃ : DecreaseDummyit -0,013 -5,310 0,000 Number of observations: 7.614 F-value: 1733,584 R-square: 0,406 Adjusted R-square: 0,406 p-value 0,000

4.2.2 Cost stickiness and CEO tenure

I argue that CEO’s with high tenure will display more cost stickiness than other CEO’s. Therefore I predict that companies with a CEO with high tenure display a smaller decrease in SG&A costs following a decrease in revenue (model 2). Subsequently this means that the β4

coefficient (interaction variable of CEO tenure Dummy on the activity decrease part of the regression) has to be significantly negative. Table 5.1 shows the summarized regression statistics

(26)

of the basic cost stickiness model extended with the interaction variable for CEO tenure. The coefficient of the tenure interaction variable in case of an increase in revenue is small (-0,024) and not significant (p= 0,420). This means that, when it comes to the management of SG&A costs, it does not matter if a company has a CEO with high tenure or not in case of activity increase. However, in case of a decrease in activity, the results display a difference; companies that have a CEO with high tenure have a 13,0 percent smaller decrease in SG&A costs compared to other companies. (β4 = -0130, p=0,009). These results prove that CEO’s with high tenure

exhibit more cost stickiness than other CEOs. Therefore I cannot find support for Hypothesis 1.

Table 5.1

Summarized regression results – CEO Tenure

β0 : Intercept 0,009 6,500 0,000

β₂ : DecreaseDummyit

* ln(Revenueit/Revenue_it-1) - -0,173 -9,091 0,000

β3 : TenureDummyit

* ln(Revenueit/Revenue_it-1) -0,024 -0,806 0,420

β4 : TenureDummyit* DecreaseDummyit

β₅ : TenureDummyit 0,004 1,211 0,226 Number of observations: 7.614 F-value: 1040,695 R-square: 0,406 Adjusted R-square: 0,406 p-value 0,000

(27)

4.2.3 Robustness check

Table 5.2 shows the summarized regression statistics of the basic cost stickiness model extended with an interaction variable for 3 years ROA (Industry adjusted). The β4 coefficient of the ROA

regression is reported significantly negative as well. The reported value is -0,177 (p = 0,001). This means that, in case of a decrease of revenue, companies who have a high ROA have a 17,7 percent smaller decrease in SG&A costs compared to other companies. These results are similar to the results in the previous paragraph.

Table 5.2

Summarized regression results – 3 years industry adj. ROA

β0 : Intercept 0,009 6,909 0,000

β₃ : ROADummyit

* ln(Revenueit/Revenue_it-1) 0,169 4,476 0,000

β₄ : ROADummyit * DecreaseDummyit

β5 : ROADummyit -0,001 -0,447 0,655 Number of observations: 7.614 F-value: 1041,787 R-square: 0,406 Adjusted R-square: 0,406 p-value 0,000

(28)

4.2.4 Future performance

To determine future performance I take the growth in revenue in two years’ time (revenue t+2

divided by revenue t). I made a dummy variable for future growth; this variable has a value of 1

when it belongs to the top 20% growth results and a value of 0 for the other observations. This dummy variable has been added as an interaction variable to the basic cost stickiness regression. Table 5.3 reports the summarized regression results of this model. The coefficient of β4 is -0,096

(p = 0,016). This means that companies that realized the highest growth in two years’ time currently display a higher level of cost stickiness. Therefore I find no support for hypothesis 2.

The previous paragraph showed that companies that have long-term behavior display significantly more cost stickiness than other companies. Based on the results of this paragraph I argue that accepting more cost stickiness than other companies is the causal bridge between CEO tenure and future growth.

Table 5.3

Summarized regression results – revenue growth t+2

β0 : Intercept 0,009 6,529 0,000

β₃ : GrowthDummyit

* ln(Revenueit/Revenue_it-1) 0,170 0,647 0,518

β4 : GrowthDummyit * DecreaseDummyit

β5 : GrowthDummyit 0,009 2,815 0,005 Number of observations: 7.614 F-value: 1042,385 R-square: 0,407 Adjusted R-square: 0,406 p-value 0,000

(29)

4.2.5 Contingencies to industry dynamism

To investigate if the above results are contingent to industry dynamism, I have divided the sample in a “stable industries” part and a “dynamic industries” part. Per subsample I have run the same regressions as in previous paragraphs, except for the ROA interaction variable. The results are shown in table 6.1 – table 6.3.

For the basic cost stickiness model (table 6.1), the result show that cost stickiness is higher in stable industries compared to dynamic industries. The value of the coefficient DecreaseDummyit * ln(Revenueit/Revenue_it-1) in stable industries is -0,202 (p = 0,000) and in

dynamic industries this is -0,083 (p = 0,021). This could imply that in dynamic industries the necessity to reduce cost stickiness is higher than in stable industries.

For the basic cost stickiness model extended with the interaction variable “CEO tenure dummy” the results are shown in table 6.2. These results show that in a stable industry the interaction variable TenureDummyit * DecreaseDummyit * ln(Revenueit/Revenue_it-1) has a value

of -0,232 (p = 0,000). In a dynamic industry this variable has a value of 0,181 (p = 0,041). These results indicate that companies that have a CEO with high tenure respond differently to a decrease in activity, depending on the industry dynamism their company operates in. In a stable industry they accept significantly more cost stickiness than other companies. In a dynamic industry they reduce significantly more cost stickiness than other companies. This implies that companies that have CEOs with high tenure are better able to reduce cost stickiness than other companies. But only if they deem this necessary. In a stable environment they do not deem this necessary and accept cost stickiness to a higher extent than other companies. However, in a dynamic environment they do deem this necessary and they reduce cost stickiness much more than other companies. This is in line with Antia (2010) who argues that high tenure should be a requisite for implementing a long-term strategy that would respond adequately to a dynamic environment. With these results in mind it is interesting to see how future growth relates to current cost stickiness levels in both subsamples.

For the basic cost stickiness model extended with the interaction variable “GrowthDummy” the results are shown in table 6.3. These results show that in a stable industry the interaction variable GrowthDummyit * DecreaseDummyit * ln(Revenueit / Revenue_it-1) has a

value of – 0,099 (p = 0,033). In a dynamic industry this variable has a value of 0,081. (p = 0,386). These results indicate that in a stable industry a CEO with high tenure seems to make the right decisions regarding the acceptance of cost stickiness. Because also companies with the highest future growth currently display significantly more cost stickiness than other companies. The

(30)

results also indicate that in a dynamic industry a CEO with high tenure also seems to make the right decision regarding the reduction of cost stickiness. Because companies with the highest future growth also currently reduce cost stickiness more than other companies (β4 = 0,081).

However this is not significant (p = 0,386).

Overall one can conclude that in a stable industry companies that have long-term behavior display significantly more cost stickiness than other companies and that future growth is significantly related with displaying more cost stickiness than other companies.

In a dynamic industry, companies that have long-term behavior display significantly less cost stickiness than other companies. Future growth is related to displaying less cost stickiness than other companies, however not significantly.

Table 6.1

Basic cost stickiness model per industry type

Stable industries Dynamic industries

Variables Coeff. t-stat p-value Coeff. t-stat p-value

β0 : Intercept 0,015 7,669 0,000 0,014 5,960 0,000

β1 : ln(Revenueit / Revenue_it-1) 0,597 38,384 0,000 0,591 27,748 0,000

* ln(Revenueit/Revenue_it-1) -0,202 -9,669 0,000 -0,083 -2,311 0,021

β₅ : DecreaseDummyit -0,014 -4,585 0,000 -0,006 -1,537 0,124 Number of observations: 5.100 2.514 F-value: 1.172,234 556,039 R-square: 0,408 0,399 Adjusted R-square: 0,408 0,399 p-value 0,000 0,000

(31)

Table 6.2

Basic cost stickiness model and CEO tenure per industry type

β0 : Intercept 0,008 4,883 0,000 0,010 4,746 0,000

β₃ : TenureDummyit

* ln(Revenueit/Revenue_it-1) 0,004 0,101 0,919 -0,074 -1,719 0,086

β₄ : TenureDummyit

* DecreaseDummyit

* ln(Revenueit/Revenue_it-1) -0,232 -3,646 0,000 0,181 2,040 0,041

β₅ : TenureDummyit 0,002 0,621 0,534 0,006 1,208 0,227 Number of observations: 5.100 2.514 F-value: 709,491 334,060 R-square: 0,411 0,400 Adjusted R-square: 0,410 0,399 p-value 0,000 0,000

(32)

Table 6.3

Basic cost stickiness model and future growth per industry type

β0 : Intercept 0,008 4,607 0,000 0,011 5,060 0,000

β₃ : GrowthDummyit

* ln(Revenueit/Revenue_it-1) 0,014 0,429 0,668 0,002 0,038 0,970

β₄ : GrowthDummyit

* DecreaseDummyit

* ln(Revenueit/Revenue_it-1) -0,099 -2,131 0,033 0,081 0,867 0,386

β₅ : GrowthDummyit 0,012 0,042 0,003 0,009 1,697 0,090 Number of observations: 5.100 2.514 F-value: 706,883 334,614 R-square: 0,410 0,400 Adjusted R-square: 0,409 0,399 p-value 0,000 0,000

(33)

5 Conclusion

Prior cost stickiness literature argues that cost stickiness for an important part can be interpreted as agency costs and has to be kept to a minimum. Prior literature shows that short-term incentives are negatively associated with costs stickiness. Conversely long-term incentives are positively associated with cost-stickiness. One research states: ”long-term incentives facilitate cost stickiness”.

I have investigated if long-term behavior leads to more cost stickiness in the short-term but also leads to better performance in the long-term. To test this I took CEO tenure as a proxy for a long-term vision. I operationalized this test by including a tenure dummy in the basic cost stickiness regression. Results show that, compared to other companies, Companies that have a CEO with high tenure significantly display more cost stickiness. These results are meaningless unless we also look at the future impact of this long-term behavior (as proxied by CEO tenure). Prior literature on CEO tenure gives mixed results on the impact of CEO tenure on future performance. I investigated how the cost stickiness levels of companies with the highest future growth compare with other companies. Results show that companies that have realized the highest growth in two years’ time currently display more cost stickiness than other companies. Therefore I argue that accepting more cost stickiness than other companies is the causal bridge between CEO tenure and future growth.

The above results are contingent to the dynamism of the industry. In a stable industry the above results are similar but stronger. In a dynamic industry, companies that have CEOs with high tenure reduce cost stickiness more effectively than other companies. In a dynamic industry there is a relation between reducing cost stickiness more than other companies and future growth but this relation is not significant. This indicates that in a stable industry cost management is a key instrument towards future growth. But in a dynamic industry cost management is less of an influence when it comes to future growth.

In prior literature cost stickiness has been interpreted as a form of agency costs. I found some support that cost stickiness can be caused by exactly the opposite of agency costs; adding value with long-term behavior.

(34)

References

Ali, A., & Zhang, W. (2012). CEO Tenure and Earnings Management. Available at SSRN 2060119.

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

Antia, M., Pantzalis, C., & Park, J. C. (2010). CEO decision horizon and firm performance: An empirical investigation. Journal of Corporate Finance, 16(3), 288-301.

Banker, R. D., Byzalov, D., & Threinen, L. (2013). Determinants of International Differences in Asymmetric Cost Behavior. Working paper, Temple University.

Bergh, D. D. (2001). Executive retention and acquisition outcomes: A test of opposing views on the influence of organizational tenure. Journal of Management, 27(5), 603-622.

Brauer, M. F. (2013). The effects of short-term and long-term oriented managerial behavior on medium-term financial performance: longitudinal evidence from Europe. Journal of Business Economics and Management, 14(2), 386-402.

Calleja, K., Steliaros, M., & Thomas, D. C. (2006). A note on cost stickiness: Some international comparisons. Management Accounting Research, 17(2), 127-140.

Cannella, A. A., & Hambrick, D. C. (1993). Effects of executive departures on the performance of acquired firms. Strategic Management Journal, 14(S1), 137-152.

Castanias, R. P., & Helfat, C. E. (1991). Managerial resources and rents. Journal of management, 17(1), 155-171.

(35)

Chen, D., & Zheng, Y. (2013). CEO tenure and risk-taking. Available at SSRN 2038064.

Covin, J. G., Green, K. M., & Slevin, D. P. (2006). Strategic process effects on the entrepreneurial orientation–sales growth rate relationship. Entrepreneurship Theory and Practice, 30(1), 57-81.

De Angelis, D., & Grinstein, Y. (2010). Pay for the right performance. Johnson School Research Paper, 03-2011.

Dekker, R. (2012). The influence of corporate governance and CEO incentive compensation on cost stickiness, Master Thesis.

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 cost behavior in private Belgian firms. The Accounting Review, 87(4), 1219–1246.

Dikolli, S. S., Mayew, W. J., & Nanda, D. (2013). CEO tenure and the performance-turnover relation. Review of Accounting Studies, 1-47.

Finkelstein, S., & Hambrick, D. C. (1990). Top-management-team tenure and organizational outcomes: The moderating role of managerial discretion. Administrative science quarterly, 484-503.

Fox, F. V., & Staw, B. M. (1979). The trapped administrator: Effects of job insecurity and policy resistance upon commitment to a course of action. Administrative Science Quarterly, 449-471.

Govindarajan, V. (1989). Implementing competitive strategies at the business unit level: Implications of matching managers to strategies. Strategic Management Journal, 10(3), 251-269.

(36)

Graham, J. R., Harvey, C. R., & Rajgopal, S. (2005). The economic implications of corporate financial reporting. Journal of accounting and economics, 40(1), 3-73.

Hambrick, D. C., & Mason, P. A. (1984). Upper echelons: The organization as a reflection of its top managers. Academy of management review, 9(2), 193-206.

Hambrick, D. C., & Fukutomi, G. D. (1991). The seasons of a CEO's tenure. Academy of Management Review, 16(4), 719-742.

Hambrick, D. C., Geletkanycz, M. A., & Fredrickson, J. W. (1993). Top executive commitment to the status quo: Some tests of its determinants. Strategic Management Journal, 14(6), 401-418.

Haspeslagh, P. C., & Jemison, D. B. (1991). Managing acquisitions: Creating value through corporate renewal (Vol. 416). New York: Free Press.

Henderson, A. D., Miller, D., & Hambrick, D. C. (2006). How quickly do CEOs become obsolete? Industry dynamism, CEO tenure, and company performance. Strategic Management Journal, 27(5), 447-460.

Hermalin, B. E., & Weisbach, M. S. (2012). Information disclosure and corporate governance.

The Journal of Finance, 67(1), 195-233.

James, H. S. (1999). Owner as manager, extended horizons and the family firm. International Journal of the Economics of Business, 6(1), 41-55.

Jensen, M. C. (2004). The agency costs of overvalued equity and the current state of corporate finance. European Financial Management, 10(4), 549-565.