Are there CEO- and CFO fixed effects on cost stickiness for SG&A costs?

Amsterdam Business School

MSc thesis

Are there CEO- and CFO fixed effects on

cost stickiness for SG&A costs?

Name: Jeroen van Assouw

Student number: 10834613

Thesis supervisor: Mr. M. Schabus MSc

Date: 15 June 2016

Word count: 12.774,00

MSc Accountancy & Control, specialization Control

Statement of Originality

This document is written by student Jeroen van Assouw, 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 study investigated the relationship between top management positions (particular CEO and CFO) and their influence on cost stickiness. Given the fact that CEO and CFO are powerful positions, those managers have managerial discretion and were able to influence firm policies. Furthermore, it is indicated that CEOs and CFOs have an influence on firm performance indicators. Therefore, I predict that those managers also have an effect on cost stickiness. The aim of this study was to figure out whether the management, based on agency considerations and the managerial discretion theory have an influence on the firm cost stickiness level. I have examined the influence of the CEO and CFO on cost stickiness. I have created a CEO sample with 5.961 (2.238 unique CEOs) cost stickiness observations and a CFO sample with 5.828 (2.230 CFOs in the sample). These samples also included: firm financial indicators and industry classification codes. From the Compustat database, I gathered S&P1500 data. SG&A costs were used as cost category to examine cost stickiness. Based on the CEO- and CFO panel, I have found that there is both a CEO- and CFO fixed effect on cost stickiness. These fixed effects are controlled for firm-, industry and time related indicators.

Contents 1 Introduction ... 5 2 Literature review ... 8 2.1 Cost Stickiness ... 8 2.2 CEO effects ... 13 2.3 CFO effects ... 15 2.4 Hypothesis development ... 16 3 Methodology ... 18 3.1 Sample selection ... 18 3.2 Econometric model ... 18 3.3 Operationalization of variables ... 21

3.4 Fixed effect and Control variables: ... 22

4 Results... 24 4.1 Sample description ... 24 4.1.1 Descriptive statistics ... 24 4.1.2 Correlation test ... 28 4.1.3 Multicollinearity ... 31 4.2 Multivariate analysis ... 32

4.2.1 CEO fixed effect on cost stickiness ... 32

4.2.2 CFO fixed effect on cost stickiness ... 34

5. Conclusion ... 36

5 Referencing and literature list... 38

1 Introduction

Recently, Chief Financial Officers (CFOs) became responsible for the accuracy and

completeness of financial statements in the US. This have been a result of the Sarbanes Oxley Act (Sox), that have been conducted in 2002 after some major financial scandals. The Sox act requires amongst other that, the Chief Executive Officer (CEO) and the CFO of US listed firms have to certify their financial statements as well as their disclosures. Whereby they have to declare that the financial statements and disclosures fulfil the requirements of material accuracy and completeness.

For the reason that the Sox implementation assumed to have a big influence for the management and firm, researchers were triggered to investigated whether the CFO role changed. Ge, Matsumoto and Zhang (2011) found that the power of the CFO has increased, after the Sox implementation. Decisions made by the management could not always be rationally declared, because executives have decision discretion. A study of Qin, Mohan & Kuang (2015) highlighted that there are different CEO types. It seemed that many CEOs are overconfidence. Within a large dataset there are arguable different CEO types, for example more- or less conservative CEOs. The individual influence of the top management on firm policies is called the managerial fixed effect. This is a net effect unregarded at which firm a executives have worked for. The managerial fixed effect is formed by several personal characteristics, like: behaviour,

(over)confidence and educational background and other factors which do not change overtime. Geiger and North (2006, p.782) used the following declaration for the individual influence of Directors: “If individuals appointed as CFOs are in a position to significantly affect the reporting of their company’s financial condition, then a change in CFO personnel may well lead to

different financial results to be reported under similar business conditions”. There have been several researches about the CEO effect on firm performance. This study will investigate the managerial fixed effect on cost stickiness. This relationship has not been investigated to the best knowledge of the author.

Anderson, Banker and Janakiraman (2003) were the first researchers, who empirically indicated the existence of this cost asymmetry phenomenon, they used cost stickiness as name for this. In the past, there was a general consensus about cost behavior, the traditional cost model was the leading model, this model contained fixed- and variable costs. Bugeja, Lu and Shan (2015) mentioned that traditional cost models assumed that costs could be directly linked to firm activity level and that costs directly depends on the cost driver.

But Bugeja et al. (2015) highlighted that this is in practice not always the case, costs not always directly change when the cost driver changes. Today, it is commonly accepted that costs do not always change proportional with the activity level and cost stickiness has been investigated several times in different institutional settings. Furthermore, authors have found several

determinants of cost stickiness. Examples of the settings and the determinants of cost stickiness are highlighted in section 2.1. To my knowledge, nobody investigated the influence of particular CEO and CFO effects (henceforth managerial effects) on cost stickiness. Moreover, this study will examine whether the contention of Geiger and North (2006) that CEOs and CFOs are equally important, also applies for cost stickiness.

To measure whether there is, and to what extent there is, a managerial fixed effect on stickiness, this research will control for firm specific factors and timing differences. My study assumes that CEOs and CFOs, have management discretion and are able to impact cost stickiness. Geiger and North (2006) concluded that CFOs have significant influence on the reported financial statements and almost have the same responsibilities as the CEO. Therefore, this research will examine whether CEOs and CFOs have influence on cost stickiness.

This research will focus on one cost category, namely: Selling, General and

Administrative (SG&A) costs. There are a couple of reasons to use this category. Firstly, US public listed firms are required to record the SG&A-costs, so there is enough available data. Secondly, these costs represent a major part of the total costs. Thirdly, SG&A costs consists of a broad range of costs, therefore are these costs not very sensitive to small firm policy

adjustments. So, my study tends to figure out whether a change of the CEO position and/or a change of the CFO position, influences the cost stickiness level, because it is assumed that management is able to influence firm policies and individual managers differ. Therefore, the research question of this study is: “Are there CEO- and CFO fixed effects on cost stickiness for SG&A costs of US listed companies”.

The underlying used theory to examine the relationship between the managerial fixed effect and cost stickiness was the agency- and managerial discretion theory. I found that both CEOs and CFOs have an impact on cost stickiness. The managerial fixed effect on cost

stickiness exists, because executives have the opportunity to influence firm policies and therefore the indicators of cost stickiness. Furthermore, it was observed that cost stickiness is industry related. I could not observe a significant effect between the chosen firm financial indicators and cost stickiness. Furthermore, there is support that the cost stickiness level is on average lower for female CEOs and CFOs than for male Executives.

To the best knowledge of the author, a comparable investigation has not been

conducted, therefore this study fulfilled the gap in the literature about the influence of directors on cost stickiness. The theoretical contribution of this study is the indicated managerial influence on cost management and specific on cost stickiness of SG&A costs in the US. Prior studies showed different causes for cost stickiness, related to firm specific characteristics. Three main influence factors of cost stickiness were detected: economic determinants, agency problems and incentives influences. It is suggested that management could be added as a determinant for cost stickiness. The practical contribution of this study for firms, stakeholders and other interested parties is the increased understanding about the influence CEOs and CFOs have. Shareholders could use cost stickiness for evaluation purposes and to monitor the CEO’- and CFO’

performance. Bugeja et al. (2015) claimed that increased knowledge about the existence and extent of cost stickiness has direct benefits for stakeholders and the economy as a whole. Furthermore, they mentioned that it provided useful information about cost control and it improved decision making.

The remainder of this paper is structured as followed. In the second section, additional analysed literature will be provided about cost stickiness theory, the CEO- and the CFO role, managerial effects on firm performance, managerial- and firm fixed effects. Furthermore, the hypothesis will be developed and substantiated in this section. In the third section, information about the research model, operationalization of variables and the sample description will be conducted. Section four contains the results and descriptive statistics. In section five the conclusions and implications will be highlighted.

2 Literature review

2.1 Cost Stickiness

This subchapter will analyze several articles from the cost stickiness literature. Moreover, the reasons to choose for SG&A costs as well as some other considerations will be mentioned. Nowadays, researchers agreed that costs not always behave proportional in relation with the activity level. Anderson et al. (2003) first explained how cost stickiness arises. They said that cost stickiness has been originated by the manager’ assessment of adjustment cost and the expected future demand. When managers expect that the decreasing demand level is temporarily,

managers will not adjust resources. As a consequence, the costs will not decline proportional if activity decrease as it increase when activity increase. When management assessed that future demand has permanently decreased, they need to take measures to avoid cost stickiness. However, there will be likely some short term adjustment costs, if there is chosen to adjust the costs. Adjustment cost are cost that arise when committed resources or assets are removed or sold, like costs of dismissal. Firms tend to maintain resources for example because they expect future growth. Amongst others, Anderson et al. (2003) claimed that the extent of cost stickiness is a management decision to maintain committed resources or not. Therefore, it is likely that individual Board members and in particular the CEO and CFO will have a major influence on cost stickiness.

Anderson et al. (2003, p.48) formulated cost stickiness as follows: “costs are sticky if the magnitude of the increase in costs associated with an increase in volume is greater than the magnitude of the decrease in costs associated with an equivalent decrease in volume”. While, Bugeja et al. (2015, p.250) provided the following definition: “costs appear to rise more with an activity increase than they fall with an activity decrease”. In the current literature, many

researchers attempted to investigate the influence of various determinants on cost stickiness. Furthermore, several studies compared cost stickiness level, within countries, industries and firm types. The following studies are examples of studies that demonstrate cost stickiness and related it to firm specific factors. The studies have measured the magnitude and the association between the determinants and cost stickiness. The determinants that have been investigated, are: firm size, profitability of firms, country origin, industry, activity change ranges, listed versus unlisted firms, cost type, empire-building behavior and incentives.

At least three researchers have investigated the relationship between firm size and cost stickiness level. They all have concluded that small firms tend to have higher volatility in sales and earnings than large firms.

Therefore, they are better and more likely able to adjust committed costs and supposed to have a lower level of cost stickiness (Fama and French, 2004; Wei and Zhang, 2006). Also in Italy, Dalla Via and Perego (2014) have discovered that cost stickiness is lower for small- and medium sized Italian companies, than for large sized Italian firms. Dierynck, Landsman and Renders (2012) had an explanation for the firm size effect on cost stickiness: larger firms do pay more attention to their labor market reputation and will less likely dismiss employees. Despite of this, larger firms as well as high profitable firms, also will likely have more resources to maintain the committed resources (Dierycnk et al., 2012). Previous studies mostly focused on one cost category and especially on SG&A- or labor costs. Despite of this, Dalla Via and Perego (2014) contributed by analyzing the cost stickiness level for three cost categories: cost of goods sold, labor costs and operating costs. They only found cost stickiness for labor costs.

Bugeja et al. (2015) figured out three determinants of cost stickiness: asset intensity, employee intensity and short term incentives. These incentives resulted into increased earnings or decreased losses, this is an example of managerial judgment or earnings management. Furthermore, they analysed whether origin country of firms influenced the cost stickiness level. They argue that there are valid reasons to assume that there are country differences, because of differences in economic structure, market competition, governance environment and firm characteristic. They have used information from Australian- and US firms and they compared those two countries. They found on average that Australian firms have a lower level of cost stickiness than US firms have. Finally, they observed that cost stickiness varies by industry. Another interesting study in the cost stickiness literature conducted by and Weidenmier (2003), investigated whether there is a threshold in activity change, that forced cost stickiness. They discovered whether and when the activity level would become the driving force for cost stickiness. They examined for specified activity change ranges the cost stickiness level, whereby activity level has been measured by total sales. They found that the activity level has to change with at least 10%, otherwise cost stickiness will not be caused by that activity change. The authors did also find significant differences per industry.

While, the earlier mentioned articles have all investigating cost stickiness for public listed firms, Dierynck et al. (2012) investigated unlisted private Belgian firms and figured out whether profitability level influenced cost stickiness. They have found costs stickiness for both high profitable- and less profitable firms. Whereas, the effect was higher for high profitable firms. So, this means that the cost stickiness level is higher in high profitable firms than in less profitable firms. The results are comparable with the studies about the firm size.

Large- or high profitable firms have more resources to maintain assets, or maintain employees that could not be fully employed within their function, when the activity level decreased, therefore those firms could have more patience to maintain resources. Finally, Dierynck et al. (2003) also investigated and found that there is an significant effect between incentives to manage and beat the zero benchmark and labor cost stickiness. For Belgian firms it is very important to avoid the situation of reporting a loss two years in a row, otherwise a going-concern declaration from the auditor is required and banks usually intervene in that situation. Therefore, the authors found that the incentive to avoid a loss leads to more labor cost adjustments when activity decreased, resulting in less cost stickiness.

While, it is commonly accepted that cost stickiness exists, it has been demonstrated that cost stickiness is not always present. For example, Zanella, Oyelere and Hossain (2015) did not found cost stickiness for labour costs in the UAE. The main explanation why Zanella et al. (2015) could not support the earlier examined findings, appears to be the lack of employment protection legislation (EPL) in the United Arab Emirates (UAE). This emphasized that cost behaviour differed on country level, it highlighted that the factor ‘country’ is an important determinant for the extent of cost stickiness. For managers in UAE, it is not a difficult decision to reduce labour expenses in an economic downturn, because the adjustment costs, in this example cost of dismissal, are relatively low. This simplifies the management decision to adjust assets or resources and that declares that cost stickiness could not be demonstrated. Anderson and Lanen (2009) also do not found cost stickiness for the following specific cost categories: Property Plant and Equipment (PPE) costs, Research and Development (R&D) costs and advertising costs. The latter one, is categorized into the SG&A costs for firms that applied International Financial Reporting Standards (IFRS) guidelines or US GAAP.

The articles referred to in this paper so far, did investigate firm specific characteristics and their influence on cost stickiness. Guenter, Riehl and Rößler (2014) made an extensive overview of the causes of cost stickiness. They mentioned reasons why managers did adjust committed resources downward, these are: legal reasons, social personnel policies, extent of adjustment costs of hiring new personnel and agency related costs like empire-building threat. Moreover, they mentioned that if the ratio of adjustment costs to holding costs increased, cost stickiness likely increased. When firms face higher adjustment costs, for example training employees to reach the desirable level, firms tend to maintain assets this combined with economic downturn demonstrated a cost stickiness increase. Furthermore, they provide an overview of factors that influence adjustment costs: actual sales level, capacity utilization, management future expectations, managerial incentives and psychological aspects.

These adjustment factors were summarized into the following categories: internal- and external factors, industry, country, technology and demand variability. Most of these factors are related to the firm fixed effect.

While, Anderson et al. (2003) proposed that cost stickiness occurred because there is rationale trade-off decision between adjustment cost and holding costs. Chen, Lu and Sougiannis (2012) concluded that agency problems are also an important factor that could influence the managerial assessment and thereby cost stickiness. The effect of the agency conflict on cost stickiness, is one of the reasons for this study to investigate the managerial influence on cost stickiness. A manager might take a decision based on self-interest instead off the best interest of the firm. Chen et al. (2012) have found that “empire building”, which means that CEOs prefer to control a larger firm over a smaller firm, that could be a result of incentives, resulting in higher cost stickiness. The incentive structure encouraged managers to perform behaviour and to decision making that firms will expand. The reason for this is that managing a larger firm will likely lead to higher annual compensation. When the CEO only act on self-interest, managers shall not reduce committed resources and this results into higher cost stickiness. Chen et al. (2012) found that the effect of empire building on cost stickiness is smaller when a firm has implemented adequate corporate governance mechanisms. They further mentioned that managers who decided to leave the firm on short term, will be less interested in short term growth or less empire building behaviour. So, it is arguably to assume that they care less about long term performance, like reputation issues. They will be more focused on cutting committed resources than CEO that have been appointed recently.

However, Kama and Weiss (2013) found that agency driven incentives to meet earnings targets diminish cost stickiness. Because some deliberate decisions depending of the underlying aim either diminish or induce cost stickiness. Managers intentionally adjusted resources to meet the net income targets which decreased cost stickiness level. Despite of the results of Chen et al. (2012) about incentives, Dierynck et al. (2012) emphasised that it is difficult to test the

association between incentive plans and cost stickiness. Because CEOs and CFOs of public traded firms will be likely incentivized by multiple measures. Therefore, it is likely that there will also be incentives that diminish cost stickiness, like reducing costs to realize a desired net income level.

This study will examine the managerial fixed effect on SG&A costs. Firms, that are listed on the S&P 1500 are required to report the total SG&A costs quarterly. The U.S. Securities and Exchange Commission (SEC) recommended firms at the S&P1500 to use US General Accepting Accounting Principles (GAAP).

The SEC is responsible for updating and maintaining US GAAP. Despite of this, also national- or regional GAAP and IFRS have been used by US firms. US GAAP is known as a rule-based set of accounting procedures. While IFRS are principle-based accounting standards. The room for interpretation within the different accounting standards could possibly have an impact on cost behavior. Weiss (2010) concluded that SG&A costs are open for managerial judgment, that may bias cost stickiness. Despite of this, all S&P 1500 firms that will be incorporated in the sample, have been audited by an independent auditing firm. Auditors have interest to comply with the standards, otherwise the auditing firm will be fined because of auditor liability that has been implemented in the US law. The auditors provide assurance about the financial statements of the firms, this improved the reliability of the reported SG&A costs.

SG&A costs is a major non-production cost, presented in the income statement. Anderson et al. (2003) and Guenther et al. (2013) mentioned that many components of SG&A costs are activity driven. According to Issing (2013, p.20), SG&A costs in US GAAP did contain: employee training, maintenance, travel, rent, salaries, stock based compensation, commissions to outside sales representatives, legal fees and consulting fees. Issing (2013) named SG&A costs, a catchall category of overhead costs. Many business units and departments are able to influence a part of the SG&A costs. But, in the end the CEO and CFO remain responsible for cost levels and net income. Furthermore, the author mentioned that growth firms are less good at managing SG&A costs than mature firms are.

The choice for using SG&A costs is based on several considerations. First of all, SG&A costs are prone to managerial discretion. Because it is catchall category and short term savings could be realized. Secondly, this cost category is present in every firm and it is not very industry- or firm specific because it is a catchall of costs. Thirdly, SG&A costs contained a considerable part of the total costs, which makes it an important consideration for a firm. Therefore, it is an important consideration to manage net income or earnings. Anderson et al. (2003) used

Compustat data, the dataset they used consisted of 7.629 firms and had a time span of 20 years. They calculated that SG&A cost made up 26,40 % of sales revenue on average. Zacharias et al. (2015) have analysed a couple of resource allocation based factors, including the SG&A ratio. This research managed to relate the influence of firm, industry and CEO characteristics to SG&A costs. It found that CEOs have an impact on the SG&A ratio of almost 13 %.

There are two main cost stickiness factors detected, firm- or industry specific indicators and agency related indicators. CEOs and CFOs have for instance different agency considerations and a different background, those characteristics will likely also result into a different cost

Further research about the influence of management based on the agency theory and

management discretion theory, might better explain the inexplicable cost stickiness results that not could be related to firm specific indicators. The individual managerial influence, will be investigated by observing the managerial fixed effects on cost stickiness, controlled for other related effects. Issues about managerial effect as well as the role of the CEO and CFO will be highlighted into the following two subparagraphs.

2.2 CEO effects

The following two subparagraphs will explain why top managers are able to affect corporate policies. This subparagraph will contain a literature overview about: CEO influence on firm performance, CEO types, CEO discretion and the agency theory. Malmendier, Tate and Yan (2011) mentioned that there are three traditional determinants for corporate financing decisions: firm, industry and market-level characteristics. Those three elements are important indicators that were also observed as determinants for cost stickiness. Recently, some specific stickiness in capital structure have been observed, that could not be related to the traditional determinants (Malmendier et al., 2011). The authors have concluded that these “unobservable factors” are related to the managerial discretion theory. Discretion means: acting on one's own authority and judgment. Managerial discretion means that managers have the possibility to take a decision whereby the underlying considerations are not clear and will not be published. Further research about the influence of the management, will explain the inexplicable cost stickiness results that could not be related to firm specific indicators.

Bugeja et al. (2015) investigated whether there is a short term effect of managerial change on cost stickiness. They have found that cost stickiness will increase on short term after a CEO replacement. Because, during the horizon the current CEO is familiar with the replacement, the CEO has no interest in the empire-building phenomenon. Also, Beatty and Zajac (1987) did research to the performance of terminating CEOs and found that firm value will decrease during the transition period of a CEO change. But, it is also arguable that appointing a new CEO will decrease the quality of decision making, resulting in higher cost stickiness. For the reason that the new CEO has less: CEO experience, firm specific knowledge and industry specific

knowledge. From another point of view, a new CEO look at the firm afresh, which enhanced firm performance on short term after the replacement.

Luo, Kanuri and Andrews (2013) investigated the effect of CEO tenure on firm performance. There are two factors that influenced the relation between CEO tenure and performance, namely: employees and customers.

When the CEO tenure increased, the employee-CEO relationship improved. But the tie between CEO and customer strengthens during tenure only for a while, after a certain time the tie

weakened and this results in performance decrease. Based on this conclusions, Luo et al. (2013) were able to calculate an optimal CEO tenure, that was 4,8 years. After this period, a CEO tend to rely too much on their accumulate knowledge and internal network. Furthermore, they are less attuned to market conditions because of the focus on firm issues and avoiding losses instead of thinking from the client perspective. Shen and Canella Jr. (2002) discovered that there is a U-shaped relationship between departing CEO tenure and operational firm performance proxies, like: Return on Assets (ROA). Both a long-tenured departing CEO and a short-tenured

departing CEO will likely impact the firm’ post succession performance negatively. A short term departing CEO tenure, means that the CEO has not been the competences or environment to significantly influence the firm, this status quo negatively impact the firm’ post succession performance. And the long-tenured departing CEO will largely focus on the short term. Shen and Canella Jr. (2002) agreed with Luo et al. (2013) about the effect of long tenured CEOs on firm performance.

Crossland and Hambrick (2007) mentioned that the CEO is the most powerful person in the organization. Therefore, it is likely that a CEO replacement have a more drastic impact on cost stickiness than a CFO replacement. Clark, Murphy and Singer (2014) mentioned factors that impacts the CEO power, namely: characteristics of task at hand, employees and the external environment. They found that the CEO effect on performance is the largest, when the governance structure and ownership weakly correspond with the institutional logic, the

institutional logic is the degree of cohesion between governance and ownership. It is likely that the more power a CEO has, the bigger the influence, the larger managerial discretion and the influence on cost stickiness. So, CEO power is comparable with managerial discretion, because firms give freedom to managers and will not monitor tightly.

The total influence of management on firm performance, could be summarized as the managerial fixed effects, which means: “the behavior and consequential decisions a CEO or CFO took whatever firm type or at which time he worked for” (Malmendier et al, 2013). While, most firm fixed effects are observable, like: business, industry and work environment. This is not the case for many managerial fixed effects. Tosi, Misangyi, Fanelli, Waldman and Yammarino (2004) did investigated one unobservable managerial determinant or characteristics, namely: charisma level. They found that the CEO’ charisma level does impact the CEO compensation level and the firm’ shareholder value. Nevertheless, they did not found significant influence of charisma on firm performance.

Shen and Canella Jr. (2002) mentioned that there are three types of CEOs, namely: followers, contenders and outsiders. When there are different types of CEOs and these CEOs have different characteristics, this likely result that decision making process differed, which impacted the cost stickiness level. Therefore, I assume that cost stickiness variation could be partly declared by executive choices.

Graham, Li and Qiu (2011) investigated the drivers for executive compensation. They found that the managerial fixed effect, consist of all (un)observable characteristics is a more important driver of executive compensation than the firm fixed effect is. Graham et al. (2011) found observable- and unobservable characteristics that determine variation in executive compensation. The authors examined the influence of some managerial traits, especially overconfidence and early life experience. They mentioned two early life experiences: educated during the great depression period and having a military background. They denominate these specific managerial factors to some of the unobservable characteristics from the management fixed effect. It seemed that both having overconfidence about future cash flows and personal experiences did impact the corporate decision making process.

Chen et al. (2012) have indicated that CEO incentives to increase the short term

performance, are sensitive to empire-building behavior, which actually increased cost stickiness. To measure the relative importance of a CEO, both the agency related issues as well as

managerial discretion theory amongst other have to be taken into account. So, based on prior literature it is suggested that (un)observable CEO characteristics impact corporate decision making and firm performance.

2.3 CFO effects

This subparagraph will contain the same elements as section 2.3, however this subparagraph relate the subjects to the CFO. A CFO is a corporate officer primarily responsible for managing financial risk, financial planning and record keeping. These three activities could be related to factors that are coherent with cost stickiness. First, the attitude to managing financial risk will likely be different per manager. For instance, scaled from risk adverse till risk seeking. Second, financial planning is also related to cost stickiness, it seemed that future expectations about the earnings forecast are an important determinant for the adjustment cost trade-off and therefore cost stickiness level. Third, applying accounting standards is sensitive to managerial judgment, there is room for interpretation to inflate or deflate the cost stickiness factors.

Geiger and North (2006) highlighted that CFOs also have significant influence on the reported financial statements. As a consequence of the Sox legislation, both the CEO and the CFO have to certify, to comply that the financial statements and disclosures fulfil the

requirements of material accuracy and completeness. Based on these requirements, Geiger and North (2006) decided to investigate the influence of CFOs on firm performance.

They concluded that the responsibility level at that moment was the same for CEOs and CFOs. “If individuals appointed as CFOs are in a position to significantly affect the reporting of their company’s financial condition, then a change in CFO personnel may will lead to different financial results to be reported under similar business conditions” (Geiger and North, 2006, p. 782). This conclusion supported the assumption that there is a managerial fixed effect on firm performance indicators. Because one CFO would possibly take other decisions than another CFO, despite of the fact that the business conditions are similar.

Ge et al. (2011) investigated the relationship between CFO individual characteristics. They pick the following characteristics, CFO style and the likelihood of accounting adjustments. They also paid attention to CFO fixed effects and CFO behaviour. A prior investigation of Hambrick and Mason (1984) found that individual CFO characteristics impact corporate decision outcomes. Another mentioned theory is, that it might be that a CEO determines the tone at the top and that this will impact the CFO style. But, this research was executed before the Sox act have been implemented, which is identified as a driver for the change of the CFO role. Ge et al. (2011) refuted the quote that tone at the top impact CFO behaviour, they claim that CFOs have discretion which impacted the firm. So, it seemed CFOs are powerful enough to eliminate this theory. Ge et al. (2011) found inter alia that the CFO personal situation did impact their behaviour and decision making choices. The CFO style could only partly declared by observable CFO characteristics. The extent of job demands did further enhance the CFO impact on accounting polies. So, the CFO power increase if there are multiple other job opportunities. Concluding, individual characteristics of CFOs differ. The CFO behaviour and the CFO fixed effect are shaped by background and other personal characteristics.

2.4 Hypothesis development

While there are mixed results about the consideration whether the CEO is more important or more powerful than the CFO, it is proposed that CEOs in general do have more power to influence corporate policies. Meaning they will likely have more influence on cost stickiness than CFOs. Despite of this, both CEOs and CFOs are able to significantly influence firm policies and decision making.

More specific to my subject, I expect that both CEOs as well as CFOs have the opportunity to influence SG&A costs and the activity level. This opportunity could be related to the managerial discretion theory.

Furthermore, SG&A is a broad cost category, that triggers managers to apply management discretion to manage earnings, by inflate or deflate SG&A costs. Moreover, I expect that there will be individual differences between different CEOs and CFOs.

These differences could be declared by time-differences, for instance: changing economic or market conditions. But also by individual differences caused by (un)observable characteristics of different managers. Therefore, this study will focus to figure out whether there is a net fixed CEO- and CFO effect on cost stickiness. Within this research, two hypothesis will be tested, compared and analyzed.

Hypothesis one: CEOs have an effect on the level of cost stickiness

Beatty & Zajac (1987) found that changing a CEO will decrease firm value. Moreover, it is arguable that changing a CEO will result in less good decision making on short term. Because a prior CEO may have more CEO experience, but more likely have more firm specific

knowledge, than the new appointed CEO, who may not have firm specific knowledge at all. This has a negative influence on firm performance, since the lack of firm specific knowledge seemed more important than CEO experience to become a successful manager (Hameri and Kayuncu , 2015). Bugeja et al. (2015) have found that changing a CEO impacts firm performance. In general, a CEO is responsible for the financial figures and also has decision making authority. Therefore, I expect that a CEO does have an impact on cost stickiness.

Hypothesis two: CFOs have an effect on the level of cost stickiness

Hameri and Kayuncu (2015) claimed that firm specific knowledge is more valuable than previous management experience. Based on that research, it is preferable to hire someone for the CFO position, that has already been working in the firm instead of someone who gained his

experience at another firm. Therefore, it is likely that a CFO replacement will result in different cost stickiness observations. Furthermore, the CFO has a powerful executive position. Because of the managerial power, the CFO has management discretion and is able to influence firm policies. Therefore, I predict that a CFO also will effect firm’ annual cost stickiness.

3 Methodology

This section contains the main considerations of the methodology. The sample selection will be illustrated, the econometric model will be highlighted, the variables will be explained and the latter subparagraph contains the CEO and CFO- fixed effect as well as the control variables.

3.1 Sample selection

The data was gathered from the Compustat Northern America database as well as from the Compustat Execucom database. Both databases contain company data from the Standard & Poors 1500 index (S&P 1500). This index contains 1500 firms listed on the NYSE or NASDAQ. It is divided into three sub-indices: the S&P 500 that contains the 500 firms with the highest market capitalization, the S&P 600 the “small cap” and the S&P 400 the “mid cap”.

The sample size is chosen based on data availability. It contains the maximum corresponding period of time. In order to separate firm- and managerial fixed effects and measure the impact on cost stickiness, it is necessary to get data from the longest period of time possible. It is important to have date for the longest time span possible in order to have data of as many firms possible that were managed by more than one CEO, otherwise it is not possible to assign the cost stickiness observations to the firm fixed effect or to the managerial fixed effect. Given the fact, that there is only data available about the CEO- and CFO code and

characteristics after 1991, the initial selected sample period is from January 1992 till December 2014. CFO information is available from 2006, so the CFO sample period is from January 2006 till December 2014. I have chosen to use the sample period 2006-2014 for both samples, to enhance the comparability.

3.2 Econometric model

The aim of this study, is to calculate the managerial effect on cost stickiness and taken into account timing differences. The timing differences do need to be taken into consideration because external circumstances like the general economic situation could change overtime. This change should not be attributed to the managerial fixed effect. The calculated managerial fixed effect is not impacted by firm characteristics and tenure differences, because a joint average of fixed effect of managers on cost stickiness is calculated.

There are two situations possible whereby the annual cost stickiness could not be calculated. First, a firm faced four quarters of increased or decreased activity change.

However, this does not impact the fixed effect of a specific manager, because observations from other years will be averaged, so that there is still a specific managerial fixed effect.

Second, when a particular manager has worked for only one firm during the full sample period. In that case it is not possible to distinct the cost stickiness, due to the managerial effect or due to the firms fixed effect. Following the approach of Anderson and Lanen (2009), my study will also exclude observations whereby the costs move into the opposite direction of the firm’ activity level. These consideration are incorporated into the cost stickiness measure, that were chosen. I decided to exclude financial institutions (firms having a SIC code between 6000-6799), because these firms are incomparable with other type of firms in the sample based on the financial ratios and their financial structure. Also, “letterbox firms” will be eliminated. To observe which firms are letterbox firms, I will drop firms having a low asset level or a sales level that is lower than one. Another consideration is to maintain only firms with ISO country code USA. This removed potential bias because of country specific differences.

To ensure to control for indicators that could not be related to unobservable managerial fixed effects, I will control for firm specific issues as well as time differences. These factors otherwise would be unfairly assigned to the managerial fixed effect on cost stickiness. The timing differences do need to be taken into consideration because external circumstances like the general economic situation could change overtime. I will use the following time variant variables as control variables: Firm size, Return on Assets (ROA), Market-to-Book ratio (MTB), Leverage and Asset Intensity. Regarding firm size, amongst others Della Via and Parego (2014) found that there is positive relationship between firm size and cost stickiness. Thus, in general larger firms have a higher level of cost stickiness than smaller firms. The latter four time variant control variables are financial ratios. It was expected that ROA positively correlates with cost stickiness, because more profitable firms likely have more resources to maintain assets. This, combined with facing activity decrease, results into cost stickiness. There is no specific expectation about the influence of MTB and Leverage on cost stickiness, but I have incorporate these control variables, to add some firm specific indicators. Moreover, based on the research of Subramanian and Weidenmier (2003), the authors mentioned asset intensity as a driver for cost stickiness. They found a positive association between asset intensity and cost stickiness.

Despite the fact that there is controlled for firm specific variables, also one time invariant variables have incorporated. Otherwise, omitted variables will be unfairly attributed to the

managerial fixed effects. The time invariant variable that will be included is: Industry. Time invariant variables assumed to be not change overtime. The industry variable will be used to examine whether the degree of the managerial effect on cost stickiness differed per industry.

The firms will be categorized into industry types, based on SIC codes. Subramanian and Weidenmier (2013) analysed industry specific differences. They discovered the highest cost stickiness level for the manufacturing industry, these firms usually have relative high asset- and employee intensity ratio’s. Following the approach of Subramanian and Weidenmier (2003), my study will initially categorize industries by using the first digit SIC codes, to create the following four industry indicators: Manufacturing firms (SIC 2 and 3), Merchandising firms (SIC 5), Services (SIC 4, 7 and 8) and Financial Services (remainder category excluded from financial institutions).

The control variable firm size is included to examine whether this might influence cost stickiness, this is also a factor that could not be attributed to the managerial effect. I have used the following approach to indicate different firm size categories. First, I have ranked the firms in the sample, based on total revenue. Secondly, I created three quintiles to categorize firm in the following firm size groups: 1. Small firms, 2. Middle firms and 3. Large firms.

To, estimate and control for time differences, I added a dummy variable for each fiscal year (FY) in the sample. I used the following two cross-sectional regression models to examine the net managerial fixed effect on cost stickiness. Whereby, model one will estimate the CEO effect and model two the CFO effect.

Model 1: Cost Stickiness (DV) = β0 + β 1 CEO FE (IV) + β 2 Asset Intensity + β3 Firm Size + β4 ROA + β5 MTB + β6 Leverage + β7 Industry-indicator + β8 Fiscal Year-indicator + ε i t

Model 2: Cost Stickiness (DV) = β0 + β 1 CFO FE (IV) + β 2 Asset Intensity + β3 Firm Size+ β4 ROA + β5 MTB + β6 Leverage + β7 Industry-indicator + β8 Fiscal Year-indicator + ε i t

3.3 Operationalization of variables

To calculate cost stickiness level, I have used the following variables: Revenue Total as proxy for SALES (Compustat #REVT) and SG&A costs as proxy for COSTS (Computstat #XSGA). Cost stickiness calculation is according to Bugeja et al. (2015): “the associating change of costs with sales revenue in periods of revenue increase, with the variation of costs with sales revenue in periods of revenue decrease”. My study will use the conventional cost stickiness measure, developed by Weiss (2010). This is a measure for calculating annual cost stickiness per firm, based on quarter sales- and cost levels. The cost stickiness measure of Weiss (2010), estimated the difference between the cost function slope in periods of upward- and downward activity (see figure one).

Figure 1: Conventional Cost Stickiness measure, developed by Weiss (2010)

The dependent variable is “Sticky”, whereby “i” indicates stickiness on firm level and “t” means the time period in years. Whereby ∆ COST will be the reported SG&A costs of the corresponding quarters. The same applies for ∆ SALE. The natural logarithm will be taken from: ∆ COST/∆ SALE. Furthermore, is the most recent (of the last four quarters) decrease in sales and is the most recent (of the last four quarters) increase in sales, such that a sales decrease and a sales increase within one year could be measured (Weiss, 2010). The first part of the calculation is the most recent quarter (Q till Q-3) of relative cost decrease of SG&A cost when the activity level decreased, while the second part is the most recent quarter that the relative costs increase when activity increase. If the outcome of the cost stickiness calculation is negative, we speak of cost stickiness. Because the relative cost increase exceeded the relative cost decrease. Therefore, the more negative the outcome variable “Sticky” is, the more intense cost stickiness. When the outcome of the “sticky formula” is positive, we speak of anti-stickiness. Thus, the relative costs function decreased more if activity decreased, than the relative cost function increase when activity level increased. To observe firm year cost stickiness, I will not remove anti-stickiness observations, because anti-stickiness could be a result of management actions and therefore a part of management discretion.

The following items have been selected from the Compustat Global Database: Global Company Key, Fiscal Year, Fiscal Quarter, Company Name, SIC Industry code, ISO Country Code Headquarter, SG&A Expense, Total Revenue, Net income, Total Assets, Total Equity, Total Liabilities, Market Price. Furthermore, the following variables have been selected from the Compustat Execucom database, to investigate CEO- and CFO identification and characteristics data: Global Company Key, Executive ID, CEOANN, CFOANN and Executive gender. Therefore, also a possible CEO gender effect could be analyzed.

3.4 Fixed effect and Control variables:

To calculate the net managerial fixed effect on cost stickiness, both time variant- and time invariant control variables have incorporated. Therefore, this study controls for significant differences in market conditions, timing differences and firm specific issues. These influences need to be detected, to avoid that omitted variables may will be unfairly attributed to the CEO or CFO fixed effect. Table one contain an explanation of the variables.

Table one: control variables in the model

Variables: Explanation:

Asset intensity: Total assets to sales revenue Firm size: Total revenue

Return on Assets: Net income divided by total assets

Market-to-Book ratio: Market value market capitalization divided by book value of equity Leverage: Debt-to-equity ratio: debt divided by equity

Industry SIC: Indicator variable for each industry group Fiscal Year: Indicator variable for each consecutive fiscal years

The first step is to calculate the annual cost stickiness level per firm. Subsequently, the executives ID codes as well as the control factors will be related to the cost stickiness firm year observations. Different managers have different (un)observable characteristics, these

characteristics will be expressed with the fixed effect option, this option makes it possible to calculate a cost stickiness level per executive. The fixed effect option, is the net effect of one specific manager on cost stickiness. So, all firm-year cost stickiness observations will be averaged and attributed to executives. Therefore, it is possible to observe individual differences between executives.

I have also added the variable executive gender, to compare average male- and female cost stickiness observations. Female CEOs comprise a disproportionately small percentage of the total CEO positions (Brady, Isaacs, Reeves, Burroway and Reynolds, 2011). Mohan (2014) summarized literature about the effect of gender on firm performance. The author observed mixed results about the influence of gender on firm performance. There were two types of researchers about the gender effect observed: stock price reaction after renouncement and long-term firm performance.

Beckmann and Menkoff (2008) indicated that females tend to be more risk-averse than males. A research of Coxbill, Sanning and Chaffer (2009) found that there is a stock price reaction, the stock price did increased more after a female appointment in comparison with that of a male appointment (3,55 % - and 2,63 % increase in the first three days respectively).

Other researchers were not able to detect significant difference between gender. While, Dezsó and Ross (2012) did found a positive relationship between more females below the CEO position and better firm performance. But, they did not found significant results for the effect of the CEO position on long-run performance. On beforehand, I do not have an expectation, whether gender did significantly impact cost stickiness and whether the cost stickiness level is higher for male or female.

4 Results

4.1 Sample description

The initial sample consists of 166.715 firm year observations. There were 15.248 different firms in the sample. So, the average number of years per firm in the sample is almost 11 years. Because there are still many firm year observations with incomplete CEO- and CFO data, those

observations needed to be eliminated. I obtain two samples, sample one is prepared for testing the fixed effect of CEOs and sample two for testing the fixed effect of CFOs.

The initial CEO sample had 13.770 CEO year observations and there were 4.210 unique CEOs in the CEO sample. On average, one CEO is related to a firm(s) for only 3,27 years. Some CEOs have been working for more than one firm in the sample. The CEO tenure is not

comparable with the optimal CEO tenure per firm measured in the article of Luo et al. (2009), which is 4,8 years for one particular firm. The average CEO tenure should be likely higher, but the reason for this is that many CEO- and CFO identification data is unavailable. In addition, the final CFO sample consist of: 5.828 observations and 2.238 unique CFO’ characteristics, this results in an average CFO tenure of: 2,60 years. Also for the CFO sample applied that a particular CFO could be worked for more than one firm. I have chosen to adjust the CEO sample period to the CFO sample period. So, also the sample period for the CEO is: 2006-2014, there are 5.961 firm year observations and 2.281 unique CEO characteristics in this final sample. To control for potential outliers, I have decided to use the Winsorizing function for the variables, that have been incorporated in the model to test my hypothesis, on a 1% level.

Winsorizing means modification of certain kind of data points at the end of the tails, to the next value. Winsorizing improved the normal distribution and is a tool for dealing with extreme observations. These extreme observations are expected to be outliers, but it could also be a: measurement-, typographical-, contamination or distribution error. Therefore, these “extreme” observations will not be removed but only modified in the initial sample. So, the outliers are not errors, but are the nature of the dataset, that could be caused by several factors.

4.1.1 Descriptive statistics

I will provide some descriptive statistics to highlight the differences and characteristics of the samples. Table two summarized main financial figures. For all the descriptive statistics and other tables applied that I have used two panels: Panel A and Panel B. Panel A contains data that has been obtained from the CEO sample and Panel B contains data from the CFO sample. In general, there are no remarkable differences between the two samples.

Although on average, periodical financial indicators are slightly larger in panel B than in panel A. Because there are some structure differences within the panels. The panels do not have an equal number of observations. Furthermore, there are also some different firms in both panels. The variables above the diagonal line are absolute values in US Dollars, while the values below the diagonal line are financial ratios.

Table 2: Firm related financial indicators (absolute values * 1 Million) Panel A (2006-2014):

Variables: #Observations: Mean: St. Deviation: Min. Max.

Assets 5.961 USD 5.532,03 9.332,89 USD 0,48 USD 38.036,50 Liabilities 5.961 USD 3.386,67 6.070,72 USD 2,47 USD 25.616,75 Equity 4.030 USD 3.607,43 11.385,79 USD -9.920,00 USD 160.580,50 Net Income 5.961 USD 737,91 1.432,44 USD -894,83 USD 5.404,00

Asset Intensity 5.961 4,86 3,86 0,58 65,40 Return on Equity 4.030 0,29 1,57 -14,24 17,19 Return on Assets 5.961 0,03 0,08 -2,54 0,26 Market-to-Book value 5.825 3,01 5,55 -47,22 51,75 Leverage 4.030 1,46 4,34 -24,42 32,61 Panel B (2006-2014):

Variables: #Observations: Mean: St. Deviation: Min. Max.

Assets 5.828 USD 5.569,82 9.372,62 USD 12,47 USD 38.036,50 Liabilities 5.828 USD 3.490,87 6.094,24 USD 2,47 USD 25.616,75 Equity 4.017 USD 3.609,59 11.401,29 USD 9.920,00 USD 160.580,50 Net Income 5.828 USD 740,77 1.436,75 USD 894,83 USD 5.404,00

Asset Intensity 5.828 4,88 3,89 0,58 65,40

Return on Equity 4.017 0,28 1,57 -14,24 17,19

Return on Assets 5.828 0,03 0,08 -2,54 0,26

Market-to-Book value 5.700 3,02 5,55 -47,22 51,75

Leverage 4.017 1,48 4,32 -24,42 32,61

For most firms, it could be assumed that they are vulnerable to economic- and market conditions. Figure three contained an overview of the average: Revenue level, SG&A costs, SG&A-to-Revenue ratio and the Stickiness Level by fiscal year. There are several potential contributors for changes overtime. As could be observed, the average Revenue- and SG&A cost level, did most significantly increased year-to-year. Overall, the Revenue level of the S&P1500 firms, did increase with 202,54%, while the SG&A cost level raised with 159,76% from 1992 to 2014.

This initial period were visualized to check whether some statements in other articles are also the case in my dataset. However, for the other tables, I have chosen to use the same sample period for both panels. Bugeja et al. (2015) mentioned that the cost stickiness level increased after the introduction of IFRS in 2001. I could only check this with the initial data of Panel A, but this finding is not visible in my sample. The cost stickiness level increased in 2001 compared with 2000. But after 2001 the cost stickiness level decreased. Note: an increase of cost stickiness means that the outcome of Stickiness formula becomes lower, or more negative.

In addition, the results of the financial crisis are visible in the revenue levels of 2007 in both samples. In the initial panel A the SG&A costs made up on average for 26,05% of firm’ total revenue. While in panel B the SG&A costs made up on average for 26,26% of firm total revenue. Those ratios are in line with the findings of Anderson et al. (2003) 26,40%. The SG&A-to-Revenue ratio is industry specific (Subramaniam & Weidenmier, 2003). My study supported this statement based on descriptive data.

Table three: Revenue, SG&A cost, Cost Stickiness and SG&A-to-Revenue (* 1M USD) Initial Panel A (1992-2014):

Fiscal Year: Revenue: SG&A Costs Cost Stickiness SG&A-to-Revenue

1992 4.571,45 1.143,10 0,0908 23,28% 1993 4.102,87 845,67 0,0341 23,91% 1994 3.276,00 680,98 0,0320 24,36% 1995 3.266,25 681,73 -0,0043 24,06% 1996 3.454,76 694,26 0,0101 24,75% 1997 3.769,78 762,25 0,0109 24,66% 1998 3.763,89 809,12 -0,0416 25,13% 1999 4.031,35 840,30 -0,0038 24,55% 2000 4.896,13 1.036,26 0,0100 30,45% 2001 4.626,50 961,68 -0,0565 26,43% 2002 5.064,25 975,29 0,0282 28,45% 2003 4.564,58 990,15 0,0621 26,32% 2004 5.563,12 1.157,75 0,0053 26,88% 2005 6.928,89 1.278,69 0,0603 26,48% 2006 6.453,94 1.297,76 0,0043 25,38% 2007 7.864,91 1.328,08 0,0721 26,10% 2008 6.428,83 1.316,98 0,0343 26,59% 2009 7.027,75 1.449,63 0,0239 28,42% 2010 6.923,17 1.507,35 0,0119 27,60% 2011 9.130,96 1.650,16 0,0443 25,51% 2012 8.879,10 1.748,75 -0,0406 26,48% 2013 9.433,37 1.848,00 0,0007 25,62% 2014 9.259,38 1.826,59 -0,0017 24,27% Mean: 5.922,16 1.180,97 0,0143 26,06%

Final Panel B (2006-2014):

Fiscal Year: Revenue: SG&A Costs Cost Stickiness SG&A-to-Revenue

2006 6.608,27 1.351,00 -0,0039 25,40% 2007 7.864,39 1.327,89 0,0721 26,08% 2008 6.420,88 1.315,38 0,0346 26,59% 2009 7.033,31 1.449,71 0,0242 28,03% 2010 6.878,28 1.517,29 0,0099 27,62% 2011 9.178,58 1.653,26 0,0439 25,57% 2012 8.873,72 1.758,12 -0,0465 26,54% 2013 9.430,24 1.846,48 -0,0032 25,64% 2014 9.247,63 1.824,01 0,0051 24,25% Mean: 7.924,62 1.554,16 0,0157 26,26%

Overall, both samples contain years with average cost stickiness, but in the majority of years there is slightly anti-stickiness. However, for both panel A and panel B holds that cost stickiness is industry related. I observe that firms that have been marked as manufacturing firms, have slight cost stickiness (panel B) or less anti-stickiness (panel A) in comparison with

Merchandising- or Service firms. These findings are similar to the research findings of Bugeja et al. (2015). They found that cost stickiness is industry related. Furthermore, Bugeja et al. (2015) also concluded that asset intensity is a main driver of cost stickiness. While, the asset intensity level in my samples is not higher for manufacturing firms than for the other industries. Weiss (2010) found cost stickiness for manufacturing firms, but did not investigated other firm types. Given the cost stickiness distribution for all industries in figure four, there are still multiple individual firm-year observations were cost stickiness could be observed.

The findings of Fama and French 2004; Wei and Zhang 2006 and Dalla Via and Perego (2014), who observed that larger firms have an higher level of cost stickiness are not visible in my samples and even the opposite applies. The next subparagraph contain correlation tests of the two models.

4.1.2 Correlation test

First, I will declare how the variables in the econometrical model have been used. Thereafter, I will present two correlation tests and a multicollinearity test for both panels. The dependent variable cost stickiness has been calculated with the conventional cost stickiness measure of Weiss (2010). The annual cost stickiness level has been calculated on a firm year level. The independent variables, CEO in panel A one and CFO in panel B are identifier variables. Executives from the Board of Directors that do not have the role of CEO or CFO have been excluded. The other independent variables from the model, are: Asset intensity, Return-on-Asset, Market-to-Book value and leverage. These variables are all quantative (straight) variables or financial ratios. Initially firm size have been categorized into three groups 1. Small 2. Middle and 3. Large. The ranking was obtained by using the “decile”-function. A method to categorize categorical variables based on a quantative variable. But the middle- and large category have been merged because of multicollinearity issues. For industry classification I have initially created four dummies, but also because of multicollinearity between services and financial services, these two categories were merged into one categorical variable. Despite the fact that service firms and financial services have been merged, both panels still contain the highest concentration of manufacturing firms, followed by merchandising firms and services.

The aim of this study was to calculate the net managerial effect on cost stickiness. To control for time differences, I have included an indicator variables for each fiscal year in the sample. The chosen variables in the two models are: firm-, industry- or time related. While, Bugeja et al. (2015) found no influence of macroeconomic growth on cost stickiness, I would like to control for time differences. Because, it is likely that macroeconomic conditions will have an impact on corporate decision making. Given, the fact that the CEOs and CFOs have worked in different periods in the sample, the net effect should not be biased with differences based on time, whereby macroeconomic conditions change overtime. To conclude whether there is a significant effect, I have chosen the significance level for my study as: α <0,10.

Table four contains the Pearson Correlation Matrix as well as the Spearman Correlation Matrix, to observe the correlation between the dependent- and independent variables. To examine multicollinearity issues, I also included the VIF- and Tolerance matrix in table five.

Table four: Correlation Matrices Spearman correlation matrix Panel A:

Variables: Cost Stickiness

Cost Stickiness 1,0000 CEO number 0,0332 Asset Intensity -0,0058 Firm Size 0,0085 ROA -0,0322 MTB -0,0106 Leverage -0,0001 Fiscal Year -0,0155 Industry Category 0,0510 Pearson correlation matrix Panel A:

Variables: Cost Stickiness

Cost Stickiness 1,0000 CEO number -0,0010 Asset Intensity -0,0194 Firm Size 0,0317 ROA -0,0087 MTB -0,0138 Leverage 0,0060 Fiscal Year 0,0029 Industry Category 0,0315 Spearman Correlation Matrix Panel B:

Variables: Cost Stickiness

Cost Stickiness 1,0000 CFO number 0,0380 Asset Intensity -0,0068 Firm Size 0,0109 ROA -0,0318 MTB -0,0108 Leverage -0,0010 Fiscal Year -0,0162 Industry Category 0,0523

Pearson Correlation Matrix Panel B:

Variables: Cost Stickiness

Cost Stickiness 1,0000 CEO number 0,0300 Asset Intensity 0,0072 Firm Size 0,0160 ROA 0,0005 MTB -0,0133 Leverage 0,0059 Fiscal Year -0,0290 Industry Category 0,0389

4.1.3 Multicollinearity

As could be observed from figure six, for both panels applies that there are no multicollinearity issues, based on the VIF-test and the Tolerance level. The Tolerance levels (1/VIF) are largely above the threshold of 0,2. While, the outcome of the VIF-test is below 5, this also indicates that the variables in my model do not have high multicollinearity.

Table five: VIF and Tolerance Panel A:

Variables: VIF: Tolerance (1/VIF):

MTB 1,19 0,8439 Leverage 1,16 0,8606 Firm Size 1,09 0,9215 CEO number 1,08 0,9292 ROA 1,07 0,9320 Fiscal Year 1,03 0,9702 Asset Intensity 1,03 0,0971 Industry Group 1,00 0,9975 Mean VIF: 1,08 Panel B:

Variables: VIF: Tolerance (1/VIF):

MTB 1,18 0,8454 Leverage 1,16 0,8596 Firm Size 1,09 0,9162 CEO number 1,08 0,9280 ROA 1,07 0,9339 Fiscal Year 1,05 0,9479 Asset Intensity 1,03 0,9706 Industry Group 1,00 0,9973 Mean VIF: 1,08

4.2 Multivariate analysis

4.2.1 CEO fixed effect on cost stickiness

The first hypothesis is: “a CEO has an effect on cost stickiness”. According to the regression analysis in figure seven, a CEO had a significant effect on the cost stickiness level. Because, the p-value (0,000) is lower than the chosen significance level (α) 0,10, therefore hypothesis one is supported.

Table six: “Regression model to test CEO fixed effect” Panel A:

Linear regression, absorbing indicators Number of observations 3.921

F ( 15, 2252) 1,59 Prob > F 0,0677 R-squared 0,5240 Adjusted R-squared 0,1714 Root MSE 0,5312

Cost Stickiness: Coefficient: Standard Error: T-value: P>|t|: [95% Confidence Interval]

Fiscal Year 2007 0,5061 0,1559 3,25 0,001 0,2004 0,8119 2008 0,3400 0,1612 2,11 0,035 0,0238 0,6561 2009 0,3681 0,1456 2,53 0,012 0,0826 0,6535 2010 0,3463 0,1456 2,38 0,017 0,0610 0,6317 2011 0,3388 0,1455 2,33 0,020 0,0535 0,6241 2012 0,2613 0,1459 1,79 0,073 -0,0248 0,5474 2013 0,3107 0,1462 2,12 0,034 0,0239 0,5975 2014 0,3100 0,1460 2,12 0,034 0,0237 0,5964 Asset Intensity -0,0126 0,0087 -1,45 0,147 -0,0296 0,0044 Return-on-Assets 0,0239 0,3198 0,07 0,940 -0,6032 0,6510 Market-to-Book value -0,0006 0,0026 -0,24 0,809 -0,0057 0,0045

Industry (Manufacturing firm) -

Merchandising firm 0,3610 0,6544 0,55 0,581 -0,9223 1,6443

Service firm -0,6994 0,7544 -0,93 0,354 -2,1787 0,7800

Firm Size (Small firm) -

Large firm 0,0353 0,0699 0,51 0,613 -0,1017 0,1724

Leverage -0,0008 0,0036 -0,22 0,828 -0,0078 0,0062

_Constant -0,2049 0,2518 -0,81 0,416 -0,6987 0,2889

The T-value of CEO number indicates that the size of the calculated differences between CEOs in this sample is: 1,478. This means that 1,478 standard deviations margins were explained by CEO influences. The cost stickiness observations were averaged and assigned to the CEOs. Therefore, the net CEO fixed effect could be observed, unregarded whether a CEO has only worked for one firm or for more firms.

I observe that the regression model has an R-square of 0,5240. This indicates that statistically 52,40% of the variance of cost stickiness could be declared by the chosen variables. I did also checked whether ROE had an significant effect and whether it increased the declared variation. But it seemed that it only slightly increase the R-square, but in fact slightly decrease the adjusted R-square. The adjusted R-square in the regression models of Weiss (2010) ranged from 7,50% to 17,60%. Weiss (2010) developed the cost stickiness measure and checked the

robustness of it with other indicated cost stickiness measures. I used the model of Weiss as benchmark. The adjusted R-square of my first panel is 17,14%. Therefore, it is almost

comparable with the highest adjusted R-square of Weiss (2010). The quantative variables in the model are: ROA, MTB and Leverage, these variables did all have a negative coefficient. This indicates that a one unit increase of those independent variables will result in a decrease of cost stickiness, the dependent variable. But, a decrease indicates to more cost stickiness. However the P-value of these variables are insignificant, because the P-value is above α of 0,10. Therefore, there is no significant link between those variables and cost stickiness.

Moreover, the probability of the F-score of the model is below α, this indicates that the model is sufficiently applied. Therefore, we could conclude that a CEO has an influence on cost stickiness. It could also be observed, that the fiscal year had a significant influence on cost stickiness. The effect of the different years incorporated all the events that have been occurred during that year. So, the economic situation in that year, likely will have a big influence on the annual cost stickiness level.

4.2.2 CFO fixed effect on cost stickiness

The second hypothesis of this research is: the CFO has an effect on cost stickiness. Table seven contained the corresponding regression analysis. The aim is to test whether there is a significant CFO fixed effect on cost stickiness.

Table seven: Regression model to test CFO fixed effect Panel B:

Linear regression, absorbing indicators Number of observations 3.915

F ( 15, 2252) 2,19 Prob > F 0,0053 R-squared 0,5415 Adjusted R-squared 0,1719 Root MSE 0,5566

Cost Stickiness: Coefficient: Standard Error: T-value: P>|t|: [95% Confidence Interval]

Fiscal Year: 2007 0,6567 0,1800 3,65 0,000 0,3037 1,0097 2008 0,5533 0,1917 2,89 0,004 0,1775 0,9292 2009 0,6008 0,1730 3,47 0,001 0,2616 0,9400 2010 0,6076 0,1725 3,52 0,000 0,2693 0,9458 2011 0,6151 0,1728 3,56 0,000 0,2762 0,9540 2012 0,5049 0,1731 2,92 0,004 0,1654 0,8443 2013 0,5701 0,1733 3,29 0,001 0,2302 0,9100 2014 0,5736 0,1736 3,30 0,001 0,2332 0,9140 Asset Intensity -0,0147 0,0078 1,88 0,060 -0,0300 0,0006 Return-on-Assets -0,0282 0,3318 -0,09 0,932 -0,6790 0,6225 Market-to-Book value -0,0014 0,0028 -0,50 0,617 -0,0069 0,0041 Industry: (Manufacturing) - Merchandising firm -0,3287 0,2403 -1,37 0,172 -0,8000 0,1426 Service firm -0,3342 0,2027 -1,65 0,099 -0,7317 0,0634

Firm Size: (Small) -

Large firm -0,0154 0,0661 -0,23 0,816 -0,1451 0,1142

Leverage 0,0007 0,0036 0,19 0,846 -0,0064 0,0079

_Constant -0,3511 0,1890 -1,86 0,063 -0,7218 0,0195