The relationship between Executive Compensation &
Firm Performance
An empirical analysis from U.S and European Airline firms in the years before and during the financial crisis.
By Tarik R. Samandar
Supervised by Silvia Dominguez Martinez
University of Amsterdam
Bachelor in Finance & Organization, Specialization in Organizational Economics Student number: 10164626
February 2nd
2015
Abstract
This paper investigates if there is a relationship between firm performance and CEO pay. This is done by considering financial performance measures, but in particular emphasizing non-financial performance measures. As a financial performance measure return on assets will be used and as a non-financial performance measure passenger load factor will be used. Furthermore this paper investigates whether the crisis years (2007-2008-2009) had an impact on the relationship between these performance measures and CEO-compensation. To conduct this research a sample of 17 Airline firms (U.S. and European) between 2004 and 2009 is used. After controlling for other determinants of CEO-compensation, this paper does not provide any evidence that there is a relationship between financial/non-financial performance measures and CEO-pay. Moreover, no evidence is found that during the crisis years (in which several airline firms were financially distressed) the relationship has changed.
Hierbij verklaar ik, Tarik Samandar, dat ik deze scriptie zelf geschreven heb en dat ik de volledige verantwoordelijkheid op me neem voor de inhoud ervan. Ik bevestig dat de tekst en het werk dat in deze scriptie gepresenteerd wordt origineel is en dat ik geen gebruik heb gemaakt van andere bronnen dan die welke in de tekst en in de referenties worden genoemd.
De Faculteit Economie en Bedrijfskunde is alleen verantwoordelijk voor de begeleiding tot het inleveren van de scriptie, niet voor de inhoud.
1.Introduction
There have been several studies that examined the relationship between executive compensation and firm performance (Jensen & Murphy, 1990; McGuire, Chiu and Elbing, 1962; Deckop, 1988; Chalmers, Koh & Stapledon, 2006). All these studies considered financial performance measures, such as return on assets, return on equity & profits when examining this relationship. However, there have not been many studies examining the relationship between executive compensation and non-financial performance measures. Recently the use of these non-financial metrics in contracting has increased (Ittner, Larcker & Rajan, 1997). Unlike financial performance
measures, non-financial performance measures are not as easy to measure and therefore not (always) comparable across industries. That is why in this study the focus is on one industry, namely the airline industry. In this industry, passenger load factor is a relevant performance indicator with regards to productivity, as stated by Schefczyk (1993).
Gilson and Vetsuypens (1993) show that financially distressed firms implement a wide array of changes in compensation of CEOs. In times of financial distress it is important to avoid bankruptcy costs and other costs that are associated with bankruptcy or financially restructuring. The CEOs are incentivized through to the bonus component of CEO compensation and particularly rewarded on traditional accounting performance measures such as profits and return on assets, to increase cash flow and so to avoid bankruptcy costs for the firm. Thus, these financial
indicators are better ways of aligning the CEOs interest with that of the firm (Gilson and Vetsuypens, 1993)
Matêjka, Merchant & Van der Stede (2009) build upon this previous research
conducted by Gilson and Vetsuypens (1993) and find that when firms are
performance measures. This complements the results of Gilson and Vetsuypens (1993) that in times of financial distress accounting performance measures are a better way of aligning the interests of the CEO with that of the shareholders.
The financial crisis of 2007-2008 also had its effects on the airline industry as
can be read from the press release of the International Air Transport Association (also known as IATA). In particular, they mention that the regions North America and Europe are the hardest hit by the crisis and will post losses of $5.2 billion and $1.8 billion respectively in 2008.
In this paper the relationship between executive compensation and firm
performance is studied by not only looking at financial performance measures, but in particular emphasizing non-financial performance measures as they are considered to be relevant in contracting. Furthermore, this paper will try to establish whether return on assets is of more importance during the crisis years and also whether passenger load factor is of less importance during the crisis years.
The sample used consists of 99 firm year observations between the (fiscal)
years 2004 and 2009 from 17 airline firms, of which 13 U.S airline firms and 4 European airline firms. For the empirical analysis the cash component of CEO-compensation, which comprises base salary and bonus, is of interest in this study (Davila and Venkatachalam, 2004). To avoid potential omitted variable bias, firm size has to be controlled for by using the natural logarithm of sales (Tosi, Werner, Katz & Gomez-Mejia, 2000).
This paper starts by discussing relevant literature and the development of the hypotheses in section 2. The third section describes the data and research method used for the empirical analysis. Section 4 provides the descriptive statistics for the
dependent and independent variables used. In section 5 the results are provided and in section 6 the paper ends with a conclusion/discussion.
2. Literature Review & Hypotheses
In section 2.1 agency theory will be explained, which leads to CEO-compensation as the answer to the agency problem. Afterwards, in section 2.2 different performance measures are discussed and in particular the relevance of non-financial measures is emphasized. In section 2.3 related literature and hypotheses are presented.
2.1 Agency Theory
The agency relationship is one of the oldest and most common codified modes of social interaction. According to Ross (1973) “an agency relationship has arisen
between two (or more) parties when one, designated as the agent, acts for, on behalf of, or as representative for the other, designated the principal, in a particular domain of decision problems” (p. 134).In this paper the shareholder is the principal and the Chief Executive Officer is the agent.
Agency theory deals with control issues resulting from conflicts of interest
between owners and top executives (Tosi et al., 2000). The conflict is that
shareholders want to maximize firm value and they want to give incentives to CEOs to take actions that do so. However, top executives have a broad discretion to follow their own objectives (e.g. empire building), which often conflict with the objectives of those of the stockholders (Tosi et al, 2000). Specifically, agency theory focuses on the mechanisms that could potentially be used to manage or reduce such conflicts of interest (Eisenhardt, 1989). Another problem, with which agency theory is concerned, is the problem of risk sharing (Eisenhardt, 1989). This problem arises when the
principal and the agent have different levels of risk tolerance. The issue at point is that the agent and the principal might prefer different actions because they have different levels of risk tolerance (Eisenhardt, 1989). Agency theory has 3 assumptions. First of all, the agent is risk averse; secondly, the agent’s interests might be different than those of the principal and last but not least, the agent is self-centered (Tosi et al, 2000).
It is a not an easy task to induce agents that are self-centered, utility
maximizing and risk-averse to act on behalf of the principals (stockholders) (Bloom & Milkovich, 1998). The manner, in which the principal addresses this problem, is by setting up a contract with the agent. First of all, this contract can include the
monitored. This way it can be ensured that the behavior and decisions of the agents are aligned with those of the stockholder (Tosi et al., 2000). Second of all, according to Garen (1994) another way to address this problem is to structure CEO
compensation to provide appropriate incentives (performance-based contract). Tosi et al. (2000) mention that if the principal has complete information about the agent’s effort, then it is most efficient to monitor the behavior of the agent. The reason is that the performance-based contract automatically imposes risk on the risk-averse agent. However, in the presence of asymmetric information and unobservable efforts of the agents, the principal has no other choice then to transfer risk to the agent. The principal does this by designing a contract that aligns incentives. More precisely, a contract based on performance outcomes that are observed. This
alternative contract is referred to as the “second best” solution to the issue of control in the agency problem (Tosi et al., 2000). According to Tosi, Katz, & Gomez-Mejia (1997) incentive alignment is the most feasible control mechanism in the presence of asymmetric information regarding the activities and the behavior of the agent. This because it gives rise to “self-monitoring” on the part of the agent (Tosi et al., 2000).
Concluding, executive compensation solves the agency problem by rewarding
CEOs with the right incentives. However, such a performance-based contract is not an ideal solution, because paying for performance automatically imposes risk on the risk-averse agent (Tosi et al., 2000).
2.2.1 The different types of performance measures
Many companies have restructured their executive compensation to link CEO pay to
firm performance (Lippert & Porter, 1997).Firm performance is measured by
performance measures and there are several types of performance measures available that can be used in an incentive plan. Examples of financial performance measures are net earnings, return on investments, unit costs (Eccles, 1991). According to Banker and Datar (1989) however, financial performance measures on their own, are not enough to provide the most efficient means to motivate Chief Executive Officers, to act in the manner desired by the shareholders of the firm. This because they do not capture all the dimensions of CEO performance and, therefore non-financial performance measures may provide additional information beyond traditional accounting measures (Banker and Datar, 1989). According to Ittner et al. (1997) a
performance measure should be included in an incentive plan if it can provide additional information content about the dimensions of CEOs action that the
shareholders wish to motivate. Non-financial performance measures include market share, efficiency/productivity, product quality, customer satisfaction and employee satisfaction (Ittner et al., 1997).
2.2.2The Relevance of Non-Financial Performance Measures.
In this section attention will be paid to non-financial performance measures.
Ittner and Larcker (2000) consider non-financial metrics and mention the advantages that these measures have over financial performance measures. First, non-financial performance metrics are better linked towards the long-term organizational strategies of the firm (compared to financial performance measures). Besides only focusing on short-term outcomes, financial performance measures don’t deal with other non-financial goals that could be important in achieving profitability. By complementing traditional accounting measures with non-financial measures firms are better able to communicate goals and incentivize their executives to address long-term strategy (Ittner and Larcker, 2000).
Second, critics of traditional accounting performance measures claim that
drivers of success in many industries are intangible assets such as intellectual capital and customer loyalty, instead of the tangible assets. Despite the fact that it is hard to quantify intangible assets in financial terms, non-financial information can provide indirect, quantitative indicators of a firm’s intangible assets (Ittner & Larcker, 2000).
Last, the choice of performance measures used in contracting should be based
on providing information about the executives’ actions and the level of “noise” in the performance measures (Ittner and Larcker, 2000). Noise refers to changes in a
performance measure that are beyond the control of the CEO or firm. These changes can be due to economic shocks or due to luck (Ittner and Larcker, 2000). Executives need to be aware of how much success is attributable do their actions. If they are not aware of this fact, they will not have the signals they need to maximize their effect on the performance of the firm (Ittner and Larcker, 2000). Because most non-financial performance measures are less prone to external noise, the use of these measures is able to improve executives’ performance by providing a more accurate evaluation of their actions. This also imposes less risk on the executives when pay is determined (Ittner and Larcker, 2000).
In the Airline Industry passenger load factor (which is the non-financial performance measure used in this study) is a common performance indicator, which measures capacity utilization (Schefczyk, 1993). According to Davila and Venkatachalam (2004) passenger load factor is an important non-financial performance measure that should be included in the annual bonus contract in addition to traditional accounting performance measures.
Because airline firms are capital intensive and have high fixed costs, the
efficiency with which they utilize their assets (capacity utilization) is an important non-financial metric (Behn and Riley, 1999). And since passenger load factor says something about current firm performance it captures operational efficiency in a manner superior to traditional accounting measures (Davila and Venkatachalam, 2004).
2.3.1 Related Literature
McGuire, Chiu and Elbing (1962) study the relationship between executive
compensation, sales and profits. To conduct their research they analyze data from 45 of the 100 largest industrial corporations in the U.S between the years 1953 and 1959. They find a positive and significant relationship between executive compensation and sales as well as a positive and significant relationship between executive
compensation and profits.
Deckop (1988) examines the determinants of Chief Executive Officer
compensation. Specifically, he focuses on measures of firm performance. He conducts his research by analyzing data on CEO-compensation from 120 companies between the years 1977 and 1981. The study also examines other potential determinants of compensation, such as CEO experience and industry effects. His main finding is that CEO compensation is positively related to profit.
Jensen & Murphy (1990) study pay for performance and top-management
incentives. To conduct their research they obtain data from over 2000 Chief Executive officers between the years 1974 and 1986. Data from these CEOs are listed in the Executive Compensation Surveys published in Forbes. Their findings suggest that chief executive compensation is related to accounting and market performance measures.
Based on the related literature suggesting that CEO-compensation and firm
performance (measured by accounting performance measures) is closely related, the following is hypothesized:
H1: the relationship between accounting performance measures and the cash
component of ceo-compensation is positive.
Banker, Potter and Srinivasan (2000) study what the implications are of including non-financial performance measures in the annual bonus contract. According to Banker et al. (2000) these non-financial metrics are better at indicating future firm performance. Therefore non-financial performance measures should be included in annual bonus contracts in addition to traditional accounting measures. In their study they focus on 18 hotels that are managed by a hospitality firm. They use time-series data for 72 months from these hotels and find empirical evidence on the impact of non-financial performance measures on firm performance. Specifically, their results indicate a positive and significant association between non-financial measures of customer satisfaction and future financial performance. They also find evidence that non-financial as well as financial performance improve after the implementation of an incentive plan that contains non-financial metrics.
Davila and Venkatachalam (2004) examine the relationship between
non-financial performance measures and executive pay. They study whether a link between passenger load factor in the U.S. Airline Industry and executive
compensation exists. Passenger load factor is a proxy for efficiency/productivity. To conduct their research they make use of a sample of 35 U.S Airline firms and 246 firm-year observations covering the years 1986 to 2000. They split the executive compensation into two components, namely cash compensation (base salary and bonus) and stock compensation. Their main finding is that passenger load factor is positively related to the cash component of CEO compensation. Furthermore, after controlling for other potential determinants of CEO- compensation, they find that a 5% increase in passenger load factor results in an increase of cash compensation of $70,000. As Davila and Venkatachalam (2004) mention, this result indicates that passenger load factor captures an important dimension of current firm performance and that it is related to the cash component of total compensation. For stock
compensation they find a negative but insignificant coefficient. As they mention, this is consistent with the fact that stock compensation is associated with long-term incentives rather than being used to reward current firm performance (Davila and Venkatachalam, 2004)
The above evidence on the importance of non-financial performance measures and the results obtained by previous research leads to the second hypothesis:
H2: the relationship between non-financial performance measures and the
cash component of ceo-compensation is positive.
2.3.2 The Choice of Performance Measures
There are some factors that might influence the relative weight that is placed on financial and non-financial performance measures (Ittner et al., 1997).
Gilson and Vetsuypens (1993) study senior management compensation in
publicly traded firms that are financially distressed. Their analysis shows that financially distressed firms implement a wide array of changes in compensation of CEOs. More precisely, the firms systematically restructure their management compensation contracts when they are experiencing severe financial difficulty. In those situations it is important to avoid the legal fees and other costs that are associated with bankruptcy or financially restructuring. The CEOs are incentivized through to the bonus component of CEO compensation and particularly rewarded on traditional accounting performance measures such as profits and return on assets, to increase cash flow and so to avoid bankruptcy costs for the firm. Thus, these financial indicators are better ways of aligning the CEOs interest with that of the firm
(shareholders).
Because of the immense losses in the Airline industry as the consequence of the financial crisis and the results obtained by Gilson and Vetsuypens (1993) it is expected that the link between accounting performance measures and CEO-compensation is stronger during the crisis. This leads to the next hypothesis:
H3: the relationship between ceo-compensation & accounting performance
measures during the financial crisis is stronger than before the crisis
Matêjka, Merchant & Van der Stede (2009) build upon previous research done by Gilson and Vetsuypens (1993). They examine in which situations more emphasis is placed on non-financial metrics with regards to the annual bonus contracts. To conduct their research they create a dummy variable, which they call “NONFIN”, which equals 1 when firms use some form of non-financial performance measures in contracting. Their results suggest that firms experiencing financial distress place less emphasis on non-financial performance measures. This complements the results of Gilson and Vetsuypens (1993) that in times of financial distress, traditional
accounting performance measures are a better way of aligning the interest of the CEO with that of the shareholders. Therefore a weaker emphasis on non-financial
performance measures is expected in times of financial distress.
Because of the results obtained by Matêjka et al., (2009) the following is hypothesized:
H4: the relationship between ceo-compensation & non-financial performance
3. Data & Research Method
In section 3.1 the data is described. In the following section, section 3.2, the research method is discussed.
3.1 Data
To conduct this research data on CEO-compensation is retrieved from 3 different sources. First, Airline firms (SIC 4512) are collected through the ExecuComp database, which is part of Compustat. This database collects data from firms that are part of, or were part of the S&P 1500. Furthermore, CEO-compensation from other US firms are collected through the EDGAR ONLINE database, which can be found at
www.sec.gov. All public U.S. firms are obliged to submit forms/filings and reports to the Securities and Exchange Commission, which gives everyone free access to this (public information). For the European airlines in this sample, CEO-compensation is retrieved from their annual report. The ultimate sample consists of 17 airline firms. The years to be used are 2004 until 2009, totaling six years. We define a pre-crisis period (2004 until 2006) and crisis period (2007 until 2009). The resulting sample contains 99 firm-year observations. An overview of the airlines that are part of this sample is included in appendix A.
The return on assets is not readily available in Compustat. It is calculated as
follows: income before extraordinary items divided by total assets. For the majority of firms, these financials are retrieved from Compustat. For the remaining two, which are EasyJet and British Airways they are hand-collected through the Annual Reports and calculated. The passenger load factor is mostly obtained from the Annual Reports in the section “Operating Statistics”. For some of the US Airline firms, the passenger load factors were given for the mainline operations and regional operations so both percentages are added and divided by two. Sales are retrieved from Compustat and the Annual Reports.
Because the final sample consists of U.S. airline firms and European airline
firms, CEO-compensation and firm financials are retrieved in three different currencies. These are the U.S. dollar, the Euro and the British pound. Because the majority of the firms in the sample are from the United States and since the majority of data available is displayed in U.S. Dollars, in this paper this currency is used to make comparisons. To convert the amounts into US dollars the OANDA website is
used. It is a foreign exchange company that provides currency conversion, online retail foreign exchange trading (forex), online foreign currency transfers and forex information. During the years that are being examined, the Brithish Pound and the Euro fluctuated against the U.S. dollar, which is the consequence of responses to market mechanisms of the foreign-exchange market. These fluctuations can be misleading for the results and inferences being made in this paper. To illustrate this, consider Lufthansa Group, which reports its amounts in euro. In 2006 and 2007 Wolfgang Mayrhuber, the Chief Executive Officer of the company earned the same salary of 700,000 euro. In those years the exchange rate was 1.31930 and 1.47185 respectively. This translates into an increase in salary of more than a hundred thousand U.S. Dollar. Because the main focus of this paper relates changes of return on assets and passenger load factor (and other independent variables) to CEO pay, other influences should be controlled for. To account for this an average exchange rate was calculated, for the years being used in the sample. After this exchange rate was calculated it was applied to all the corresponding six years that are used in this sample.
3.2 Research Method
In this section the research method will be discussed. To test the hypotheses in this study, regression analysis will be performed.
The dependent variable is CEO-compensation. In this paper attention is paid
to the cash component, which is the annual salary and annual bonus. The reason why only the cash component is considered is because Davila and Venkatachalam (2004) mention that current performance is related to the cash component of total
compensation.
As the first independent variable we define return on assets, which is the
proxy for an accounting performance measure. Based on previous research done by Chalmers, Koh & Stapledon (2006) and their findings, return on assets is expected to be positive and significant.
The second independent variable is passenger load factor, which is the proxy
for a non-financial performance measure. Following previous research done by Davila and Venkatachalam (2004), passenger load factor is expected to be positively
This leads to the following two empirical models tested by a simple linear regression:
CEO-compensation= α + β1*Roa +ε (1)
CEO-compensation= α + β1*Plf +ε (2)
The models above (incorrectly) leave out an important determinant of
CEO-compensation. Firm size has to be controlled for as is evident from prior research. As Tosi et al. (2000) mention, greater size may be used to legitimize higher
CEO-compensation by appealing to rationalizations to justify a size premium. One of these rationalizations may include greater organizational complexity, as argued by Kostiuk (1990). To control for firm size, the natural logarithm of (Revenue) Sales is used. This is based on prior research done by Davila and Venkatachalam (2004), who also
consider this proxy to control for firm size. Taking into account the results of Tosi et al. (2000), it is expected that the natural logarithm of sales is positively related to CEO-compensation. Because the above models leave out this important determinant, this might result in biased estimates. The models compensate for the missing
determinant by over- or underestimating the effect that the independent variable has on CEO-compensation.
This leads to the following three empirical models to be tested by a multiple regression:
CEO-compensation= α + β1*Roa + β2,*LnSales +ε (3)
CEO-compensation= α + β1*Plf + β2,*LnSales + ε (4)
It is not (necessarily) the case that Chief Executive Officers are rewarded the same in Europe as in the United States. Actually as can be read from Forbes, they say that it is like comparing apples to oranges, for U.S. firms operate differently from their
international counterparts, particularly in terms of corporate governance. Typically, in the United States a larger fraction of total compensation is awarded in the form of stock options and restricted shares. To control for this fact, a dummy variable is created named “Europe”. This binary variable equals 1 for European airline firms and 0 otherwise. It is expected to be positive since CEOs in Europe are rewarded higher cash amounts compared to the United States. In the latter, cash amounts are
considered to be only a small fraction of total CEO-compensation.
This leads to the following empirical model to be tested by a multiple regression:
CEO-compensation= α + β1*Roa + β2*Plf + β3*LnSales + β4*Europe + ε (6)
Next, the crisis is incorporated in the empirical model. To do this, a dummy variable is created named “Crisis”. This binary variable equals 1 for 2007, 2008 and 2009 and 0 otherwise. According to Core, Holthausen and Larcker (1998) the board of directors is in charge of structuring the compensation package for the CEO. As Core et al. (1998) mention the CEO has a lot of power over the board of directors. Furthermore, the CEO and the Chairman of the Board are frequently the same person. Therefore the CEOs might not experience a decrease in their compensation, even if the firms are hit by the crisis. Gilson and Vetsuypens (1993), however, find that when firms are in financial distress the CEOs experience salary and bonus reductions. Because of these findings it is not sure what the coefficient will be for the dummy variable “Crisis”.
“RoaCrisis” is an interaction term added to the model to estimate the
relationship between return on assets and CEO-compensation during the crisis years as compared to non-crisis years. This interaction is expected to be positive. As can be read by a press release from the IATA, Airline firms experienced severe financial distress because of the crisis. In those situations it is important to avoid bankruptcy costs and other costs that are associated with bankruptcy or financially restructuring (Gilson and Vetsuypens, 1993). The CEOs are incentivized through to the bonus component of CEO compensation. In particular they are rewarded on performance
measures such as profits and return on assets, to increase cash flow and so to avoid bankruptcy costs for the firm. Thus, these financial indicators are better ways of aligning the CEOs interest with that of the firm (shareholders) (Gilson and Vetsuypens, 1993).
“PlfCrisis” is another interaction term added to the model to estimate the
relationship between passenger load factor and CEO-compensation during the crisis years as compared to non-crisis years. Matêjka et al. (2009) build upon previous research done by Gilson and Vetsuypens (1993) and find that in times of financial distress less emphasis is placed on non-financial performance measures. This complements the results of Gilson and Vetsuypens (1993) that when firms are experiencing financial distress, traditional accounting performance measures are a better way of aligning the interest of the CEO with that of the shareholders. Therefore it is expected that the interaction term is negative.
This leads to the last empirical model with all the relevant variables included:
CEO-compensation= α + β1*Roa + β2*Crisis + β3*RoaCrisis+ (7) β4*Plf +β5*PlfCrisis + β6*LnSales + β7*Europe + ε
α: constant
Roa: return on assets Plf: passenger load factor
Crisis: dummy variable that equals 1 for crisis years and 0 otherwise
PlfCrisis: passenger load factor during the crisis years
LnSales: natural Logarithm of (Revenue) Sales to control for firm size
RoaCrisis: return on Assets during the crisis years
Europe: dummy variable that equals 1 for European airline firms and 0
otherwise
4. Descriptive Statistics
Part 1 of table 2 provides descriptive statistics for the components of
CEO-compensation used in this paper. The mean of base salary, annual bonuses and total of base salary and annual bonuses (all in thousands) is respectively $531, $385 and $916. In the sample there is a minimum of zero, this is in fact zero and are not missing values. From the definitive proxy statement of Allegiant Travel Company, it can be read that the chief executive officer, Maurice J. Gallagher, Jr., has a substantial equity position. Therefore, he has chosen to serve without any compensation
whatsoever and expects to continue to serve without compensation into the future. Concluding, Mr. Gallagher does not receive base salary, does not participate in our annual cash bonuses and does not receive equity incentive grants.
Part 2 of Table 2 presents descriptive statistics for the independent variables used. Of particular interest is the return on assets, with a mean of -2.04%. The minimum of the sample is -255.34% and is not an outlier. In 2004 Hawaiian Holdings Inc. posted a loss of $7.262 million and recorded total assets being worth $2.844 million, which gives rise to such a huge negative return on assets.
Table 2. Descriptive Statistics
Part 1: Descriptive Statistics for CEO compensation (N=99)
(in thousand of $) Mean Standard DeviationMin Max
Base Salary 531.51 288.13 0 1278.58
Bonus 385.40 454.01 0 1909.17
Total Cash Compensation 916.91 630.35 0 2963.65 Part 2:Descriptive statistics for independent variables
Mean Standard DeviationMin Max
Roa (2.04) 30.50 (255.34) 90.17
Plf 78.14 4.94 68.60 90.40
LnSales 8.24 1.34 4.89 10.49
Note: Roa is the return on assets calculated by dividing Income before Extraordinary items by Total assets.
Plf stands for passenger load factor and is calculated by dividing revenue passenger miles by available seat miles. LnSales is the natural logarithm of Sales, that is used as a proxy for firm size.
Table 3 presents a correlation matrix for the independent variables used in this study, to check for multicollinearity.
5. Results
In this section the results are presented in table 4 (page 18). The return on assets, which is the proxy for firm performance (financial performance measure) is not significant when considering a simple regression of CEO compensation on Return on Assets (refer to equation 1). However after controlling for firm size (equation 3), return on assets is positive and significant at the 5% level. After controlling for both firm size and passenger load factor (equation 5), return on assets remains significant at the 5% level. From equation 6 onward when “Europe” is added to the regression, return on assets is not significant.
Passenger Load Factor, which is the proxy for firm performance
(non-financial performance measure) is significant at the 5% level when considering a simple regression of CEO-compensation on passenger load factor (equation 2). After controlling for firms size (equation 4) passenger load factor is significant at the 1% level. After controlling for both firm size and return on assets (equation 5), passenger
Table 3. Correlation Statistics
Roa Plf LnSales Europe Crisis
Roa 1
Plf 0.0021 1
LnSales -0.2008 -0.1077 1
Europe 0.0826 0.2340 0.3146 1
Crisis -0.0106 0.2568 0.1247 -0.0000 1
Note: Roa is the return on assets calculated by dividing Income before Extraordinary items by Total assets. Plf stands for passenger load factor and is calculted by dividing revenue passenger miles by available seat miles. LnSales is the natural logarithm of Sales that is used as a proxy for firm size.
Europe is a dummy variable which equals 1 for European Airline firms and 0 otherwise. Crisis is a dummy variable which is 1 for 2007, 2008 and 2009 and 0 otherwise.
load factor remains significant at the 1% level. From equation 6 onward, like return on assets, passenger load factor is not significant.
The natural logarithm of Sales, which is a proxy for firm size, is of vital
importance as far as a determinant of the CEO-compensation concerns. It is positive and significant in all the regression done. This means that Chief Executive Officers of larger firms have a higher compensation level. This supports the theory of Gabaix and Landier’s (2008). They state that the rationale for a positive size-pay-relationship is reflected by the growth in complexity and difficulties of managing large firms. Because of this increased complexity there is a demand for higher quality CEOs who are paid higher remuneration. Also, as Tosi et al (2000) mention in their paper, firm size is very relevant as far as a determinant of CEO-compensation concerns.
In equation 6 the dummy variable “Europe” is introduced. It is positive and
significant at the 1% level (in equation 6 as well as 7). This result indicates that Chief Executive Officers of European airline firms are paid higher cash amounts compared to Chief Executive Officers of U.S. Airline firms. This result is consistent with the fact that in the United States cash amounts are considered to be only a very small fraction of total compensations. In the United States Chief Executive Officers are paid incredibly high stock compensation.
In equation 7, the dummy variable “Crisis” is added. It has a negative (but
insignificant) relationship with the cash-component of CEO-compensation, which means that there is no evidence that the crisis period affected the amounts that the Chief Executive Officers got paid. As Core et al. (1998) mention, the board of directors structures the Chief Executive Officer’s compensation package. However, the CEO has a lot of influence over the board of directors. Moreover, in many cases the CEO and the Chairman of the board are the same person. For these reasons, reductions in the remuneration of the CEOs are not common (Core et al., 1998). This could be reason for why the dummy variable is insignificant.
The interaction term “RoaCrisis” is positive, but insignificant, which means that this finding does not support prior research done by Gilson and Vetsuypens (1993) who find empirical evidence on the fact that when firms are experiencing financial distress, more emphasis is placed on traditional accounting performance measures.
The interaction term “PlfCrisis” is positive, but insignificant, which means that this study does not support previous research done by Matêjka et al. (2009). They provide empirical evidence that when firms are dealing with financial distress they place less emphasis on non-financial performance measures, because financial performance measures are a better way of aligning the interest of the CEO with that of the stockholders.
There are some extreme values for CEO-compensation and return on assets in
this dataset. These have been removed to verify the robustness of the results. The differences in results obtained by the new regressions are not big. Specifically, these results, do not impact the conclusions made based on the dataset without these extreme values. An overview of the new regressions is included in Appendix B.
Table 4. OLS Estimates.
Dependent Variable
Total Cash Compensation
Independent Variable [1] [2] [3] [4] [5] [6] [7] Roa 0.675 6.139** 6.536** 3.333 3.034 [1.105] [3.180] [3.137] [2.691] [3.295] Plf 29.363** 35.431*** 35.867*** 17.159 9.013 [13.899] [13.064] [13.288] [10.780] [15.114] LnSales 192.853*** 191.233*** 207.588*** 113.41*** 98.925** [46.918] [47.627] [48.516] [36.156] [43.764] Crisis -670 [1837.971] RoaCrisis 0.0014 [0.0115] PlfCrisis 10.052 [24.311] Europe 774.836*** 798.094*** [145.776] [149.070] N 99 98 99 98 98 98 98
Note: Robust standard errors are reported between brackets. * indicates significance at the 10% level, ** indicates significance at the 5% level and *** indicates significance at the 1% level. Roa is the return on assets calculated by dividing Income before Extraordinary items by Total assets. Plf stands for passenger load factor and is calculted by dividing revenue passenger miles by available seat miles.
LnSales is the natural logarithm of Sales that is used as a proxy for firm size.Crisis is a dummy variable which is 1 for 2007, 2008 and 2009 and 0 otherwise. RoaCrisis is an interaction term added to the regression to estimate the relationship between return on assets and CEO-compensation during the crisis years as compared to non-crisis years. PlfCrisis is an interaction term added to the regression to estimate the relationship
between passenger load factor and CEO-compensation during the crisis years as compared to non-crisis years.
6. Discussion
This paper empirically analyses the relationship between firm performance and CEO cash compensation and the impact the recent financial crisis might have had on this relationship. Both financial and non-financial performance measures are considered in this study. The proxy for the financial performance measure used is return on assets. The proxy for the non-financial performance measure is passenger load factor, which is a common measure for productivity in the Airline Industry. This is done for U.S and European Airline firms between the fiscal years 2004 through 2009 based on panel data. The final sample consists of 17 airline firms, resulting in 99 firm-year observations.
According to the results obtained return on assets and passenger load factor
are positive, but not significant after all the variables are included in the regression. This is not consistent with prior research done (McGuire et al., 1962; Deckop, 1988; Davila & Venkatachalam). However, Tosi et al (2000) examine the relationship between firm size, firm performance and CEO-compensation. Their findings are that firm performance accounts for less than 5% of the variance in CEO-compensation, suggesting that there is almost no link between firm performance and CEO-pay. Kerr & Bettis (1987) study top management compensation and even find a non-existing relationship between executive compensation and firm performance. Another reason for the results obtained might be the small sample size in this study, which leads to large standard errors and consequently insignificant results.
The interaction terms “RoaCrisis” and “PlfCrisis”, are positive but also
insignificant, which means that not a lot can be said about whether the relationship between return on assets /passenger load factor and CEO-compensation has changed during the crisis years as compared to the non-crisis years. There might be 2 reasons for this result. First of all, it is not necessarily the case that all the airline firms in this sample were experiencing financial distress during the crisis years or that they were in financial distress at the same time. Looking at Lufthansa’s Annual Report of 2008 for example, it can be read that they had immense profits. As the Chairman of the
Board/CEO Wolfgang Mayrhuber mentions, Lufthansa reported operating profits of EUR 1.35 billion, which is the second best result in the history of the Company. Second, because the sample size is small this leads to large standard errors and consequently insignificant interaction terms.
Furthermore, because of the small sample size only the natural logarithm of Sales (LnSales) is included as a control variable. This seemed to be the most relevant to include as a control variable based on prior research done (as mentioned in the above paragraph). In all regressions LnSales is positive and significant, which means that it served its purpose. However there are other determinants of CEO-pay, which are not included. According to Bushman, Indjejikian & Smith (1996), CEO tenure is also an important determinant of CEO-compensation. CEO tenure captures the quality of the CEO as well as CEO power. Chief executive officers with greater power to influence the board of directors are more likely to obtain a higher pay (Davila & Venkatachalam, 2004). According to Smith & Watts (1992) firms with more growth possibilities have higher executive pay. Because these determinants are not included, the problem of omitted variables is not completely ruled out. If important variables are omitted from the regressions, the estimates of the coefficients might be biased. The reason for this is that the model compensates for the missing determinants by over- or underestimating the effect of the coefficients of the included variables.
As mentioned in the beginning of this discussion, panel data is used and
regression analysis is applied. However, there is an important difference between the so-called panel data assumptions and the least squares assumptions. Specifically, assumption 2 of the panel data assumptions states that the variables are independent across firms, but makes no such restriction within a firm. This also has implications for the standard errors. Specifically they can be autocorrelated. The explanation is that what happens this year tends to be correlated with what happens the year after. If the standard errors are correlated then this may lead to false hypothesis tests (Stock and Watson, 2012, p. 405).
Reverse causality is a phenomenon also worth mentioning in this study.
Reverse causality means that the dependent variable (Y) is also able to exert a causal effect on the independent variable (X), in addition to the effect of the independent variable (X) on the dependent variable (Y). In this paper, firm performance is the independent variable (X) and CEO-compensation is the dependent variable (Y). However, it is also possible that CEO-compensation effects firm performance. Abowd (1990) provides evidence that an additional payment of 10% bonus (rewarded for good economic performance) is associated with a 0.6% increase (on average) in firm performance (measured by economic and market performance measures) in the year
after. The result of reverse causality is that the independent variable (X) is correlated with the error term in the equation, which has as a consequence that the estimates of the coefficients are biased.
As was mentioned earlier in this discussion, not all airlines experienced
financial distress during the crisis years, which might be a reason for why
insignificant coefficients were obtained for the interaction terms. For future research it might be interesting to consider only a sample of airline firms that are financially distressed and examine whether the relationship between return on assets/passenger load factor and executive compensation changed.
Appendix A
Table&1.&Sample&of&Airline&firm&to&be&used&in&this&study
2. EasyJet
3. Ryanair
5. Skywest Iinc
8. AMR Corp
Note:!*!indicates!that!for!these!Airline!firms!data!for!1!fiscal!years!was!not!available16. Allegiant Travel Co*
17. Frontier Airlines Holdings*
10. United Contiental Holdings Inc
11. Airtran Holdings Inc
12. Jetblue Airways Corp
13. Pinnacle Airlines Corp
14. Hawaiian Holdings
15. Mesa Airgroup Inc*
9. Southwest Airlines
1. Lufthansa Group
4. British Airways
6. Delta Airlines Inc
7. Alaska Airgroup
Appendix B
Table 4. OLS Estimates.
Dependent Variable
Total Cash Compensation
Independent Variable [1] [2] [3] [4] [5] [6] [7] Roa 3.393 6.175* 6.383** 3.144 2.857 [2.943] [3.149] [3.074] [2.736] [3.367] Plf 31.619** 35.828*** 36.227*** 17.444 10.656 [13.899] [13.119] [13.286] [10.821] [14.997] LnSales 178.659*** 175.239*** 192.359*** 95.518** 73.23 [51.275] [51.668] [52.755] [38.917] [48.889] Crisis (521.546] [1855.533] RoaCrisis 0.004 [0.012] PlfCrisis 7.660 [24,604] Europe 778.834*** 798.486*** [146.669] [150.376] N 96 96 96 96 96 96 96
Note: Robust standard errors are reported between brackets. * indicates significance at the 10% level, ** indicates significance at the 5% level
and *** indicates significance at the 1% level. Roa is the return on assets calculated by dividing Income before Extraordinary items by Total assets. Plf stands for passenger load factor and is calculted by dividing revenue passenger miles by available seat miles.
LnSales is the natural logarithm of Sales that is used as a proxy for firm size.Crisis is a dummy variable which is 1 for 2007, 2008 and 2009 and 0 otherwise. RoaCrisis is an interaction term added to the regression to estimate the relationship between return on assets and CEO-compensation during the crisis years as compared to non-crisis years. PlfCrisis is an interaction term added to the regression to estimate the relationship
between passenger load factor and CEO-compensation during the crisis years as compared to non-crisis years.
References
Abowd, J. M. (1990). Does performance-based managerial compensation affect corporate performance? Industrial and Labor Relations Review, 43 (3), pp. 52-
73.
Banker, R. D. & Datar, S.M. (1989). Sensitivity, precision and linear aggregation of Signals for performance evaluation. Journal of Accounting research, 27, pp.
21-39
Banker, R. D., Potter, G. & Srinivasan, D. (2000). An imperical investigation of an incentive plan that includes nonfinancial performance measures. The
Accounting Review, 75, pp. 65-92.
Behn, B. K. & R. A. Riley. (1999). ‘‘Using Non-financial Information to Predict Financial Performance: The Case of the U.S. Airline Industry.’’ Journal of
Accounting, Auditing and Finance 14, 29–56.
Bloom, M. & Milkovich, G. T. (1998). Relationships among risk, incentive pay, and organizational performance. Academy of Management Journal, 41 (3), pp.
283–297.
Bushman, R. M., Indjejikian, R. J. & Smith, A. (1996). ‘‘CEO Compensation: The Role of Individual Performance Evaluation.’’ Journal of Accounting and
Economics, 21, 161–193.
Core, J. E., R. W. Holthausen and D. F. Larcker. (1999). ‘‘Corporate Governance, Chief Executive Officer Compensation, and Firm Performance.’’ Journal of
Financial Economics, 51, pp. 371 – 406.
Chalmers, K., Koh, P. S. & Stapledon, G. (2006). The determinants of CEO compensation: rent extraction or labour demand? The British Accounting
Review, 38 (3), pp. 259 – 275.
Davila, A. & Venkatachalam, M. (2004). The Relevance of Non-financial
Performance Measures for CEO Compensation: Evidence from the Airline Industry. Review of Accounting Studies, 9, pp. 443-464.
Deckop, J. R. (1988). Determinants of Chief Executive Officer Compensation. Industrial and Labor Relations Review, 41 (2), pp. 215 – 226.
Eccles, R. (1991). The performance measurement manifesto. Harvard Business
Review, January-February, pp. 131-7.
Eisenhardt, K.M. (1989). Agency Theory: An assessment and review. The academy of
Management Review, 14 (1), pp. 57-74
Gabaix, X. and Landier, A. (2008). Why has CEO Pay Increased so Much? Quarterly
Journal of Economics, 123 (1), pp. 49 – 100.
Garen, J. E., (1994). Executive Compensation and Principal-Agent Theory. Journal of
Political Economy, 102 (6), pp. 1175 -1199.
Gilson, S. C., & Vetsuypens, M. R. (1993). CEO compensation in Financially Distressed Firms: An Empirical Analysis. The Journal of Finance, 48, pp.
425-458.
Ittner, C. D., Larcker, D. F. & Rajan, M. V. (1997). The choice of performance measures in annual bonus contracts. The Accounting Review, 72 (2), pp. 231-
Ittner, C. D. & Larcker, D. F. (1998). Are Nonfinancial Measures Leading Indicators of Financial Performance? An Analysis of Customer Satisfaction. Journal of
Accounting Research, 36, pp. 1-35
Ittner, C. D. & Larcker, D. F. (2000). Non-Financial Performance Measures: What Works and What Does Not. Knowledge@Wharton. Retrieved 4th of January
2015, from http://knowledge.wharton.upenn.edu/article/non-financial- performance-measures-what-works-and-what-doesnt/
Jensen, M.C., Murphy, K.J., 1990. Performance pay and top-management incentives. Journal of Political Economy, 98 (2), 225-264.
Kerr, J. L., & Bettis, R. A. 1987. Boards of directors, top management compensation and shareholder returns. Academy of Management Journal, 30 (4): 645–664. Kostiuk, P. F. (1990). Firm size and executive compensation. Journal of Human
Resources, 25 (1), pp. 90–105.
Kroll, M., Simmons, S., & Wright, P. (1990). Determinants of CEO compensation following major acquisitions. Journal of Business Research, 20, pp. 349–366. Lavin, D. (1994). Chrysler aids to get bonuses equal to salaries. Wall Street
Journal (February 22) : A3
Lippert, R.L &Porter, G. (1997). Understanding CEO pay: a test of two pay-to- performance Sensitivity measures of alignment and influence. Journal of
Business Research, 40, pp. 127-138.
Matêjka, M., Merchant, K. A., & Van der Stede, W. A. (2009). Employment horizon and the choice of performance measures: empirical evidence from annual bonus plans of loss-making entities. Management Science, 55 (6), pp. 890-905
McGuire, J.W., Chiu, J.S.Y., Elbing, A.O. (1962). Executive incomes, sales and profits. The American Economic Review, 52 (4), pp. 753-761
Monsen, R. Joseph and A., Downs (1965). A theory of large managerial firms. Journal of Political Economy, 73 (3), pp. 221-236.
Ross, S. A., (1973). The Economic Theory of Agency: The Principal's Problem. American Economic Review, 63 (2), pp. 134-139.
Schefczyk, M., (1993). Operational Performance of Airlines: An Extension of Traditional Measurement Paradigms. Strategic Management Journal, 14 (4),
pp. 301-317.
Smith, C. W. & Watts, R.L. (1992). ‘‘The Investment Opportunity Set and Corporate Financing, Dividend and Financing Policies.’’ Journal of Financial
Economics 32, pp. 262–292.
Tosi, H. L., Katz, J. P., & Gomez-Mejia, L. R. 1997. Disaggregating the agency contract: The effects ofmonitoring, incentive alignment, and CEO tenure on managerial decision making. Academy of Manage-ment Journal, 40 (3), pp.
620– 638.
Tosi, H. L., Werner, S., Katz, J. P., & Gomez-Mejia, L. R. (2000). How Much Does Performance Matter? A Meta-Analysis of CEO Pay Studies. Journal of
