國立臺灣大學管理學院財務金融學研究所 碩士論文
Graduate Institute of Finance College of Management
National Taiwan University Master Thesis
董事會成員的飛航經驗對於企業投資決策之影響 Pilot Directors and Corporate Investment
李嘉軒 Chia-Hsuan Li
指導教授：張景宏 博士 Advisor: Ching-Hung Chang, Ph.D.
誌謝
在這段研究的航程中，首先，我要感謝我的指導教授，張景宏老師，一路以來
無私的給予幫助，無論是針對學術領域上的研究建議與寶貴經驗，抑或者是在問題 解決上給予的指導與鼓勵，都使我獲益良多並且有所成長，不僅能在研究中運用所 學，也進一步開拓新領域；每每與景宏老師討論後都會有新的啟發，而能持續向前 邁進。感謝口試委員劉心才教授以及戚永苓教授所提供的建議，使本論文在修訂後 得以更加完整。
在學期間，感謝老師們傳遞各領域的專業知識與做學問的嚴謹態度，這些都成
為了研究中的重要燃料。感謝李哲銓學長、莊博翔學長、孔德蓉學姊、李嘉勛學姊 在研究工具上的協助與學術寫作上的建議，讓我得以從中學習並順利完成這次的 研究。謝謝學長姐以及同學、朋友們的參與，為這趟旅程增添多樣的色彩。
最後，謝謝我摯愛的人們一直以來的陪伴與包容，以及在研究所生涯中給予的 鼓勵與支持，因為有你們，我的碩士學位別具意義。滿滿謝意溢於字裡行間，謹以 最誠摯的祝福，獻給我愛的人與愛我的人；也期許自己保有勇於嘗試又審慎周全的 飛行員精神，繼續啟航。
李嘉軒謹致 2021.02.05
中文摘要
我們經實證研究發現，董事會成員中具飛航經驗的董事數量與比率，皆與企業
不效率的投資行為呈現顯著的負相關性，亦即飛行員董事成員的數量或佔比愈高 時，則企業愈少發生過度投資或投資不足的情形。本論文以飛行員的性格特徵作為 衡量具飛航執照之董事的風險態度指標；縱然飛航經驗意味著勇於挑戰、追求冒險 的態度，但作為董事會成員進行投資決策時，最終係以飛行員董事嚴謹、循規的人 格特質影響大過於對風險的追求，而選擇避免過度投資。
關鍵詞：飛行員、董事、投資決策、過度投資、人格特質。
ABSTRACT
We find empirical evidence that directors with the experience of flying airplanes are associated with keeping firms from inefficient investment, including overinvestment and underinvestment. This thesis proposes using pilot status as a measure of directors’
personality that captures both audacity and prudence. Our results show that pilot directors’
cautiousness and careful planning may be more powerful than sensation seeking in making corporate resolutions. This finding is additionally supported by robustness tests.
Index Terms: Pilot, Director, Investment, Overinvestment, Personality.
CONTENTS
口試委員會審定書 ... #
誌謝 ... i
中文摘要 ... ii
ABSTRACT ... iii
CONTENTS ... iv
Chapter 1 Introduction ... 1
1.1 Backgrounds and Motivations ... 1
1.2 Literature Reviews ... 2
1.3 Contributions ... 5
Chapter 2 Sample Construction and Summary Statistics ... 6
2.1 Sample Construction ... 6
2.2 Variable Measurement ... 8
2.2.1 Measuring Overinvestment ... 8
2.2.2 Measuring Pilot Directors and Other Control Variables ... 8
2.3 Summary Statistics ... 12
Chapter 3 Empirical Results ... 15
3.1 Difference in Means ... 15
3.2 OLS Regressions ... 17
LIST OF TABLES
Table 1 Four perspectives on boards of directors ... 4
Table 2 Variable definition ... 10
Table 3 Descriptive statistics ... 13
Table 4 Correlation coefficient matrix ... 14
Table 5 Difference in means between firms with and without pilot directors ... 16
Table 6 OLS regressions examining the impacts of pilot directors on the corporate investment decisions ... 20
Table 7 Robustness tests for OLS regressions examining the impacts of pilot directors on the different investment expenditure elements with year fixed effects ... 23
Table 8 Robustness tests for OLS regressions examining the impacts of pilot directors on the inefficiency of corporate investment policies ... 25
Table 9 Robustness tests for OLS regressions examining the impacts of pilot directors on the R&D expenditure ... 27
Table 10 Robustness tests for OLS regressions examining the impacts of pilot directors on the corporate investment decisions (excluding aviation) ... 29
Chapter 1 Introduction
1.1 Backgrounds and Motivations
Determining the definition and attributes of personality, Martinussen and Hunter (2010) observe that “personality is a sweeping construct”. This assertion about personality appears less exaggerated if one considers both psychology and reality. It is known that most people construct a perception of reality and then engage in behaviors based on their personality traits. This also applies to decision-makers (King, 2014). For example, people tend to be risk averse or act more conservatively if perceived by either others or themselves as cautious, careful, and organized. Conversely, people tend to have higher risk tolerance and be more innovative when they have certain characteristics, such as being bold, aggressive, and adventurous. (Galasso and Simcoe, 2011; Hirshleifer, Low, and Teoh, 2012).
Boards of directors are one of the most influential decision makers in corporate management (Adams and Ferreira, 2007). Forbes and Milliken (1999) argue that understanding the nature of effective board functioning is among the most important areas of management research. Boards select managers, decide the compensation policy of managers (Fama and Jensen, 1983) and are involved in almost every significant strategic decision, including capital investment decisions. The investment policies of a company strongly affect its financial performance. These decisions reflect decision makers’ risk
preference on a corporate level, this thesis chooses directors from S&P 500 companies who used to be pilots or are holding pilot licenses issued by the Federal Aviation Administration (FAA) as the main factor of the independent variable. It then examines whether a company has the tendency to overinvest if there is/are pilot director(s) whose attitudes and personality traits affect decision making (Hunter, 2002) on its board.
Researchers often describe pilots as sensation seekers, who pursue novelty and stimulation (Kish and Donnenwerth, 1969; Mittelstaedt, Grossbart, Curtis, and DeVere, 1976; J. Lopez-Bonilla and L. Lopez-Bonilla, 2012). Though gratified by the danger and thrills of flying, pilots must be careful conformists who are able to abide by rules meticulously (Retzlaff and Gibertini, 1987; Callister et al., 1999; Hormann and Maschke, 1996). Based on the studies on the personality traits of pilots and the importance of the board of directors to company’s development, we expect pilot directors’ audacity as well as prudence to affect their corporate investment activities. The purpose of this thesis is to observe the net effect of both personality components on the investment decision and make an important contribution to the field using an empirical study.
1.2 Literature Reviews
This thesis is related to two strands of literature. First, it is related to the body of empirical literature that investigates the effect of personal characteristics on firms’
investment decisions (Malmendier and Nagel, 2011; Malmendier, Tate, and Yan, 2011;
Cronqvist, Makhija, and Yonker, 2012; Graham, Harvey, and Puri, 2013; Davidson, Dey, and Smith, 2015). The reason for choosing pilots as the indicator is that numerous studies have been conducted on pilots (Martinussen and Hunter, 2010). The Temperament Structure Scale (TSS), which is a multidimensional personality questionnaire with 234
items, has been shown to be an effective measurement for pilot selection. The TSS parses many personality characteristics into eight dimensions to analyze an individual’s interpersonal behavior, emotions and feeling, and work and achievement (Hormann, 1996). Thus, it is possible to capture the desire for both adventure and caution of directors who are pilots (hereafter “pilot directors”) by adopting their disclosed experiences or preferences for piloting (Sunder, Sunder, and Zhang, 2017; Callister et al., 1999). This will enable determination of whether and how pilot directors are related to overinvestment or underinvestment.
Second, this thesis builds on studies of boards of directors’ function and influence on corporate policies (Chi, Tzu, Liao and Huang, 2017). Boards of directors are generally regarded as the principals of the company’s shareholders, and serve as the supervisors inspecting the performance of managers and protecting the investors (Henn, 1974).
According to Table 1 (Zahra and Pearc, 1989, p. 293), many studies have shown that boards of directors are able to exert direct influence on the firm’s investment activities whether from the perspectives of legalism, resource dependence, class hegemony, or agency theory. In addition to the supervisory role mentioned above, boards of directors often give suggestions to managers (Brickley and Zimmerman, 2010), and thus have influence over corporate policies.
Table 1
Four perspectives on boards of directors
PERSPECTIVES
Legalistic Resource Dependence Class Hegemony Agency Theory
Board Role
Represent and protect shareholders' interest.
Manage the corporation without interference in day-to-day operations.
Extract resources vital to company performance.
Serve a boundary spanning role and enhance organizational legitimacy.
Perpetuate the power and control of the ruling capitalist elite over social and economic institutions.
Reduce transaction cost.
Monitor actions of agents to ensure their efficiency and protect principals’
interests.
Make strategic decision.
Theoretical
Origins Corporate law Organizational Theory &
Sociology Marxist Sociology Economics & Finance
Representative Studies
Berle & Means (1968) Chaganti, et al. (1985) Mace (1971)
Molz (1988) Williamson (1964)
Pfeffer (1972) Pfeffer (1973)
Pfeffer & Salancik (1978) Proven (1980)
Zald (1967)
Domhoff (1969) Mills (1956) Ratcliff (1980)
Baysinger & Butler, (1985) Fama & Jensen (1985) Kosnik (1987)
Empirical
Support Moderate Strong Limited Moderate
Note. Reprinted from “Boards of Directors and Corporate Financial Performance: A Review and Integrative Model”, by Zahra, S. A., and Pearce, J. A., 1989, Journal of Management, 15(2), p. 293.
1.3 Contributions
The literature on pilots focuses mainly on the extraverted sides of pilots’ personality (Sunder et al., 2017), exploring traits such as their sociability, expressiveness, aggressiveness, ambitiousness, and audacity. Studies in psychology have also shown that pilots are more accepting of risk, which in turn leads to pilot CEOs being more creative and inclined to invest more in research and development (R&D) spending (Carretta and Tee, 1994; Zuckerman, 1971; Zuckerman, S. B. Eysenck, and H. J. Eysenck, 1978; King, 2014). However, it is worth noting that although pilots are found to be risk seekers, being prudent and principle-minded is also essential to successfully becoming a pilot (Hoermann and Maschke, 1993), which may result in pilot directors taking conservative measures. Furthermore, most of the studies in corporate finance on boards of directors place greater emphasis on the influence of directors’ expertise, experience, and on the independence of boards rather than directors’ personality traits. Our study adds to the large literature discussing the effects of pilot personality characteristics. We conjecture that pilot directors have the inclination for both overinvesting and underinvesting since they possess both risk-seeking and risk-averse personality traits (Martinussen and Hunter, 2010). As a result, we apply pilots’ character traits as an alternative to infer pilot directors’
risk preferences.
The remainder of this study is organized as follows. Chapter 2 describes the sample selection and reports summary statistics. Chapter 3 presents our empirical results.
Chapter4 provides robustness tests. Chapter 5 concludes this study.
Chapter 2 Sample Construction and Summary Statistics
This chapter describes the sample and provides the definition and summary statistics of the main variables.
2.1 Sample Construction
The sample examined in this thesis is constructed mainly from three resources taken from different domains. First, the historical constituent lists for the S&P 500 Index are taken from the Compustat database, as are the financial statement data for 2000 to 2010 for these firms. Second, the BoardEX database provides information about millions of global business leaders. From this database, we are able to identify the boards of directors for S&P 500 companies in 2000 and then collect their company name, company ID, director name, director ID, and a partial of the directors’ year of birth. The reason for using S&P 500 directors listed in 2000 instead of spanning several years (from 2000 to 2010) is that fixing on the beginning year allows us to obtain a constant analysis over time.
Third, we use the airmen inquiry page on the Federal Aviation Administration (FAA) computer system to determine whether directors possess a pilot certification recognized by the FAA. This webpage provides the name, certificate level, and rating information for all pilots. The address and medical information for pilots willing to declare may be found here as well. Airmen who do not want their addresses released are not included in the downloadable file due to privacy policies. Though the information in that file is updated every month, there are no way for us to ensure the coverage of all the pilots in the database.
To address this issue, we wrote a web crawler using Python to confirm whether any of
the more than 59,400 directors’ names are in the airmen database instead of downloading the dataset directly. The web crawler records all the airmen certification information shown in the search results as long as the directors are authorized with any rating of piloting, including student pilot, private pilot, commercial pilot, or mechanic. To better identify the airmen corresponding to those on the boards of directors list, we add another condition. Only airmen older than sixty years old are defined as pilot directors in our study since Bloomberg notes that about 64% of directors from S&P 500 companies are over sixty years old and the average age among S&P 500 boards is sixty-two. In light of the law of large numbers, we are still able to obtain a picture of the effect on the outcome although this method is not very accurate.
Finally, we merged the pilot directors with financial statement data of firms from Compustat to explore the firms’ investment activities. The resulting sample covers 4,701 firm-years between 2000 and 2010, including 2,866 boards with at least one pilot director and 1,835 otherwise.
2.2 Variable Measurement
2.2.1 Measuring Overinvestment
We obtain the investment expenditures of the firms from the Compustat database.
This database provides annual financial statement information. The amount of investment is constructed based on capital expenditure, acquisitions, and increase in investment.
Overinvestment, by definition, is investment in excess of the optimal investment. It is the portion of the reported investment that is greater than the optimal investment (Fu, 2010).
The firms are then clustered by their two-digit Standard Industrial Classification (SIC) codes for each fiscal year in order to calculate the industry average and median by fiscal years as the optimal investment. Because it is difficult for us to indicate all individual firms’ optimal investment, the average or median measured by industry and fiscal years serves as an appropriate benchmark to assess overinvestment size. This approach also considers both the industrial characteristics and the business cycle across different years.
Thus, the degree of overinvestment in a firm-year is measured by subtracting the industry average or median from the investment expenditure. It is denoted OverI_Avg and OverI_Md in this thesis.
2.2.2 Measuring Pilot Directors and Other Control Variables
We construct our explanatory variables, which are mainly comprised of pilot directors, through the Compustat database. Pilot information is obtained from the FAA online airmen inquiry website. It is widely known that pilots in the United States are required to obtain a pilot certificate regulated by the FAA, a branch of the U.S.
Department of Transportation (USDOT) (Sunder et al., 2017). In this thesis, there are two types of explanatory variables used to measure the influence of pilot directors, PilotNum
and PilotRatio. The former is the natural logarithm of (one plus) the raw number of the pilot directors; the latter is the number of pilot directors divided by total number of directors. Measures above rely on the fact that relative and absolute numbers have different power.
This thesis is related to literature on corporate finance through its connection with the measurement of overinvestment. Studies have shown that certain variables should be controlled when estimating overinvestment (Goh, 1966). Therefore, to better understand the relationship between the dependent variables examined, we add a set of controls that have been identified as important determinants of firm investment: firm size, Tobin’s Q, debt leverage, cash holdings, and R&D expenses (Fu, 2010; He and Tian,2013; Jiang, 2016; Khémiri and Noubbigh, 2020). We add to these studies by bringing the board size into the control variables in order to compare the difference in impact stemming from the absolute count and relative ratio of pilot directors among the board. In addition, we also take board independence as one of our controls in order to manage the monitoring power from independent directors.
As mentioned above, we control for a vector of time-varying firm characteristics that are influential determinants of investment activities. We compute all variables for each fiscal year of a firm. Firm control variables are firm size (the natural logarithm of assets), Tobin’s Q (the market value of assets divided by the book value of assets), leverage (the debt-to-equity ratio), cash (the cash holding divided by the book value of assets), and R&D spending (the R&D expense divided by the book value of assets). We also include
Table 2
Variable definition
This table describes the definition of the dependent, independent, and control variables.
Variable Definition
Measures of investment and overinvestment
RawI Standardized investment expenditure
= (Capital expenditures + Acquisitions + Increase in investments) / Total assets OverI_Avg Amount of overinvestment in each sample firm using industry average sorted
by two-digit SIC codes and fiscal years
= RawI – Industry average investment expenditures by fiscal years (set to zero if negative)
OverI_Md Amount of overinvestment in each sample firm using industry median sorted by two-digit SIC codes and fiscal years
= RawI – Industry median investment expenditures by fiscal years (set to zero if negative)
Measures of pilot directors and other variables
PilotNum Log of the total number of pilot directors in each sample firm
= ln (#Pilot directors + 1)
PilotRatio Ratio of pilot directors to the board of directors in each sample firm
= #Pilot directors / # Total directors
FirmSize Log of total assets measured at the start of the year
= ln (Total assets) TobinQ Market to book ratio
= (Market value of equity + Total assets - Book value of equity - Deferred taxes) / Total assets
Leverage Debt-to-equity ratio
= (Short-term debt + Long-term debt) / Book value of equity Cash Standardized value of assets that can be easily converted into cash
= Cash and short-term investments / Total assets RD Standardized research and development expenditure
= Research and development expense / Total assets
BoardSize Total number of directors on the board of each sample firm
= #Total directors
BoardIndep Degree of board independence of each sample firm
= #Independent directors / #Total directors Note.
*Total assets refer to the book value of assets.
*Capital expenditure, acquisitions, increase in investment, deferred taxes, short-term debt, long-term debt, cash and short-term investments, and research and development expenditure are set to zero if missing.
Table 2 (continued)
Variable Definition
Measures of dependent variables in robustness tests
OverCAPX_Avg Overinvestment in capital expenditures (CAPX) based on industry average
= CAPX – Industry average CAPX by fiscal years (set to zero if negative)
OverCAPX_Md Overinvestment in capital expenditures (CAPX) based on industry median
= CAPX – Industry median CAPX by fiscal years (set to zero if negative)
OverAQC_Avg Overinvestment in acquisitions (AQC) based on industry average
= AQC – Industry average AQC by fiscal years (set to zero if negative)
OverAQC_Md Overinvestment in acquisitions (AQC) based on industry median
= AQC – Industry median AQC by fiscal years (set to zero if negative)
OverIVCH_Avg Overinvestment in increase in investments (IVCH) based on industry average
= IVCH – Industry average IVCH by fiscal years (set to zero if negative)
OverIVCH_Md Overinvestment in increase in investments (IVCH) based on industry median
= IVCH – Industry median IVCH by fiscal years (set to zero if negative)
InefficI_Avg Amount of investment expenditure above or below the industry average
= | RawI – OverI_Avg |
InefficI_Md Amount of investment expenditure above or below the industry median
= | RawI – OverI_Md |
OverRD_Avg Overinvestment in R&D spending based on industry average
= RD – Industry average RD by fiscal years (set to zero if negative)
OverRD_Md Overinvestment in R&D spending based on industry median
= RD – Industry median RD by fiscal years (set to zero if negative)
2.3 Summary Statistics
To minimize the effect of outliers, all control variables are winsorized at the 1st and 99th percentiles. Panel A of Table 3 reports descriptive statistics for the major variables used in this study. Based on our sample, about 25% (75th percentile) of examined firm- years include more than one pilot director (exponent of 1.099 and minus one approximately equals to two) and the overall pilot-director ratio is 8.5%. On average, a firm has book value assets of $11.27 (exponent of 9.33) billion, a Tobin’s Q of 1.9, a debt- to-equity ratio of 1.2, a cash-to-assets ratio of 12.6%, and an R&D-to-assets ratio of 2.5%.
The average board has twelve directors, ranging from two to twenty-three, with most boards having between ten to fourteen members, and 72.5% of the directors are independent. Panel B presents the industry distribution of the examined sample. As shown, most of the examined S&P 500 firms belong to the manufacturing industry. The industry with the highest average ratio of pilot directors is “Mining” at about 76.05%, followed by
“Finance, Insurance and Real Estate”, and “Transportation, Communications, Electric, Gas and Sanitary Services.” Table 4 tabulates the results of the Pearson correlation test.
The Pearson correlation matrix shows that there is no problem of multicollinearity between the control variables since all the estimated coefficients have values of less than 0.80 (Khémiri and Noubbigh, 2020).
Table 3
Descriptive statistics
Panel A of this table reports the summary statistics for the variables used in the empirical analysis.
These variables are constructed based on the sample of S&P 500 firms during a period from 2000 to 2010. Panel B presents the industry distribution of the examined sample.
Panel A: Descriptive statistics
Variables Mean STD P25 Median P75 N
PilotNum 0.580 0.524 0.000 0.693 1.099 4,701
PilotRatio 0.085 0.090 0.000 0.077 0.143 4,701
FirmSize 9.330 1.474 8.198 9.255 10.243 4,701
TobinQ 1.919 1.496 1.115 1.468 2.199 4,701
Leverage 1.205 3.776 0.262 0.614 1.329 4,701
Cash 0.126 0.141 0.029 0.071 0.175 4,701
RD 0.025 0.043 0.000 0.000 0.032 4,701
BoardSize 12.016 2.944 10.000 12.000 14.000 4,701
BoardIndep 0.725 0.189 0.632 0.778 0.867 4,701
Panel B: Industry distribution SIC
codes Industry title Firm-years Firm-years with
pilot director(s) Ratio
01-09 Agriculture, Forestry and Fishing 0 0 0
10-14 Mining 167 127 76.05%
15-17 Construction 46 21 45.65%
20-39 Manufacturing 2,277 1,344 59.03%
40-49 Transportation, Communications, Electric, Gas and Sanitary Services
567 409 72.13%
50-51 Wholesale Trade 94 63 67.02%
52-59 Retail Trade 423 229 54.14%
60-67 Finance, Insurance and Real Estate 637 467 73.31%
Table 4
Correlation coefficient matrix
This table reports the correlations of variables for firms listed on S&P 500 between 2000 and 2010. Correlation is significant at the 5% level.
RawI OverI_Avg OverI_Md PilotNum PilotRatio Firm
Size TobinQ Leverage Cash RD Board Size
Board Indep
RawI 1.000
OverI_Avg 0.971* 1.000
OverI_Md 0.966* 0.985* 1.000
PilotNum -0.053* -0.056* -0.057* 1.000
PilotRatio -0.040* -0.041* -0.042* 0.934* 1.000
FirmSize -0.011 -0.041* -0.041* 0.124* 0.011 1.000
TobinQ 0.072* 0.074* 0.077* -0.046* -0.005 -0.281* 1.000
Leverage -0.007 -0.020 -0.017 0.030* 0.004 0.254* -0.109* 1.000
Cash 0.112* 0.134* 0.135* -0.135* -0.071* -0.267* 0.366* -0.067* 1.000
RD 0.075* 0.095* 0.103* -0.122* -0.061* -0.307* 0.330* -0.129* 0.541* 1.000
BoardSize -0.076* -0.088* -0.089* 0.321* 0.100* 0.513* -0.194* 0.137* -0.305* -0.282* 1.000
BoardIndep -0.002 -0.003 -0.001 -0.143* -0.151* 0.059* -0.082* -0.031* 0.027 0.017 0.015 1.000
Chapter 3 Empirical Results
In this chapter, we explore whether and how pilot directors affect corporate investment activities.
3.1 Difference in Means
Table 5 reports the difference in means between firms with and without pilot directors. To compare firms with and without pilot directors by fiscal year, we classify the sample based on the directors’ pilot credentials and report the observed numbers and means of all variables.
Then, the difference in means with p-value is performed for each variable between the two groups. Surprisingly, about 60% (2,866 divided by 4,701 and multiply 100%) of our sample include at least one pilot director. On average, firms with pilot directors have 1.59 (exponent of 0.951 and minus one) pilot directors and 14% of directors awarded an FAA certificate on the boards. The differences in the means of RawI, OverI_Avg and OverI_Md between firms with and without pilot directors are all statistically significant at the 1% level. Thus, we find that firms without pilot directors tend to have higher positive abnormal investment. The standardized investment expenditure of firms without pilot directors is approximately 30%
(0.046 divided by 0.137 and multiply 100%) higher than that of other firms.
Table 5
Difference in means between firms with and without pilot directors
This table provides t-tests results (Wilcoxon–Mann–Whitney tests) conducted to test for differences between the means of firms with and without pilot directors on the board of directors. The dummy variables of firms without any pilot director are set to zero. In contrast, others are set to one. ***, **
and * denote significance at the 1%, 5% and 10% level, respectively.
Valuables
Without pilot director (dummy = 0)
With pilot director
(dummy = 1) Mean in
Difference p-value
N Mean N Mean
RawI 1,835 0.182 2,866 0.137 0.046*** 0.002
OverI_Avg 1,835 0.079 2,866 0.047 0.032*** 0.001
OverI_Md 1,835 0.109 2,866 0.066 0.044*** 0.001
PilotNum 1,835 0.000 2,866 0.951 -0.951*** 0.000
PilotRatio 1,835 0.000 2,866 0.140 -0.140*** 0.000
FirmSize 1,835 9.170 2,866 9.431 -0.261*** 0.000
TobinQ 1,835 1.966 2,866 1.888 0.078* 0.080
Leverage 1,835 1.125 2,866 1.257 -0.132 0.241
Cash 1,835 0.142 2,866 0.115 0.027*** 0.000
RD 1,835 0.030 2,866 0.022 0.007*** 0.000
BoardSize 1,835 11.083 2,866 12.614 -1.531*** 0.000
BoardIndep 1,835 0.749 2,866 0.709 0.040*** 0.000
3.2 OLS Regressions
To assess how pilot directors affect corporate investment decisions, this thesis estimates the following models using the ordinary least squares (OLS) method:
Ii,t = ⍺ + βPiloti,t + γ1FirmSizei,t + γ2TobinQi,t + γ3Leveragei,t + γ4Cashi,t + γ5RDi,t + γ6BoardSizei,t + γ7BoardIndepi,t + ⍵Yeart + εi,t,
where i indexes firm and t indexes time. The dependent variable, I, captures a firm’s investment inclination: RawI (the standardized value of investment expenditure), Over_Avg (overinvestment based on the industry average by fiscal years), and Over_Md (overinvestment based on the industry median by fiscal years). We examine the mean and the median because the mean considers the data as a whole but is liable to be affected by outliers while the median is barely affected. Further, the median is more stable but less sensitive to outliers. The main independent variable, Pilot, refers to PilotNum or PilotRatio in different regressions and is measured for firm i over its fiscal year t. Other variables control for firm and industry characteristics that can affect a firm’s investment decisions, as discussed in Section 2.2.2. Year indicates year fixed effects and its coefficient, ⍵, is a dummy.
Table 6 presents the results for the relation between pilot directors and corporate investment activities as estimated via OLS models. Table 6 begins with regressions without year fixed effects in Panel A. We find that the relation between the investment amount and pilot
more pilot directors on a board, the lower the investment expenditure and the lower the probability of a firm overinvesting. Between these two alternative independent variables, we prefer PilotRatio as reported in the even-numbered columns to PilotNum for the following two reasons. First, most regulations of the board voting process are based on the relative ratio instead of the absolute number. Second, ratios have higher volatility and therefore make the effect be more observable.
Next, year fixed effects are added to absorb aggregate time effect. Results are reported in Panel B of Table 6. The coefficient estimates of PilotNum and PilotRatio continue to be negative and become significant at the 5% level with a slightly larger magnitude. As mentioned previously, since pilot directors are bold, aggressive and adventurous, many studies find that they have a tendency to take risks. However, the other side of their personality is also an issue worth attention and investigation: pilots’ cautiousness, meticulousness, and self-discipline may lead them to perform more conservatively.
In regard to the control variables, firms that are larger, with higher Tobin’s Q, lower leverage, and more cash tend to invest more than the optimal expenditure. As can be seen, the estimated coefficients on FirmSize are positive and significant at the 1-5% level and those of TobinQ are also positive, showing that firms with larger size, lower cost of tangible assets,
higher market value, or higher value of intangible assets have more investment opportunities and a slight tendency to overinvest. Higher leverage, which means higher debt-to-equity ratio, is associated with tighter financial constraints and higher default risk for a firm. Thus, overinvestment rarely occurs in firms. Cash holdings are strongly and positively related to overinvestment since they allow firms to obtain immediate funding easily, consistent with the literature. Furthermore, the coefficient estimates of BoardSize are negative and statistically significant at the 1% level since a larger board of directors is expected to have more monitoring function and power. Board independence is also negatively related to overinvestment.
In sum, according to the OLS regressions results and the psychology literature, the net effect of pilot directors’ personality traits on firms’ investment tendency is to reduce overinvestment. This finding suggests that overinvestment is less likely to happen if the board of directors includes one or more pilot directors. In addition, we found that investment expenditure goes up while R&D spending increases. The advanced examination related to R&D spending associated with pilot directors is performed in chapter 4.
Table 6
OLS regressions examining the impacts of pilot directors on the corporate investment decisions
This table presents the results from OLS regressions of investment expenditure on pilot directors.
Columns 1 and 2 report the estimates taking the raw value of investment expenditure as the dependent variable. Columns 3 and 4 report the estimates taking the amount of overinvestment based on the industry average for each fiscal year as the dependent variable. Columns 5 and 6 report the estimates taking the amount of overinvestment based on the industry median for each fiscal year as the dependent variable. PilotNum, which is the independent variable in columns 1, 3 and 5, is the number of pilot directors on the board of directors. PilotRatio, which is the independent variable in columns 2, 4 and 6, is the number of pilot directors divided by the total number of directors on the board. Variable definitions are provided in Table 2. Regressions in Panel A are without year fixed effects. Regressions in Panel B include year fixed effects. The t-statistics are reported in parentheses. ***, ** and * denote significance at the 1%, 5% and 10% level, respectively.
Panel A: Without year fixed effects
(1) (2) (3) (4) (5) (6)
RawI RawI OverI_Avg OverI_Avg OverI_Md OverI_Md
PilotNum -0.025* -0.017* -0.022*
(-1.74) (-1.70) (-1.74)
PilotRatio -0.154* -0.102* -0.140**
(-1.94) (-1.88) (-1.97)
FirmSize 0.021*** 0.021*** 0.008* 0.008* 0.010* 0.010*
(3.55) (3.55) (1.92) (1.92) (1.96) (1.95)
TobinQ 0.013** 0.013** 0.006 0.006 0.008* 0.008*
(2.48) (2.48) (1.62) (1.62) (1.76) (1.76)
Leverage -0.000 -0.000 -0.001 -0.001 -0.000 -0.000
(-0.17) (-0.16) (-0.54) (-0.53) (-0.27) (-0.26) Cash 0.287*** 0.287*** 0.240*** 0.239*** 0.295*** 0.294***
(4.62) (4.61) (5.67) (5.66) (5.30) (5.29)
RD 0.172 0.172 0.170 0.170 0.343* 0.342*
(0.86) (0.86) (1.25) (1.25) (1.91) (1.91)
BoardSize -0.010*** -0.011*** -0.006*** -0.007*** -0.008*** -0.009***
(-3.43) (-3.90) (-3.00) (-3.45) (-2.97) (-3.42)
BoardIndep -0.021 -0.021 -0.016 -0.017 -0.016 -0.017
(-0.54) (-0.57) (-0.63) (-0.65) (-0.46) (-0.49)
Year FE No No No No No No
N 4,701 4,701 4,701 4,701 4,701 4,701
R2 0.019 0.019 0.023 0.023 0.023 0.024
Adjusted R2 0.02 0.02 0.02 0.02 0.02 0.02
Table 6 (continued)
Panel B: With year fixed effects
(1) (2) (3) (4) (5) (6)
RawI RawI OverI_Avg OverI_Avg OverI_Md OverI_Md
PilotNum -0.030** -0.019* -0.025*
(-2.00) (-1.81) (-1.87)
PilotRatio -0.179** -0.110** -0.152**
(-2.17) (-1.96) (-2.06)
FirmSize 0.021*** 0.021*** 0.008** 0.008** 0.011** 0.011**
(3.62) (3.62) (1.98) (1.98) (2.01) (2.01)
TobinQ 0.012** 0.012** 0.006 0.006 0.008* 0.008*
(2.19) (2.18) (1.55) (1.54) (1.66) (1.65)
Leverage -0.001 -0.001 -0.001 -0.001 -0.001 -0.001
(-0.27) (-0.26) (-0.60) (-0.59) (-0.34) (-0.34) Cash 0.307*** 0.307*** 0.245*** 0.245*** 0.301*** 0.301***
(4.85) (4.85) (5.69) (5.69) (5.32) (5.32)
RD 0.133 0.133 0.152 0.152 0.323* 0.322*
(0.66) (0.66) (1.11) (1.11) (1.79) (1.79)
BoardSize -0.010*** -0.011*** -0.006*** -0.007*** -0.008*** -0.009***
(-3.42) (-3.97) (-3.05) (-3.55) (-3.02) (-3.52)
BoardIndep -0.008 -0.008 -0.021 -0.021 -0.024 -0.024
(-0.17) (-0.16) (-0.64) (-0.63) (-0.55) (-0.54)
Year FE Yes Yes Yes Yes Yes Yes
N 4,701 4,701 4,701 4,701 4,701 4,701
R2 0.022 0.022 0.025 0.025 0.026 0.026
Adjusted R2 0.02 0.02 0.02 0.02 0.02 0.02
Chapter 4 Robustness Tests
We conduct a set of robustness tests for baseline results and discuss the details of these tests in this chapter.
The baseline tests in this thesis show a negative relation between pilot directors and overinvestment. We then proceed to decompose investment expenditure into CAPX, AQC, and IVCH, which represent capital expenditures, acquisitions, and the increase in investments, respectively. By U.S. and Canadian GAAP definition, an increase in investments represents funds used to increase a company’s long-term investments, such as long-term receivables, or investments in unconsolidated subsidiaries. OLS regressions are conducted to catch a glimpse of which investment element is most affected by pilot directors. The results are delivered in Table 7. Although overCAPX_Avg and overCAPX_Md increase with PilotRatio increases, the estimated coefficients on overIVCH_Avg and overIVCH_Md are negative and statistically significant at the 1% level. In addition, overAQC_Avg and overAQC_Md are also negative though they are not significant. The positive and significant coefficient between pilot director ratio and overinvestment in capital investments implies that pilot directors’ aggressiveness and ambition may drive them to overinvest. Nevertheless, the relation between pilot directors and overinvestment is negative overall. These circumstances imply that the main deduction in section 3.2 derives from the acquisitions and increase in investments rather than capital expenditures. In other words, pilot directors do not restrain the capital expenditures which boosts long-term growth but they are concerned with the amount of increase in investments.
Table 7
Robustness tests for OLS regressions examining the impacts of pilot directors on the different investment expenditure elements with year fixed effects
This table presents the results from OLS regressions regarding to the three elements of investment expenditure on pilot directors. Investment expenditure is decomposed into CAPX, AQC and IVCH, which represent capital expenditures, acquisitions, and the increase in investments, respectively. For better interpretation, we magnify IVCH by ten thousand times. The dependent variables in columns 1 and 2 are the amount of CAPX deducted by the industry average and median by fiscal year, respectively.
Variables are set to zero if negative. The same methodology is applied to AQC in columns 3 and 4, and IVCH in columns 5 and 6, to get the overinvestment amount in CAPX, AQC and IVCH. PilotRatio, the independent variable, is the number of pilot directors divided by the total number of directors on the board. Variable definitions are provided in Table 2. Regressions include year fixed effects. The t- statistics are reported in parentheses. ***, ** and * denote significance at the 1%, 5% and 10% level, respectively.
(1) (2) (3) (4) (5) (6)
overCAPX _Avg
overCAPX _Md
overAQC _Avg
overAQC _Md
overIVCH _Avg
overIVCH _Md PilotRatio 385.936** 410.860* -128.023 -119.751 -0.489*** -0.489***
(1.99) (1.81) (-0.61) (-0.51) (-3.85) (-3.85)
FirmSize 347.937*** 435.380*** 207.612*** 235.673*** 0.049*** 0.049***
(25.22) (26.84) (13.95) (14.05) (5.38) (5.38)
TobinQ 67.765*** 80.536*** 20.683 23.843 -0.006 -0.006
(5.39) (5.45) (1.53) (1.56) (-0.67) (-0.67)
Leverage -20.054*** -27.021*** -7.259 -8.149 -0.002 -0.002 (-4.42) (-5.06) (-1.48) (-1.48) (-0.67) (-0.67) Cash -368.410** -481.133*** -317.092** -470.089*** 0.075 0.075
(-2.49) (-2.77) (-1.99) (-2.62) (0.77) (0.77)
RD 2357.295*** 3181.859*** 2652.210*** 3414.450*** -0.206 -0.206
(5.01) (5.75) (5.22) (5.97) (-0.67) (-0.67)
BoardSize 16.889** 18.272** 5.565 1.813 -0.011*** -0.011***
(2.52) (2.32) (0.77) (0.22) (-2.60) (-2.60)
BoardIndep -3.938 10.050 -187.518 -185.078 -0.142* -0.142*
(-0.03) (0.07) (-1.51) (-1.33) -0.489*** -0.489***
Year FE Yes Yes Yes Yes Yes Yes
N 4,701 4,701 4,701 4,701 4,701 4,701
R2 0.168 0.183 0.060 0.060 0.014 0.014
Next, we aim to capture the net effect of pilot directors on firms’ overall investment efficiency. Investment efficiency is measured by subtracting the industry average/median by fiscal year from the individual firm’s investment expenditure, and then taking the absolute value. It is assumed that the higher the amount of abnormal investment, the worse the investment efficiency. Table 8 shows the OLS results from regressing pilot directors on firms’
investment efficiency, which is measured via InefficI_Avg in Panel A and InefficI_Md in Panel B. Comparing across columns, the results echo the finding in this thesis that pilot directors keep firms from overinvesting and underinvesting. Furthermore, they reduce inefficient investment outcomes as well.
Table 9 presents the OLS results from regressing R&D spending on PilotNum and PilotRatio, respectively. As noted earlier, we also include year fixed effects in the regressions
to account for systematic variations in dependent variables across years. overRD_Avg is the R&D spending above the average measured by fiscal years and industry, defined at the two- digit SIC level. overRD_Md is constructed in the same way, in which the industry median is used to substitute the industry average. The estimated coefficient on both of the independent variables is negative as expected, supporting the results in the Panel B of Table 6 while all of the other controls remain the same. R&D spending and some control variables, such as firm size, leverage, board size and board independence, move in the opposite direction. This may result from such firms’ spending being subject to greater scrutiny by the financial supervision authorities or the firms’ boards of directors. Consistent with the literature, we find that the higher the Tobin’s Q (i.e., more growth opportunities), the higher the R&D spending. Further, firms engage more in R&D when they have a larger amount of cash and cash equivalents.
Table 8
Robustness tests for OLS regressions examining the impacts of pilot directors on the inefficiency of corporate investment policies
This table presents the OLS regressions results of the effect of pilot directors on investment inefficiency. The measures of inefficiency are computed with the industry average and the industry median, respectively. InefficI_Avg, the independent variable in Panel A, is the absolute value of investment spending deducted by the industry average by fiscal year, and therefore represents the amount above or below the average. The similar is applied to the independent variable InefficI_Md in Panel B, which is deducted by the industry median instead. PilotNum, the independent variable in columns 1, 2, 5 and 6, is the number of pilot directors on the board of directors. PilotRatio, the independent variable in columns 3, 4, 7 and 8, is the number of pilot directors divided by the total number of directors on the board. Variable definitions are provided in Table 2. Regressions in odd-numbered columns include year fixed effects while in even- numbered don’t. The t-statistics are reported in parentheses. ***, ** and * denote significance at the 1%, 5% and 10% level, respectively.
Panel A: Measures of inefficiency based on the industry average
(1) (2) (3) (4)
InefficI_Avg InefficI_Avg InefficI_Avg InefficI_Avg
PilotNum -0.019* -0.026**
(-1.72) (-2.19)
PilotRatio -0.112* -0.142**
(-1.80) (-2.19)
FirmSize 0.004 0.004 0.004 0.004
(0.79) (0.92) (0.78) (0.92)
TobinQ 0.007* 0.006 0.007* 0.006
(1.66) (1.32) (1.66) (1.32)
Leverage -0.000 -0.001 -0.000 -0.000
(-0.19) (-0.33) (-0.18) (-0.32)
Cash 0.263*** 0.274*** 0.262*** 0.274***
(5.36) (5.51) (5.36) (5.51)
RD 0.268* 0.249 0.268* 0.250
(1.70) (1.58) (1.70) (1.58)
BoardSize -0.008*** -0.008*** -0.009*** -0.009***
(-3.35) (-3.34) (-3.83) (-3.95)
BoardIndep 0.005 0.009 0.005 0.009
Table 8 (continued)
Panel B: Measures of inefficiency based on the industry median
(5) (6) (7) (8)
InefficI_Md InefficI_Md InefficI_Md InefficI_Md
PilotNum -0.024* -0.028**
(-1.83) (-2.00)
PilotRatio -0.149** -0.165**
(-2.04) (-2.17)
FirmSize 0.011** 0.011** 0.011** 0.011**
(2.00) (2.06) (1.99) (2.05)
TobinQ 0.008* 0.008 0.008* 0.008
(1.72) (1.54) (1.72) (1.53)
Leverage -0.002 -0.002 -0.002 -0.002
(-0.96) (-1.07) (-0.95) (-1.06)
Cash 0.307*** 0.317*** 0.306*** 0.317***
(5.36) (5.46) (5.35) (5.46)
RD 0.255 0.229 0.254 0.228
(1.38) (1.24) (1.38) (1.23)
BoardSize -0.008*** -0.008*** -0.009*** -0.009***
(-2.98) (-3.01) (-3.45) (-3.55)
BoardIndep -0.011 -0.016 -0.012 -0.016
(-0.32) (-0.36) (-0.35) (-0.35)
Year FE No Yes No Yes
N 4,701 4,701 4,701 4,701
R2 0.022 0.026 0.022 0.026
Adjusted R2 0.02 0.02 0.02 0.02
Table 9
Robustness tests for OLS regressions examining the impacts of pilot directors on the R&D expenditure
This table presents the results from OLS regressions of R&D spending on pilot directors. Columns 1 and 2 report the estimates taking the raw value of R&D spending as the dependent variable. Columns 3 and 4 report the estimates taking the amount of R&D spending exceeding the industry average for each fiscal year as the dependent variable. Columns 5 and 6 report the estimates taking the amount of R&D spending exceeding the industry median for each fiscal year as the dependent variable. PilotNum, the independent variable in columns 1, 3 and 5, is the number of pilot directors on the board of directors.
PilotRatio, the independent variable in columns 2, 4 and 6, is the number of pilot directors divided by the total number of directors on the board. Variable definitions are provided in Table 2. Regressions include year fixed effects. The t-statistics are reported in parentheses. ***, ** and * denote significance at the 1%, 5% and 10% level, respectively.
(1) (2) (3) (4) (5) (6)
RD RD overRD_Avg overRD_Avg overRD_Md overRD_Md
PilotNum -0.003** -0.002*** -0.003***
(-2.36) (-2.80) (-3.51)
PilotRatio -0.013** -0.009** -0.012***
(-2.21) (-2.52) (-3.10)
FirmSize -0.003*** -0.003*** -0.001*** -0.001*** -0.001*** -0.001***
(-7.46) (-7.46) (-5.69) (-5.69) (-5.14) (-5.14) TobinQ 0.003*** 0.003*** 0.001*** 0.001*** 0.001*** 0.001***
(8.65) (8.65) (4.49) (4.50) (5.26) (5.27)
Leverage -0.001*** -0.001*** -0.000*** -0.000*** -0.000*** -0.000***
(-4.16) (-4.14) (-2.69) (-2.66) (-3.07) (-3.04) Cash 0.138*** 0.138*** 0.053*** 0.053*** 0.062*** 0.062***
(33.63) (33.63) (22.36) (22.35) (22.96) (22.95) BoardSize -0.001*** -0.001*** -0.000** -0.000*** -0.000 -0.000**
(-3.38) (-4.04) (-2.29) (-3.00) (-1.60) (-2.45)
BoardIndep 0.011*** 0.011*** 0.002 0.002 0.003 0.003
(3.19) (3.19) (1.02) (1.01) (1.48) (1.46)
Year FE Yes Yes Yes Yes Yes Yes
N 4,701 4,701 4,701 4,701 4,701 4,701
R2 0.344 0.344 0.188 0.188 0.194 0.194
AdjustedR2 0.34 0.34 0.19 0.19 0.19 0.19
Table 10 displays the OLS results of non-aviation firms’ investment expenditure on pilot directors. Compared with Table 6, these two tables share the same variables but use different samples. We make the results shown in Table 6 more robust by removing firms in the aviation industry to address the concern that firms in this industry are more likely to elect directors with pilot licenses. As can be seen, there is a very similar pattern for the coefficient estimates in Table 6.
All in all, the main finding of this thesis is supported through various robustness tests in this chapter. These tests begin with decomposing the investment expenditure for a micro analysis, and then move on to the measuring investment efficiency of the firms to obtain a macroscopic view. After that, we explore how pilot directors affect corporate investment decisions in intangible assets, R&D spending. Finally, OLS regressions similar to those in section 3.2 are conducted. All variables are the same but in the examined sample, the aviation industry is removed in this robustness test.
