Faculty of Economics and Business BSc
Does CEO tenure directly influence firm value of telecommunication
and technological companies within North America in the most
recent decade?
Name: Wanlin Guo
Student Number: 10799176 Bachelor Thesis
Specialisation Finance and Organisation
BSc Economics and Business, University of Amsterdam First supervisor: Timo Klein
Statement of Originality
This document is written by Wanlin Guo, who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
This paper concerns about the relationship between CEO tenure and firm value. As the adoption of Huawei’s short-CEO-tenure rotating system is followed by an outstanding achievement in telecommunication industry as well as other related technological industries such as computer software, whether there is a causal relationship between CEO tenure and firm value is a topic. Based on previous studies that were done at least ten years ago, an argument that CEO tenure is reversely related to firm value especially in dynamic industries is widely discussed. Considering the pace of updating and reforming in technological industry is at an above-average level, arguments should be updated using new data in recent years. In this paper, log-linear regression analysis is employed with 286 samples of North American listed companies in telecommunication, computer software and data processing industries in the year of 2006-2015. There is not enough evidence to support the inverse relationship if exclude industry effect. After including industry factors, it is found that every additional year of tenure is associated with 0.52% increase in firm value, this result is significant at 10 percent. In other words, a positive relationship is found eventually and the assumption of a negative relationship is rejected.
Table of Content
I. INTRODUCTION 5 MOTIVATION 5 EXISTING CONTROVERSY 5 PAPER STRUCTURE 6 II. LITERATURE 7 MILLER’S THEORY 7CEO PARADIGM THEORY 9
THEORY OF FIRM VALUE,CEO TENURE AND CEO TURNOVER 10
III. HYPOTHESIS 12 PROPOSITION 12 IV. METHODS 14 DEPENDENT VARIABLE 15 INDEPENDENT VARIABLE 16 CONTROL VARIABLES 16 - CEO COMPENSATION 17
- S&P CREDIT RANKING 17
REGRESSION AND METHOD 18
V. EMPIRICAL DATA 22
VI. ANALYSIS 28
RECAP ON HYPOTHESIS 28
FINDINGS IN EMPIRICAL DATA 28
EXPLANATION OF RESULT 29
VII. CONCLUSION 31
LIMITATIONS 31
SUGGESTIONS 32
I. Introduction
motivation
Recently, with the increasing influential power and remarkable achievement of a Chinese emerging technological company Huawei, an innovative rotating CEO system came in sight. Breaking the stereotype of “heroic leader” in a company, Huawei’s board members selected three chairmen acting as the rotating and acting CEO for tenure of six months (Cremer & Tao, 2015). As now Huawei has become the leader in the telecommunication and technological industry, such as data processing and software, under such a unique short tenure system inspired by the four-year presidential election in the US. It might be a time to re-think whether there exists casual relationship between the rotating system and Huawei’s outstanding firm performance. Going broader aspired by this, we might have reason to think whether there would be a clear linkage between CEO tenure and firm value in the technology-related industries in recent years.
existing controversy
According to Berk and DeMarzo (2014, pp.962-981), the separation of control (CEOs) and ownership (shareholders) is an essential start for a sufficient corporate management, especially for those large companies, and the separation even influences the prosperity or decline of a firm. Because these two parties might pursue different goals: CEOs have more incentives to focus on short-term operating profit, which decides main part of their compensations under most corporate compensation systems, thus risk-shifting investments could appear with less scrupulousness; while shareholders focus more on overall firm value and long-term development. Due to CEOs are appointed by shareholders, it is essential for shareholders to figure out what optimal factors and regulations should be considered to reach their long-term expectation on development. Apart from the education or experience of CEOs per se, as CEO is one of the most
Back to ten years ago, Henderson et al. (2006) document the importance of CEO tenure in dynamic industries using corporate paradigms. They tracked 228 CEOs in a highly dynamic computer industry, and find that the best firm performance steadily declined associated with CEO’s increasing tenure. The reason is, explained by them, the firm paradigms in such kind dynamic industry grew obsolete more quickly. Following in this sense, a shorter CEO tenure should be preferred.
However, Brookman and Thistle (2009) select 1472 firms and 2325 CEOs in their samples, and examine the change of each firm’s value after every CEO’s tenure terminated. In their findings, there are no significance results of firm value change after CEO’s leaving and appointing, thereby going to a conclusion that CEO tenure is not a case for firm value. Because they appoint out that a decreasing firm value can come from CEO’s increase negotiation power to persuade shareholders to keep them in position longer, while increasing firm value can be resulted from a good match between CEO and shareholder. There is risk in a termination of current CEO, but the risk is not economically important. Therefore, they conclude that the relationship between CEO and firm value is not strongly associated with each other.
paper structure
According to above, whether CEO tenure will influence firm value is brought to the table. To find the relationship and give response to the controversy, in this paper, new firm data in the most recent decade 2006-2015 would be collected. As inspired by Huawei, the data of study object is telecommunication, computer software and data processing industries. In the second section, arguments and findings on this issue in previous publications or journals are given. Based on the other researchers’ work, hypothesis in the paper is addressed upfront of new data analysis in the third section. In the fourth and fifth part, new regression with respect to interest variables is given and the method and datasets used are explained in details. In the end, an answer to the question in the beginning are given.
II. Literature
A mass of existent literatures reveal that firm value heavily relies on capital structures, investment decisions, financial options and so on. It seems that financial elements are regarded as the most important elements in consideration of corporate development. Nevertheless, many researchers also believe the decision on CEO tenure can be a key aspect of successful corporate governance, more importantly, the length of CEO tenure influences on the change of firm value (Bushman et al., 2010).
In this paper, telecommunication and technological industries are the object of study. Unlike other industries, such as medicine or aircraft, technological industries have high level of mobility. Not only that, with the consensus of globalization tendency, more firms attach importance to the convergence with outside world. Not only develops structure and internal management, firms also focus on changes of outside environment, either rival firms or the whole industry. Telecommunication as well as other technological industries, relating to all walks of life, requires an open mind to new knowledge and innovative inspiration to keep up with continuous enhancement of needs. A new concept of open innovative approach nowadays is introduced as a product by combining inflows and outflows of knowledge, requires an interact between firm and the outsides (Bigliardi et al., 2012). The interexchange between individuals encourages information mobility as well as dynamics of industry. In this sense, the extreme dynamic of industry makes it more important but also difficult for shareholder to take an optimal decision on CEO tenure. Therefore, investigating the relationship between CEO tenure and firm value becomes more necessary when refer to this industry. This sense brings more value to the work in this paper.
important in an uncertain setting, that is, dynamic industries in which surroundings change quickly and unexpectedly, because CEO faces more possibilities and firm burdens more risk. In his research, long-tenured CEO has more problem in meeting the outside change in an unstable industry, therefore deviates from requirements of external environment. More specifically, long CEO tenure has more chance to produce failure of decisions, so long tenure is inversely related to firm financial performance. In the other way around, in his opinion, obtaining short CEO tenure is beneficial to gain a higher firm value.
Firstly, longer tenure can enhance CEO autonomy thereby decreasing firm value. CEO’s personal connection with important clients increases when tenure increases, the ever-increasing personal power gives CEO chance to resist new investment decisions and structure changes, which threaten their own benefits. Even though the change is good for overall advancement of firm. In such a dynamic industry, flexibility of corporate policy is required to be confronted by outside challenges, so CEO’s resistance towards changes in management is likely to have negative impact on firm value.
Secondly, a longer tenure is usually treated as a compliment for CEO’s good past behaviour, which should be a good thing, but in many cases, it turns out that this compliment encourages overconfidence problem for CEO in decision making. More failure is likely to be generated when CEO is in the position for a long time. Therefore, firm value may be negatively affected.
Thirdly, a mature organisational structure experiences re-organized and reform when a new CEO takes office, the process of completing a new structure is called “gestalt” in Miller’s theory. Once a gestalt becomes full-fledged, CEO’s perception for environment decreased as their tenure increased, only when new CEOs sprang up the obsolete gestalt can be changed. Therefore, once CEO tenure is too long, it is possible that the gestalt is too obsolete to response to the dynamic in technological industry. Following this idea, a short tenure is preferred.
CEO paradigm theory
McClelland, Baker and Oh (2012) propose a theory of CEO paradigm regarding to tenure. There are two types of paradigm, fixed and flexible, and different type leads different outcome in firm value.
Hambrick and Fukutomi (1991) document that in the early stage, the paradigm is more flexible, as CEO is more eager to new knowledge and innovations in the first period of incumbency in order to capture more possible resources and build new firm structure. This argument is supported by Gabarro’s (1987) findings, which shows that the first 2.5 years of new CEO appointment is a peak season for generation of new firm-structure modification or major changes. As tenure increases, CEO is gradually unwilling to accept new change, due to new change might imply threat to their power and position. In this way, CEO paradigm turns from flexible to fixed. Also, Hermalin and Weisbach (1998) add that the impact of CEOs on board decisions becomes prominent and their negotiating power increases sharply with tenure extends. For this reasons, CEO paradigm becomes more fixed when tenure is long. As McClelland et al. (2012) maintains, a fixed paradigm that results from enriched CEO’s personal power with extended tenure, hurts firm’s future performance, so fixed paradigm has negatively impact on firm value. This problem is more obvious in a dynamic industry, because fixed paradigm is persistent to an unchanged structure that prevents CEO from updating concepts regularly, thus more misalignment with outsides is produced, and then reduces firm value. Thus, in a dynamic industry, a flexible paradigm with short tenure is highly demanded.
Accordingly, in an unsteady and dynamic industry, the frequent and irregular change within the industry requires a high level of consistency with outside environment. The flexibility of CEO paradigm is required. In the theory, the flexibility is more likely to be achieved with a short CEO tenure. Therefore, short CEO tenure is usually in line with a higher firm value.
Theory of firm value, CEO tenure and CEO turnover
Many researchers maintain that the key driven factor of changes in CEO tenure can be expressed by CEO turnover, and these changes can influence on firm value. According to Dikolli et al. (2014), when CEO tenure is long, due to owners’ trust on executives after a long-term relationship, CEO turnover is relatively low. In their perspective, CEO tenure has casual relation with CEO turnover and they are inversely correlated with each other. Looking into underlying concept, CEO turnover depend partially on the dynamics of industry.
Because of uncertainty in dynamic industries, such as telecommunication and technological industries, firms are exposed to more risks. Bushman et al. (2010) establish a fundamental linkage between firm performance risk and CEO turnover in different industrial conditions. In their research, the key point of corporate governance is embodied in a proper decision of appointing and replacing CEO at a proper time. The performance risk, representing noise with respect to observing CEO’s actions, plays a crucial role. The concept is, if variance in performance result connects closely to unobservable CEO talent (idiosyncratic risk), firm performance is diagnostic about such talent, thereby allowing shareholders to assess CEO and decide optimal tenure length to replace incumbent, in this case, CEO turnover is high, so CEO tenure is shorter. In the other way around, if variance in performance results is unrelated to CEO talent, more influenced by outside uncertainty (systematic risk) such as dynamic industries, the ability of making right decision is limited, in this situation, CEO turnover is low, so CEO tenure is longer. That is, in a stable industry, the output variance depends more on capability difference of executives, while outside variance is more significant in dynamic industry. Following this idea, empirical outcomes prove that CEO turnover increases when idiosyncratic risk is outstanding and decreases when systematic risk is prominent. In other words, to capture more firm value, turnover rate should be low, therefore tenure is long, in a dynamic industry.
Apart from the previous case that turnover difference resulted from industry specialization, there is also a case that turnover difference can be generated from
industry similarity. There are enough evidences to show that CEO is easily to be replaced by a new successor from another firm instead of from internal appointment, if these two firms share high level of similarity (Zhang & Rajagopalan, 2010). Identically, Parrino (1997) argues that the likelihood of CEO turnover caused by outside succession is high when firms are in the similar industries or the involving industries share a common knowledge. Because high level of shared information could decrease the failure of appointing new outside CEO, thereby reducing the cost of replacing CEO, besides, appointing an outside successor can bring more innovation and reforms, which can generate more firm value. Their studies imply that firm value is likely to increase when CEO turnover is high and CEO tenure is short, in telecommunication and technological industries.
Based on the two opposite arguments about CEO tenure in dynamic industry in previous two paragraphs, no conclusion can be exerted but they generate a broad way of thinking when analyze CEO tenure in telecommunication and technological industries.
III. Hypothesis
Proposition: firm value is oppositely related to CEO tenure in telecommunication industry: firm value increase with shorter CEO tenure and decrease with extended CEO tenure.
Based upon the previous section, many researches have discussed the dynamics in telecommunication or related technological industries. Unlike other stable environment, technological industry is an information-shared industry, the basic equipment and operation channel has high level of similarity, thus requires greater outputs of innovation and creation. Therefore, CEO should think, plan and implement in accordance with the outside knowledge reformation so that cater to high demand of updating systems, software or even by-products. In this sense, a flexible structure and open-to-change pattern is beneficial to increase firm value in telecommunication industry. Additionally, considering the CEO paradigm is likely to turn to be fixed as CEO tenure grows: on the one hand, CEO has less incentives to explore new innovations as they do in the first 2.5 years so more misalignments and deviations with outside environment change may appear in their longer tenure; one the other hand, CEO autonomy develops and CEO is unwilling to risk making change to threat their rights, they are likely take advantage of negotiable power to influence shareholders’ decision on appointing CEO to stay longer. In short, to acquire a flexible paradigm, CEO tenure should be shorter. According to Weng and Lin (2012), as tenure increases, CEO is less likely to jump out of the fixed pattern, all new knowledge will be born in refining past experience rather than in growing new skills. Following, Wu, Levitas and Preim (2005) display the evidences to show greater technological output is generated when CEO tenure is short and suggest shorter tenure can explore more potentials. Integrating the above points, an assumption is concluded that high firm value needs a shorter CEO tenure in telecommunication or related technological industries.
However, literatures also reveal that variance of firm performance is closely related to uncertainty of the whole industries rather than CEO’s talent. The outside
uncertainty fluctuation is much harder to measure, compared to assess CEO talent. Thus, shareholders have less evaluation criteria in predicting future firm value and less reason to change CEO frequently. Which means, CEO turnover ratio is low so that CEO tenure is high in reality. But Parrino (1997) and Zhang and Rajagopalan (2010) assert that the cost of replacing CEO with outside successors is cut down in an industry where firms have strong connection with each other, as a result, turnover ratio caused by outside replacement will be high when internal replacement ratio unchanged. In total, the average turnover ratio is increased. Under this assumption, CEO tenure should be shorter. Combining these two conclusions, it is ambiguous to give clear statement on the optimal tenure length in increasing firm value. But these two opposite arguments with respect to CEO tenure in technological industry bring more value to the following research and new findings in this paper.
In all, based upon the dynamics of telecommunication and technological industry and CEO paradigm theory, a negative relationship between firm value and CEO tenure is hypothesized in this paper, that is, firm value increases while CEO tenure decreases.
IV. Methods
As prior studies state, telecommunication industry and technological industries, such as data and computer software, share a comprehensive universality. Due to the research question in this paper is inspired by Huawei’s success, the interest industries are selected from industrial categories to which Huawei belongs. According to Huawei’s annual report (2015, pp.1-5), through a way of creating value to communication technology, Huawei also dedicates to related computer software and data processing innovation. Besides, considering the necessity of sufficient sample size as well as the extensiveness of outcome, apart from telecommunication industry itself, technological industries are also included, which can be shown below.
- 48: communications (telephone communications, etc.) - 4911: electric services
- 1731: electrical work (telecommunications equipment and contractors, etc.) - 357: computer and office equipment
- 366: communications equipment
- 5045: computer and computer equipment and software
- 5065: electronic parts and equipment (telephone equipment & system, etc.) - 5415: computer systems design and related services
- 5734: computer and computer software stores
- 737: computer programming, data processing, and computer related services - 8243: data processing schools (computer training, internet service, etc.)
In total, the number of 2486 samples are collected from Wharton Research Data Services (WRDS) database, which covers all listed North American companies within the above industries in the period of 2006-2015. To be more concrete, variables regarding CEO elements (CEO tenure and CEO compensation) are collected from Compustat Executive Compensation database, and firm factors (firm value and credit ranking) are collected from Compustat –Capital IQ database. After a merger of two databases by filtering data with firm’s Standard Industry Classification (SIC) code and
corresponding year and dropping out unclear data, the final amount of 286 samples are left, from the following industries:
Table 1 – Industry Observation
SIC code 1731 4911 5045 5065 5734
observations 21 182 53 13 17
Dependent variable
The way to evaluate firm value and performance is various, for instance, McCelland et al. (2012) examined future Return on Asset of firms when evaluates the firm value, Henderson et al. (2006) calculated weighted average profitability of return of sales, assets and invested capitals. However, in this paper, Market Value is as the firm value measurement for several reasons.
First, unlike McCelland et al. (2012) research, which emphasizes on exploring the future firm performance and predicting potential future change in the effect of CEO tenure, the focus in this paper is based on historical actual data and incurred impact of CEO tenure on firm value. Considering the fact that tax effect is different in different states as well as the methods of recording Return data is not the same in different firms, ROA is not an optimal way in this research.
Second, unlike Henderson et al. (2006) research, which mixes computer industry and food industry together to evaluate the relation between CEO tenure and firm performance. In this paper, telecommunication industry can be seen as a branch of broad technological industry like other computer software industry etc. There is no need to get a weighted profitability to compare their firm performance. All the data in this study is in the same region and same broad industry and collected from the same database, market value measurement is eligible and more straightforward with less possibility of artificial errors or deviations caused from record intentions.
Last but not least, market value is a good indication of both internal and external investors’ perception of firm performance. As Denis and Denis (1995) document, a significant poorer firm performance appears when management changes. They indicate that this poorer firm performance reflects on stock price change obviously, accordingly, market value (calculated via share price multiplying number of share outstanding) is a good measurement when investigate firm value fluctuation with CEO tenure change.
Therefore, in this paper, Market Value will be the measurement of firm value, the same as the one adopted in Denis and Denis’s (1995) study.
Independent variable
Identifying CEO tenure straightforward is not easy in process of collecting data. For this reason, CEO tenure in this paper is given by a simple calculation of “Date of leaving as CEO – Date of becoming as CEO”, measured in years. The after-calculated tenure only indicates the time in position, no matter whether some CEOs still work in the firm after being replaced by successors. Until the date of collecting data, some CEOs are still in the position, thus their “Data of leaving as CEO” data are unknown so far. For this reason, this group of sample is eliminated from the sample size of 2486. This restriction reduces the sample to 286 observations in the end.
Control variables
In the light of previous researchers’ experience, when analyse the firm value and CEO tenure, two other variables – CEO compensation and firm’s credit ranking – should not be ignored. To minimize omitted variable bias as much as possible, two control variables CEO compensation and S&P credit rating are introduced.
- CEO compensation
Prior studies show a clear relationship between CEO compensation and firm value. Mehran (1995) examines 153 randomly-selected firms and provides an evidence to show there is positive relationship between CEO compensation and firm value. More compensation encourages higher firm value.
Besides, Rose and Shepard (1997) argue that better compensation reflects a better ability of CEO to match the objectives of firm, so the turnover ratio will decrease, which ultimately reduce the chance of succession and increase the tenure to some extent.
- S&P credit ranking
According to Berk and DeMarzo (2014, pp.187-191), credit ranking expresses the likelihood of firm default. Higher the ranking is, less likely the firm fails to payoff debts. It is a sufficient way to make firms’ creditworthiness information available to investors. A good ranking therefore encourages firm’s enrichment and investment. In this way, a higher ranking induces likelihood of creating more firm value.
On the other hand, higher the ranking is, less possibility of firm default, more chances that CEO tenure gets extended. First, a well-performed firm with less likelihood to be default implying a comparably longer firm age, there is more possibility for CEO to stay at his position; second, a firm with satisfactory credit usually has good firm performance and a satisfactory sales outcome, which come from well-pleasing management, in this case, there is no strong incentive for shareholders to change CEO, thus tenure will increase.
In short, to testify the individual effect of CEO tenure on firm value, compensation and credit ranking should be included in the regression, due to these two control variables
Regression and Method
To estimate the relationship between firm value and CEO tenure, a Log-linear Regression Function is normally used. On the one hand, the logarithm of market value is a frequently used method in corporate governance sector (Al-Matari et al., 2014). Besides, Stock and Watson (2011) state, “labour economists typically model earnings using logarithms” (p.272). On the other hand, nearly all researchers, who are mentioned in previous sections, measure tenure in years. Because this way conforms to practical situation in management. In reality, a precise assessment system applies to different level of credit, from A+ to D. To be more concise, in this paper, combine A+, A, A- into highest rank named Rank1, B+, B, B- into second rank named Rank2, C into third rank named Rank3 and D into fourth rank named Rank4. Moreover, due to the constant term is included, Rank1, Rank2, Rank3 and Rank4 are highly correlated, which leads to multicollinearity issue. To avoid the problem, the highest rank Rank1 is eliminated beforehand.
The following table displays all variables to be used in regression analysis:
Variable Description
lfirmvalue Dependent variable: Natural logarithm of market value of firm in millions of U.S dollars in the current year.
Tenure Independent variable: CEO tenure, measured in years.
Compensation Control variable: CEO compensation measured in millions of U.S dollars in the current year.
Rank2 Dummy variable equals 1 if the credit ranking is B+, B or B-; 0 if otherwise.
Rank3 Dummy variable equals 1 if the credit ranking is C; 0 if otherwise.
Rank4 Dummy variable equals 1 if the credit ranking is D; 0 if otherwise.
SIC4911 Industry dummy variable equals 1 if the firm is from industry with SIC code 4911; 0 if otherwise.
SIC5045 Industry dummy variable equals 1 if the firm is from industry with SIC code 5045; 0 if otherwise.
SIC1731 Industry dummy variable equals 1 if the firm is from industry with SIC code 1731; 0 if otherwise.
SIC5065 Industry dummy variable equals 1 if the firm is from industry with SIC code 5065; 0 if otherwise.
SIC5734 Industry dummy variable equals 1 if the firm is from industry with SIC code 5734; 0 if otherwise.
Therefore, the following regression is estimated: Regression 1: CEO effect
𝑙𝑓𝑖𝑟𝑚𝑣𝑎𝑙𝑢𝑒 = 𝛽,+ 𝛽.⋅ 𝑇𝑒𝑛𝑢𝑟𝑒 + 𝛽2 ⋅ 𝐶𝑜𝑚𝑝𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛
In this regression, 𝛽, represents constant term. 𝛽. is coefficient of interest variable Tenure, representing the relationship between firm value and CEO tenure in a way of an additional year change in tenure is associated with a 100×𝛽.% change in firm value. Following the hypothesis, tenure is expected to inversely related to firm value. For this reason, 𝛽. as the coefficient of tenure is expected to be statistically significant, and the expected sign is negative. 𝛽2 is coefficient of control variables Compensation.
This regression examines the influence of CEO factors on firm value, regardless of firm factors. In this way, it is obvious to see how much CEO elements can influence firm value, furthermore, this regression shows the importance of governance on CEO and explains the motivation of this paper deeply in an empirical way.
Regression 2: Total Effect, excluding industry dummies
𝑙𝑓𝑖𝑟𝑚𝑣𝑎𝑙𝑢𝑒 = 𝛽,+ 𝛽. ⋅ 𝑇𝑒𝑛𝑢𝑟𝑒 + 𝛽2⋅ 𝐶𝑜𝑚𝑝𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛 + 𝛽9⋅ 𝑅𝑎𝑛𝑘2 + 𝛽=⋅ 𝑅𝑎𝑛𝑘3 + 𝛽? ⋅ 𝑅𝑎𝑛𝑘4
To include firm factors into consideration, 𝛽9, 𝛽=, 𝛽? are introduced and represent coefficients of control variables Rank2, Rank3, Rank4 respectively. Rank2 equals to 1, indicating the focused firm ranks at B level in credit ranking system; equals to 0, if otherwise. The same idea applies to Rank3 and Rank4 in which Rank3 expresses C credit level and Rank4 stands for D credit level. When Rank2, Rank3, Rank4 equals to 0 at the same time, firm value of a company that stays at Rank1 can be shown.
Because five industries are analysed, even though they are all selected from related technological (telecommunication, computer software, data processing), there is not industry adjusted factors included in the regression, therefore, it is still necessary to evaluate under different industries to explore whether industry factor also has impact on firm value when analyse the connection between tenure and firm value. In this way, CEO tenure influence can be analysed in a more decent and convincing way.
According to Keller (2012), “In many practical situations, a sample size of 30 may be sufficiently large to allow us to use the normal distribution as an approximation for the sampling distribution of X” (p.306). From table 1, only SIC code 4911 and 5045 contains observations more than 30. Hence, separate regression analysis taken from these two industries only. Regression 3 and regression 4 shows the regression under SIC 4911 and SIC 5045 industries respectively as follows.
Regression 3: Separate regression in industry with SIC code 4911
𝑙𝑓𝑖𝑟𝑚𝑣𝑎𝑙𝑢𝑒 = 𝛽,+ 𝛽. ⋅ 𝑇𝑒𝑛𝑢𝑟𝑒 + 𝛽2⋅ 𝐶𝑜𝑚𝑝𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛 + 𝛽9⋅ 𝑅𝑎𝑛𝑘2 + 𝛽=⋅ 𝑅𝑎𝑛𝑘3 + 𝛽? ⋅ 𝑅𝑎𝑛𝑘4
Regression 4: Separate regression in industry with SIC code 5045
𝑙𝑓𝑖𝑟𝑚𝑣𝑎𝑙𝑢𝑒 = 𝛽,+ 𝛽. ⋅ 𝑇𝑒𝑛𝑢𝑟𝑒 + 𝛽2⋅ 𝐶𝑜𝑚𝑝𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛 + 𝛽9⋅ 𝑅𝑎𝑛𝑘2 + 𝛽=⋅ 𝑅𝑎𝑛𝑘3 + 𝛽? ⋅ 𝑅𝑎𝑛𝑘4
After concluding all variables, all dummy variables are shown in one regression together with independent variable and control variables:
Regression 5: Total regression, including all industry dummies
𝑙𝑓𝑖𝑟𝑚𝑣𝑎𝑙𝑢𝑒 = 𝛽.⋅ 𝑇𝑒𝑛𝑢𝑟𝑒 + 𝛽2⋅ 𝐶𝑜𝑚𝑝𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛 + 𝛽9⋅ 𝑅𝑎𝑛𝑘2 + 𝛽=⋅ 𝑅𝑎𝑛𝑘3 + 𝛽? ⋅ 𝑅𝑎𝑛𝑘4 + 𝛽B ⋅ 𝑆𝐼𝐶4911 + 𝛽H⋅ 𝑆𝐼𝐶5045 + 𝛽K ⋅ 𝑆𝐼𝐶1731 + 𝛽M⋅ 𝑆𝐼𝐶5065 + 𝛽.,⋅ 𝑆𝐼𝐶5734
V. Empirical data
In this section, the data collected from the database are shown. And how these data are cleaned or chosen is explained in details.
Table 2 - Descriptive Statistics of the variables in total industries
Variable Mean Std. Dev. Min. Max. Observation
lfirmvalue 8.0980 1.1444 5.4749 10.8964 286 Tenure 8.06 5.08 0.34 20.75 286 Compensation 0.8757 0.4753 0.0481 3.5040 286 Dummy Rank2 0.7937 0.4054 0 1 286 Rank3 0.0315 0.1749 0 1 286 Rank4 0.0210 0.1436 0 1 286
Due to natural logarithm is adopted for dependent variable (firm value) throughout the research, values of Mean, Min. and Max. are also in natural logarithm for firm value. With a mean of 8.0980 (around $3287.89 in millions), the lowest firm value is 5.4749 (around $238.63 in millions) while the highest firm value is 10.8964 (around $53981.68 in millions).
For independent variable, Tenure, average length of tenure is 8.06 years. The shortest tenure is less than one year, while the longest tenure is over twenty years. The standard deviation is high compared to other variables in this regression, which indicates that the tenure data are spread out over a wide range in the selected samples.
For control variable, Compensation, average amount of 0.8757 million dollars are distributed to CEO every year, while the highest is 3.5040 million and the lowest is 0.0481 million.
For dummy variables, Rank 2-4, around 79% in the sample have a B level credit ranking, 3.15% have C level, and 2.10% have D level. Firms that have A level shares the reminder 15.75%. From this, more than two-thirds firms hold at least B level credit ranking, only less than ten percent firms have a C or D credit ranking.
Table 3 - Descriptive Statistics of the variables in separate regressions SIC4911 industry
Variable Mean Std. Dev. Min. Max. Observation
lfirmvalue 8.4199 1.1933 5.4749 10.8964 182 Tenure 7.83 3.6978 0.74 16.17 182 Compensation 0.8648 0.3808 0.0481 1.98 182 Dummy Rank2 0.7308 0.4448 0 1 182 Rank3 Rank4 0.0330 0.1790 0 1 182 SIC5045 industry
Variable Mean Std. Dev. Min. Max. Observation
lfirmvalue 7.1438 0.6558 5.7513 8.6196 53 Tenure 7.36 6.82 0.325 20.75 53 Compensation 0.6542 0.4872 0.0750 3.2280 53 Dummy Rank2 0.8113 0.3950 0 1 53 Rank3 0.1698 0.1698 0 1 53 Rank4
From above, variables are analysed under two different industries. The mean values of firm value and tenure are very close in different industries. More specifically, the firm with highest firm value within overall sample group is in SIC4911 industry, as the total group and SIC4911 industry group share the same maximum value of firm value. Following the same idea, the firm with lowest firm value is also in SIC4911 industry. Besides, the standard deviation of SIC4911 industry is higher than overall group as well as SIC5045 industry. Both length of tenure in SIC4911 and SIC5045 is slightly less than overall average length. And the firm with longest CEO tenure is in SIC5045.
The level of compensation in SIC4911 industry is closer to the overall average level compared to SIC5045 industry, however the lowest level of compensation is in
compensation is much lower than the overall industry average.
In SIC4911 industry, no C level credit ranking firm is recorded and most of firms rank at B level. 23.62% firms rank at A level, higher than overall A-level ratio, which means firms in SIC4911 performs well in credit ranking in an overall comparison. In SIC5045 industry, no D level credit ranking firm is recorded and most of firms rank at B level.
Table 4 - Correlation between variables in the regression model
Tenure Compensation Rank2 Rank3 Rank4 SIC4911 SIC5045
Tenure 1 Compensation 0.099 1 Rank2 -0.020 0.062 1 Rank3 0.234 -0.126 -0.354 1 Rank4 -0.175 0.016 -0.287 -0.026 1 SIC4911 -0.060 -0.031 -0.206 -0.239 0.111 1 SIC5045 -0.065 -0.223 0.021 0.378 -0.069 -0.631 1
From the table 4, the correlation between individual variable is checked. As shown above, strong linear relationship between variables is not found, thereby decreasing the possibility of multicollinearity problem.
Table 5
Regression of Firm Value on CEO Tenure and other variables
This regression indicates the relationship of firm value and CEO tenure. In total, 286 samples contain interest variable of North American public technological companies in the period of 2006-2015. The dependent variable in this regression is Firm Value, expressed by natural logarithm of total market value in that fiscal year. Tenure indicates CEO tenure, measured in years for all time a person served as CEO in the firm. Compensation stands for total CEO compensation in millions of US dollars, as reported in the fiscal year. Rank2,3,4 is dummy variable, showing Standard & Poor’s Quality Ranking in that fiscal year. The dummy variable Rank 2,3,4 equals to 1, implying the firm ranking B,C,D respectively. To avoid multicollinearity problem, Rank1 implying A level credit ranking is eliminated from regression and can be explained when Rank2, Rank3, Rank4 equal to 0 at the same time. All standard errors are heteroskedasticity in this regression.
CEO Effect CEO Effect+ Ranking Effect Separate regressions in
different industries Total Regression SIC4911 SIC5045 (1) (2) (3) (4) (5) Tenure 0.0036 (0.797) 0.0139 (0.344) 0.0023 (0.928) 0.0318*** (0.001) 0.0052* (0.0732) Compen-sation 1.1385*** (0.000) 1.0891*** (0.000) 2.2301*** (0.000) 0.3881*** (0.000) 1.1712*** (0.000) Rank2 -0.6614*** (0.000) -0.3167** (0.016) -1.2432*** (0.000) -0.3452** (0.036) Rank3 -1.6299*** (0.000) -1.8754*** (0.000) -0.6611*** (0.009) Rank4 -0.4440 (0.186) -0.5849 (0.209) -0.5206 (0.131) SIC4911 7.6356*** (0.000) SIC5045 6.7314*** (0.000) SIC1731 7.4093*** (0.000) SIC5065 6.3041*** (0.000) SIC5734 6.7115*** (0.000) Constant 7.0720*** (0.000) 7.6178*** (0.000) 6.7235*** (0.000) 7.9831*** (0.000) Observation 286 286 182 53 286 Adjusted R2 0.2253 0.2935 0.5229 0.3458 0.9887
*, **, *** denotes statistical significance at 10 percent, 5 percent, 1 percent, respectively.
P-value is in the parentheses, two-sided.
(1) CEO effect caused by tenure and compensation factor is included.
(2) Total Effect contains both CEO Effect and Separate Credit Ranking Effect in the regression.
(3)-(5) show difference of coefficients under different selected industries and overall industry.
Based on the information of coefficients in the table, the estimated regression can be shown as:
Regression 2: 𝑙𝑓𝑖𝑟𝑚𝑣𝑎𝑙𝑢𝑒 = 7.6178 + 0.0139 ⋅ 𝑇𝑒𝑛𝑢𝑟𝑒 + 1.0891 ⋅ 𝐶𝑜𝑚𝑝𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛 − 0.6614 ⋅ 𝑅𝑎𝑛𝑘2 − 1.6299 ⋅ 𝑅𝑎𝑛𝑘3 − 0.4440 ⋅ 𝑅𝑎𝑛𝑘4 Regression3 (SIC4911): 𝑙𝑓𝑖𝑟𝑚𝑣𝑎𝑙𝑢𝑒 = 6.7235 + 0.0023 ⋅ 𝑇𝑒𝑛𝑢𝑟𝑒 + 2.2301 ⋅ 𝐶𝑜𝑚𝑝𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛 − 0.3167 ⋅ 𝑅𝑎𝑛𝑘2 − 0.5849 ⋅ 𝑅𝑎𝑛𝑘4 Regression4 (SIC5045): 𝑙𝑓𝑖𝑟𝑚𝑣𝑎𝑙𝑢𝑒 = 7.9831 + 0.0318 ⋅ 𝑇𝑒𝑛𝑢𝑟𝑒 + 0.3881 ⋅ 𝐶𝑜𝑚𝑝𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛 − 1.2432 ⋅ 𝑅𝑎𝑛𝑘2 − 1.8754 ⋅ 𝑅𝑎𝑛𝑘3 Regression5: 𝑙𝑓𝑖𝑟𝑚𝑣𝑎𝑙𝑢𝑒 = 0.0052 ⋅ 𝑇𝑒𝑛𝑢𝑟𝑒 + 1.1712 ⋅ 𝐶𝑜𝑚𝑝𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛 − 0.3452 ⋅ 𝑅𝑎𝑛𝑘2 − 0.6611 ⋅ 𝑅𝑎𝑛𝑘3 − 0.5206 ⋅ 𝑅𝑎𝑛𝑘4 + 7.6356 ⋅ 𝑆𝐼𝐶4911 + 6.7314 ⋅ 𝑆𝐼𝐶5045 + 7.4093 ⋅ 𝑆𝐼𝐶1731 + 6.3041 ⋅ 𝑆𝐼𝐶5065 + 6.7115 ⋅ 𝑆𝐼𝐶5734
In the above table, each coefficient of independent and control variables is shown. As stated before, coefficient of dummy variable Rank1 is intentionally eliminated from regression due to multicollinearity issue, but can be expressed when Rank2, Rank3, Rank4 equal to 0 at the same time. In SIC4911 industry, none of firms rank at C level of credit; in SIC5045 industry, none of firms rank at D level.
Column (1) shows the individual CEO effect on firm value change, namely, the marginal effect of CEO tenure and compensation on firm value change. As the coefficient implies, every additional year of CEO tenure is associated with 0.36% increase of firm value. Column (2) shows CEO effect plus separate credit ranking effect on firm value change, which indicates how much tenure can impact firm value at different level of credit ranking. To be specific, when considering the effect of credit ranking, a year of CEO tenure increases with 1.39% growth of firm value. Column (3) and (4) show firm value is predicted to increase 0.23% and 3.28% for each additional year of CEO tenure in separate industry, the coefficient of tenure is statistically significant at 1 percent under SIC5045 industry. In Column (5), the coefficient of tenure is 0.0052, significant at 10 percent. The overall regression indicates that when all
variables are included, adjusted R square is 98.87%, which means 98.87% sample variance of firm value can be explained by the variables in the regression.
VI. Analysis
Recap on hypothesis
Based on previous studies, considering the dynamics of technological industry and flexible CEO paradigm theory, CEO tenure is assumed to influence firm value directly and has a negative relationship: firm value increases while CEO tenure decreases.
Therefore, the coefficient of tenure, 𝛽., is expected to be statistically significant and shows a negative sign in the regression.
Findings in empirical data
According to the empirical data, with total 286 observations excluding the industry dummies, p-value for 𝛽. is 0.797 when only CEO effect is taken into account while p-value equals to 0.344 when firm effect is added. In both conditions, there is no enough evidence to show 𝛽. is statistically significant. Following this idea, the empirical data cannot support the hypothesis that tenure has direct relationship with firm value.
The coefficient of compensation, 𝛽2, is positive and statistically significant at 1 percent under either pure CEO effect or both CEO plus firm effect, which implies that compensation increase is associated with firm value increase. The coefficients of Rank2,3,4 - 𝛽9, 𝛽=, 𝛽? - show a negative sign, implying a fact that Rank1 induces more firm value compared to Rank2,3,4 when CEO tenure and compensation are unchanged. Because the firm value of Rank1 is under the condition that Rank2,3,4 equals to 0 at the same time. This finding is consistent with the literatures stated before that higher ranking induces likelihood of creating more firm value.
When industry dummies are included, 𝛽. is positive. But in SIC4911 (electric service) industry, there is not sufficient evidence to show 𝛽. is significant, namely, no enough evidence to show the direct relationship between CEO tenure and firm value in electric service industry. In SIC5045 (computer and software) industry, with a p-value of 0.001, 𝛽. is significant, which means it is reasonable to say the relationship between
CEO tenure and firm value can be found but there is a positive relationship that an additional year of CEO tenure corresponds with 3.18% increase in firm value, which is not in line with the hypothesis. But it implies that CEO tenure links much closer to firm value in software industry than technology service industry. This result gives a perspective that different product orientation industries influence the relationship between tenure and firm value, even though these industries all come from broad technology category. And compensation coefficient 𝛽2 is significant at 1 percent in both separate industries. When all industry dummies are included, 𝛽. is significant at 10 percent with a positive sign, showing that one year increase in CEO tenure corresponds with 0.52% increase in firm value.
In all, there is no enough evidence to show a direct relationship between firm value and CEO tenure when only consider CEO and firm effect. Only when industry effect involves, a positive relationship between CEO tenure and firm value can be found. Accordingly, the hypothesis in this paper cannot be supported by empirical data.
Explanation of result
As stated above, the empirical results do not support the hypothesis, there are some reasons that can explain: Initially, the sample size is small, due to a large number of ambiguous data are artificially omitted to improve the accuracy, also, the introduction of specific region restriction (North American) and time period (2006-2015) limits the number of firms on record. Secondly, the time period covers a severe financial crisis occurring in 2008, the market uncertainty increases sharply and the cost of changing CEO is enormous, therefore most of firms are unwilling to replace CEO even though they had planed to do so. In this sense, CEO tenure increases on a large scale within all industries. Not only that, firm value decreases significantly during finance crisis, and experiences a recover period afterwards, thus the data of firm value is also unusual during and after the financial crisis.
Above all, in this paper, less than enough evidence can support the proposition that CEO tenure has a negative relationship with firm value in telecommunication and technological industries within North American during 2006-2015.
VII. Conclusion
In this paper, the central topic of relationship between firm value and CEO tenure is inspired by a newly emerging short-tenure CEO rotating system that is adopted by a successful telecommunication company Huawei. Based upon a large number of prior literatures, hypothesis is given that firm value negatively correlates with CEO tenure in telecommunication, computer software and data processing industries. After analysing the dynamics of telecommunication and technological industries within North American, it is found that an extensive degree of sharing information builds up an intercommunity over these industries.
Including equivalent SIC codes and companies, 286 samples are picked out into a log-linear regression analysis. When excluding industry effect, the coefficient of tenure, 𝛽., is 0.0036 and 0.0139 with p-value of 0.797 and 0.344 respectively, which indicate non-significant results. However, after adding industry effect into regression, 𝛽. is 0.0052 with p-value of 0.0732, displaying a significant positive relationship at 10 percent. Also, it is found that the relationship between tenure and firm value is significantly positive in computer software industry while it is non-significant in electric service industry. A guess that service or product orientation can influence the relationship between CEO tenure and firm value is given in the end of research. To sum up, there is no sufficient evidence to show that the negative relationship between firm value and CEO tenure significantly exists.
Limitations
In this paper, the investigated data covers the period of financial crisis in 2008, which leads to direct impact on both firm value and tenure, thus the data are likely to be abnormal. Therefore, it is necessary to analyse this period specifically. However, the regression does not specify this period of time.
from dataset, so just 286 samples are left. Because many successful telecommunication and related technological companies outside North American are also worth investigating, expanding the regional area and increasing the sample size can increase the power of the study.
Third, when searching the comparable industries with Huawei, a more reliable method of filtering industries should be employed and explained. Alternatively, when different industries are included in the regression, an adjusted indicator of industries can be used to display the similarity of these industries to increase the uniformness of outcomes.
Suggestions
First, more regional areas can be included. Because of tendency of globalisation, telecommunication and technological industries are no longer restricted within a closed area. The connection between different technological companies increases remarkably, and many companies in Europe and Asia get a great achievement as well. Therefore, expanding the regional area, increasing the sample size, can be considered in later studies.
Second, in the empirical data part, it is found that different service or product orientation of industry can drive different result, despite they are all in the same broad category, for instance, even though electric service and computer software industries are all under technology category, the relationship between tenure and firm value is found significant for computer software industry rather than electric service industry. Therefore, a suggestion is that service or production orientation can be also an element in analysis for further studies.
Reference
Al-Matari, E.M., Al-Swidi, A.K., & Fadzil, F.H. (2014). The Measurement of Firm Performance’s Dimensions. Asian Journal of Finance & Accounting, 6(1). Berk, J. & DeMarzo, P. (2014). Corporate Finance (pp. 962-981). Edinburgh: Laura. Bigliardi, B., Dormio, A. I., & Galati, F. (2012). The adoption of open innovation
within the telecommunication industry. European Journal of Innovation Management, 15(1), 27-54.
Brookman, J., & Thistle, P. D. (2009). CEO tenure, the risk of termination and firm value. Journal of Corporate Finance, 15(3), 331-344.
Bushman, R., Dai, Z., & Wang, X. (2010). Risk and CEO turnover. Journal of Financial Economics, 96(3), 381-398.
Cremer, D. D., & Tao, T. (2015). Leadership Innovation: Huawei's rotating CEO system. The European Business Review.
Denis, D. J., & Denis, D. K. (1995). Performance Changes Following Top Management Dismissals. Journal of Finance, 50(4), 1029-1057.
Dikolli, S., Mayew, W., & Nanda, D. (2014). CEO tenure and the performance-turnover relation. Review of Accounting Studies, 19(1), 281-327.
Gabarro, J. J. (1987). The dynamics of taking charge. Boston: Harvard Business School. Hambrick D.C., & Fukutomi G.D.S. (1991). The seasons of a CEO's tenure. Academy
of Management Review, 16, 719–42.
Henderson A.D., Miller D., & Hambrick D.C. (2006). How quickly do CEOs become obsolete? Industry dynamism, CEO tenure, and company performance. Strategic Management, 27, 447–60.
Henderson A.D., Miller D., & Hambrick D.C. (2006). How Quickly Do CEOs Become Obsolete? Industry Dynamism, CEO Tenure, and Company Performance. Strategic Management Journal, 27, 447-460.
Keller, G. (2012). Managerial Statistics 9th
edition (p.306). Global: Joe Sabatino. McClelland P.L., Barker III V.L., & Oh W. (2012). CEO career horizon and tenure:
Future performance implications under different contingencies. Journal of Business Research, 65, 1387-1393.
Mehran, H. (1995). Executive compensation structure, ownership, and firm performance. Journal of Financial Economics, 38(2), 163-184.
Miller D. (1991). Stale in the saddle: CEO tenure and the match between organization and environment. Management Science, 37, 34–52.
Miller D., & Friesen P. H. (1980). Archetypes of Organizational Transition. Science Quarterly, 25, 268-299.
Parrino, R. (1997). CEO turnover and outside succession A cross-sectional analysis. Journal of Financial Economics, 46(2), 165-197.
Rose, N. L., & Shepard, A. (1997). Firm Diversification and CEO Compensation: Managerial Ability or Executive Entrenchment? Journal of Economics, 28(3), 489-514.
Stock, J. H. & Watson, M. W. (2011). Introduction to Econometrics third edition (p.272). Boston: Michael Joyce.
Weng, D.H., & Lin, Z. (2014). Beyond CEO tenure: the effect of CEO newness on strategic changes. Journal of Management, 40(7).
Wu, S., Levitas, E., & Preim, R. L. (2005). CEO tenure and company invention under differing levels of techno- logical dynamism. Academy of Management Journal, 48, 859-873.
Zhang, Y., & Rajagopalan, N. (2010). Once an outsider, always an outsider? CEO origin, strategic change and firm performance. Strategic Management Journal, 31, 334-346.
