A Study on the association of the use of Forward-Looking Sentences and VEGA
with managerial geographical empire building
Name: Sella Baykara Student number: 10438793
Thesis supervisor: Dr. Reka Felleg Date: 20 June 2016
Word count: 12,291
MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam
Statement of Originality
This document is written by Sella Baykara 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
In this study I examine whether US-Firms’ use of forward-looking sentences and US-firms with a higher VEGA are more likely to engage in geographical empire building in the post SFAS 131 period. I define geographical empire building by an increase in foreign sales growth and a simultaneous decrease in foreign profit margin. I find, although significant statistical results for a negative association between US-firms’ use of forward-looking sentences and foreign profit margin and a negative association between US-firms with higher VEGA and foreign profit margin, no sufficient evidence in support for my hypotheses. My results indicate, given the sample used, that I do not find evidence in support of US-firms engaging in geographical empire building by the use of forward-looking sentences or US-firms with a higher VEGA engaging in geographical empire building.
Table of Content
1. Introduction 5
2. Prior research and hypothesis development 7
2.1 Agency Theory 7
2.2 Information Asymmetry 9
2.3 Disclosures 11
2.3.1 Financial Disclosures 11
2.3.2 Forward-Looking Disclosures 11
2.3.3 Considerations with regard to Forward-Looking Disclosures 12
2.3.4 Foreign and Domestic Operations Disclosures 13
2.4 Empire Building and Incentives 14
2.5 SFAS 131 12
2.5.1 Empire Building Expectations 13
2.6 VEGA 19
2.6.1 VEGA Expectations 21
3. Research Method 23
3.1 Sample 23
3.2 Model & Variables 25
3.2.1 Hypothesis 1 25
3.2.2 Hypothesis 2 26
3.2.3 Hypothesis 3 27
4. Results 28
4.1 Descriptive Statistics 28
4.2 Results of Hypothesis Test 31
4.2.1 Geographical Empire Building and Forward-Looking Sentences 31
4.2.2 Geographical Empire Building and VEGA 33
4.2.3 Effect of VEGA on FLS and Geographical Empire Building 34
5. Robustness Test 35
6. Conclusion 37
1. Introduction
One of the oldest and most common modes of social interaction is agency theory. As Ross (1973) mentions, an agency relationship will arise when there is a contract between two or more parties whereas one is designated as the agent who acts for, as representative for or on behalf of the other who is designated as the principal. To further enlighten this relationship, Hope and Thomas (2008) give a more understanding definition. According to Thomas and Hope (2008), the agency theory is a description of a natural conflict between shareholders (who can be seen as the principal) and managers (who can be seen as the agent). Jensen and Meckling (1976) discuss the arising conflict between shareholders and the agent where individuals choose actions to maximize their own utility. According to Jensen and Meckling (1976), this suggests that managers do not always act in the best interest of the shareholders.
Furthermore, Ross (1973) emphasizes that many problems of moral hazard are concerned with problems that are raised by agency. Since the manager can observe his own actions, whereas the shareholder cannot, this leads to moral hazard problems. One way to solve this conflict between the shareholder and the agent is through monitoring the agent (Bushman and Smith, 2001). They mention one obvious monitoring system, which are financial disclosures by the firm. As disclosures can serve as a monitoring mechanism, managers are incentivized and disciplined to act in the best interest of the shareholders. However, as Jensen’s (1986) study predicts, managers can make self-maximizing decisions when the quality of the disclosures is reduced.
An inclusion of these suboptimal decisions is a phenomenon known as “empire building”, which according to Jensen (1986) decreases operating performance and reduces firm value. Hope and Thomas (2008) argue that some managers engage in empire building. An explanation for this occurrence is that when the quality of monitoring mechanisms, such as financial disclosures, is reduced, investors are less capable of linking managerial decisions to firm performance.
Moreover, Coles, Daniel and Naveen (2006) provide with their paper empirical evidence of a strong causal relation between the structure of managerial compensation and value-critical managerial decisions derived from both investment policy and debt policy. In their study they compare the sensitivity of CEO wealth to stock return volatility (Vega). Coles, Daniel and Naveen (2006) find that CEO’s compensation with higher Vega provides managers with incentives to both invest in riskier assets and implement more aggressive debt policy.
As Muslu, Radhakrishnan, Subramanyam and Lim (2015) state, it is required from public companies that they file MD&A sections as an integral part of their 10-K filings, although the majority of the content is voluntary. Furthermore, the SEC emphasized companies to disclose forward-looking disclosures because of investor’s greater need. Consequently, companies are guided by the SEC to present known events or plans that might materially affect the company’s liquidity, capital resources or future operations. One way of doing so is by disclosing forward-looking sentences. However, by disclosing forward-looking sentences, firms and their managers might have incentives other than to inform the market since manager’s interests are not aligned with that of principals. Using forward-looking sentences might therefore be used as a tool for managers to easier build empires or influence their compensation. Consequently, I argue that the more a firm discloses forward-looking sentences, the more likely they are to build empires. Furthermore, since managers might be able to influence their compensation, this may have an effect on the amount of forward-looking sentences disclosed. This leads me to the research question whether firms that choose to disclose forward-looking sentences and firms with a higher VEGA are more likely to engage in geographical empire building.
In order to test my expectations I participate in the research project of Dr. Wim Janssen together with several other students at the University of Amsterdam. I follow the same models Thomas and Hope (2008) used for measuring geographical empire building by hypothesizing a positive relationship between foreign sales growth and forward-looking sentences (and VEGA) and at the same time a negative relationship between foreign profit margin and forward-looking sentences (and VEGA).
My statistical results first show in all my regressions that firm’s foreign sales growth increases due to the use of forward-looking sentences, and that foreign profit margin decreases, however the findings are not significant. Second, I find that foreign profit margin decreases for firms with a higher VEGA with significant results in all my regressions. However, I do not find support for my expectations for the relationship between VEGA and foreign sales growth.
Although empire building is of potential influence for investors, limited research exists on the role of forward-looking sentences disclosed. Answering the research question will increase our understanding of manager’s incentives of forward-looking disclosures and how this is influenced by CEO compensation contracts. Furthermore, this Master thesis contributes to prior literature since it provides insight in companies engaging in empire
building by researching a relationship both influenced by forward-looking sentences and VEGA in one study.
The remainder of this thesis is organized as follows. The next section will review prior literature whereas the phenomenon known, as empire building will be further discussed and prior literature about CEO compensation contracts enlightened. Also my hypothesis will be developed in this chapter. Section 3 describes my research method and section 4 presents my results. In section 5 some robustness tests are run and section 6 offers a conclusion.
2. Prior research and hypotheses
In this section of my thesis I will elaborate upon the theoretical construct and the main literature. First, in order to give the reader a better understanding of the fundamental theory used in this thesis, the agency theory will be explained in more detail. Second, the phenomenon known as Empire building will be explained and a connection to the theory will be made. Third, management incentives will be explained so as to gain a better understanding of why some managers act differently with relation to investments. Following the incentives, a paragraph is devoted to firm disclosures so as to understand how these may impact the managers’ actions and impact the ability of shareholders in monitoring managers. The information provided in these sections will help me develop my first hypothesis. In the last paragraph of this chapter, prior literature regarding Vega will be discussed in order to develop my second hypothesis regarding the managerial incentives and compensation schemes.
2.1. Agency Theory
As early as in the 1960s and 1970s, risk-sharing among groups or individuals was explored by economists (Eisenhardt, 1989). The literature which explored risk-sharing described the occurrence of this problem by parties who cooperate but have different attitudes toward risk. The inclusion of the agency problem broadened the risk-sharing literature, where the origin of the agency problem takes place in the different goals and division of labour among cooperating parties (Jensen & Meckling, 1976; Ross, 1973). As Ross (1973) explains in his article, agency is seen as a problem which results from compensation contracting, or which can also be stated as an incentives problem. Here, it is of the principals concern to select an appropriate incentive mechanism in order to get the desired behaviour by the agent (Ross, 1973).
Jensen & Meckling (1976) later proposed an agency theory of the firm, where they integrate the theory of property rights, the theory of finance and the theory of agency into one theory namely, the ownership structure of the firm. Jensen & Meckling (1976) describe that a firm can be seen as a set of contracts among factors of production. Furthermore, the authors explain that each factor is motivated by self-interest. As this suggestion by Jensen & Meckling (1976) is shared by Fama (1980) where each factor is motivated by self-interest, Fama (1980) also states that the enabling of the entire firm performance is the result of competition of other firms where firms are disciplined by. Since managers can be seen as a factor of production, these markets where firms face competition, provide opportunities for managers and therefore, as Fama (1980) states, incentives can also be regarded as tools to retain managers.
In their article, Jensen & Meckling (1976, p.310) divide the traditional functions of the entrepreneur into management and risk bearing. The definition the authors give for both terms are as follows:
The agent is the “one party who is common to all the contracts of joint inputs” and the one “who has the right to renegotiate an input’s contract independently of contracts with other input owners”
The principal is addressed as the one ‘who holds the residual claim” and ‘who has the right to sell his central contractual residual status’ Jensen & Meckling (1976, p.320).
More specifically, agency theory explains the universal agency relationship in which one party delegates work to another who performs that work (Eisenhardt, 1989). In this case, the party who delegates work is called the principal, whereas the party who performs the work is called the agent. According to Jensen & Meckling (1976), agency theory attempts to describe this relationship, where the separation of the ownership (the ones who bear the risk) and the control over the firm (management) by using the metaphor of a contract.
The optimal contract is the one where the person who is controlling the firm (the agent) will perform tasks on behalf of the ones who bear the risk (the owner, or principal). Next to the optimal contract, Jensen & Meckling (1976) argue that both the management and principles are utility maximizers, which leads to the assumption that the manager will not always act in the best interest of the principal.
The broadly known concern of the agency theory is regarded with solving two problems that can occur in agency relationships. The first problem is concerned with the agency problem, which arises when the desires or goals of the agent and principal are not the same and the principal finds it hard or expensive in verifying what the agent is actually doing.
This leads to the problem where the principal cannot find out as to whether the agent has behaved appropriately. The second problem is the problem of risk-sharing. This problem arises when there is a different risk appetite between the principal and the agent, which can lead to different, preferred actions by the two parties (Eisenhardt, 1989).
2.2. Information Asymmetry
According to agency theory everyone maximizes his or her own interests. Managers are supposed to act in the best interests of the principals, however they are mainly driven my self-interest (Jensen & Meckling, 1976). In this situation managers have more and better information regarding the value of the firm and their own behaviour compared to the principals of the firm, which causes a problematic situation. In this situation where the agents have advantage information wise compared to the principals is called information asymmetry (Myers & Majluf, 1984).
The information asymmetry which occurs between the agent and the principal causes two kind of problems for the principal and differs in whether the principal has already made an investment decision or not (Wallace, 1980). The first kind of problem for the principal is called an adverse selection problem. This problem occurs when a principal finds himself in a situation where he or she has to decide whether or not to invest in the firm, no matter which type of external financing is used (debt or equity) (Wallace, 1980). The principals base their decision to invest in the firm on their assessment of the expected ability of the firm to generate future cash flows and the ability to repay its liabilities. But since managers have more and better information compared to the principals, principals might find themselves in positions with wrong assessments, which leads to poor investment decisions. Put bluntly, agents take advantage of the better and more information they have at the expense of the principals (Scott, 2011).
Principals have incentives to reduce the information asymmetry between them and the agents, which is created by the adverse selection problem. On the other hand, agents want to diminish the information asymmetry as well (Wallace, 1980). Principals consider the return and the risk of the investment in their decision to invest. When the information asymmetry is greater, this will lead to greater difficulties for the principals to make a proper assessment of the firm’s expected ability to generate future cash flows and its ability to repay its liabilities. Yet, agents have incentives to reduce this information asymmetry when they consider their firms as attractive investment opportunities so they can communicate the difference between
their ‘good type’ firms from ‘bad types’ (Akerlof, 1970). When information asymmetry is not reduced as an effect from communication by the agents, consequences are that ‘good type’ firms will be undervalued (lower share price) because they are now pooled with overvalued ‘bad type’ firms. In addition, in the case of a reduced information asymmetry, share prices increases due to the lower degree of investment risk and leads to principals requiring a lower return on their investment in terms of interest and dividend (Wallace, 1980). Thus, the requirement for a lower return on the investment and a higher share price are favourable for agents and for the principal. Concluding, both agents and principals have incentives to reduce the information asymmetry with relation to the adverse selection problems because agents can communicate their ‘good’ type firms to the market whereas principals can make better assessments of the firm leading to a lower required return on their investment due to reduced risk (Akerlof, 1970; Wallace, 1980).
The second kind of problems are moral hazard problems. These kinds of problems occur in the situation where a principal has already made an investment in the firm (Wallace, 1980). On the one hand, principals want agents to work in their best interests, but on the other hand, principals do not have the ability to continuously monitor the behaviour of the agents. As a consequence, this leaves agents with more and better information on their own behaviour and performance relative to the information principals get when monitoring the agents. Because of this advantage, as Scott (2011) mentions, agents can exploit this advantage at the expense of their principals. An example given by Scott (2011) is an agent reducing his effort or by engaging in behaviour, which is not aligned with the best interests of the principal. As with the adverse selection problems, principals have also incentives to diminish the information asymmetry in relation to moral hazard problems. In a moral hazard situation, the principal has difficulties to monitor whether the agent is behaving in the principals’ best interest, which is caused by the information asymmetry. Therefore, the principal assumes that the agent’s behaviour is not in his or her best interests if monitoring the agent’s behaviour cannot be done adequately (Wallace, 1980). As a consequence, principals propose lower agent compensation when anticipating the inefficient behaviour of the agents. The lowered agent compensation is an incentive for the agent to reduce the information asymmetry (Wallace, 1980). Hence, both the agent and the principal have incentives to reduce the information asymmetry with regard to moral hazard problems.
In summary, separation of management and ownership of a firm creates an information asymmetry between the agents and the principals. Within this information asymmetry agents have an information advantage compared to their principals, which causes
both adverse selection and moral hazard problems. However, regardless this advantage, both the principals and the agents have incentives to reduce the information asymmetry.
2.3 Disclosures
2.3.1. Financial Disclosures (backward looking)
More recent research is concerned with investigating the conditions under which moral hazard problems are more severe. An example of such a study is that of Jiraporn et al. (2006). In their article, they suggest that managers are more likely to make decisions in order to fulfil their own interests when these managers are less accountable to the firm’s investors. This consequently leads to less profitable firm performances and in turn to loss of shareholder value (Jiraporn et al., 2006). According to Thomas & Hope (2008), a way to make managers who act in this way more accountable is by disclosing financial disclosures in order to better monitor the actions of the managers. In order to reduce moral hazard problems between investors and managers, investors seek high-quality disclosures (Thomas & Hope, 2008). Hence, as investors obtain better disclosures, they improve their ability to monitor management because they are able to better relate managerial decisions to firm performance and reduce moral hazard problems (Lombardo & Pagano, 2002).
2.3.2. Forward looking disclosures
An important part of the annual report is the Management Discussion and Analysis (from here on forward (MD&A) section. In this section managerial commentary about a firm’s current state and future prospects are included (Muslu, Radhakrishnan, Subramanyam & Lim, 2015). As an integral part of their 10-K filings, companies are required to file MD&A sections but the content of these MD&A sections remain largely voluntary (Muslu et al., 2015). Put differently, the disclosure setting of the MD&A section is quasimandatory or quasivoluntary (Muslu et al., 2015).
The SEC has periodically provided guidance with relation to the content of MD&A disclosures with the intention that MD&A disclosures provide investors with the opportunity to see the company through the eyes of the management (SEC, 1987; Garmong, 2007). The SEC has guided to present any known factors, which are likely to materially affect the
company’s liquidity, future operations or capital resources (SEC, 2003). However, the SEC is of opinion that MD&A disclosures are deficient in general (Garmong, 2007).
In early studies, MD&A disclosures are examined by using small samples (Muslu et al., 2015). For example, Pava and Epstein (1993) find evidence that MD&A sections mostly describe past performance after they selected 25 random companies. In more recent studies, researchers make use of computer intensive techniques in order to examine MD&A disclosures (Muslu et al., 2015). One of such a study is that of Hussainey and Walker (2009) where they show that investors are able to form better expectations about the changes in cumulative earnings during the next three years. Likewise, Li (2010) assesses the tone of forward-looking MD&A disclosures between 1994 and 2007 using a Bayesian machine-learning algorithm. She finds that optimistic (pessimistic) tone is positively (negatively) related to future earnings. Therefore, the way the market reacts on forward-looking disclosures can be influenced by managers as they can change the tone into more optimistic or pessimistic sentences.
2.3.3. Considerations with regard to Forward-Looking Disclosures
According to prior evidence regarding voluntary disclosures, the quality of forward-looking disclosures seems to be associated with useful information with regard to stock prices (Muslu et al., 2015). Yet, it is also possible that MD&A disclosures are not useful for several reasons. Muslu et al. (2015) use the timeliness, lack of evidence with regard to usefulness of MD&A disclosures as reasons why MD&A disclosures might not be informative to the capital markets.
With regard to voluntary disclosures, theories seem to be ambivalent about whether managers disclose useful information. According to the ‘informativeness perspective’, managers will disclose value relevant information (Muslu et al., 2015). In order to mitigate adverse selection problems, early signaling models argue that all value-relevant information is disclosed (Muslu et al., 2015). By using subsequent models, which impose costs and derive selective disclosure strategies, managers disclose good news in general (Muslu et al., 2015). However, in accordance with the litigation risk hypothesis, firms voluntarily disclose bad news. According to an independent strand of literature, managers meet investors’ information demands or align investor’s expectations with their own by disclosing either good or bad news.
On the other hand, the opportunism perspective argues that managers’ motives largely shape the company disclosures. In order to ‘hype’ the stock, especially prior to raising capital or extracting rents, managers disclose good news to the market (Lang and Lundholm, 2000). Furthermore, when the news is good, managers seem to leak this on a timely fashion. However, when the news is bad, managers seem to withhold the news based on their career concerns (Kothari, Shu and Wysocki, 2009).
Overall, forward-looking disclosures will be informative regardless if these are selective according to the informativeness perspective. In contrast, the opportunism perspective is of opinion that the market can be mislead in the short-term by voluntary disclosures. As a consequence, the empirical question whether forward-looking MD&A disclosures communicate useful information to the capital markets still remains.
2.3.4. Foreign Operations and Domestic Operations Disclosures
According to Thomas and Hope (2008), segment disclosures play an important role to investors. They use as example a citation from paragraph 44 of SFAS 131:
Segment data is vital, essential, fundamental, indispensable and integral to the investment analysis process. Analysts need to know and understand how the various components of a multifaceted enterprise behave economically. Even in the absence of weakness, different segments will generate dissimilar streams of cash flows to which are attached disparate risks and which bring about unique values. (Thomas and Hope, 2008, p.597)
Furthermore, there are several studies which provide evidence of higher information asymmetry when companies engage in foreign operations compared to domestic operations (Thomas, 1999; Denis, Denis and Yost (2002); Callen, Hope and Segal (2005). According to these studies and Hope and Thomas’ (2008), managers, by moving into unprofitable or less profitable foreign markets, can grow the firm via sales or assets.
Under SFAS 131, issued in June 1997, firms are still required to disclose geographic sales and long-lived assets. However, it is no longer required to disclose geographic earnings (Thomas and Hope, 2008). Firm’s disclosure with regard to geographic earnings is voluntary when operating segments are defined on any basis that is not geographic area. Consequently, most firms choose not to disclose (Thomas and Hope, 2008). SEC Regulation §210.4-08
states that all firms are still required to disclose ‘total’ foreign income (Thomas and Hope, 2008). This means that the profits from disaggregated foreign segments can be aggregated into one total foreign income number. Consequently, the authors discuss that in the case of declining total foreign profits, investors are no longer capable of determining whether these lower profits occur because of the result of existing foreign operations or because of decisions of managers that are related to expanding foreign operations. Hence, firms who do not disclose make it more difficult for investors to hold managers accountable for geographic-specific decisions (Thomas and Hope, 2008).
2.4. Empire Building and Incentives
Prior literature argues that some managers engage in empire building (Hope and Thomas, 2008). They state that it is long recognized by researchers that the allocations of resources by managers may not be efficient and can therefore destroy shareholder value. Schumpeter (1934) postulated that managers are empire builders. This assumption is shared by other researchers such as Aggarwal and Samwick (2006), Jensen (1986) and Kanniainen (2000). Thomas & Hope (2008) argue that since the recognition of the above problem, it has become one of the pillars of the literature on corporate governance where it is assumed that managers turn into empire builders when they are not reined in by some tight form of governance.
According to Schumpeter (1934) and Aggarwal and Samwick (2006), managers have incentives to continue investing in projects, even after all net present value investments have been taken. As Jensen (1986) and Thomas and Hope (2008) state, managers have incentives to engage in empire building, which means growing the firm beyond its optimal size.
According to Thomas and Hope (2008) there are two forms of empire building. These are excessive growth and excessive investment. Furthermore, managers engage in empire building because increasing firm size or the diversification of operations can serve their private interests in various ways (Thomas and Hope, 2008). General variables as to why managers are motivated to build empires are the managers’ hunger for status, power, compensation and prestige (Jensen, 1986; Williamson, 1974). As Thomas & Hope (2008) and Dominquez-Martinez, Swank and Visser (2006) therefore conclude, empire building exists because of the misalignment in the preferences between the board of directors (representing investors) and the executives (management). This in conjunction with the lack of observability is a typical moral hazard problem.
With regard to the lack of observability, Kanodia and Lee (2004) develop a theoretical model using enhanced firm disclosures to mitigate the excessive investment problem, which is a form of empire building. Furthermore, Bens and Monahan (2004) find a significant positive relation between the level of disclosure and a measure of the excess value of diversification. The authors use the monitoring effect of disclosures as reason for this finding. Accordingly, Bens and Monahan (2004) conclude that greater firm disclosures lead to a decreased tendency for investing in assets that destroy shareholder value.
As mentioned in the previous paragraph, the reason why managers may engage in empire building is to serve their private interests in various ways, such as executive’s hunger for status, power, prestige and compensation (Marris, 1964; Williamson, 1974; Jensen, 1986). In the widely cited article of Jensen (1986), he presents a ‘free cash flow’ theory. Here, firms with low investment opportunities but high free cash flows, have incentives to grow beyond their optimal size.
Managers gain by increasing the resources under their control and their prestige when engaging in empire building (Stulz, 1990). Furthermore, it is also possible for managers to increase their compensation when engaging in empire building since compensation is often tied to firm size, sales growth and diversification (Murphy, 1985; Jensen and Murphy, 1990). Also, growing the firm beyond its optimal size decreases the unemployment risk of managers, creates additional middle manager promotions and makes the manager more indispensable to the firm which are incentives for managers for engaging in empire building according to the theory of ‘managerialism’ presented by Amihud and Lev (1981).
Narayanan (1985) attempts to investigate managerial incentives for making decisions that are at the expense of the shareholders’ interests. The reason for this investigation as Narayanan (1985) states is because of the expressed concern about American managers making decisions, which yield short-term results at the expense of the long-term interests of the shareholders. According to Narayanan (1985) this behaviour exists because of the influence of institutional investors. In order to retain and attract clients, these institutional investors are put under pressure to show short-term profits (Narayanan, 1985). However, this pressure on institutional investors puts pressure on corporate managers to show short-term profits and influence the share price of the firm. Narayanan (1985) argues that wage schemes also influence the short-term bias because they base executive bonuses on annual profits.
As Narayanan (1985) mentions these explanations for the short-term bias of managers, they can fall into two categories namely; first, the short-term view of stockholders who want quick profits is reflected in the decisions of managers and second, managers tend to be quick
profit – oriented because of profit sharing (this means that they make a part or their whole pay contingent on performance) (Narayanan, 1985). The findings of the study show that managers might make decisions at the expense of the long-term interests of the firm, which yield only short-term profits. This is only the case if the manager possesses private information that is not available to the investors. According to Narayanan (1985) the incentive behind these decisions arise because the manager hopes to increase his reputation faster which will lead to higher wages.
Furthermore, Narayanan (1985) finds that a manager is less likely to opt for short-term profits when he/she is more experienced. The duration of the manager’s contract is another finding, where the longer the duration of the contract lowers the probability that the manager will choose quicker-return projects that are suboptimal. The interpretation behind this finding is that the manager is less incentivized to sacrifice the long-term benefits from future cash flows for short-term benefits in the case of longer contract duration. Narayanan (1985) concludes with discussing how the frequent turnover of managers’ tasks can also lead to short-term bias. Since managers’ reputation would in this case depend only on his/her performance in the new assignment, the manager would neglect the fate of the old assignment.
In the study of Thomas and Hope (2008), the authors expect managers to engage in ‘geographical’ empire building (expansion of foreign operations, reduced performance of foreign operations, and an overall reduction in firm value) in the case if nondisclosure of geographic earnings reduces the ability of shareholders in monitoring the managers. Thomas and Hope (2008) find that firms who do not disclose geographic earnings are associated with a significant increase in foreign sales growth and a significant decrease in foreign profit margin. Because nondisclosure leads to constraining shareholders in properly monitoring managers’ actions, managers are more willing to build geographical empires despite the fact that it leads to a lower firm performance (Thomas and Hope, 2008). Controlling for domestic operations, Thomas and Hope (2008) do not find a significant increase in domestic sales growth or significant drop in domestic profit margin. Thus concluding that firms who choose not to disclose are more willing to engage in geographical empire building (Thomas and Hope, 2008).
The findings from Thomas and Hope (2008) study suggest that principal’s ability to monitor manager’s behaviour decreases when managers decide to disclose less disclosures about foreign earnings. Because these disclosures about foreign earnings are backward looking, managers have the ability to conceal the bad investment decisions they have taken
when the numbers show that the investments in foreign operations were not as profitable as expected. Moreover, managers have also the ability to disclose forward-looking disclosures about the firm’s expectations in the short-term. Since these disclosures are overall voluntary, managers may use these disclosures to influence future investment choices of prospective principals.
2.5. SFAS 131
There is evidence from prior research that investors, managers and analysts experience an increase in the complexity of information processing when firms expand into international markets (Hope, Kang, Thomas and Vasvari, 2009). The complexity increases due to differences in cultures, growth opportunities, competition, labor relations, business practices, governmental regulations and market conditions across countries (Hope et al., 2009). Furthermore, the content and structure of the location portfolio of firms is becoming more critical to their global competitive positions due to an increased geographical dispersion of firm assets. Hope et al. (2009) argue that therefore it is becoming more important to gather information that is related to the foreign operations of the firm. Hence, this means that for managers and investors it is becoming more critical in understanding better the impact of foreign operations on the overall performance of the firm. Given the differences in accounting practices worldwide, it is complicated to acquire high quality financial information about foreign operations of US- based multinational enterprises (Hope et al., 2009). For these multinationals, geographical segment disclosures are an important source of information that is provided in the firm’s annual reports. Given the importance of segment disclosures provided by firms, Hope et al. (2009) investigate the potential impact of SFAS 131 Segment Disclosure Requirement on US-based multinationals enterprises from an investor’s perspective. As discussed earlier by Thomas and Hope (2008), segment disclosures play an important role to investors. With SFAS 131 issued in 1997, there are two changes with regard to disclosure of geographic information. First, before SFAS 131, most firms could disaggregate their foreign operations into broad regions such as ‘West Europe’ or continents (e.g., Asia). However, after SFAS 131, firms are allowed to aggregate immaterial countries into a single “Other Foreign” segment (Hope et al., 2009). Second, prior to SFAS 131, firms were required to disclose information about sales, assets and earnings by geographic area. However, under SFAS 131 the requirement for disclosing earnings by geographic area is no longer mandatory (Hope et al., 2009). As Hope et al., (2009) argue, there are differences in
risk, profitability and growth potential with regard to geographic operations. This may lead to hinder the investor’s ability in understanding foreign operations when firms do not disclose geographic earnings (Hope et al., 2009).
In summary, from the studies mentioned above it is clear that managers have incentives to engage in empire building. Serving their own private interests at the cost of the shareholders’ and increasing their compensation are incentives for managers to grow the firm beyond its optimal size. Furthermore, managers are put under pressure for showing short-term results, which are at the cost of the shareholders’ long-term interests. Firms choosing not to disclose create more moral hazard problems leading to managers that are more willing to engage in geographical empire building. However, managers who are more experienced or with a longer duration of their contract seem to be less likely to engage in empire building. Also, with SFAS being effective for fiscal years after 1997, firms are no longer required to disclose geographic earnings. With increasing complexity of information processing and firms not disclosing information about geographical earnings, investors may be hindered in their ability to understand foreign operations, thus leading to adverse selection problems for investors when deciding whether or not to invest in a firm.
2.5.1 Empire building expectations
As mentioned before, information asymmetries between agents and principals can lead to moral hazard and adverse selection problems depending on the situation of the prospective principal (before or after making an investment decision). Although information asymmetries arise because of conflicts of interest between the agent and principal, both have incentives to reduce these problems.
Furthermore, because it is no longer required for firms to disclose geographic earnings under SFAS 131, one can question whether this leads to an increased opportunity for managers to mislead the market in the short-term using forward-looking sentences on non-US operations. Because forward-looking sentences are voluntary I am interested in why managers choose to disclose these sentences with regard to foreign operations and investments. One explanation of these forward-looking sentences might be geographical empire building. Since the issuance of SFAS 131 and the complexity of information processing for investors and analysts with regard to foreign operations, managers have higher opportunity to hide their true incentives in why they disclose forward-looking disclosures. Using SFAS 131 and the complexity of information processing for foreign operations and the opportunism perspective,
this may lead to managers from US-firms engaging in geographical empire building in the post SFAS 131 period when disclosing forward-looking sentences on foreign operations and investments. Therefore for my first hypothesis I hypothesize that US-firms that disclose forward-looking sentences on non-US operations and investments are more likely to engage in geographical empire building than US-firms that do not disclose. My hypothesis is stated below:
H1: US-firms that disclose forward-looking sentences on non-US operations and investments are more likely to engage in geographical empire building than US-firms that do not disclose forward-looking sentences on non-US operations and investments.
2.6 Vega
Coles, Daniel and Naveen (2006) provide empirical evidence for a strong causal relation between the structure of managerial compensation and value-critical managerial decisions derived from both the investment policy and debt policy. The authors consider the sensitivity of CEO wealth to stock return volatility as their primary characteristic of compensation (from here on VEGA). They find that higher prior VEGA implements riskier policy choices, which include relatively more investment in R&D and less in PPE. Furthermore, Coles et al. (2006) provide evidence in support of a higher sensitivity of the managerial compensation scheme giving executives incentives to both invest in riskier assets and implement more aggressive debt policy.
Guay (1999) finds that stock-return volatility is positively related to the convexity of compensation provided to managers. Since convexity in managerial compensation schemes makes risk more valuable to managers, hence a higher Vega should implement riskier investment and financial policies.
The sensitivity of CEO wealth to stock price (form here on DELTA) is seen as aligning the incentives of managers with the interests of shareholders (Coles et al., 2006). This leads to the assumption that a higher delta will lead to managers working harder or more effectively for the shareholders because managers share gains and losses with shareholders.
Next to these findings, Kanniainen (2000) suggests that the information between the agent and principal is asymmetric. As Kanniainen (2000) mentions, the shareholders’ (principals) best response to managers future announcements regarding future investment policy as a given. In his paper, Kanniainen (2000) studies the consequences of separation of
ownership and control on corporate investment thereby providing proof of Jensen’s proposal that independent corporate managers fail to maximize the wealth of corporate owners since they tend to overinvest when they are operating under a linear incentive contract. Furthermore, Kanniainen (2000) finds that the empire building motive is related to the managerial compensation scheme.
Hagendorff & Vallascas (2011) analyse the structure of executive compensation and how this affects the risk choices made by bank CEOs. As they mention, it is widely believed that the use of incentive pay in the banking world, has led to excessive risk-taking behaviour by executives (Hagendorff & Vallascas, 2011). Controlling for CEO pay-performance sensitivity they find that CEOs with higher pay-risk sensitivity engage in risk-inducing mergers. From their findings they offer evidence in support of the causal link between financial stability and the risk-taking incentives embedded in the executive compensation contracts at banks.
Liu and Mauer (2011) elaborate on the work of Jensen & Meckling (1976) and the works of Coles, Daniel and Naveen (2006) where they examine the effect of CEO compensation incentives on corporate cash holdings and the value of cash. The objective of their study is to better understand how compensation incentives designed to enhance the alignment of manager and shareholder interests could influence stockholder-bondholder conflicts. In examining their research question, they make use of three hypotheses where they explain the relation of VEGA to cash holdings. First, the alignment hypothesis makes use of the findings of Jensen and Meckling (1976) where firms align risk-averse managers with equity-holders by using a compensation scheme such as stock options. Also as mentioned by Coles et al. (2006), an increase in a CEOs compensation leads to riskier investments. Using these findings, Liu and Mauer (2011) hypothesise that an increase in VEGA leads to a decrease in cash holdings, since investing in cash lowers the overall risk of the firm.
Second, Liu and Mauer’s(2011) costly external finance hypothesis predicts a positive relation between VEGA and cash holdings. The explanation used is that firms with a high VEGA compensation can face trouble raising external capital when these firms encourage greater risk-taking. Furthermore, in a study of Kim, Mauer and Sherman (1999) and Opler, Pinkowitz, Stulz and Williamson (1999), there is support for Liu and Mauer’s (2011) second hypothesis since in the studies mentioned, the findings suggest that significantly large cash balances are present in companies with substantially growth opportunities. Interpretation of these findings lies in the assumption of high growth firms compensating managers with stock options in order to conserve cash when they face higher costs of external funds.
The third hypothesis, costly contracting hypothesis, predicts a positive relation between VEGA and cash holdings of firms. Liu and Mauer (2011) explain this by anticipating that managers with high VEGA compensation are more likely to implement riskier policy choices, which in turn requires covenants in order to limit the risk-taking behaviour. An example to do that is by maintaining sufficient liquidity (Liu and Mauer, 2011).
Liu and Mauer (2011) find that the VEGA is significantly positively related to a firm’s cash holdings, suggesting that greater CEO risk-taking incentives encourage greater liquidity. Liu and Mauer (2011) provide two explanations. First, as the costly contracting hypothesis asserts, high VEGA incentive firms hold more cash since bondholders expect more risk-taking behaviour by the CEOs. This expectation requires additional cash as a cushion against possible losses. Second, as the costly external finance hypothesis asserts, high VEGA incentive firms build excess cash because they have a greater probability in facing financial constraints (Liu and Mauer, 2011). Liu and Mauer’s (2011) overall findings are consistent with their third hypothesis; costly contracting hypothesis asserting that bondholders require a higher liquidity in higher VEGA firms because they expect more risk-taking behaviour from CEOs.
2.6.1 Vega Expectations
Using the work of Coles, Daniel and Naveen (2006) and the several other others’ findings regarding VEGA, one can rationally expect a strong relation between a higher VEGA and the implementation of a riskier investment policy. Riskier investment policies can be linked to empire building when the true motive of managers is not certain. When firms have a higher VEGA, CEO wealth is sensitive to stock return volatility. Furthermore, since manager’s compensation is based on the stock return volatility ( in case of firms with high VEGA), this might give managers an incentive to increase their investments to signal the markets leading to higher stock prices and more compensation even when these investments seem to be unprofitable in the end. Therefore it is interesting to assess whether firms with a higher VEGA are more likely to engage in geographical empire building. My second hypothesis is stated below:
H2: US-firms with a higher VEGA are more likely to engage in geographical empire building.
Using SFAS 131 and VEGA as an incentive for managers to overinvest, it is interesting to question whether there is an effect of VEGA on the relation between forward-looking sentences and geographical empire building. Because firms with a higher VEGA are related to a riskier investment policy and under SFAS 131 where firms are no longer required to disclose geographic earnings, firms with a higher VEGA might now have the opportunity to easier engage in empire building by making use of forward-looking sentences. Therefore for my third hypothesis the relationship between forward-looking sentences and geographical empire building is more positive for firms with a high VEGA.
H3: The relationship between forward-looking sentences and geographical empire building is more positive for firms with a high VEGA.
In order to provide an overall overview of the hypotheses figure 1 is added below showing the hypothesized relations.
3. Research Method
In this section I discuss the methodology of the research conducted. First I explain my sample and the sample selection procedure. Second, I elaborate on the models and the dependent/independent variables used to test my hypotheses.
3.1 Sample
To test my hypotheses I use forward looking sentences that are derived from US firms during the 1998 – 2000 period with information on foreign operations and investments. These operations include sales, revenues, profits, costs and investments. For my raw data, which includes the timespan from 1996 – 2000, I participate in the project of Dr. Wim Janssen at the University of Amsterdam, together with 8 other students from the university. Each student is provided with a set of circa 750 Forward-Looking Sentences that are derived from US-Firms during the 1996 to 2000 period with information on foreign operations. In order to double-check the sentences, each student has an overlap of half the amount of their sentences with another student (not known which students are a pair). The sentences are derived from the MD&A sections of 10-K filings using automated language-processing technique. Because these sentences are qualitative in nature, and derived using an automated technique, first the data needs to be manually classified as a forward looking sentence on foreign operations and investments under certain criteria.
First, each student, provided with their own sample of sentences is required to classify a sentence as a forward looking sentence on foreign operations and investments when:
1 – The sentence is forward looking (words as: will, plan to, expect)
2 – The sentence is on disaggregated foreign operations outside the US (Europe, Asia, Specific country name)
3 – Does the sentence consist of Sales, Costs, Profits or Investments?
After each student manually specifies the sentences, the sentences are checked by Dr. Wim Janssen in order to find differences between classification of the same sentence by the pair of students. Afterward, the pair of students would sit around the table with Dr. Wim Janssen to discuss the sentence to figure out whether to eliminate the sentence or correct it. After each pair of student discusses the differences in classification with Dr. Wim Janssen, the final data is provided by Dr. Wim Janssen with additional data derived from Compustat. A limitation of the final dataset provided by Dr. Wim Janssen is that the cross check between the students
was not done. This means that sentences classified by one student as a forward-looking sentence, may have been classified as non forward-looking sentence. Furthermore, a second limitation of my dataset is that it only contains data from 1996 to 2000. Since I am interested in the post SFAS 131 period, I can only use data from 1998 to 2000. This would not be a limitation in the first place because I could extract additional data from other databases however, since I make use of Forward-Looking sentences, my data needs to be matched. Therefore, extracting additional data from Compustat is not possible because I miss the additional manually classified sentences.
The final dataset, which is provided to all participating students, consists of 2,090 sentences and 418 firms. Each firm comes back in each year from 1996 to 2000. Since the cross check between the students was not done, the first thing I do is delete the sentences which are classified as a Forward-Looking Sentence but do not meet any of the criteria: 1 a forward-looking sentence, 2 a sentence on disaggregated foreign operations outside the US and 3 a sentence that consists of either sales, costs profits or investments. Using filters, I drop 10 sentences leaving me with 2,080 observations. Second, I drop all sentences where there is no information about the domestic and foreign pre-tax income (see explanation in “model & Variables” section) leading to a drop of 333 sentences leaving me with 1,747 observations. For the second part of my data I consult the site http://astro.temple.edu/~lnaveen/data.html where I can download data about VEGA. Using STATA14, I merge my data, which consists of 1,747 observations with my VEGA data using GVKEY and FYEAR codes as merging variables. Furthermore, because I am interested in the post SFAS131 period, I exclude observations from 1996 and 1997 leaving me with 1046 observations. Last, because I need all the required information about each firm in each year from 1998 to 2000, firms that don’t meet this requirement are excluded from my observations. This means dropping 44 observations leading to my final dataset consisting of 1,002 observations. My sample selection procedure is summarized in table1.
Table 1: Data Sample Selection procedure.
Data Sample Selection
procedure observations
Original raw dataset from Dr. Wim Janssen 2090
Exluding misclassified sentences that do not meet the criteria -10
Excluding observations that miss foreign and domestic pre-tax income -333
Merging dataset with VEGA data No drop
3.2 Model & Variables
For my research I use an empirical research approach using archival data. My first hypothesis assesses whether US firms that disclose Forward-Looking sentences are more likely to engage in geographical empire building. For testing this hypothesis I follow the model which Thomas and Hope (2008) used in their study for examining empire building in the pre and post SFAS 131 period (where firms are not required to disclose geographical earnings after adoption of SFAS 131). In order to test empire building, Thomas and Hope (2008) hypothesise a growth in foreign sales and at the same time a decline in foreign profit margin and decline in firm value for non-disclosing firms after adoption of SFAS 131.
3.2.1. Hypothesis 1
Since I am interested in the relationship between forward-looking disclosures and geographical empire building, I hypothesise the same expectations as Thomas and Hope (2008), however instead of non-disclosing firms, I use firms that disclose forward-looking sentences as my independent variable. In order to test my first hypothesis I will use two formulas from Thomas and Hope’s (2008) study, which I adjust with my own variable.
ForSG = + 1(FLS) + 2(DomSG) + 3(Size) + 4(Follow) + 5(ForeignPerc) +
6(ForSeg +
ForPM = + 1(FLS) + 2(DomPM) + 3(Size) + 4(Follow) + 5(ForeignPerc) +
6(ForSeg +
Where:
ForSG is the growth in foreign sales. It is defined as the natural log of the year-over-year change in foreign sales.
ForPM is the foreign profit margin. It is defined as the SEC required foreign earnings divided by the total foreign sales. Foreign pre-tax earnings are used as required foreign earnings.
FLS stands for forward-looking sentence. It is the number of forward-looking sentences a US-firm discloses on non-US operations and investments if criteria as mentioned above are met.
DomSG is the growth in domestic sales. It is defined as the natural log of the year-over-year change in domestic sales.
DomPM is the domestic profit margin. It is defined as the SEC required foreign earnings divided by the total domestic sales. Domestic pre-tax earnings are used as required domestic earnings.
Size is used since it is a standard proxy for the overall disclosure level of a firm and controls for other disclosures that are provided by firms. It is defined as the natural log of total assets of a firm(Thomas and Hope, 2008).
Follow is the control variable for the firm’s information environment. It is defined as the number of analysts that follow the firm.
ForeignPerc is the control variable for the relative importance of foreign operations across my sample firms. It is defined as the percentage of foreign sales to total sales.
ForSeg is the control variable for foreign segments wherein the firm operates.
Finding a positive relationship between FLS and ForSG, and a negative relationship between FLS and ForPM will be evidence for US-firms that disclose Forward-Looking sentences engaging in geographical empire building.
3.2.2. Hypothesis 2
For testing the second hypothesis where I am interested in the relation between firms that engage in geographical empire building and VEGA I will also use the model from Thomas and Hope (2008), but using VEGA as my independent variable. This will lead to the following two models:
ForSG = + 1(VEGA) + 2(DomSG) + 3(Size) + 4(Follow) + 5(ForeignPerc) +
6(ForSeg +
ForPM = + 1(VEGA) + 2(DomPM) + 3(Size) + 4(Follow)+ 5(ForeignPerc) +
6(ForSeg +
Where:
VEGA is the change in the dollar value of the executive’s wealth for a 0.01 change in the annualized standard deviation of stock returns
For the second hypothesis I expect to find a positive relationship between ForSG and VEGA and a negative relationship between ForPM and VEGA. Finding the expected relationships will provide evidence that US-firms with a higher VEGA are more likely to engage in geographical empire building.
3.2.3. Hypothesis 3
For the third hypothesis where I examine whether there is an effect of VEGA on the relationship between Forward-Looking sentences and geographical empire building, the following model is provided by manipulating Thomas and Hope (2008) model.
ForSG = 1(FLS) + 2(VEGA) + 3(VEGA*FLS) + 4(Size) + 5(Follow) +
6(ForeignPerc) + 7(ForSeg +
ForPM = 1(FLS) + 2(VEGA) + 3(VEGA*FLS) + 4(Size) + 5(Follow) +
6(ForeignPerc) + 7(ForSeg +
For the third hypothesis I expect to find a positive relationship between ForSG and β3 and a negative relationship between ForPM and β3 using a Chi^2 test providing me with evidence that the relationship between forward-looking sentences and geographical empire building is more positive for firms with a higher VEGA.
Because my data is specified as panel data, this means that each firm has to be present in each year from 1998 to 2000 leading to a strongly balanced dataset without repeated time values. Therefore, in order to draw the regression, I cluster the firms and acquire robust p-values.
4. Results
4.1 Descriptive statistics
Panel A: Descriptive statistics by fiscal year
Fiscal year MD&A sentences Forward-Looking sentences Forward-Looking intensity (%) 1998 401 175 43.6% 1999 366 114 31.2% 2000 364 104 28.6%
Panel A reports the number of MD&A sentences, looking sentences and forward-looking intensity across the three fiscal years (post SFAS131 period). The number of MD&A sentences decreases from 401 sentences in 1998 to 364 sentences in 2000, a drop of 9.2%. The number of looking sentences also decreases. In 1998 there were 175 forward-looking sentences whereas in 2000 there are 104 forward-forward-looking sentences. This is a drop of 40.6%.
Table 2: Descriptive statistics for dependent and independent variables
(1) (2) (3) (4) (5)
VARIABLES N mean sd min max
FLS 1,002 0.392 0.766 0 5 Follow 1,002 7.376 9.399 0 45 ForSeg 1,002 3.485 1.863 1 21 ForSG 1,002 0.137 0.388 -0.833 1.368 ForPM 1,002 0.061 0.105 -0.358 0.410 Size 1,002 6.848 1.859 1.627 12.99 ForeignPerc 1,002 0.426 0.175 0.0224 0.961 VEGA 1,002 48.58 72.79 0.246 461.9 VEGAxFLS 1,002 19.43 64.19 0 461.9 Number of firms 334 334 334 334 334
Table 2 shows the descriptive statistics on the dependent and independent variables, which I used in my analysis for the post SFAS 131 period. The statistics from the table provide me with some preliminary insights to about the hypotheses of this study.
As shown in table 2, the variable FLS has a mean value of 0.392. This indicates that the average firm in my sample, given the total amount sentences from the MD&A section, discloses on average 39.2 % forward-looking sentences. As shown in panel A, this is in line with the average forward-looking intensity. Furthermore, variable ForSG has a mean of 0.137, indicating that the average firm in my sample grows their foreign sales by 13.7% on average.
Moreover, variable ForPM has a mean value of only 0.061, which indicates that the foreign profit margin for the average firm in my sample is 6.1%. These statistics show that on average, foreign sales grow more than foreign profit margin.
In table 7, the correlation matrix is shown for al variables used. As shown, the only correlation, which requires attention is the correlation between the variables Size and Follow. Since the variable Follow is defined as the number of analysts following the firm in order to control for the information environment and Size as a control for the overall disclosure level, it is not surprising to find a correlation between the two variables as also is found in prior studies (Thomas and Hope, 2008). Furthermore, in order to ensure that my analysis is not affected by multicollinearity problems, I perform a multicollinearity test using STATA for all regressions. In panel B the results from this test is shown. From the test, the greatest variable inflation factor is for variable VEGAxFLS, which is equal to 2.11. This is likely to be the case since the variable itself is a combination of two single variables that are also present as single variables (FLS and VEGA). For the remaining variables, all the variable inflation factors are lower than 2.0. Hence, showing that my analysis does not suffer problems arising from multicollinearity.
Panel B: Multicollinearity results:
Variable VIF 1/VIF
VEGAxFLS 2.11 0.474216 FLS 1.62 0.619230 VEGA 1.47 0.683428 Size 1.46 0.684383 Follow 1.44 0.695909 ForSeg 1.11 0.917798 ForeignPerc 1.11 0.919644 DomSG 1.05 0.953319 Mean VIF 1.42
Table 7: Correlation matrix
ForSG ForPM DomSG DomPM FLS vega VEGAxFLS Follow ForSeg Size ForeignPerc ForSG 1000 ForPM 0.0514 1000 0.1038 DomSG 0.2909 0.1218 1000 0.0000 0.0001 DomPM -0.0314 0.3278 0.1517 1000 0.3211 0.0000 0.0000 FLS 0.0270 -0.0530 -0.0069 -0.0218 1000 0.3933 0.0939 0.8262 0.4910 vega 0.0498 -0.0958 0.0496 -0.0052 0.0372 1000 0.1151 0.0024 0.1167 0.8702 0.2391 VEGAxFLS 0.0431 -0.0963 -0.0096 -0.0481 0.5501 0.4867 1000 0.1733 0.7620 0.1281 0.0000 0.0000 0.0000 Follow 0.0813 0.2702 0.0853 0.1953 0.0127 -0.0156 0.0264 1000 0.0100 0.0000 0.0069 0.0000 0.6892 0.6210 0.4046 ForSeg 0.1998 0.0097 0.1065 0.0322 0.0598 0.0487 0.0538 0.0869 1000 0.000 0.7599 0.0007 0.3090 0.0585 0.1235 0.0886 0.0059 Size 0.0334 0.2748 0.0820 0.1909 0.0671 -.0.598 -0.0243 0.5459 0.1294 1000 0.2903 0.0000 0.0094 0.0000 0.0338 0.0774 0.4425 0.0000 0.0000 ForeignPerc 0.1235 0.0120 -0.1168 -0.0348 0.0585 0.1009 0.1151 0.0624 0.2578 0.0547 1000 0.0001 0.7047 0.0002 0.2712 0.0643 0.0014 0.0003 0.0482 0.0000 0.0835
4.2 Results of Hypothesis Tests
4.2.1 Geographical Empire Building and Forward-Looking sentences
Table 3: Results H1
(1) (2) VARIABLES ForSG ForPM
FLS 0.008 -0.008* (0.544) (0.094) DomSG 0.291*** (0.000) Follow 0.003** 0.001*** (0.043) (0.007) ForSeg 0.029*** -0.003 (0.000) (0.132) Size -0.011 0.011*** (0.122) (0.000) ForeignPerc 0.264*** 0.012 (0.000) (0.631) DomPM 0.071*** (0.000) Constant -0.045 -0.015 (0.404) (0.519) Observations 1,002 1,002 Number of firms 334 334 Wald Chi^2 113.31 58.94
Robust P-value in parentheses *** p<0.01, ** p<0.05, * p<0.1
My first hypothesis, H1, is accepted if forward-looking sentences are significantly positively related to the foreign sales growth and when forward-looking sentences are significantly negatively related to foreign profit margin. If this is the case, then sufficient evidence is provided for firms that disclose more forward-looking sentences are more likely to engage in geographical empire building.
Table 3 shows the results for H1. The first column refers to the relation between Forward-Looking sentences and foreign sales growth. Here I find a positive relation between my dependent and independent variable of interest (FLS), however this result is not significant. The results in the second column of H1 show a negative relationship between foreign profit margin and forward-looking sentences as expected. Furthermore, the result is significant with p-value: -0.094. In order to provide sufficient evidence about geographical empire building, both relationships should be significant. I find statistical significant evidence
for forward-looking sentences being negatively associated with foreign profit margin. However, since I find no statistical significant evidence for the relation between forward-looking sentences and foreign sales growth, I must reject H1.
With regard to the control variables I used in my analysis, I find positive and significant associations in the first column between ForPerc and DomSG with ForSG. These findings are in line with the findings from Thomas’ and Hope (2008) in the post SFAS 131 period, both in direction of the association and significance. In the second column my results show that Follow, Size and DomPM are all positively related to ForPM and are also significant. These findings are also in line with Thomas and Hope (2008), however, I find a positive and significant relationship between Size and ForPM whereas Thomas and Hope (2008) do not.
4.2.2. Geographical Empire Building and VEGA
Table 4: Results H2
Robust P-values in parentheses *** p<0.01, ** p<0.05, * p<0.1
My second hypothesis, H2, is accepted if firms with a higher VEGA are significantly positively related to foreign sales growth and when VEGA is significantly negatively related to foreign profit margin. If this is the case, I can accept H2 where firms with a higher VEGA are more likely to engage in geographical empire building.
Table 4 as shown above, shows the results for H2. The first column shows that VEGA is positively associated with foreign sales growth. However, this result is not significant. The second column shows that VEGA is negatively associated with foreign profit margin as expected. Furthermore, this association is significant with p-value: 0.015. Since both regressions must show significant results for my H2 to be accepted, the results show that only VEGA is significantly negatively related to foreign profit margin. Although VEGA is positively associated to foreign sales growth, the association is not significant. Therefore I must reject H2. With regard to the control variables used in my analysis, I find the same relationships between my control variables and my dependent variable. In the first column is shown that DomSG, ForeingPerc and ForeinSeg are positively related to ForSG and these results are also significant. As mentioned for the control variables for my first hypothesis, these are also in line with the results from Thomas and Hope (2008) research for the post
(1) (2)
VARIABLES ForSG ForPM
Vega 0.000 (0.693) -0.000** (0.015) DomSG 0.290*** (0.000) Follow 0.003** 0.001*** (0.044) (0.006) ForSeg 0.029*** -0.003 (0.000) (0.118) Size -0.010 0.010*** (0.139) (0.001) ForeignPerc 0.262*** 0.015 (0.000) (0.542) DomPM 0.072*** (0.000) Constant -0.048 -.0.011 (0.372) (0.649) Observations 1,002 1,002 Number of firms 334 334 Wald Chi^2 114.37 59.78
