!
MSc Business Economics, Finance track
Master Thesis
Economic Cycles, Financial Constraints and CEO Turnover
Author: Xianjing Cai
Administration number: 10874089
Supervisor: Dr.F.S.Peters (Florian)
STATEMENTOF ORIGINALITY
THISDOCUMENTISWRITTENBY XIANJING CAI, WHODECLARESTOTAKEFULLRESPONSIBILITYFOR THE CONTENTSOFTHISDOCUMENT.
I DECLARETHATTHETEXTANDTHEWORKPRESENTEDINTHISDOCUMENTISORIGINALANDTHAT NOSOURCESOTHERTHANTHOSEMENTIONEDINTHETEXTANDITSREFERENCESHAVEBEENUSED INCREATINGIT.
THE FACULTYOF ECONOMICSAND BUSINESSISRESPONSIBLESOLELYFORTHESUPERVISIONOF COMPLETIONOFTHEWORK, NOTFORTHECONTENTS.
ABSTRACT
This paper represents that boards tend to fire CEOs after bad firm performance, however firm-specific performance is not the only factor involved in boards’ decision making. Both industry performance and financial constraints influence a forced CEO turnover in the economic cycle. Based on some previous literature, I use forced CEO turnover samples from 1993 to 2010 to confirm the relationship between forced CEO turnover and economic cycle, CEO dismissal and peer group performance and financial constraints.
Table of Content
1. Introduction...1
2. Literature review...5
2.1. Economic situation and CEO turnover changed...5
2.2. Relative performance evaluation and CEO turnover...6
3. Hypothesis and Methodology...12
4. Sample and Data...16
5. Results...19
5.1. Testing stage one...19
5.2. Testing stage two...23
5.3. Boards evaluation on CEO turnover...25
5.4. Robustness tests...26
6. Conclusions...27
References...29
1. Introduction
Corporate governance is most often understood as the mechanisms, processes and relations by which corporations are controlled and directed. Due to some financial scandals and questionable business practices all over the world, people paid increased attention to corporate governance. In a firm, it is very important for the board of directors to shape corporate governance practices. According to a firm, generally one of the main responsibilities of the board of directors is to limit self-serving behavior of managers, and to evaluate whether a CEO should be replaced or kept (Dharmadasa, Gamage and Herath, 2014). At the same time, boards also spend a lot of time on controlling and advising the management on the strategic direction of the firm. Moreover, the board has the ultimate decision-making authority and is empowered to set company’s policy and overall direction. Previous literature is focused on CEOs’ performance according to the financial crisis, and their databases generally cover a short period without a complete business cycle. Then, the
influence of corporate governance on the “Boardroom” is still an attractive topic. Additionally, the financial crisis of 2007 and 2008 resulted in great recession in global economics, it may be worth it to discuss about the business cycle and firms, especially how firms best respond to them.
The literature correlated to the study topic is limited, and only some empirical literature on the relationship between board decisions and business cycles are used in this paper. While few other papers can offer some support to my hypothesis. For example, in the paper written by Philippon (2006) he built a model to evaluate the preferenced behavior of managers and shareholders in good times and provided empirical evidence that “badly governed firms respond more to aggregate shocks than do well governed firms”. This research is related to the microeconomic literature on governance conflicts between managers and shareholders, and discusses these governance conflicts in the standard dynamic macro model. This paper is important and related to my thesis because my thesis is tends to study the relationship between business cycle and corporate governance.
Then, Jenter and Kanaan (2006) examined whether CEOs are fired after bad performance caused by factors beyond the CEO’s control. The article is very helpful and offered several hypotheses and could support my idea of boards’ action that related to CEOs. In the same paper, they documented that firm performance affects CEO replacing strongly when the firms is not doing well. They
pointed out that performance in recessions is more informative about CEO quality than performance in booms, and CEOs are significantly more likely to be fired after bad industry or market
performance. They also found that boards tend to blame the CEOs for performance caused by factors out of their control, moreover, underperforming CEOs are more possible to be affected by attribution errors than outperforming CEOs. Besides, the article pointed out that stock returns relative to the overall market may a better predictor of CEO dismissals rather than absolute performance. Also, boards of directors filter industry and market from firm performance when deciding whether to dismiss their CEO. Moreover, they discussed several extensions to the basic CEO turnover and that one of the modifications shows that performance in recessions is more informative about CEO quality than performance in booms. A paper by Schoar and Washington (2010) explained that in corporate governance structure, in respect to firm performance, managers and shareholders tend to perform differently. Also, person-specific styles of CEOs or other top managers affect the operations of the firms, but the quality of the managerial labor market affects the CEO’s career path and management style, these styles are then brought into their companies (Schoar and Zou, 2011). They concluded that the early environment would likely affect the management styles into heterogeneity and this could lead to potential mismatches if the economic conditions change radically.
Thus, I would like to discuss the relationship between corporate governance and economic cycle. More specifically the change of boards decisions according to the economic cycle. Much research has been done on performance and decisions about corporate governance, and how economics
affects firm value or firm policies. As a consequence of the financial crisis, the effects of the financial crisis and how firms respond to them became a hot issue. Therefore, the research question of my thesis is: Will governance itself changes with the business cycle?
In order to answer the question above, I would like to focus on the sub-questions as following: How firms best respond to influences caused by the financial crisis? Do boards tend to perform
differently than before? For example, change manage team, dismiss CEO, or have higher risk aversion in busts than booms? Will different types of firms have varies respond to the economic cycle?
Most existing literature has examined that CEO turnover would be effected by the financial
recession, but few researches offer a detailed detection on the board decisions and economic cycles. Besides, even though currently the main idea is that forced CEO turnover would increase when the firm or industry is in bad times. Other literature still argues that more CEOs are fired by boards in good times rather than in bad times. In order to confirm this question, it is important for me to make establish a relationship between CEO turnover and economic cycles. To fully conduct the research I divide the research into two stages. First stage is to determine the percentage of forced CEO
turnover during booms or in busts. Beside voluntary retire, CEO remaining or dismissing is managed by the board of directors. Several papers believed that firm performance or industry conditions would affect the boards’ evaluation on CEO performance. I predict that bad firm performance would increase the possibility of forced CEO turnover, and that the whole industry environment matters as well. At the same time, I predict that there is a difference between firm performance and industry performance (named as firm specific performance). I will also examine how these factors affect CEO turnover. In conclusion, in stage one I will describe the relationship between CEO turnover and GDP, as well as the relationship between CEO turnover and both firm-specific and peer group performance.
The research of second stage would be based on the regression model of the first stage but will have some added parameters. In this stage we will add some financial constraints to the firms as another variable to see whether it will have interaction with the time period that can affect decisions made by boards on CEO retentions. The financial crisis, or just bust times of economies may have
different impacts on firms. For example, big companies such as IBM, Walmart, Citigroup and so on, may not be impacted much by these bust times. Because even in a bad time, they still have ways to finance or invest due to their high credit. However, this is different for small companies. When they face bad industry conditions, it is harder for them to find financing. Therefore, with financial constraints included, we will know which factors are more important to firms that would affect boards’ decision on CEO turnover. Moreover, I would like to emphasize that the study in the thesis will focus on boom and bust period from the year 1993 to 2010, which include a complete
economic cycle. Good time of economies or bad time of economies will be a variable and I will measure it with stock return or GDP data. Even though part of the research in the thesis is based on the study by Jenter and Kannan (2006), the factors of financial constraints make this thesis different from previous literature and will also make a contribution.
This paper proceeds as follows. Section two presents an overall literature review about the prior research that relates to corporate governance, board characteristics and firm performance, as well as some prior discussion about CEO performance and board behavior. Subsequently section three formulates the research hypotheses and outlines the research method. Section four lists data sources and provides some descriptive statistics about database. Section five presents and discusses the results of this study, and robustness test will also be included. Lastly, the conclusion section summarizes the whole research and discusses the limitations for this study.
2. Literature review
An economic cycle consist of boom and bust cycle. A boom occurs when real national output is rising at a rate faster than the trend rate of growth. During a boom the market brings high returns to investors and the economy performs well. A recession, or a bust, means a fall in the level of national output, and during the bust the economy shrinks and incomes or profits also contracts. Klein (1990) pointed out that the classical business cycle is defined as “Recurring expansions and contractions in the absolute level of aggregate economic activity and growth cycles, which consist of fluctuations in the rate of growth of economic activity”. Since 1993, the most rigorous boom happened in 2000 and a recession happened in 2007, thus taking the period between 1993 and 2010 contains both a bust and a boom, therefore it is an economic cycle.
2.1. Economic situation and CEO turnover changed
Since the 1990s, the US economy persistently maintained a positive growth rate. Robust personal consumption, rising stock price and relative low interest rates sustained a good economy
environment. From 1994 to 2000, the growth of real output and technological innovation of the computer revolution supported the development of economies. At the same time, the rising
emerging economies provided lots of cheap products to the United States to stimulate consumption. However, around 2000 this economic bubble burst, an economic depression lead to a sharp
downturn in 2001. Economic deterioration in 2001 with only 0.3% national output growth ,sharply raised unemployment and bankruptcies. Even though from 2002 the United States tried to recover their economy, a financial crisis happened again around 2007 and they entered a subprime crisis. Under such a volatile economic environment, a variety of industries- and firm performances were also influenced. In order to protect their own benefits, boards of firms tended to adjust all kinds of
policies to adapt to the situation. CEO ability seemed important to them and several factors affect boards’ decisions on CEO turnover.
Early research showed that CEO turnover was increased in the 1990s (Jensen et al, 2004). Then Kaplan and Minton (2012) studied CEO turnover in U.S. companies from 1992 to 2005. They found that from 1992 to 2005 the annual CEO turnover was 15.6%, while since 1998, CEO turnover increased to 17.4%. Most CEOs who were in place in 1992 were fired or quit their job during the following years, only 36% of them remained CEO in 1997. The ratio of retention of CEO then decreased to 25% during the period of 1998 to 2003. Including the condition of takeovers and bankruptcies, CEO retention appears more uncertain than before especially after 1997 (Kaplan and Minton, 2012). Some other literature in early years combine research of Kaplan and Minton (2012) 1
showing that the relationship between turnover and performance has changed, and CEO tenures became more sensitive to market condition and also CEO pay. The decline of CEO tenures tends to imply that boards attitude towards CEO retention may change when they consider different factors for decision-making. Not only firm performance, but other factors would also become important elements that could affect boards decision-making.
2.2. Relative performance evaluation and CEO turnover
Standard economic theories suggest that the board of directors should filter out exogenous shocks from firm performance when they evaluate the quality of their CEO. However, Jenter and Kanaan (2006) pointed out that CEOs are more likely to be fired after negative performance, and exogenous shocks are also involved when the board make decisions on CEO retention. In their research, they
Mikkelson and Partch (1997) compared complete management turnover in two period: 1984 to
1
1988 and 1989 to 1993; Huson, Parrino, and Starks (2001) studied four six-periods during 1971 to 1994.
used 3365 CEO turnovers as the sample and documented the relationship between stock or market return and CEO turnovers: Lower market returns leads to higher a frequency of forced CEO turnovers. Also, firm performance affect their evaluation of the CEO, even though it is because of bad market performance that may affect the whole industry, these exogenous shocks still influence their CEO retention decisions. Jenter and Kanaan (2006) also pointed out that according to most of the theoretical literature , the prediction follows that forced CEO turnover should happen in good 2
times rather than in bad times. However their empirical results strongly reject this prediction.
Moreover, they concluded that a theoretical point is that mostly CEO turnover happened because firms tend to choose the “right” person and prefer to select a CEO with matching quality instead of 3
the case of pay-for-performance. They then explained that boards may mistakenly blame CEOs for exogenous performance shocks, and during financial crisis, the firm-specific performance affects CEO turnover more than in booms. In their research, Jenter and Kannan (2006) emphasize two ideas: Firstly, forced CEO turnover happened more frequently in bust than in boom. Secondly, both bad industry performance and firm-specific performance have strong effects to the boards when they consider about the CEO retention decision, and both industry- and firm-specific performance would increase the probability of a CEO dismissal.
These ideas are supported by Peters and Wagner (2014). According to their research, they first make the idea clear that CEOs are fired by boards for bad firm performance. Then they further concluded
The theoretical literatures are about CEO dismissals and mentioned in the paper “CEO Turnover
2
and Relative Performance Evaluation”. Jenter and Kannan (2006) also annotated examples: Hermalin and Weisbach (1998, 2003), Warther (1998), Adams and Ferreira (2007), and Taylor (2010).
The idea is that pay-for-performance scheme could not be a credible manner, because if the board
3
dismiss the CEO after bad performance, while the fired CEO may still be suitable for the job, but the board cannot find it.
that CEOs have higher likelihood to be dismissed when companies are in turbulent industry
conditions. They also indicated that the outside options of firms are changed by industry shocks, so for boards it is advisable to consider industry conditions in their decision of CEO retention . Peters 4
and Wagner (2014) argued that a different managerial skill set may be required when industry condition changed, and the match between the requirements in industry and the CEO’s skill set tend to be weaker than before. As for the mismatch between requirements and actual skills, the board would fire the current CEO.
The idea mentioned above seems to be the main idea about the relationship between CEO turnover and firm performance. According to most literature, even though CEOs are more likely to be forced out because of their poor performance, overall industry performance matters too (Eisfeldt and Kuhnen, 2013). Eisfeldt and Kuhnen (2013) then concluded that when industry conditions effect the outside options of firms and manager, the shock will drive managerial turnover. Evidence from Jenter and Kanaan (2006) illustrated that when industries have performed badly, CEO dismissals becomes more common. When the board of directors considers the CEO retention decision,
industry performance cannot be ignored. Besides, when industry conditions change, a CEO is more likely to be fired by the board (Peters and Wagner, 2014). All above literature supports the idea that the boards tends to dismiss CEOs in a bust rather than in boom.
Firm performance is effected by economic situations, which leads boards to have different evaluations on CEO performance. The research of Eisfeldt and Rampini (2008) illustrated the relationship between output and CEO turnover and executive compensation of the economic cycle. In their study, the result showed that correlation between CEO turnover and output is highly significant. There is no doubt that CEO turnover would change with the economic cycle because
A competitive assignment model that provided by Eisfeldt and Kuhnen (2013).
boards who are in a turbulent industry situation tend to adapt to the environment and try to better respond to the business cycle. However, in their study they presented a different conclusion. The article was not focused on the relationship between CEO turnover and business cycle, but was focused on the problems of CEO compensation and capital reallocation and agency problems between owners and managers. They pointed out that a CEO can have a higher average cash compensation in good times, but have lower compensation in bad times. Because it is the investor who needs to finance the bonuses and payments. Hence, to the investors, reallocation and associated productivity gain is too costly that they would prefer to forgo these actions. Eisfeldt and Rampini (2008) found in their results a connection between CEO turnover, executive compensation and the business cycle. They found a correlation between CEO turnover and output, a correlation between executive compensation and output, and a correlation between CEO turnover and executive compensation which were all significant. However their results showed that in good times, for example in 2000, GDP is very high, and at the same time, CEO turnover is high too. While from 2002 to 2003 the economy was in a recession with low GDP, and CEO turnover also decreasing. According to their research, because of the countercyclical agency costs and aggregate capital reallocation, CEO turnover and managerial compensation should be procyclical. In good times boards can expect high productivity managers, with a better CEO they will receive more additional capital but have relatively lower agency costs. In bad times the pressure of reallocation is big to boards which leads to less efficient deploy of capital. Therefore, boards tend to fire CEO in good times rather than in bad times because they can gain more benefit (Eisfeldt and Rampini, 2008).
Although the research result from Eisfeldt and Rampini (2008) is different from the main ideas, it is still necessary for us to confirm the relationship between CEO turnover and economic cycle, which is a part of my thesis. To measure the economic cycle, I consider to use not only GDP but also stock returns. Stock returns could reflect firm or industry performance and they are factors that boards
would refer to in CEO performance when making the final decision. Kaplan and Minton (2012) once distinguish CEO turnover as internal turnover and external turnover, in which internal turnovers are initiated by boards while external turnovers are related to acquisitions. They then found that external turnover are only affected by industry environments and market situations, internal turnover is more complicated. It is known that the CEO usually makes most major decisions, so different individuals would affect decision-making that leads to different consequences for the firm and the role that boards play is very important (Adams et al., 2005).Compared to the early 1980s, in order to evaluate a CEO performance and decide CEO retention, boards tend to consider more factors that measure CEO performance. Kaplan and Minton (2012) explained that according to their research, boards become more sensitive to three
components of stock performance, which are poor industry-adjusted stock performance, bad
industry performance and lower overall market performance. Similarly, Eisfeldt and Kuhnen (2013) proposed that performance indeed affects CEO turnover. Although CEOs are mostly fired after poor firm performance, overall industry performance cannot be ignored. Because industry conditions would “determine the most desirable managerial skill set” and “impact the outside options of matched firms”, thus industry shocks naturally drive CEO turnover (Eisfeldt and Kuhnen, 2013).
My paper is closely related to CEO turnover and relative performance evaluation (Jenter and Kanaan, 2014) the first part of my research is similar to their literature. Similar to their research, I also focus on the relation among CEO turnover, firm-specific and peer group performance. On the other hand, Jenter and Kanaan (2014) used two-stage regression approach to estimate the sensitivity of CEO turnover to peer performance. In the first stage they use peer group performance and firm-specific performance to evaluate CEO ability. Then in the second stage they use “the estimated peer group component and the estimated residual component of firm performance” to predict the
from 1993 to 2009. This thesis is based on their paper to find the relationship between CEO turnover and economic cycle, however I also take into account financial constraints and will give further explanation on the financial constraints as well.
3. Hypothesis and Methodology
In this paper, I would like to detect the response of the board of directors to dynamic economic cycle, especially their different performances during the recession period rather than in booms. As for the main responsibility of the board of directors is to evaluate CEO’s performance and
determine whether a CEO should be replaced or not, this paper would determinate the behavior of the board by detecting the relationship between forced CEO turnover and economic cycles. In order to finally examine the research question of the thesis, I would like to proceed the research into two stages.
First stage is to determinate whether the recession period or boom period will affect the firm performance or the board decisions and what are the main factors that would impact boards’ decision on CEO retention. The theory of standard CEO turnover model predicted that exogenous 5
shocks to firm performance should not be a factor that would affect CEO dismissals. Jenter and Kanaan (2006) criticized that the model is too simple that only takes firm-specific component of firm performance into account when they evaluate CEOs performance. As a result, the model does not reflect the true reaction of boards to turbulent economic situation in the real world.
Based on several researches I mentioned in the previous section, I would like to take not only firm performance but also industry situation into account. My hypothesis is as follows:
The decisions made by the board of directors tend to be sensitivity to not only firm-adjust performance, but also industry performance and market condition. Economic cycle influence
This model argued that because CEO ability would not change with the economic cycle or other
5
exogenous shocks, according to the boards the ability of CEO should be the same in either boom or bust. Therefore, the likelihood of CEO dismissal could not be predicted by the performance of the peer group (Jenter and Kanaan, 2006).
industry environment and then boards are more likely to fire the incumbent CEOs in recession than in good times.
Black and Bhagat (1999) indicated that firm performance usually improves after a CEO is replaced. This statement seems to support the idea that boards prefer to replace the CEO in bad economic times, especially when the firm performance is very poor. However, some factors, such as age, may also affect CEO replacement. Therefore I should eliminate those factors from the database and only keep the data that indicates forced CEO turnover. More details about sample data will be discussed in the following section. Here, in essence the formula looks as follow:
1) Prob (CEO dismissal) = Φ(α + β1* GDPt +
ε
i,t )2) Prob (CEO dismissal) = Φ (α + β1*
r
jt + β2* (r
it -r
jt) +ε
i,t)Both regressions will use a probit regression model. The first equation evaluates the relationship between CEO turnover and output. GDP is an indicator that reflects the condition of the
macroeconomics (whether the economic is experiencing boom or recession period) and the percentage of CEO dismissal is the percentage of CEOs that are fired by the boards. It would be record as 1 if the CEO turnover was forced, otherwise it will equal to 0. This formula is used to confirm whether forced CEO turnover is relative to the macroeconomic condition. Besides, my estimation of output is relative to the forced CEO turnover, which shows that the macroeconomic environment would affect boards’ decision on CEO dismissal.
The second formula uses the model from Jenter and Kanaan (2006). They developed a model to estimate the sensitivity of CEO turnover to peer performance. They used a two stage regression approach to test the probability of CEO dismissal related to firm performance and reference group performance. The model I use would directly predict the likelihood of a CEO dismissal using
firm-specific performance and peer group component, where
r
jt stands for the average industry stockreturns at time t,
r
itrepresents the stock returns of firms and then the deviation between the twovariables means the estimated residual component of firm performance.
After confirming the hypothesis in the first stage, I can further consider the second stage to better answer the research question. In the second stage we take firm characteristics into account when we consider the relationship between CEO dismissal and economic cycle. The hypothesis is as follow:
Boards’ decision on CEO retention not only affect by firm and industry performance, but also related to their firm characteristics. Firms with more financial constraints would feel more difficult in recession than in boom, and boards in these firms are more likely to fire CEOs after bad
performance.
Then the formulation is as below:
1) Prob(CEO dismissal) = Φ (α + β1*
r
jt + β2* (r
it -r
jt) + β3*ω
it + β4*[r
jt *ω
it] +
εi,t )2) Prob (CEO dismissal) = Φ (α + β1*
GDP
+ β2*ω
it +β3*[GDP
*ω
it] +
εi,t )In equation (1) I take financial constraints into account. The first two parts
r
jtand r
it -r
jt stands for the same meaning as they are in first stage, whereω
it implies financial constraints. In the formula, (r
jt *ω
it)
indicate the weights on different firms. The second formula is used to detect therelationship between CEO turnover and GDP weighted by financial constraints, where GDP is log GDP per capita,
ω
it stands for financial constraints and measured by KZ index. GDP* ωit impliesinteraction of GDP and financial constraints. As we know that when economic conditions are good and stable, all firms would feel easy to finance their ongoing operations, and under this situation big firms or small firms may not have too much difference. However, even though all firms go through financial recession, to different firms it is more difficult to pass the recession than for other firms.
For example, companies such as Apple or IBM may still have ease in financing their operations because they are big companies with enough credit. However other median or small companies would have more difficulties finding financing and may even have the risk of bankrupcy. Therefore, the financial constraints should have affect on boards from different companies in varying degrees, and the industry performance and firm performance may have varying impacts on different boards on forced CEO turnover. Then a CEO with professional skills or good ability may become more important to some firms than others. Additionally, I predict that because of financial constraints, CEOs are more likely to be fired in downturns than in good times.
Coming back to the equation itself. Here
ω
it is an important variable. I will use Kaplan-Zingales Index to identify this variable. Companies with a higher KZ-Index score are more likely to 6experience difficulty when the economic situation is under bust or recession. A higher KZ-Index also implies that companies would have more difficulty in financing their operations and the economic cycle should have more effect on them. The original formula of KZ-Index shows as below:
KZ Index(t-1) = -1.001909 x Cash Flows(t-1)/ K(t-2) + 0.2826389 x Q(t-1) + 3.139193 x Debt(t-1)/ Total
Capital(t-1)+ '-39.3678 x Dividends(t-1) / K(t-2) + -1.314759 x Cash(t-1) /K(t-2)
The five variables in the index are: cash flow to total capital, market to book ratio, debt to total capital, dividends to total capital and cash holdings to capital (Lamont et al., 2001). According to the index, K stands for total property, plant and equipment, which is lagged data.
The KZ-Index is a relative measurement of reliance on external financing. Note: “1) The model is
6
a probabilistic model, so it will not perfectly predict which companies are financially constraintsed and which are not, but it should help to raise warning flags. 2) The synthetic KZ-Index that we use was developed only for non-financial firms, so it is safest to disregard financial firms' KZ-Index scores.” ( Financial Glossary Index)
4. Sample and Data
The data information of all firms and CEO turnover are gathered from Standard & Poors
ExecuComp database in the period of 1993 to 2010. The ExecuComp sample includes information on the top executives of all firms in the S&P 500, S&P MidCap, and S&P SmallCap indexes. The final sample records 911 forced CEO turnover during the nearly twenty years. Part of the data comes from a merge version of data collected by Jenter and Kanaan (1993-2001) and Peters and Wagner (2001-2010). They classified each CEO turnover according to whether the turnover was forced or voluntary. The determination of the standard of classification of CEO turnovers into forced or voluntary follows Parrino (1997): The departures of CEOs that are above and including age 60 are determined as voluntary, departures that are caused by policy differences or pressure, for example the CEO is fired, forced out or resigns, are classified as forced. Additionally, they exclude CEO turnovers associated with mergers and spin-offs from the analysis.
Furthermore, the state of the economy is measured by the Fama-French industry classification. All accounting information referred to firm characteristics are annual files, while equal-weighted and value-weighted average stock returns are data information that identified industry performance and collected through monthly CRSP tapes. On the other hand, GDP can reflect the state of the
economy, especially when exogenous shocks happened. These macroeconomic data come from Federal Reserve Economic Data (FRED). The data of stock returns and accounting data comes from the Center for Research on Securities Prices (CRSP) and CRSP /Compustat Merged. All data are focus on the United States market.
More specific, table 1 presents a summary of the CEO turnover data set. Panel A reports the frequencies of forced and voluntary turnovers. The final sample includes 6590 firms with 29741 firm-year observations from 1993 to 2010, in which there are 852 forced CEO turnover. This CEO
turnover data is recorded as 1 if a CEO turnover is classified as forced, otherwise it will recorded as a 0. The mean of CEO turnover is about 0.028, which indicates that during these years about 3% CEO are fired by boards. Panel B indicates GDP dataset from 1993 to 2010. I use annual real GDP and annual population data to calculate log GDP per capita, which record as GDPPC=ln (GDPC1/ POP*1,000,000). Then I separated the time series into trend and cyclical components because trend components may be non-stationary while cyclical components are stationary and are driven by stochastic cycles at a range of periods. Furthermore, Panel C and Panel D represents all information about stock returns, where Panel C is the basic stock returns information that reflect firm
performance and industry performance and Panel D illustrates the probability of forced CEO turnovers by industry performance quintile. In Panel C “Firm-specific returns (VW returns)” is the difference between firm performance and industry performance using value-weighted industry stock returns and “Firm-specific returns (EW returns)”. EW returns has the same definition as the VW returns but uses equal-weighted industry stock returns. Then, similar to what Jenter and Kanaan (2006) did in their research, the result in Panel D suggests that comparing to good industry
performance, forced CEO turnover is more likely to happen in industries that have performed badly. Using average value-weighted industry stock return, the likelihood of CEO dismissal is 3.18% in lowest industry performance quintile but is 2.62% in highest industry performance quintile. The results obtained through equal-weighted industry stock return seems a little bit instable and slightly weaker than the previous results. This is different from the research of Jenter and Kanaan (2006) because they found that EW industry stock return should have been stronger. However the general result here also supports the idea that badly industry performance drives boards to fire their current CEO.
Additionally, Panel E summarizes all accounting variables that are used as financial constraints. As the financial constraints are identified by KZ Index, which is:
KZ Index = -1.001909 x Cash Flows / K + 0.2826389 x Q + 3.139193 x Debt / Total Capital - 39.3678 x Dividends / K + -1.314759 x Cash /K
Based on the research of Lamont et al. (2001), and Gorodnichenko and Weber (2013), I define Cash Flows as the sum of Income Before Extraordinary Items and total Depreciation and Amortization at time t, K defined as property, plant and equipment at time t-1. Q = (Market Capitalization + Total
Shareholder's Equity - Book Value of Common Equity - Deferred Tax Assets) / Total Shareholder's Equity, where all items are use data at time t. Then I determine Debt = Total Long Term Debt + Current Portion of Long Term Debt. Again, all items are at time t. Among these items, Dividends is
the sum of preferred dividend and common dividend. Market Capitalization is the number of common shares outstanding times the closing price.
5. Results
5.1. Testing stage one
In this section, I test the relationship between forced CEO turnover and economic output, and also the impact of industry performance and firm-specific performance to boards’ decisions on CEO retention. It is known that boards expect CEO with good ability such as better organizational skill or good leadership, and can make “right” decisions to bring more benefit for the company. Boards tend to evaluate a CEO based on his or her performance, which can be reflected by firm performance. However, according to an economic cycle, it seems that the ability of a CEO is more important in bad times than in good times. Except firm performance, I would like to estimate whether peer group performance and firm-specific performance will affect boards’ decisions.
I predict that when boards consider CEO retention, they would take industry performance into account. That is, when the whole industry performs badly that leads to a worse firm performance than before, boards would prefer to blame the CEO and may fire the current CEO. However from this idea we could not simply infer that the likelihood of forced CEO turnover would be higher in bad times than that in good times. Therefore I will first confirm the relationship between CEO dismissal and GDP, and then determine the sensitivity of boards’ decision making to exogenous effects from industry and firm-specific performance.
I use a simple probit regression to estimate the relationship between output and CEO turnover. The model I use to test CEO turnover and peer group and firm-specific performance is based on the model built by Jenter and Kanaan (2006). According to their research they test both strong-form and weak-form relative performance evaluation regress forced CEO turnover on peer group
include any dummy variable as Jenter and Kanaan (2006) did before because I would not take “likely retirements” or “firm equity owned by CEOs” into account.
Then panel A of table 2 represents relationship between forced CEO turnover and log GDP per capita. The coefficient of GDP per capita is about 0.58 with standard error of 0.17. The p value of GDPPC is 0.003, which is highly significant. The result shows that GDP can indeed predict the possibility of CEO turnover, however the relationship between GDP and forced CEO turnover is positive. This result is concordant with research of Eisfeldt and Rampini (2008), where they also concluded that CEO turnover has a positive relationship to the output. The result in table 2 indicates that CEOs are more likely to be fired when there is a higher GDP. Panel B uses GDP cyclical instead. I use the Hedrick-Prescott high-pass filter to separate the time series into GDP trend components and GDP cyclical components. In this case GDP cyclical is not significant anymore. The coefficient is 0.008 and standard error is 0.005, but p value is 0.134.
Table 3 represents the sensitivity of forced CEO turnover to peer group performance and firm-specific performance. I using average value-weighted industry stock returns and average equal-weighted industry stock returns separately. The returns measure industry performance, and the difference between holding period returns and average VW or EW returns refer to firm-specific performance. Firstly, column (1) records the probit regression of forced CEO turnover using average value-weight returns which shows that the coefficient of average value-weighted returns is -0.0220592 with a standard error of 0.0086916 and p value is 0.01. The coefficient of firm-specific returns is -0.0014598 with a standard error of 0.0008226, and p value is 0.076. The result pointed out that forced CEO turnover is negative to both firm-specific performance and peer group performance. That is, when firm performance or industry performance is bad, the possibility of CEO dismissal increases. Besides, both industry performance and firm-specific performance show significant correlation with forced CEO turnover. The correlation between peer group performance
and CEO dismissal is -2.54, while correlation between firm-specific performance and CEO dismissal is -1.77, both are significant. We can then conclude that both firm and industry
performance can predict forced CEO turnover. In other words, the effect of industry performance on the percentage of forced CEO turnover is obvious, and boards indeed consult industry performance to evaluate a CEO performance when they need to make a CEO retention decision. Industry
performance strongly affects forced CEO turnover . At the same time, firm-specific performance can be used to predict the frequency of CEO dismissals, and boards decision making on CEO turnover would also rely on this factor.
Column (2) presents a similar result of column (1) but uses average equal-weighted industry stock returns. The coefficient is -0.013193 with a standard error of 0.007, and has a correlation of -1.86. The average equal-weighted industry stock returns can also be used to predict the frequency of CEO dismissals. Compared to the result in column (1), column (2) shows that firm-specific returns are still negative to the percentage of CEO turnover and firm-specific returns can be to predict the frequency of CEO dismissals with lower significance. However, the results pointed out that VW industry stock returns have stronger predictive power to forced CEO turnover than EW returns. Because the correlation with VW stock returns is highly significant under 1% significance level. Because the results of the regression of GDP and the regression of industry performance seem to have a contradictory statement, I then decided to make another regression. I use GDP cyclical and industry performance as independent variables to confirm the relationship between CEO turnover and the economic cycle to make the result clearer. Because a stationary cyclical component is driven by stochastic cycles at a range of periods, and trend component of GDP may be non-stationary, I use GDP cyclical instead of log GDP per capita in this regression. Panel B shows the result of this regression with both equal-weighted and value-weighted industry stock returns. When using VW returns, it shows that GDP cyclical has a positive relation to CEO dismissal, but not
significant at all, and average VW stock return is negative to forced CEO turnover with -0.017 coefficient and it still significant.
According to these regression, the results suggest that the percentage of forced CEO turnover usually increases after bad industry performance, but boards are more likely to replace their current CEO in good times rather than in bad times. Firstly, generally speaking, CEO turnover will be effected by economic cycle and boards are more likely to change CEO in good times rather than in bad times. There are two explanation that could interpret this phenomenon. One reason is the existence of agency problem between stockholders and managers leading to several conflicts. The boards expect CEO to gain more benefit for the company, but sometimes CEO may focus on their own interests. Usually replacing a CEO does not only mean changing one person, since it has happened that other managers have also been replaced. On the other hand, several previous researches have proved that CEO turnover risk is higher in recession while the risk is positively associated with compensation . When going through boom periods of business cycles, higher 7
average cash compensation reduces the reallocation costs borne by boards, and boards can expect high productivity CEO to bring more benefits. But in bad times the expected compensation is low and boards need to finance the payment of agency costs themselves. To boards, changing CEO in bad times seems more costly than in good time. The other reason is that in this regression GDP is the only independent variable without any other controlled variable. Even though CEO turnover and GDP are indeed correlated, we cannot simply conclude that boards tend to fire CEO in good times rather than in recessions because many other factors may be more important to boards when they are considering about CEO retention.
Besides, other results indicate that both firm and industry performance have strong correlation with forced CEO turnover. Boards may mistakenly blame CEOs for exogenous performance shock and
See Eisfeldt and Rampini (2008), Eisfeldt and Kuhnen (2013), Peters and Wagner (2014).
conclude bad firm performance as underperforming of CEO. When boards consider about retention of a CEO, they use firm-specific performance to evaluate performance or ability of CEO, at the same time they could not filter out exogenous industry shocks.
5.2. Testing stage two
In this stage my main purpose is to detect the impact of financial constraints to boards’ decision making. In stage one I already confirmed that industry- and firm performance would indeed have influence on boards’ evaluation of CEO ability and further affect possibilities of forced CEO turnover. However the influence may be different to different firms in varied industries. I use KZ index to weigh financial constraints on industry stock returns. Lamont and Saa-Requejo (2001) conclude five variables and signs of their coefficients in KZ index, which are cash flow to total capital (negative), market to book ratio (positive), debt to total capital (positive), dividends to total capital (negative) and finally cash holdings to capital (negative). Then we can detect the influence of those constraints to firms.
Panel A of table 4 represents probit regression using average EW industry stock returns (column (1)) and VW industry stock return separately (column(2)). In column (1), coefficient of industry returns, firm-specific returns, financial constraints and industry returns weighted by financial constraints are -0.015, -0.0012, 0.0012 and 0.0006. Their standard error are 0.0088, 0.0009, 0.0006 and 0.0003 respectively. Here we only focus on the impact of financial constraints and their weight to firm-specific performance to predict the probability of forced CEO turnover. Industry
performance still matters, and both of these two variables also show strong correlation to CEO dismissal under 10% and 5% significance level separately. Moreover, consistent with the result in stage one, industry performance is negative to forced CEO turnover, while financial constraints have positive relationship with CEO dismissal. That is, the more financial constraints, the higher
possibility of CEO dismissal. At the same time, firms with higher financial constraints would have stronger correlation with CEO turnover.
Column (2) is regress with average VW industry stock returns, which has the same result as column (1). However financial constraints under value-weighted industry stock returns have higher
correlation to CEO dismissal than using equal-weighted returns, which is 1.89, and financial constraints on firm-specific performance also shows stronger impact.
The results indicates that financial constraints would truly affect boards’ evaluation on CEO retention. KZ index points out that firms with higher KZ index are more restricted. Higher KZ index also means that firms would have higher debt and lower cash or dividends, which makes companies especially vulnerable in bad times. Moreover, under the same situation, CEOs are more likely to be fired by firms with higher KZ index because financial constrains may impel boards to put more emphasis on firm performance.
Therefore, companies have lower cash flow to total capital, higher market to book ratio and debt to total capital, or lower dividends or cash would have more difficulty financing their project. The results then illustrate that it is more common for CEOs in firms with higher financial constraints to be fired by boards after bad performance. Bad industry performance and high KZ index would increase the possibility of forced CEO turnover.
Then panel B represents the relationship between CEO dismissal and financial constraints on GDP. I use log GDP per capita, KZ index and GDP weighted by financial constraints as independent variables. The result shows that financial constraints have very strong correlation with CEO dismissals, at the same time, the financial constraints and GDP have interaction on forced CEO turnover. Both variables are highly significant at 1% significance level with a p value of 0.004. On the other hand, financial constraints are positive to CEO dismissal and both GDP per capita and GDP weighted by financial constraints are negative to forced CEO turnover. Consequently, CEOs
have higher possibility to be fired in recessions than in booms, and CEOs in firms with more financial constraints are more likely to be replaced after bad performance.
5.3. Boards evaluation on CEO turnover
Generally, the results of the regression I did are in accordance with ideas by several pervious researches. With the change of economic cycle, boards tend to change their behavior to respond to the cycle. The two stage approach confirmed my hypotheses in section three that both firm- and industry performance would affect boards evaluation on CEO. Besides, financial constraints are also an important factor that obviously influences boards’ judgment and decision making.
When boards consider whether to fire a current CEO, the most direct reference is firm performance. However, exogenous shocks of industry or market should also be taken into account. Jenter and Kanaan (2006) and some other researches proposed several possible reasons to explain this phenomenon. Firstly, it happened that firms performance are improved after fire underperforming CEOs in recession, then boards tend to consider dismissal of CEO as an efficient method. In downturns the ability of the CEO seems more important in booms because the macroeconomic environment is turbulent. Therefore, recessions would change the requirement skills of CEOs and boards may expect a CEO with a better skill set. Secondly, Fisman et al. (2014) pointed out that recession would lead to greater pressure on boards so that they may improperly blame the
underperformance of firms on underperforming CEOs. Thirdly, in the 1990s and 2000s, CEOs with limit abilities may have been hiding behind good industry performance (Bebchuk and Fried, 2004). Besides, financial constraints offer different weight to boards when they consider whether a CEO should be retained. This can happen because firms with different characteristics have various degrees of abilities to respond to the economic cycle. Firms with higher cash or dividends, or lower debt would have less issues to finance their projects even in downturns. For example, big
companies may have higher credit than small companies. In contrary, firms which have more constraints would be weighted more on industry performance, therefore their boards would also care more about industry performance when they evaluate their CEO’s ability.
5.4. Robustness tests
I then did some robustness check, and all results are shown in table five.
For the robustness check, I firstly repeat the estimation as I did in previous section. Then I found that the results are similar to the ones in table 3 and table 4. Peer group performance can still reflect forced CEO turnover.
Secondly, I use accounting returns instead of stock returns to measure firm performance to examine whether industry performance and financial constraints have the same predictive power for forced CEO turnover. Panel A presents results of the regression between forced CEO turnover and industry- and firm-specific performance. Column (1) using value-weighted stock returns but column (2) using equal-weighted stock returns. The results illustrate that both industry and firm have strong predictive power to forced CEO turnover. Both of these two variables are significant under 1% significant level. Panel B then run the regression with financial constraints. I then found that the results here are similar to the results in stage two. Both firm- and industry- performance could determine the possibility of forced CEO turnover. Firm-specific returns weighted by financial constraints have a strong correlation with CEO dismissal under 5% significant level.
6. Conclusions
In this thesis I focus on the relationship between boards’ attitude and behavior to CEO turnover according to the economic cycle. Many researches have been discussed about corporate governance and firm policies, however, it is also worth to pay attention on how firms’ best respond to financial recession in the economic cycle. According to the standard economic model, boards may become overoptimistic in good times, but overly pessimistic in bad times. Besides, the characteristics of economic cycles push firms closer to default in downturn than in boom. These features would affect boards’ level of risk aversion, in that boards would have higher aversion in downturns and lower risk aversion in good times.
To summarize, I have documented that not only firm performance, but also industry performance would affect the frequency of forced CEO turnover. The results of my research showed that when boards consider CEO retention, firm-specific performance would not be the only factor that boards need to take into account. According to boards, industry or market performance are also important. Therefore, some exogenous shocks from industry or uncertain macroeconomic environments would also affect boards’ decision on CEO turnover. Especially after bad firm or industry performance, the possibility of forced CEO turnover would increase.
Besides, through the test of effect of financial constraints, I then can conclude that financial constraints have strong effect on CEO dismissal. Firms with more financial constraints would care more about exogenous shocks, because compared to firms with lower constraints, they have more difficulties to get through a financial recession. Financial constraints lead to varied attitudes of companies to the economic cycle and they tend to make different decisions on CEO turnover. However there are also some limitations in the thesis. Firstly, I generalized boards. The constitute of boards (insider or outsider) may also be a factor in their behavior. Secondly, I did not pay more attention on the agency problems between boards and CEOs, which should also be an important
factor which influences boards’ decision in different periods of the economic cycle. Also, more causes of industry or market performance which have effect on CEO turnover should be identified and analyzed. In conclusion, the results of my thesis supports ideas of previous literature that boards are influenced by industry and market performance when they evaluate the performance or abilities of a CEO, and in bad times CEOs are more likely to be fired by boards than in good times. Lastly, financial constraints cannot be ignored.
References
Adams, R. B., Almeida, H., & Ferreira, D. (2005). Powerful CEOs and their impact on corporate performance. Review of Financial Studies, 18(4), 1403-1432.
Bhagat, S., & Black, B. (1999). The uncertain relationship between board composition and firm performance. The Business Lawyer, 921-963.
Dhamadasa, P., Gamage, P., & Herath, S. K. (2014). Corporate Governance, Board Characteristics and Firm Performance: Evidence from Sri Lanka. South Asian Journal Of Management, 21(1), 7-31.
Eisfeldt, A. L., & Kuhnen, C. M. (2013). CEO turnover in a competitive assignment framework. Journal of Financial Economics, 109(2), 351-372.
Eisfeldt, A. L., & Rampini, A. A. (2008). Managerial incentives, capital reallocation, and the business cycle. Journal of Financial Economics, 87(1), 177-199.
Gorodnichenko, Y., & Weber, M. (2013). Are Sticky Prices Costly? Evidence from the Stock Market. Working paper, University of California, Berkeley.
Jenter, D., & Kanaan, F. (2006). CEO turnover and relative performance evaluation (No. w12068). National Bureau of Economic Research.
Kaplan, S. N., & Minton, B. A. (2012). How has CEO turnover changed?. International Review of Finance, 12(1), 57-87.
Lamont, O., Polk, C., & Saa-Requejo, J. (2001). Financial constraints and stock returns. Review of financial studies, 14(2), 529-554.
Peters, F. S., & Wagner, A. F. (2014). The executive turnover risk premium. The Journal of Finance, 69(4), 1529-1563.
Philippon, T. (2006). Corporate governance over the business cycle. Journal of Economic Dynamics and Control, 30(11), 2117-2141.
Schoar, A., & Washington, E. (2010). Are the Seeds of Bad Governance Sown in Good Times. Working paper, MIT.
Schoar, A., & Zuo, L. (2011). Shaped by booms and busts: How the economy impacts CEO careers
Appendix
Table 1
Summary statistics
This table presents an overview of all datasets involved in my regressions. Panel A simply shows th total observations and means of forced CEO turnover in the sample. As for this variable is record as 0 or 1, item “means” illustrates the frequency of forced CEO turnover. Panel B reports all related data on Gross Domestic Product (GDP) and separate GDP per capita into a GDP trend component and a GDP cyclical component. Panel C describes statistic summary of mean, median and standard error about stock returns that can reflect firm-specific and industry performance. Panel D illustrates frequency of forced CEO turnovers by peer group performance quintile. Additionally, Panel E reports all information about accounting data that is needed to calculate KZ index, which is
calculated as follows: KZ Index = -1.001909 x Cash Flows / K + 0.2826389 x Q + 3.139193 x Debt / Total Capital - 39.3678 x Dividends / K + -1.314759 x Cash /K
Panel A: Frequency of forced CEO turnovers
Total observations Means
Panel B: Summaries Gross Domestic Product (GDP) Total
observations Mean Std.Dev. Min. Max.
Real GDP 29844 12835.13 1746.458 9521 14873.8 Total National Population 29844 287454.5 14915.8 260146 309761 GDP per capita (log) 29844 10.69821 0.0911308 10.5077 10.80501 GDP Trend Component 29844 1069.789 8.809958 1053.71 1082.944 GDP Cyclical Component 29844 0.0315885 2.872636 -5.797895 3.692811
Panel C: Summaries stock return (Firms and Industries)
Mean Median Std.Dev. Min. Max.
Holding period returns 0.9085422 0 18.948 -90 490 Average Value-Weighted Returns 0.63779 1.119659 1.726904 -3.870574 2.379601 Average Equal-Weighted Returns 1.053458 1.528827 2.143854 -4.471142 4.553268 Firm-specific returns (VW returns) 0.2683041 -0.6870254 18.86083 -89.8796 493.8705 Firm-specific returns (EW returns) -0.1475336 -1.102709 18.81391 -89.60017 494.4711
Panel D: Probabilities of forced CEO turnover by industry performance EW industry stock return in the
year before the CEO turnover Probability of a forced CEO turnover
VW industry stock return in the
year before the CEO turnover Probability of a forced CEO turnover Quintile [S.E.] 1 3.04% [0.172] Quintile [S.E.] 1 3.18% [0.175] 2 2.97% [0.169] 2 [0.174]3.13% 3 2.48% [0.155] 3 [0.152]2.38% 4 2.49% [0.156] 4 [0.167]2.87% 5 3.18% [0.176] 5 [0.159]2.62%
Panel E: KZ-Index information
Observations Mean Median Std.Dev. Min Max
Total Assets 28813 11846.2 1555.892 63531.93 3.432 2264909 Cash and Short-Term Investments 28805 1228.105 102.769 9366.44 -0.636 368149 Common Shares Outstanding 28804 168.5111 53.649 468.6462 1.167 10862 Debt in Current Liabilities 28779 1706.642 14.79 17700.17 -882 499293 Long-Term Debt 28727 2098.457 253.611 14105.18 0 286876 Depreciation and Amortization 27911 220.5004 44.496 811.2345 0 21577 Total Dividends 28726 109.8183 6.1995 510.2722 -665.665 36968 Income Before Extraordinary Items 28808 262.1132 55.3975 1540.383 -99289 45220 Total Liabilities and Stockholders’ Equity 28813 11846.777 1555.892 63531.93 3.432 2264909 Notes Payable Short-Term Borrowings 28749 1391.777 0 15404.62 -1072 479777 Total Property, Plant and Equipment 26052 3266.942 499.064 11510.62 0 373938 Total Stockholders’ Equity 28814 2206.362 573.8245 6852.775 -86154 231444 Net Deferred Tax Asset 18989 417.696 65.8 1757.126 -9.042 51823 Total Common Equity 1683 2495.696 1360 2927.13 -456.076 26102 Price Close- Annual Fiscal 28816 31.7026 26.64 30.86285 0.09 983.02
Table 2
Testing Stage One (Part A): The first stage regression aims to detect whether GDP and firm or industry performance has any influence on boards when they determine whether a CEO should be replaced. Table 2 presents the results of regression on GDP and forced CEO turnover. Panel A uses log GDP per capita as independent variable and forced CEO turnover as dependent variable. The log GDP per capita is calculated as: GDP per Capita = ln (Real GDP/Population*1,000,000). Then I separate GDP per capita as GDP trend component and GDP cyclical component. Panel B then shows the result of regression on GDP and GDP cyclical component. *** indicate 1% significant level.
Panel A: CEO dismissal and GDP
Percentage of Forced CEO
turnover Coef. Std.Err Z P > |z| [95% Conf. Interval]
GDP per
capita (log) 0.504252 0.1678157 3.00 0.003*** 0.1753394 0.8331647
_cons -7.300343 1.796541 -4.06 0.000*** -10.8215 -3.779187
Panel B: CEO dismissal and GDP cyclical
Percentage of Forced CEO
turnover Coef. Std.Err Z P > |z| [95% Conf. Interval]
GDP Cyclical
Component 0.0077963 0.0052016 1.50 0.134 -0.0023987 0.0179913
Table 3
Testing Stage One (Part B): Table 3 presents results of regression on forced CEO turnover and both firm-specific and industry performance. Panel A shows estimated percentages of CEO dismissal that are predict by firm-specific performance and peer group performance. Column (1) uses average value-weighted industry stock return to reflect firm performance, and column (2) uses average equal-weighted industry stock return to measure peer group performance. Panel B uses stock returns and GDP cyclical component as independent variables to further explain the
relationship among forced ECO turnover, firm and industry performance and GDP. Because both GDP and industry performance can be viewed as an exogenous shock that may affect CEO dismissal.
Panel A: Forced CEO turnover on firm performance and industry performance
(1) (2)
Percentage of Forced CEO turnover Percentage of Forced CEO turnover
Constant -1.889182*** (-116.39)
Average VW industry stock return in
year t -0.0220592*** (-2.54)
Firm-specific performance in year t -0.0014598* (-1.77)
R-squared 0.0009
Constant -1.891331*** (-112.68)
Average EW industry stock return in
year t -0.013193** (-1.86)
Firm-specific performance in year t -0.0014999* (-1.81)
Panel B: Forced CEO turnover on GDP cyclical and industry performance
(1) (2)
Percentage of Forced CEO turnover Percentage of Forced CEO turnover
Constant -1.893326***
(-111.46) Average EW industry stock return in
year t -0.0113131* (-1.53) GDP cyclical Component 0.0058464 (1.05) R-squared 0.0006 Constant -1.890895*** (-115.59) Average VW industry stock return in
year t -0.0203272** (-2.27)
GDP cyclical Component 0.0051844
(0.94)
Table 4
Testing Stage Two: Stage two uses financial constraints as additional variable to further detect how those variables would affect boards’ decision making on CEO retention. Financial constraints are measured by Kaplan-Zingales Index. Panel A presents results of regression based on the model in stage one, but also estimates the predictive power of financial constraints and interaction of
financial constraints and firm-specific performance. The difference between columns (1) and (2) is that regression in column (1) uses equal-weighted stock return but column (2) uses
average-weighted stock return. Similar to Panel A, Panel B reports the relationship between forced CEO turnover and GDP weighted by financial constraints.
Panel A: Forced CEO turnover and Industry performance with financial constraints
(1) (2)
Percentage of Forced CEO turnover Percentage of Forced CEO turnover
Constant -1.845373
(-94.99) Average EW industry stock return in
year t -0.0157019 (-1.78)
Firm-specific performance in year t -0.0011596 (-1.19) Financial constraints 0.0011842
(1.01) Industry performance under financial
constraints 0.0006269 (2.38)
R-squared 0.0026
Constant -1.861743
(-92.19) Average VW industry stock return in
year t -0.0193239** (-1.75)
Firm-specific performance in year t -0.0011407
(-1.18)
Financial constraints 0.0012408**
(1.89) Industry performance under financial
constraints 0.0008496* (2.51)
Panel B: Forced CEO turnover and GDP with Financial Constraints
Percentage of Forced CEO
turnover Coef. Std.Err Z P > |z| [95% Conf. Interval]
GDP per capita (log) -0.009341 0.2898195 -0.03 0.974 -0.5773768 0.5586948 Financial constraints 0.3900915 0.1348547 2.89 0.004*** 0.1257811 0.6544019 GDP with Financial constraints -0.0362193 0.0125373 -2.89 0.004*** -0.0607919 -0.0116466 _cons -1.76937 3.100225 -0.57 0.569 -7.86138 4.322639
Table 5
Robustness tests: For robustness tests I use accounting returns instead of stock return to examine the relationship between CEO dismissal and firm- and industry performance (Panel A), and also inspect the predictive power of financial constraints (Panel B). The accounting returns are return on assets, which calculated as: ROA= Net Income/ Total Assets. All z-statistics are calculated with robust standard error.
Panel A: Forced CEO turnover on firm performance and industry performance
(1) (2)
Percentage of Forced CEO turnover Percentage of Forced CEO turnover
Constant -1.996564*** (-105.41)
Average VW industry stock return in
year t -0.64628*** (-11.43) Firm-specific performance (using
Return on Asset) -0.6265541*** (-11.08)
R-squared 0.0161
Constant -1.993004*** (-96.09)
Average EW industry stock return in
year t -0.6639533*** (-7.33)
Firm-specific performance (using
Return on Asset) -0.651523*** (-7.49)
Panel B: Forced CEO turnover and Industry performance with financial constraints
(1) (2)
Percentage of Forced CEO turnover Percentage of Forced CEO turnover
Constant -1.865243***
(-94.13) Average VW industry stock return in
year t -0.8987019*** (-10.42) Firm-specific performance (using
Return on Asset) -0.8926262*** (-10.31) Financial constraints 0.0002762
(0.72) Industry performance under financial
constraints 0.0006558** (2.29)
R-squared 0.0256
Constant -1.859319***
(-90.66) Average EW industry stock return in
year t -0.9039182*** (-10.49)
Firm-specific performance (using
Return on Asset) -0.8934881*** (-10.39)
Financial constraints 0.0002592
(0.69) Industry performance under financial
constraints 0.0005225** (2.30)
