Master thesis – Business Economics: Finance
Does firm location affect CEO turnover?
Student Name: Shaobo Li
Student Number: 11086874
Thesis supervisor: Dr. Torsten Jochem
July 2016
Statement of Originality This document is written by Shaobo Li
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
I study the empirical association between remoteness of firm headquarters and CEO turnover rate using the sample of U.S firms from 1995 to 2015 and test three possible theory related to this issue. I analyze the correlation between overall CEO turnover rate and the remoteness of firm headquarters and correlation between CEO turnover rate in each year and remoteness of firm headquarters using OLS and Panel regression respectively. I didn’t find evidence to support that firms with headquarter located in remote districts are more likely to have a higher CEO turnover rate compared to their peers situated in big cities, even when I control for arrays of firm and CEO level characteristics such as firm age, firm size, CEO age, CEO compensation as well as time and industry fixed effects. This finding is also consistent with the finding of robust check. The robust check indicates that there is not any difference between the remotely located firms and centrally located to have a succession plan. I find that small firms located in a remote area have a higher CEO turnover rate. This finding is also significant when I control for arrays of firm and CEO level control variables. Results didn’t show that remoteness increase the likelihood that young CEO(less than 60 years old) leave current position within five years. However, findings show that many other factors like CEO age, firm size do have a strongly influence on CEO turnover rate. These findings are consistent with the prior literature in corporate governance.
Content
1. Introduction...5
2. Literature review and theoretical background...10
2.1 Literature review...10
2.2 Theoretical background...12
2.2.1 labor market segmentation...12
2.2.2 Cost of searching...13
2.2.3 Geographic preference...14
2.2.4 Hypothesis ...16
3.Data and descriptive statistics...17
3.1 Data source...18
3.2 Sample construction ...18
3.3 Key measures ...19
3.4 Control variables ...21
4. Methodology and Result ... 24
5. Robustness check...36
6. Summary and conclusion ...38
Tables………...………...41
Appendix A... 48
1.Introduction
Chief executive officer (CEO) turnover is one of the most important changes for a corporation. The turnover of CEO is always linked to changes in the strategy of firm's operations. An incidence of CEO turnover can cause the unexpected fluctuation of the short-term performance of the firm in the stock market, or change the long-term operation strategy of the company. In some extreme cases, this could cause the situation worsen. If the CEO is the founder or someone who deeply shapes the reputation of the firm, CEO turnover can have a large effect. In the real world, CEO turnover is becoming more and more frequent. Kaplan and Minton (2012) find that CEO turnover was 15.6% in 23 years from 1992 to 2005, which implies an average tenure for CEOs of less than seven years. Moreover, from the year of 1998 average turnover has increased to 17.4%, which means an average tenure of less than six years. Many frictions can lead to the turnover of CEO, these frictions can be from internal elements (driven by board) or external factors (through merges and acquisitions). For example, a conflict between the management and shareholders may lead to the forced turnover of CEO. The shareholders may be more likely to focus on the long-term value of the company while the CEO may care more about the short-term performance in the stock market because the CEO’s compensations are always related to the stock price, or CEO may try to entrench themselves to make them indispensable. Other possibilities of CEO turnover could be CEO’s stepping down after a merge or acquisition activities. A lot of the existing literature demonstrates that there are many factors with a close correlation to CEO turnover,
for example outside options through networks, CEO’s age, short-term and long-term firm performances, CEO total compensation etc. However there is currently very little research referring to the geographic factors influencing CEO turnover, for example the geographic location of the firms headquarters.
In this article I focus on some of the geographic factors, especially the remoteness of a firm’s headquarters that could possibly have a significant influence on the CEO turnover rate. Although the operations and management of a firm could involve many different districts or even countries, I focus on the location of firm headquarters, because the majority of important decisions made in corporate governance are made at the location of headquarters. Geographic factors can affect the CEO turnover in many ways, for example, rural areas have a smaller local labor market than the labor market in metropolises, this can affect the cost of searching for and replacing a CEO, thus the location may have an influence on the long-term CEO turnover rate.
I analyze whether a firm’s location contributes to its turnover rate of CEOs. More specifically, I analyze several approaches that argue that the remoteness of a firm’s headquarters and the CEO turnover rate might be related. These channels are not mutually exclusive, for example the segmentation of local labor markets can also affect the cost of searching for a new CEO. One possibility is that geographic segmentation can affect the size of the local market of CEOs and other executive talent. There is a lot of evidence to show that geographic segmentation of labor markets is a key factor affecting corporate governance. Corporate Governance is
affected by the ease with which a board can replace its top executives. It may more difficult to find a replacement in a district where there is a small local managerial labor market than in a metropolis such as New York or London. Boards of firms in remote locations may be more lenient towards the CEO, even if the firm has a poor performance because of this. Geographic remoteness in the location of a firm is one possible source of friction in the labor market for CEOs. Another possibility could be the preference of a particular geographic area. Young CEOs(less than 60 years old) may enjoy the working environment in big cities rather than remote area due to geographic attractiveness. Working in big cities enables them to have easy access to outside networks, business resources and more exposure to the public. Having more network connections as a CEO usually means more outside employment options. Outside connectedness has a great influence on the CEO turnover rate, especially among the CEOs with poor performances, the probability that CEOs leaving for other positions is increased(Liu,2014).
In this article, I will propose three different explanations for geographic factors that influence changes in CEO turnover rate. I will illustrate in more detail the geographic segmentation theory, the cost of searching theory and the geographic preference theory in the theoretical background section.
As is discussed in more detail in the conclusion, the findings of this article suggest several implications.
Firstly, when using the overall turnover rate from 1995 to 2015 as a dependent variable with whole samples, I did not find any evidence to support the idea that
remotely located firms have a lower turnover rate than centrally located firms. However, interestingly, when I tested the centrally located firms and remotely located firms respectively, in both separate groups, the remoteness do have a negative correlation with the CEO turnover rate. This means there may be a segmentation of executive talent market in district level rather than in national-wide level, but this still needs further testing.
Secondly, when I analyzed the association between remoteness and CEO turnover rate in each individual year of the 20-year period from 1995 to 2015, the results indicate that there no correlation between remoteness and CEO turnover rate. This finding is not significant at the 10 percent confidence level and the economic effect is negligible.
Thirdly, I find that small firms that are located in remote area are more likely to be affected by cost of searching for a replacement, whilst big companies are less or not at all affected by the cost of searching for new CEO. This finding is consistent with the cost of searching theory.
Fourth, I tested the correlation between young CEOs sitting in remotely located firms and CEO turnover rate. The results show that the CEO’s age has a significant influence on CEO turnover. More specifically, young CEOs have a lower turnover rate than their peers. However, the finding didn’t suggest that young CEO’s (less than 60 years old) in remotely located firms are more likely to leave current position in the next five years compared to their peers in centrally located firms. In this test I used a dummy variable that indicates whether the CEO turnover in the next five years is or is not the dependent variable. The results didn’t suggest that a geographic
preference exists in the labor market, the result is not significant and its influence is minor.
To find more evidence about the correlation between geographical factors and CEO turnover I did an additional test to find evidence of whether remotely located firms are more likely to hire insiders when the CEO leaves (instead of outsiders) than centrally located firms. In each year of the 20-year period from 1995 to 2015, I defined a firm more often to have a succession plan in place if there is an executive with a “COO” title within the company. The result of this test is consistent with the findings of the test of the geographic segmentation theory. The results indicate that the correlation between COO existence and remoteness is nearly zero, which means that there is not any difference between the remotely located firms and centrally located firms in having a succession plan.
This paper contributes to a large amount of existing literature on CEO turnover (See Kaplan and Minton, 2012; Leggett and Nagel, 2015) and gives some suggestions to understanding the key factors that drive the CEO turnover. The findings of this paper are useful for a number of recent papers that focus on geographic factors and the labor market (see Yonker, 2015; Kedia and Rajgopal, 2009). I show that the location of the firm headquarters and other factors like CEO age, firm size have an impacts on CEO turnover rate. This paper also contributes to the area of corporate governance literature. It will help to discover the methods needed to reduce frictions in top executive labor markets and improve the efficiency of finding a fit CEO candidate for companies.
The remainder of this paper proceeds as follows. In section 2, I position the paper in the related literature and describe the theoretical background. Section 3, describes the sample selection process and the primary data. In Section 4, I present empirical results of regressions. In Section 5 I report the results of additional testing and in section 6 I summarize the results and discuss the implications in more detail.
2.literature review and theoretical background
2.1 literature review
Geographic factors such as firm location have gained more and more attention recently in both economic and finance literature because many findings of recent research show that geographic factors have a great influence on corporate governance issues. including firm operation, management of talents and firm performance. These factors may help to explain the difference in corporate governance and corporate finance policies, the size of local managerial market , liquidity of stock shares, capital structure, CEO power, board composition and so on (see Gao, 2011; Yonker, 2015; Knyazeva, 2008; Francis, John and Waismann, 2007). Geographic factors may also be possible determinants that affect the CEO turnover rate. In this paper I focus on the determinants of CEO turnover rate that are related to the geographic factors. One key determinant of CEO turnover rate is the size of labor market of executive talents, a widely accepted assumption in literature related to the labor market is that the executive market is national-wide rather than geographically segmented (Kedia and Rajgopal, 2009). However there are also
different perspectives about this. One implication of the existence of geographic segmentation in the market for CEOs is the existence of local labor markets (Yonker, 2015). If there do exist a geographic segmentation of the labor market and the size of the local market for top executives, it helps to explain the difference in turnover rate between remote located and centrally located firms. For example, managers of funds in big cities, increase the their stay in the same city and their tenure in the same company if there is a good market performance(Christoffersen and sergei., 2009), but this is only the case in financial firms, to make it more general I will also test for other type of firms. For firms that are located in remote cities the search cost for a new CEO may be higher than peer firms located in big cities (Bebchuk and Fried, 2004), thus the board may more willing to have an insider succession plan or even be more lenient towards the CEO. Firm location also has a significant effect on other factors that attract new CEOs. More executives t are attracted to big cities with more employment opportunities where they find better jobs more easily and the compensation packages can be much better (Glaeser 2011). Since there are more opportunities in big cities than in remote area, top executives are more easily able to find a better position in metropolises. CEO compensation also plays a crucial role in the labor market, Leggett and Nage (2015) find that CEO’s compensation has a positive correlation with the amount of CEOs in the same area, one possibility is that social pressures can affect CEO compensation. Location of companies’ headquarter and the cluster of the same type of firms, in general, have an significant influence on the level and structure of CEO compensation. I also test for other factors that may affect the turnover rate such as firm size or CEO age.
2.2 Theoretical background
In this section I outline several theories of how and why remoteness of firm headquarters may affect the turnover rate of CEOs. The basic idea is that the imbalance of labor market can lead to friction of CEO matching progress and CEO turnover. This imbalance can be driven by many factors like cost of searching, labor market segmentation, geographic preference or other geographic related factors.
2.2.1 labor market segmentation theory
Many studies show evidence that there exist local labor markets. Yonker(2015) finds that the CEO labor market is geographically segmented and a number of theories of local hiring appear to be playing a substantial role in the matching process. Moreover, the findings of Leggett and Nagel (2015) indicate that the more amount of firms within 60 miles of the firm's location is, the higher compensation of top executives is. In addition, some studies show that the the average compensation of local market can also affect CEO compensation .
For rural districts, the local labor market for CEOs is likely to me smaller than in metropolises. Consequently, the cost and time of searching for a CEO may be higher even if the firm has a worse performance it is difficult for remotely located firms to find a replacement in a short time. Additionally, since the local executive labor pool in remote areas could be so small and limit the choices, if they want to extend the searching pool for CEO, the cost of search may outweigh the gain of replacement in case the candidate pool is extended.
In reality, the CEO turnover can be much more complicated, because this may reflect the imbalance of the demand and supply of top executives. This imbalance can lead to local matching bias and affect the cost of searching for executive talents. The imbalance can be driven by many factors, such as an agency problem: even if there are many talented candidates in the local labor market, the boards may prefer local candidates, because they can be controlled more easily or the board members have more information to justify whether the candidates are fit for the long-term benefit of the company. As Yonker (2015) write “ when considering only CEOs hired externally to the firm and to the industry, local hiring decisions are nearly three times more likely than expected if CEO geographic origin were random; in addition, the tests of CEO turnover reveal that unforced turnover is about 20% less likely for local than similar non-local CEOs.”
If there does exist labor market segmentation rather than a single national wide labor market, then we would expect firms located in remote area to be more likely to have a lower turnover rate than the centrally located firms. In addition, we might expect that the remotely located firms are more often to have a succession plan in place due to the small size of local labor market.
2.2.2 Cost of searching theory
The cost of searching for CEO is related to the size of local labor market, as well as many other factors. For instance, the search cost may stem from the limited access to top talent pool: if firms have limited access to top talent pool, they need to extend
the options of fit candidates. The process of extending can be costly and sometimes may even outweigh the gain of replacement. On one hand, in order to conduct the search for a CEO, they need to relocate part of the resources away from the production process. On the other hand, hiring an executive search firm or enhancing the internal human resources department may help to improve the efficiency of finding a fit candidate but can also be costly. Both Bebchuk and Fried (2004) and Shivaram (2012) discuss the role and the prevalence of executive search firms in the CEO hiring process. Furthermore, the cost of searching is also subject to the size of the firm. For large firms the cost of searching for top talent accounts for a small part of the budget or could even be negligible compared to the loss of a CEO or a poorly performing CEO. In contrast, small firms located in rural district suffer more from the cost of searching or replacing. This can be attributed to two main reasons. Firstly, small firms have a more limited budget compared to large companies or are unable to change the funding structure in the short term. Secondly, small firms located in remote districts face a limited pool of qualified candidates, and thus lose the barging power when negotiating with the desirable candidate. According to cost of searching theory, I expect that small firms located in remote areas will have a higher CEO turnover rate.
2.2.3 Geographic preference
Many executives, especially younger ones, may prefer to work in big cities rather than remote area due to geographic attractiveness, which includes more job opportunities, better compensation packages and more business sources. Many studies show that the geographic desirability of the firm location plays an important
role in deciding CEO compensation. Deng and Gao (2013) find a strong substitute effect between good living environment and CEO compensation. If a company is located in a area with poor quality of environment, this can be high crime rate, heavy pollution, etc. A higher monetary compensation need to be paid to CEO as a compensation. Many existing studies show that the living environment is an important consideration people take into account when making career choices. Roback (1982) shows that people are more likely to live in area with a higher quality of life, thus if a company located in an area with poor quality they must provide better compensation package to remain the same quality employees . However, there are also other explanations as well: central areas can also have a higher average compensation due to the high competition between CEOs. Both of the two perspectives show that geographic factors can affect the compensation.
Another geographic preference matter is that CEOs may view their current job as temporary and have an incentive to find a better place. Additionally, CEOs can have an easy access to business resources like a business network in a metropolitan area, because more business leaders and board members live there. As a consequence, a CEO with a better external network, can more easily find a better position or bid up for a higher compensation package. Many studies on networks and labor markets show that network connections play a role in reducing the frictions of searching for top executive talent in the labor market - more network connection for CEOs usually means more outside employment options. The findings of Liu (2014) indicate that outside contentedness increases the probability of CEOs either leaving for other
outside positions, or retiring and taking part-time positions elsewhere. This effect is significant positive within all ranges of performance, especially for poor performers. From the two dimensions of geographic preference theory we can expect that firms located in big cities might have higher CEO compensation because big cities have higher crime rate and more severe air pollution. Although big cities have a lot of advantages, the big companies can also give their CEOs with better offer to retain the talent. Thus, I expect that young CEOs in small firms are more likely to be attracted to big cities compared to their peers.
2.2.4 Hypothesis
In this paper I propose three main hypotheses and an additional one. First of all, according to the geographic segmentation theory, remotely located firms will have a lower CEO turnover rate than centrally located firms due to many reasons, such as smaller size of executive talent market which makes it more difficult to find a replacement in a short time. The second hypothesis is that small firms whose headquarters are located in a remote district tend to have higher turnover rate than any other types of firms. According to the cost of searching theory I proposed, small firms in rural area are most likely to suffer from the cost of searching for a replace due to the shortage of resources. For small firms located in remote area, the cost of searching may be much higher. The third hypothesis is that young CEOs (less than 60 years old) in remotely located firms have a higher tendency to leave their current position. This can be explained by many reasons. For instance, one possibility is that young CEOs are more eager to gain access to outside networking or gain business resources that can expend their future career options. Another possible explanation
may be that young CEOs are simply more fond of the vibrant and convenient city life than their older peers.
In the last part of the paper I also test the additional hypothesis to gain more insights into the correlation between remoteness and CEO turnover. I examine if remotely located firms are more likely to have a succession plan (a COO title in company) than firms located in big cities. In consistency with the geographic segmentation theory, the correlation between a tendency to have a succession plan in place and the remoteness would be positive: since it is not easy for remotely located firms to find a replacement, so a succession plan would be an approach to reduce the friction.
In summary, the aim of this paper is to provide direct evidences on the association between remoteness of firm headquarters and turnover rate of CEO. There is plenty of literature that focuses on the role of geography in corporate governance and corporate finance, and this paper can contribute to the prior research as well as enhance the understanding of the importance of local labor market for CEOs and the key determinants of the turnover of CEOs.
3.Data and descriptive statistics
In this part, I describe the data sources, sample construction, control variables and then discuss the construction of two important measures of remoteness and CEO turnover.
I construct a dataset with 1672 public listed firms in United State for the years 1995 through 2015. This dataset includes basic annual information on companies, as well as the personal information and annual compensation of CEOs. To construct the dataset I used two different sources and merge them together, as explained below. Basic annual accounting information on public listed firms comes from Compustat North America which contains selected accounting information on more than 24,000 active and inactive public companies. These records include the gvkey, sic code, total asset, book value, location of headquarters, company initial public offering date etc. To construct CEO turnover and CEO characteristics to be included as control variables in regressions, I use personal information and annual compensation of CEOs from Execucomp. ExecuComp includes detailed data on executive compensation such as annual salary, bonus, and stock and option grants for top executives within the firm. The records I used include the personal identification, the annual title, compensation, age etc.
3.2 Sample construction
The initial sample comes from a merging of the data from Compustat and Execucomp datesets based on gvkey and fiscal year. To be more specific, I begin with the sample of firms covered by the Compustat database between 1995 and 2015. At the same time I choose the item of ID number of the executives, name of CEO, CEO age and total compensation of CEO from Execucomp. Then I merge these two datasets together (joinby gvkey and fyear). This provides us with a sample of 3,424 firms and a total of 36569 firm-year observations. I keep firms whose headquarters are located in the United States because a focus on only US firms helps to control for
any differences caused by different countries, these firms are subject to similar regulation. I use the headquarter of the firms as a proxy of a firm location. Although some firms may be situated in more than one location, especially the multinational companies, most of the important decisions are made in the headquarter and CEO offices are generally situated there also. In this paper I focus on the turnover rate of CEOs, thus it is reasonable to choose the firm headquarter as a proxy. Also I exclude from the sample firms of financial and utility firms (with Standard Industrial Classification(SIC) codes 4900-4999 and 6000-6999) because financing policies of financial and utility firms are more often affected by regulation. This leaves us with a sample of 2,390 firms and 26,179 firm-year observations. To avoid the potential bias of the firms just below or above the threshold of central and remote, I also exclude 718 firms that situated in the joint of urban and rural area, this leaves us with a sample of 1,672 firms and 17,627 firm-year observations.
3.3 Key measures
In this article, there are two key measures needed. The first one is to measure the location of the firms; I use GPS distance and corresponding zip codes on locations of corporate headquarters. More specifically, I download the zipcode of US firms through Compustat, then I assign the corresponding GPS value(longitude and latitude) of the zipcode to each firms. I also collect the zipcode of the largest 10 metropolises according to latest US population census(2010), These include New York City, Los Angeles, Chicago, Washington-Baltimore, San Francisco, Philadelphia, Boston, Detroit, Dallas, and Houston . I then calculate the distance between the location of headquarters and the nearest big city by using the Stata
command vincenty for distances between two GPS locations (the GPS of firm location and the GPS of top-10 metropolises location). Since it is quite common that one metropolitan area has several zipcodes, I just choose the zipcode of the city center as proxy. To avoid the potential bias of the firms just below or above the threshold, a company is defined as centrally located if its distance to the ten largest metropolitan area are among the top 40% of the distant range (Ascending order by distance). Similarly, a company is defined as remotely located if its distance to the ten largest metropolitan areas are among the bottom 40% of the distant range (Ascending order by distance).
Another key measure is the measurement of turnover rate of CEO. In this paper, to keep it simple, I only consider total turnover. Total turnover means a company remains publicly listed during the period from 1995 to 2015, but the CEO next year is different from the current CEO (which means the current CEO leaves in the current year). I didn’t distinguish between forced turnover and unforced turnover. Firstly, we denote the indicator equal to 1 if the ID number of the CEO is different from the ID number of the CEO next year, otherwise it is equal to zero. Thus turnover equal to one means the Current CEO resign in this specific year. Secondly, I calculated the total times of the CEO turnover during the period from 1995 to 2015 as the overall turnover. The higher the sum of the indicator the more frequently the CEO turns over.
Table1 describes the summary statistics of firm and CEO characteristics that can potentially affect the CEO turnover. The firm-characteristic statistics clearly show that the average firm age of the samples is 12.51 years, with an average market value of $7.3 billion, an average return on asset of 0.08 and an average market to book ratio of 2.04. In this paper I use return on asset as a proxy of short term performance and use market to book ration as a proxy of long term performance. From the CEO-characteristic statistics, we can see that the average age of a CEO is 55.81, there is a wide range from 28 to 96 years old. The annual total compensation is from 0 million to 655.4 million which on average is 4.98 million for the whole sample. The tenure of a CEO is also quite different, from 0 year to 62 year, and they have an average tenure of 10.64 year. The maximum turnover rate for a company of the full sample is 6 times, which is nearly four times than the average during the 20 year-period from 1995-2015, and the average is 1.62 times during this 20 years period.
3.4 Control variables
Aside from measures of the CEO turnover rate and measures of remoteness, I also use a lot of accounting variables that may potentially affect the CEO turnover rate and construct ratios as is typically done in the other corporate finance literature. Market to book ratio: I use the market to book ratio as a proxy of the long-term performance of corporations. The Market to Book ratio is calculated as follows, the total market value is defined as the total market value of equity, which is equal to
the number of common equity outstanding times the annual close price of common shares. Then, I add the book value of total assets and minus the number of common equity minus deferred taxes. Finally, I compute the market to book ratio equal to market value of the common equity divided by book value of the total asset.
Return on asset: I use the return on asset as a proxy of short-term performance of corporations. The return on asset ratio is calculated as follows: In each year of the 20-year period from 1995 to 2015, I take the operating income before depreciate subtract the depreciation and amortization. Then this is divided by Assets.
Firm age: In each year of the 20-year period from 1995 to 2015, the firm age is defined as the number of years since the firm listed in one of the stock exchanges. I take the date of the current fiscal year minus the company initial public offering date; I then normalize the resulting firm’s age from the unit of day to the unit of year using the measure of 365 day per year.
Firm size: In each year of the 20-year period from 1995 to 2015, I use the measure of total assets as a proxy of the measure of firm size. As is usual done in other corporate finance literature, I use the natural logarithm term of total asset. To define whether a firm is a small or large firm, first I rank all of the firms in the dataset according to the firm size. Then I define a firm I as small if the asset of the company belongs to the lowest 20 percent of the sample. Correspondingly, a firm is defined as large if the size of the company belongs to the top 20 percent of the sample.
CEO tenure: The CEO tenure is defined as the duration of the CEO stay in office. To be more specific, the CEO tenure is equal to the date left as CEO title minus the date became the CEO. I then normalize the resulting CEO tenure from the unit of day to the unit of year using the measure of 365 day per year.
CEO age: In each year of the 20-year period from 1995 to 2015, I define a CEO as a young CEO if the age is less than 60 years old, correspondingly the CEO older than 60 years old is defined as an old CEO.
CEO compensation: In each year of the 20-year period from 1995 to 2015, the CEO compensation is the total compensation for the individual year, this is comprised of the following: Salary, Bonus, Other Annual, Total Value of Restricted Stock Granted, Net Value of Stock Options Exercised, Long-Term Incentive Payouts, and All Other Total.
TABLE 2-SUMMARY STATISTICS
Table 2 clearly shows the difference between centrally located firms and remotely located firms. I divided the sample into those firms located in remote areas and those located in central areas and it reports the differences in average values. From the summary statistics we can see that there are not too many differences between these two types of firms in variables like Firm size, market to book ratio, return on asset and CEO tenure. However the other characteristics like market value can be very different.
As is shown in Table 2, a firm located in a central district has an average market value of 8.2 billion which is 1.8 billion more than the remotely located firms. CEOs sitting in centrally located firms also have a much higher compensation than their
peer in remotely located firms. This is consistent with the findings of the prior literature. On average, the CEOs in centrally located firms have a compensation of 5.68 million, while the CEOs in remotely located firms have 4.28 million per year, the difference is 1.4 million. Although a firm is defined as remote if its distance to nearest metropolis is more than 99 mile, the average distance can be much larger. In general, remotely located firms have an average distance of 351 mile, and the range varies from 99 mile to 935 mile, which is more than 65 times more than the average distance of centrally located firms. The centrally located firms have an average distance of only 14 mile, which is about 4% of the distance of remotely located firms. This can also lead to potential friction within the remotely located firms due to the wide range of distances.
4. Methodology and results
In this section, I describe the base analyses of the empirical association between remoteness and the CEO turnover rate using the sample of U.S firms from 1995 to 2015. I begin in Section 4.1 with the analysis of overall CEO turnover rate and remoteness to gain some intuitive insights. Then in section 4.2 multivariate tests control for a number of firm and CEO level characteristics that are potentially correlated with firm location and CEO turnover rate. These additional variables are firm size, firm age, return on asset, market to book ratio, CEO age and CEO tenure. I also consider the year and industry fixed effect, all regressions use industry effects defined at the SIC level. More specific, in 4.2.1 Regression analysis between CEO turnover rate and remoteness in each year from 1995 to 2015. Then in 4.2.2
Regression analysis between CEO turnover rate and firm size. Final, in 4.2.3 Regression analysis between CEO turnover rate and CEO age.
4.1 Analysis between overall CEO turnover rate and remoteness.
Begin the analysis with univariate comparisons of CEO turnover rate in both centrally located and remotely located firms.
In this paper, a company is defined as centrally located if its distance to the ten largest metropolitan areas is among the top 40% distance range which is less than 32 mile. Similarly, a company is defined as remotely located if its distance to the ten largest metropolitan area is among the bottom 40% distance range which is larger than 99 mile.
Table3
In this analysis I divide the sample into three different sub-samples, first the full sample, then centrally located firms and remotely located firms.
Table 3 indicates the correlation between remoteness and overall CEO turnover rate during the 20-year period from 1995 to 2015. The dependent variable is the total number of times of CEO turnover during the 20-year period.
Column 1 begins with the regression using the full samples in my dataset. The coefficient of distance is insignificant at any percent confidence level. Thus we need to do further tests to see whether the correlation between overall CEO turnover rate
and remoteness do exist. In column 1, I also find that the coefficient of average the firm size is strongly significant at the 1 percent confidence level and the economic effect is large. This means that an average firm size increase by 1 million would increase 0.22 in CEO turnover rate, because CEOs of large firms are more likely to leave can various. One possibility is that CEOs with work experience in big companies are more popular in the executive labor market and they have more options outside the current position. Thus, they are more likely to leave if there are better positions or when facing too much pressure from the boards. I also found that the long term performance had a negative correlation with CEO turnover. The coefficient of market to book ratio is negative and significant at a 5 percent confidence level which means that the better long term performance of a company, the lower the CEO turnover rate. One explanation is that CEOs may gain higher benefits if the firm has a better performance, because part of their compensation exists as stock option and this is related to the stock price. If the firm has better long term performance, they can have a better compensation package or have more decision making power which enables them to fully extend their potential or entrench themselves, so they prefer to stay in current position rather than to leave.
Columns (2) presents regression results for the sub sample of firms that are located less than 32 mile (among the top 40% distance range) to the 10 largest metropolis according to the 2010 US population census ( New York City, Los Angeles, Chicago, Washington-Baltimore, San Francisco, Philadelphia, Boston, Detroit, Dallas, and Houston ). I found that the coefficient of distance becomes negative and significant at 10 percent confidence level. The coefficient is negative, which means that within the type of centrally located firms, the further away from city center, the lower
turnover rate of CEO. This can be for various reasons. One possibility is that many people enjoy the convenience of living in central area; while at the same time they try to avoid living too central due to pollution or higher crime rates. A lot of the existing literature shows that living environment is an important consideration when people make their career choices. On the one hand living in central area can help CEOs to gain easy access to business resources and networking. On the other hand they also prefer a quiet living environment. This finding is consist with the geography preference theory. I also noticed that the coefficient of the average firm size remains significant at a 1 percent confidence level and therefore economically influential. From column (2) we can see, when considering only centrally located firms, the coefficient of the average market to book ratio increases to -0.12, which means that the long term performance has a significant influence on CEO turnover rate.
In Column 3, I analyze for the sub sample of firms that is located further than 99 miles (among the bottom 40% distance range) to the 10 largest metropolises. Again, the coefficient of distance is negative. The coefficient becomes much smaller than in column 2, the explanation of the coefficient may also be different from column 2 considering the wide range of distances to the nearest big cities. This can be consistent with theory of geographic segmentation of labor market. The main difference between the distances within these two types of firms is that the distance range of remotely located firms is much wider, from 99 mile to 935 mile. For rural areas where the distance is quit far away from big cities (e.g. More than 500 mile) they are more likely to have a smaller local labor market for CEOs than the labor market in metropolises. This situation may lead to a higher cost of searching for a new CEO, even if the firm has a worse performance it is difficult for remotely located
firms to find a replacement in a short time. So the coefficient of distance is negative within the type of firms that are located in remote area. This means that the further away a firm is from the metropolis, the lower the CEO turnover rate will be. However, this finding is not economically influential, so there may exist other variables that can affect the outcome.
In summary, from table 3, I didn’t find the evidence to support for the theory of geographic segmentation when using the overall turnover rate and the full sample of my dataset. However, within each type of firms there may exist evidence of segmentation of the CEO labor market. I will do further analysis in this paper.
4.2 Multivariate regression
Multivariate tests control for a number of firms and CEO characteristics that are potentially correlated with firm location and CEO turnover rate, and others such as firm size, firm age, return on asset, market to book ratio, CEO age, CEO compensation and CEO tenure. Since industry affiliation and time period can have a significant effect on the choice of location of firm headquarter (e.g., geographic clusters of energy, automotive, and technology firms) and turnover rate, all regressions use industry effects defined at the SIC level. The main idea of this part is to test the three main hypotheses respectively.
4.2.1 Regression analysis between CEO turnover rate and remoteness of
firm location in each year
According to labor market segmentation theory, for rural districts there is more likely to be a smaller local labor market for CEOs than the labor market in metropolises. This situation may lead to a higher cost of searching for a new CEO, even if the firm has a worse performance it is difficult for remotely located firms to find a replacement in a short time. Thus I expect remoteness of firm location to have a negative influence on the CEO turnover rate. In 4.2.1 I analyze this correlation through panel regression.
Table4
Table 4 indicates relations between the remoteness of firm location and the CEO turnover rate, in each year of the 20-year period from 1995 to 2015. The dependent variable is the CEO turnover rate in each year. The CEO turnover dummy equals to 1 if the CEO turnover in this year, and 0 otherwise. The remote dummy equals to 1 if the firm is defined as remotely located. Similarly, the remote dummy equals to 0 if the firm is defined as centrally located.
Column (1) begins with the basic regression without any additional control variables and without controlling year and industry fixed effect. The table didn’t show that remotely located firms have a lower turnover rate than centrally located firms. The coefficient is insignificant at the 10 percent confidence level and the economic effect is also negligible. From column (1) we can not know if firm location affects the CEO turnover rate or not.
In column (2), I control for year and industry fixed effect for further tests, industry fixed effects are at the SIC level.
The coefficient increases to 0.0033, but it is not statistically different from 0.0023 found in column 1. Again it is still insignificant at the 10 percent confidence level. From Column (3), I control for many firm and CEO characteristics that may potentially affect the CEO turnover rate. Similarly to the prior analysis, these additional variables are firm size, firm age, return on asset, CEO age and CEO tenure. I also control for the year and industry fixed effects. The coefficient of the remote dummy remains insignificant.
Column (4) replicates the estimation performed in column 4 using the long-term firm performance. In this paper, I use market to book ratio as a proxy of long-term firm performance. There is no any different in the coefficient of remote dummy in column 3 and column 4.
The results in column 4 show that the coefficient remains stable. This means, in general, long-term performance and short-term performance tend to have the similar trend.
Column 5 replicates the estimation performed in column 3 using both short-term and long-term firm performance. The coefficient is almost the same as well as insignificant.
From table 4 we can find that the coefficients of the remote dummy in Column 1 to Column 5 are not significant at the 10 percent confidence level and the economic effect is very small. This table also shows us that the coefficient of CEO age and CEO tenure are strongly significant. There is a positive correlation between CEO age and the CEO turnover rate. The CEO turnover rate increases with the increase of age. One possibility is that older CEOs may prefer to leave current position to enjoy a quiet life or this is simply due to older CEO reaching the retiring age, in this situation
they have to leave. From the table I also find that CEO tenure is negatively correlated with the CEO turnover rate, this is obvious and consistent in logic. The longer the CEO tenure the lower the CEO turnover rate.
4.2.2 Regression analysis between CEO turnover rate and firm size.
According to cost of searching theory. For large firms the cost of searching for new CEO accounts for a small part of the budget or even the cost is negligible compared to the loss of without a CEO or with a poor-performed CEO. In general, small firms that located in rural district suffer more from the cost of searching or replacing. Thus I expect small firms in remote area will have a higher turnover rate. In 4.2.2 I analysis this correlation through panel regression.
Table 5
Table 5 indicates the correlation between the CEO turnover rate and the firm size in each year of the 20-year period from 1995 to 2015. The dependent variable is the CEO turnover rate in each year. The CEO turnover dummy equals to 1 if the CEO turnover in this year and is 0 otherwise. Small dummy equals to 1 if the firm is defined as small, otherwise it is equal to 0.
Column (1) begins with the basic regression without any additional controls and without controlling the year and industry fixed effect. The coefficient of remote and small dummy are both insignificant at the 10 percent confidence level and the economic effect is negligible.
In column (2), however, when I control for year and industry fixed effect, the coefficient of small dummy becomes negative. The coefficient of small dummy decrease to -0.0056 and become significant at the 10 percent confidence level. Then Column(3) replicates the estimation performed in column(2) adding the interact term Remote×Small into a regression to test whether CEOs in small firms that are located in remote area are less likely to leave due to the higher cost of searching for a replacement. This is consistent with the theory of the cost of searching. The coefficient of interacting terms is negative and significant at the 5 percent confidence level which means that CEOs in small firms located in remote areas are less likely to leave current positions than their peers. More specifically, a remotely located small firm has a 0.028 lower turnover rate than other type of firms. In column 3 you can also see that in general small firms have lower CEO turnover rate , this finding is consistent with table 3. One reason could be that CEOs sitting in small firms have less outside options than the CEOs in large companies, thus they tend to stay in their current position for a longer time.
Column (4) replicates the estimation performed in column 3 as well as controls for other firm and CEO characteristic variables. I control for firm size, firm age, market to book ratio, CEO age and CEO tenure. I find that the coefficient of the interacting term remains significant and the coefficient becomes larger than in column 4 from -0.028 to -0.054.This means a remotely located small company has a 0.054 lower turnover rate than other types of firms in a specific year. Again, the finding is consistent with column(3) and table3, showing that small firms have a lower turnover rate. Column (4) indicates that CEO age has a positive correlation with CEO
turnover rate and CEO tenure has a negative correlation with CEO turnover rate, this is consistent with my prior findings.
From table 5 we can find that there is a negative correlation between CEO turnover rate and the interacting term of Remote×Small. The coefficient is both significant and economically influential. This means small firms that are located in remote areas are more likely to be affected by cost of searching for a new CEO. The higher cost of searching can be attributed to many different factors, for example, small firms have limited funding or the CEO positions in remotely located firms are less attractive. They tend to retain the current CEO, even if the CEO has a bad performance history, it is better to give him/her a chance than to find a replacement, thus the CEO turnover rate of small firms located in remote district is lower.
4.2.3 Regression analysis between CEO turnover rate and CEO age
According to geographic preference theory, many executives, especially in the younger generation, prefer to work in big cities rather than remote areas due to geographic attractiveness. These attractions can be increased job opportunities, better compensation packages and more business sources. Increased outside network connection for CEOs usually means further outside employment options. I expect that younger CEOs in remotely located firms are more eager to gain access to outside networking or business resources, thus the CEO turnover rate will be higher among younger CEOs in remotely located firms. On the one hand, young CEOs have more urges to fulfill their own potential, on the other hand, remote districts have limited business resources and the social networking in big cities is more convenient than in rural area.
Table 6
Table 6 indicates relations between the CEO turnover rate and the CEO age in each year of the 20-year period from 1995 to 2015. The dependent variable is the CEO turnover rate in each year. The CEO turnover dummy equals to 1 if the CEO turnover in this year, and is 0 otherwise. The young dummy equals to 1 if the CEO is less than 60 years old. Similarly, the young dummy equals 0 if the CEO is older than 60 years old.
Column (1) begins with the basic regression without any additional controls and without controlling year and industry fixed effect. Young CEOs have a lower turnover rate than elderly CEOs. The coefficient is -0.17, this is strongly significant at the 1 percent confidence level and the economic effect is large, which means that if a CEO is less than 60 years old, he/she has 0.17 lower turnover rate than his/her peers that older than 60 years old in a specific year. This finding seems to suggest a different result to the generally held opinion, which says that the young generation tend to have a higher turnover rate. However, we need to notice that in this paper, the definition of young differs from the generally used definition, CEOs who are less than 60 years old are still defined as young, while a person whose age is 60 is defined as old. Young CEOs tend to have a lower turnover rate, which is consistent with the findings in table 4 and table 5.
In column(2), when I control for the year and industry fixed effect, the coefficient of the young dummy remains stable. The coefficient remains at -0.17 and it is still strongly significant at the 1 percent confidence level thus the economic influence is significant. If a CEO is young the turnover rate is 17% lower than the turnover rate for CEOs older than 60 years old.
Then Column (3) replicates the estimation performed in column 2 by adding the interact term Young×remote into the regression to test whether young CEOs in remotely located firms are more likely to leave due to their urge to get access to business resources or networking in big cities. The coefficient of interacting term is negative. While It is insignificant at the 10 percent confidence level and the economic effect is small. However the coefficient of the young dummy remains stable and strongly significant, which means that CEO age does have a strong influence on the CEO turnover rate.
Column (4) replicates the estimation performed in column 3 as well as controls for other firm and CEO characteristic variables. I then control for firm size, firm age, market to book ratio, CEO age and CEO tenure. I found that the coefficient of the interacting term remain insignificant and the coefficient is only 0.0019, which is statistically negligible. Again column (4) indicates that young CEOs negatively correlate to the CEO turnover rate which is consistent with column (1), column (2) and column (3).
From table 6 I cannot prove that young CEOs in remotely located firms have a higher turnover rate than their peers. However, the findings in table 6 clearly show that the CEO age is strongly correlated with the CEO turnover rate, the coefficient are both significant and economic influential.
5. Robustness check
According to the labor market segmentation theory, for rural districts there is likely to be a smaller local labor market for CEOs than in metropolises. Due to this theory, I expect that remotely located firms will have a lower turnover rate. However I did not find evidence to support this opinion in my prior analysis.
In this section I do an additional test to find evidence that suggests if remotely located firms are more likely to hire insiders when the current CEO leaves (instead of outsiders), compared to centrally located firms. Do remotely located firms more often to have a succession plan in place? In each year of the 20-year period from 1995 to 2015, I defined a firm more often to have a succession plan in place if there is an executive with a “COO” title within the company. In general the chief operating officer is usually the second in command within the company, especially if the highest ranking executive is the Chairman and CEO. In many companies, one of the primary reasons to set up a position of COO is to cultivate the candidate of CEO. Routinely in firms the Chief operation officer will be the apparent heir to the position of chief executive officer. The COO can be chosen from individuals that have worked for many years in the company or CEOs can be recruited from an outside company. Either way, the position of a COO is used as a training and testing ground for the next CEO. My findings suggest that the correlation between the remoteness and CEO turnover rate is related to the size of local executive labor market. If the remote districts do have a smaller pool of top talent, remotely located firms would be more likely to find a replacement in the earlier stages to avoid a sudden and unexpected
change in management. If the remotely located firms do have a succession plan, this can support our suggestion that remote area has a smaller labor market for CEOs.
Table 7
Table 7 indicates the relations between COO existence and the remoteness of the firm’s headquarters in each year of the 20-year period from 1995 to 2015. The dependent variable is COO existence. The dependent variable equals 1 if there exists a COO title in the company, and is 0 otherwise. The remote dummy is consistent with the former analyses. The remote dummy equals to 1 if the firm is defined as remotely located and it equals to 0 if the firm is defined as centrally located.
Column (1) begins with a basic regression without any additional controls and without controlling year and industry fixed effect. The coefficient is 0.00082, which is neither significant nor economically influential.
Then in column 2, I replicate the estimation performed in column (3) but control for the year and industry fixed effect, the coefficient of the independent variable become negative. As in column 1, the coefficient remains small (-0.00024) and insignificant at a 10 percent confidence level. This means that the location of a firm’s headquarters has little influence on the existence of a COO, thus I cannot find evidence to support the correlation.
Column 3 controls for the firm and CEO level characteristics that may potentially affect the outcome. More specific, I control for firm size, firm age and market to
book ratio. I find that the coefficient of remote dummy become much larger to 0.0011. However, the coefficient remains insignificant at 10 percent confidence level. The findings of column 1 to column 3 in table 7 shows that the correlation between COO existence and remoteness is nearly zero and the coefficient are all insignificant. Which means I can’t find evidence to support the hypothesis that remote located firms are more often to have a succession plan.
6. Summary and conclusions
For an organization, CEO turnover means a great change in the management and the short-term or long-term changes in strategy of operations, which will have influential effects on a firms’ business. In this article, I explore several channels that suggest that geographic factors can potentially affect the CEO turnover. These factors help us to gain a deeper understanding of what drives CEO turnover and what cause frictions in the top executive labor market. Prior findings in corporate governance literature indicate that internal and external reasons lead to the turnover of CEOs, this paper adds to this literature by suggesting that geographic factors, especially location of the firm’s headquarter, are also important determinants which need to be considered.
I analyze the correlation between the overall CEO turnover rate and the remoteness of a firm’s headquarters and the correlation between the CEO turnover rate and the remoteness of firm headquarters using OLS and Panel regressions respectively. Unfortunately, I did not find any evidence to support the idea that firms with headquarters located in remote districts are more likely to have a higher CEO
turnover rate compared to their peers situated in big cities. Even when I control for an array of firm and CEO level characteristics such as firm age, firm size, CEO age, CEO compensation, as well as time and industry fixed effects, the correlations are insignificant. This can due to many reasons, for example, there is not a correlation between the remoteness of a firm’s headquarters and the CEO turnover rate. Another possibility is that, because in this article I use the total turnover rate rather than forced turnover rate or unforced turnover rate, it may lead to some bias and thus affect the outcome. This finding is also consistent with the finding of the robust check. Additional tests show that the correlation between COO existence and remoteness is nearly zero, which means that there is not much difference between the remotely located firms and centrally located in regards to having a succession plan. I also find that small firms located in remote area have a higher CEO turnover rate. This finding is also significant when I control for arrays of firm and CEO level control variables. The results did not show that remoteness increases the likelihood that young CEOs (less than 60 years old) leave current position within five years. However, the findings show that CEO age does have a strong influence on the CEO turnover rate. This finding is consistent with prior literature in corporate governance. The effects of the remoteness of a firm compared with the CEO turnover has important implications for research about the reasons which drive the CEO turnover and is useful when creating methods to reduce the frictions of searching for a fit candidate in the CEO labor market.
Finally, the findings of this paper are useful for a number of recent papers that focus on the factors driving CEO turnover. I show that the location of the firm’s headquarters impacts the frequency of CEO turnover and many other factors like
CEO age, firm size have a significant influence on CEO turnover. One caveat, however, is that the results do not address the differences between forced and unforced turnover rate, this is an issue which awaits further study.
TABLE 1-SUMMARY STATISTICS
Variable Min Mean Median Max Std.dev Obs Firm characteristics Firm size 0.00 7.25 7.12 13.17 1.58 17619 Market value 0.09 7300.90 1372.46 504239 24814 15871 Market/book 0.29 2.04 1.57 147.35 2.05 16056 Return on asset -32.00 0.08 0.09 1.27 0.33 17558 Firm age 12.51 12.04 6.56 7.49 16.75 8278 Distance 0.14 199.92 31.88 935.45 239.66 17627 Average turnover rate 0.00 1.62 1.00 6.00 1.30 17627 CEO characteristics
CEO age 28.00 55.73 56.00 96.00 7.65 17055 Total Compensation 0.00 4983.35 2913.33 655448.00 9052.23 17565 Tenure 0.00 10.64 8.75 61.45 8.20 10280
Notes: Firm size is defined as the logarithm of total revenue. Market value is defined as (common shares outstanding ×price close-fiscal +asset total- common equity-total- deferred tax ) . Market/book is defined as the market/value of firm (common shares outstanding ×price close-fiscal +asset total- common equity-total- deferred tax ) normalized by their book value. Return on asset (operating income before depreciation minus depreciation divided by asset total. Firm age is defined as the number of years since IPO. Distance is defined as the least distance to the 10 largest metropolis according to the 2010 US population census ( New York City, Los Angeles, Chicago, Washington-Baltimore, San Francisco, Philadelphia, Boston, Detroit, Dallas, and Houston ). Overall turnover rate is defined as the number of CEO turnover during the 20-year period form 1995 to 2015. Total compensation is defined as total compensation for the individual year comprised of the following: Salary, Bonus, Other Annual, Total Value of Restricted Stock Granted, Net Value of Stock Options Exercised, Long-Term Incentive Payouts, and All Other Total. Tenure is defined as the duration of CEO stay in office((date left as CEO-date become CEO)/365).
TABLE 2-SUMMARY STATISTICS
Variable Obs Mean Median std.dev Obs Mean Median std.dev Diff in means 1: Central 2: Remote 1 minus 2 Firm size 8821 7.31 7.23 1.63 8798 7.18 7.02 1.51 0.29*** Market value 7918 8220.01 1498.54 26282.13 7953 6385.83 1250.09 23226.37 1834.18*** Market to book 8031 2.07 1.61 1.57 8025 1.99 1.52 2.43 0.08*** Return on asset 8785 0.08 0.09 0.14 8773 0.08 0.09 0.45 0.00*** Firm age 4267 12.18 11.71 6.53 4011 12.85 12.33 6.58 -0.67*** Distance 8823 14.04 13.22 9.36 8804 386.19 351.04 213.51 -371.79*** CEO Age 8521 55.81 56.00 7.56 8534 55.65 56.00 7.73 0.16*** Compensation 8782 5685.83 3269.81 11249.54 8783 4280.95 2592.69 6030.03 1404.88*** Tenure 5129 10.82 8.73 8.34 5151 10.47 8.80 8.07 0.35*** Notes: A firm is defined as central if the distance to the nearest largest city less than 32 mile(top 40% among the distance range). A firm is defined as remote if the distance to the nearest largest city larger than 99.18 mile(bottom 40% among the distance rang). Firm size is defined as the logarithm of total revenue. Market value is defined as (common shares outstanding ×price close-fiscal +asset total- common equity-total- deferred tax ) . Market/book is defined as the market/value of firm (common shares outstanding ×price close-fiscal +asset total- common equity-total-deferred tax ) normalized by their book value. Return on asset (operating income before depreciation minus
depreciation divided by asset total. Firm age is defined as the number of years since IPO. Distance is defined as the least distance to the 10 largest metropolis according to the 2010 US population census ( New York City, Los Angeles, Chicago, Washington-Baltimore, San Francisco, Philadelphia, Boston, Detroit, Dallas, and Houston ). Overall turnover rate is defined as the number of CEO turnover during the 20-year period form 1995 to 2015. Total compensation is defined as total compensation for the individual year comprised of the following: Salary, Bonus, Other Annual, Total Value of Restricted Stock Granted, Net Value of Stock Options Exercised, Long-Term Incentive Payouts, and All Other Total. Tenure is defined as the duration of CEO stay in office((date left as CEO-date become CEO)/365) *, ** and *** indicate statistical significance at the 10%, 5% and 1% levels, using two sample t-test (mean comparison test)
Table3
Table3 Regression analysis of the relation between remoteness and overall CEO turnover rate during the 20-year period from 1995 to 2015. Dependent variable is the overall turnover rate of CEO turnover during the 20-year. Overall turnover rate is defined as the total number of times of the CEO turnover during the period from 1995 to 2015. The regressions in columns (1) is conducted at the firm level for the full sample in my dataset. Columns (2) present regression results at firm level for the sub sample of firms with only centrally located firms. Columns (3) present regression results at firm level for the sub sample of firms with remotely located firms. All variable definitions can be found in Appendix 1. Industry fixed effects are at the SIC level. T-statistics based on standard errors clustered by gvkey and robust to heteroskedasticity are reported in the parentheses below the estimated coefficients. *, ** and *** indicate statistical significance at the 10%, 5% and 1% levels, respectively.
(1) (2) (3)
Full sample Central Remote Distance 0.00019 -0.011* -.000026*
(1.1) (-1.7) (-.11)
Average Firm Size 0.22*** 0.22*** 0.18***
(7.1) (5.2) (4.1)
Average Market/book -0.075** -0.12** -0.064
(-2.1) (-2.5) (-1.6)
Constant 0.650 -1.30*** 0.98*
(1.6) (-3.7) (1.9)
Year fixed effects Yes Yes Yes Industry fixed effects Yes Yes Yes Observations 17,672 8,718 8,702
Table4
Table4 Regression analysis of the relation between remoteness of firm location and CEO turnover rate In each year of the 20-year period from 1995 to 2015. The dependent variable is CEO turnover rate in each year. CEO turnover dummy equals to 1 if the CEO turnover in this year, and 0 otherwise. Remote Dummy equals to 1 if the observation belongs to the bottom 40% distance range(remotely located), and equal to 0 if the observation belongs to the top 40% distance range(centrally located). All variable definitions can be found in Appendix 1. Industry fixed effects are at the SIC level. T-statistics based on standard errors clustered by gvkey and robust to heteroskedasticity are reported in the parentheses below the estimated coefficients. *, ** and *** indicate statistical significance at the 10%, 5% and 1% levels,
respectively.
（1） （2） （3） （4） （5） Remote Dummy 0.0023 0.0033 0.017 0.017 0.018
(0.48) (0.63) (1.3) (1.3) (1.3) Firm level characters
Firm Size -0.0064 -0.0069 -0.0056 (-1.1) (-1.2) (-0.91) Firm Age 0.0019 0.0016 0.0015 (1.6) (1.3) (1.2) Lag ROA -0.043 -0.045 (-1.1) (-1.1) Marker/Book -0.0047 -0.004 (-1.3) (-1.1) CEO level characters
CEO Age 0.0089*** 0.0086*** 0.0086*** (9.6) (8.9) (8.9) CEO tenure -0.0073*** -0.007*** -0.0069*** (-8) (-7.6) (-7.4) Constant 0.11*** 0.15*** -0.2* -0.25*** -0.26*** (33) (9.5) (-1.9) (-3.5) (-3.5) Year fixed effects No Yes Yes Yes Yes Industry fixed effects No Yes Yes Yes Yes Observations 17,627 17,627 4,397 4,049 4,044 Pseudo R2 0.00 0.0247 0.1032 0.1054 0.1058
Table5
Table5 Regression analysis of the correlation between CEO turnover rate and firm size In each year of the 20-year period from 1995 to 2015. The dependent variable is CEO turnover rate in each year. CEO turnover dummy equals to 1 if the CEO turnover in this year, and 0 otherwise. Remote Dummy equals to 1 if the observation belongs to the bottom 40% distance range(remotely located), and equal to 0 if the observation belongs to the top 40% distance
range(centrally located). Small Dummy equals to 1 if the observation belongs to the top 20% firm size range, and 0 if the observation belongs to bottom 20% distance range. All variable definitions can be found in Appendix 1. Industry fixed effects are at the SIC level. T-statistics based on standard errors clustered by gvkey and robust to
heteroskedasticity are reported in the parentheses below the estimated coefficients. *, ** and *** indicate statistical significance at the 10%, 5% and 1% levels, respectively.
（1） （2） （3） （4） Remote Dummy 0.0023 0.0034 -0.0027 0.0017 (0.49) (0.65) (-0.47) (0.11) Small Dummy 0.006 -0.0056 -0.019** -0.054** (1) (-0.85) (-2.2) (-2.2) Remote×Small -0.028** - 0.054* (-2.2) (-1.9) Firm Size -0.015* (-1.9) Firm Age 0.0013 (1.1) Marker/Book -0.0049 (-1.4) CEO Age 0.0085*** (8.8) CEO tenure -0.0069*** (-7.5) Constant 0.11*** 0.15*** 0.16*** -0.19** (31) (9.5) (9.6) (-2.1)
Year fixed effects No Yes Yes Yes Industry fixed effects No Yes Yes Yes Observations 17627 17627 17627 4049 Pseudo R2 0.0001 0.0248 0.0250 0.1067
Table6
Table6 Regression analysis of the correlation between CEO turnover rate and CEO age In each year of the 20-year period from 1995 to 2015. The dependent variable is CEO turnover rate in each year. CEO turnover dummy equals to 1 if the CEO turnover in this year, and 0 otherwise. Remote Dummy equals to 1 if the observation belongs to the bottom 40% distance range(remotely located), and equal to 0 if the observation belongs to the top 40% distance range (centrally located). Young Dummy equals to 1 if the CEO’s is less than 60 years old, and 0 otherwise. All variable definitions can be found in Appendix 1. Industry fixed effects are at the SIC level. T-statistics based on standard errors clustered by gvkey and robust to heteroskedasticity are reported in the parentheses below the estimated coefficients. *, ** and *** indicate statistical significance at the 10%, 5% and 1% levels, respectively.
（1） （2） （3） （4） Remote Dummy -0.011 -0.012 0.0042 -0.076 (-1.1) (-0.54) (0.14) (-1.4) Young Dummy -0.17*** -0.17*** -0.16*** -0.13*** (-15) (-11) (-7) (-2.8) Young×remote -0.022 0.0019 (-0.71) (0.032) Firm Size 0.001 (0.072) Firm Age -0.00067 (-0.089) Marker/Book -0.0049 (-1.4) CEO Age 0.0061** (2.3) CEO tenure -0.015*** (-6.6) Constant 0.11*** 0.15*** 0.16*** -0.19** (31) (9.5) (9.6) (-2.1)
Year fixed effects No Yes Yes Yes Industry fixed effects No Yes Yes Yes Observations 10233 10233 10233 2753 Pseudo R2 0.0001 0.0248 0.0250 0.1067
Table7
Table7 Regression analysis of the relation between likelihood to have a COO title in place and remoteness In each year of the 20-year period from 1995 to 2015. The dependent variable is COO existence dummy, COO existence equals 1 if there exits a COO title in this specific year, and 0 otherwise. Remote Dummy equals to 1 if the observation belongs to the bottom 40% distance range(remotely located), and equal to 0 if the observation belongs to the top 40% distance range (centrally located). All variable definitions can be found in Appendix 1. Industry fixed effects are at the SIC level. T-statistics based on standard errors clustered by gvkey and robust to heteroskedasticity are reported in the
parentheses below the estimated coefficients. *, ** and *** indicate statistical significance at the 10%, 5% and 1% levels, respectively. （1） （2） （3） Remote Dummy 0.00082 -0.00024 0.0011 (0.22) (-0.047) (0.13) Firm Size 0.0055 (1.6) Firm Age -0.00099 (-1.3) Marker/Book 0.00026 (0.22) Constant 0.064*** 0.0049 -0.032 (24) (0.48) (-0.97)
Year fixed effects No Yes Yes Industry fixed effects No Yes Yes Observations 17627 17627 7599 Pseudo R2 0.0000 0.0364 0.0568
