Excessive Corporate Cash Accumulations : the understanding from the CEO perspective

(1)

Universiteit van Amsterdam

Amsterdam Business School

Master thesis

Excessive Corporate Cash Accumulations: The understanding

from the CEO perspective.

Student Name: BSc. Lucaci Valeriu-Constantin Student Number: 10604367

Supervisor: Prof. Dr. Arnoud Boot Date: 15th of December 2014

Programme: MSc Business and Economics, specialization Finance Faculty: Faculty of Economics and Business, University of Amsterdam

(2)

Statement of Originality

This document is written by Student Valeriu-Constantin Lucaci who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

Acknowledgement

I would like to take this opportunity to thank my thesis supervisor Prof. Dr. Arnoud Boot for the guidance, advice and patience throughout the whole course of this Master Thesis. I would also like to thank Dr. Florian Peters for the courtesy of sharing the data on forced CEO turnovers. In addition, I would like to thank the co-reader of this work for reviewing my Master Thesis and being the second examiner.

Valeriu-Constantin, Lucaci

(4)

This paper tests the effect of CEOs’ myopia on firms’ likelihood of holding excess cash. Employing a baseline model to classify normal and excess cash reserves of companies in the US, this study finds that corporates are more likely to hold excess cash when the top manager has a shorter decision horizon compared to the industry. Higher probability of holding too much cash is also associated with CEO pay cuts and the number of turnovers within the firm while instances of executive forced turnover and experience in other industries are associated with a lower probability of cash hoarding. The results are consistent with the intuition that executive characteristics are associated with the likelihood of corporates holding excess liquidity.

1. Introduction

During recent years, accumulating cash reserves in excess has been a continuously growing trend among corporates around the world. From 2008 until 2012, non-financial members of the global S&P 1200 index, i.e. 975 of the world’s largest corporations, increased

their cash levels by 62%, to USD 3.2 trillion from USD 1.95 trillion1. In the U.S., as of the

end of the second quarter of 2013, corporate cash holdings amounted to USD 1.8 trillion (11% of the GDP)2, giving rise to pressure from investors, media and politicians alike to invest these overabundant cash stockpiles and stimulate economic growth.

As a consequence, this increased pressure raises the dilemma for managers to choose between spending the free cash flow and stockpiling it as excess cash reserves. From a rational standpoint, a self-interested manager will trade off the benefits of spending today, whether on investments or payouts to shareholders, and the costs associated with potential for discipline related to holding excess cash (Harford et al., 2008). While the quality of the investments made is often visible in the future, with delay, the large accumulation of cash can attract the immediate attention of shareholders. Kirk Kerkorian’s attack on Chrysler in 1995 or the more recent call by activist Paul Singer to Juniper Networks at the end of 2013, to return USD 3.5 billion to investors, demonstrated that excessive cash reserves can quickly become a focal point for activist shareholders and threaten the managers’ positions. In addition, Faleye (2004) brings evidence that proxy contests are increasing with excess cash and following these contests, CEO turnover increases as well as cash payouts to shareholders,

1_{“Pressure mounts for corporates’ cash piles to be put to work”, Financial Times January 21 2014.}

http://www.ft.com/intl/cms/s/2/1e1b9952-794f-11e3-91ac-00144feabdc0.html#axzz3LskF63V4. Accessed June 14th 2014.

(6)

2

while cash stockpiles considerably decline. Therefore, there are both anecdotal and empirical evidence suggesting that executives in the US should avoid visibly excessive cash reserves. However, holding cash also comes with benefits, such as increased flexibility in financing future investment opportunities or a comfortable cushion against cash shortfalls (Harford et al. 2008). This adds nuance to the managers’ trade-off who can also balance benefits of spending with the benefits of holding excess cash.

In the past 25 years average executive tenures have decreased from around 8 years to little less than 4 years (Miller and Le Breton-Miller, 2006). This signal exerts extra pressure on the executive managers to focus on delivering quick results as opposed to longer-term investments. From the managers’ perspective, bigger scale, long-term investments are associated with a higher risk and they directly consume internal resources. Therefore, there may be reasons to avoid investments that do not pay off in the short run. For instance, there is the risk that a long-term, resource-consuming project can go bad which will ultimately threaten their job security. Furthermore, the pay-offs from longer term investments, which can occur after the CEO who initiated them departs, provide and generate benefits only for their successors (James, 1999). This outlines a manager decision horizon nuance to the agency conflict. As executives’ claims on the firm are limited only to their tenure, this induces a “myopic” approach on firm investments and capital structure, in which the short-term performance of the firm is the main focal point. Consequences of managerial myopia are that potentially beneficial projects may be avoided (Antia et al., 2010).

The present research continues in the line of this managerial short-term vision investigation while focusing on the corporate excess cash levels and on the top manager of the firm. The broad question that we posit is: How do the characteristics of CEOs influence

the likelihood of the firm to hold excess cash? We investigate several CEO related

characteristics, namely the CEO’s decision horizon, their age and tenure, the instances they suffered a pay cut (which reflects their lack of effort), the instances they were fired (which reflects their lack of skill) and whether they have accumulated experience in more than one industries. These characteristics are placed in a probabilistic context of the firms they run to hold excess cash. The novelty of this approach is that it focuses strictly on the interaction between the CEO and the excess cash levels of the firm. This niche fits in the larger body of literature studying corporate governance issues and finds a bridging point with the literature on corporate cash holdings. Therefore, we contribute to the literature with a deeper

(7)

3

understanding of the reasons behind the accumulation of corporate excess liquidity by also investigating the managerial component, not only firm specific predictors.

We conduct this study on the US market which is characterized by strong shareholder protection mechanisms and strong enforcement of this protection, a reason for which many studies in favor of Jensen’s (1986) agency theory found evidence that this theory does not hold. Financial and CEO data is collected for the period of 1992-2013. Firstly, a base line model for predicting normal cash levels is developed following Opler et al. (1999). The sample used contains 1,518 unique US publicly listed firms with 26,479 firm-year observations over the period 1992-2013. The model yields a classification of firms either as cash-rich or cash-normal by looking at the residuals from the regressions. More specifically, the sample is divided into three quantiles, ordered by residuals and the firms that are in the top third of this ranking for a given year are classified as cash-rich. Secondly, CEO data is collected for a sample of 1,731 executives and 11,186 individual-year observations. This sample is then merged with the company financial information. Using this new data sample, univariate tests are conducted between the firms identified as holding excess cash and the rest of the firms in the sample over the firm specific characteristics and CEO characteristics. Lastly, using 1993-2010 data, a logistic regression analysis is conducted to determine the relationships between the executives’ characteristics and the likelihood of the firms they run to accumulate excess cash.

Overall, the results in this research indicate that there are significant relationships between the CEO characteristics that have been observed and the likelihood of a firm holding excess cash. Firstly, and most importantly, the CEO’s decision horizon affects his perspective on corporate cash holdings. We find that firms run by CEOs with longer decision horizons relative to the industry, i.e. with a higher expected tenure, are 1.1% less likely to hold cash in excess as opposed to firms run by CEOs with a shorter decision horizon relative to the industry. Further, firms replacing poor skilled CEOs have a lower tendency to hold cash in excess, by roughly 30%-32%. A rather surprising result is observed in the instances when the CEO is disciplined internally, via a pay cut. On average, in these cases, the CEO feels a higher pressure to deliver results which incentivizes him/her to adopt a myopic perspective on firm performance. Firms where a CEO pay cut occurs are more likely to accumulate cash in excess, although the evidence is weak. Last but not least, CEOs with experience in more than one industry prove to have a greater inclination towards spending the cash resources, thus a firm run by such a CEO is less likely to hold cash in excess. Although these

(8)

4

characteristics are indicative for the CEOs decision making with respect to the cash balances and capital budgeting, the fact that the recent financial crisis has induced a significant amount of precaution in managing liquidity is uncontestable. This nuance is captured in our results, as after 2008, firms are 36% to 43% more likely to accumulate excessively large cash balances.

The remainder of this thesis is structured as follows. Section 2 covers the existing literature on corporate cash holdings and corporate governance and provides the theoretical framework on which the empirical hypotheses are built. Section 3 outlines the data collection process and describes the samples. Section 4 presents the methodology employed in testing the hypotheses. The results and their interpretation are documented in Section 5. Section 6 concludes, describes the limitations and provides avenues for future research.

2. Theory and empirical hypotheses

In this section of the paper, we will discuss the existing literature with regards to top management propensity towards a “myopic” perspective on corporate performance and capital structure and build our hypotheses related to excess cash holdings as we go along.

Perhaps one of the most interesting debates surrounding corporate cash levels is centered around the question “Why do firms hold excessively large cash balances?”. If we were to adopt the perspective of a world of perfect capital markets, in which capital, be it in the form of debt or equity, can be transferred without frictions from one party to another, then holding cash would have no apparent purpose since firms will always be able to finance themselves from outside investors, at a fair price and at any moment in time (the concept of perfect capital markets was first introduced by Modigliani and Miller, 1958). Far from the utopic assumption, the reality in which enterprises operate is subject to many imperfections: transaction costs, taxes, information asymmetries, moral hazard problems and agency conflicts, which all raise new hurdles for firms to access external capital. If these hurdles are too high, companies might be constrained to give up raising outside capital, reduce investments and, ultimately pass up on otherwise value-adding investment projects. Consequently, firms accumulate cash because it offers them the assurance to engage in future lucrative investments or navigate through times of turmoil, while reducing their dependence on the outside capital markets for financing.

At the end of 2013, Esco Technologies Inc., a US technology firm, held USD 42 million in cash and cash equivalents and USD 1 billion in assets, or a 3.8% cash-to-assets ratio. Its rival,

(9)

5

Zebra Technologies Corp., held at the end of 2013 USD 413 million on an asset base of USD 1.1 billion, thus having a cash-to-assets ratio of 37.5%3. This striking contrast of cash holdings between very similar firms indicates that the determinants of corporate cash holding are significantly reliant upon firm characteristics. Since we study the idiosyncratic effects of managers on the firm’s preference for holding excess cash, it is of paramount importance to first determine the factors that drive the expected levels of cash of individual firms.

The existing literature on cash holdings relies on a number of traditional theories to explain the levels of cash holdings and can be divided into two different lines of research. The first stream of research aims to identify the most significant determinants of corporations’ demand for cash and liquid assets. Opler et al. (1999) explore the main motivations and rationales management has behind corporate cash balances: (1) the precautionary motive, (2) the transactions cost motive and (3) managerial self-interest. Simply said, firms may accumulate cash as a measure of precaution against possible future cash shortfalls, thereby avoiding the high transactions costs associated with external financing. Thus, the precautionary and transactions cost motives are beneficial and serve shareholders’ interest. From a different perspective, a firm’s management might stockpile cash for the sole purpose of following personal objectives (i.e. pet projects and investments, acquisitions, corporate perks etc.) at the expense of the shareholders. The managerial self-interest rationale is based on Jensen’s (1986) free cash flow theory4.

The second stream of literature deals with the effects of corporate liquidity on firm value and performance and attempts to find out how corporations employ these highly liquid assets. Lang et al. (1991), Blanchard et al. (1994) and Harford (1999) show that cash rich firms systematically overinvest their resources and these investments are sometimes value destroying. In contrast, in later studies, Mikkelson and Partch (2003) and Fresard (2010) argue that high cash balances are associated with greater operating performance and higher growth. The main conclusion that stems from these studies is that there is an ongoing debate on whether excess cash holdings are more harmful rather than beneficial for the firms who possess them.

In this paper, we focus on the first strand of literature described above which will contribute to the theoretical framework of assessing a firm’s relative cash-richness. More

3_{The figures are taken from the sample of 26,479 firm-year observations from this research} 4_{Also referred to as the agency theory.}

(10)

6

specifically, it will provide a set of fundamental determinants for constructing a baseline model to predict the normal levels of corporate cash.

2.1. The determinants of cash levels.

Considering that this paper aims to study the relationship between CEO characteristics such as decision horizon or disciplinary measures such as pay cuts and outings and the likelihood of the firm holding excess cash, it is crucial to first determine the factors that drive a company’s expected levels of cash. The causes and consequences of corporate cash levels have been extensively studied and the existing literature provides valuable resources in this direction. Initially, transaction costs were seen as the major cause of cash build-ups and firms with a greater marginal cost of cash shortages were expected to have more cash5. A considerably broader view on the determinants of corporate cash holdings was brought forward by Opler, Pinkowitz, Stulz and Williamson (1999) who developed two main explanations for cash holdings: (1) the tradeoff theory and (2) the financing hierarchy theory. In the set-up of the tradeoff theory, firms balance the costs and benefits of holding cash, thus aiming for an optimal level of cash holdings. In this view, they consider the transaction costs motive mentioned above as well as the implications of asymmetric information and the agency costs of external financing on their preference for holding cash. The financing hierarchy theory assumes that there is no

optimal level of cash using arguments similar to the pecking order6 theory of capital structure, in

which cheaper internal funding is preferred over more costly outside capital (e.g. debt and equity). While these two theories assume that managers act in the best interests of their shareholders, Jensen (1986) introduces the (3) free cash flow theory which is centered around the principal-agent conflict. Accordingly, excessive cash levels allow for greater managerial discretion and therefore managers are tempted to stockpile cash for purposes that serve their personal interest instead of the owners’.

In view of Myers and Majluf (1984), the presence of asymmetric information in the capital markets, between managers, who, naturally poses more information about the assets under their control, and investors, increases the appeal of cash over external financing. As less informed investors try to avoid purchasing overpriced assets they will require a premium on their expected return in order to mitigate this risk. As a consequence, raising funds for new projects becomes more costly for the firms and may even lead to impossibility to make value adding

5_{See Miller and Orr (1966)}

(11)

7

investments. To avoid resorting to this more expensive outside capital, firms prefer the cheaper internal funding. Therefore, information asymmetries financially constrain firms and the higher this asymmetry is around a firm, the larger its demand for cash and highly liquid assets will be. Myers and Majluf (1984) argue that firms with large growth opportunities are subject to more information asymmetries because growth prospects are hard to assess and highly intangible.

However, the intensity of information asymmetry and growth opportunities can vary by certain firm characteristics. Larger, more developed firms are associated with a lower degree of informational concerns as opposed to smaller firms, as stated by Brennan and Hughes (1991). Investors find it harder to assess the riskiness and growth opportunities of smaller, less established firms and, consequently, they discount their securities more thoroughly, or, in the worst case, deny them access to the capital markets (Petersen and Rajan, 2004). Considering that firm size can affect the cost of external funding, due to presence of informational inequality, we would expect smaller firms to hold relatively higher amounts of cash and cash equivalents compared to their larger peers.

Nonetheless, Aboody and Lev (2000) argue that firm size is a fairly noisy proxy for measuring information asymmetry and growth prospects. A better alternative, in their view, is the research and development (R&D) expense. The reason is that R&D expense information is highly difficult to observe by outsiders of the firm due to its unique nature and, therefore, gauges the informational inequalities more accurately. The R&D expense has also been used as a proxy for growth opportunities and informational asymmetries by Bates, Kahle and Stulz (2006) and Opler et al. (1999). They suggest that firms with large R&D expenses are more exposed to the risk of financial distress and, thereby, will tend to hold higher amounts of cash as opposed to firms with relatively lower R&D investments. It is therefore expected that R&D expenses are positively related to cash levels.

The demand for cash is also believed to be affected by the presence of agency costs of debt (Opler et al., 1999). These agency costs occur when the interests of the firm’s shareholders and its debt holders (e.g. bondholders, banks) diverge. Debt holders expect a safe return on their investments and the firm’s engagement in risky projects, which are more beneficial for the shareholders, exposes them to the risk of financial distress or default. To mitigate this risk, bondholders could impose stricter covenants on the firms which can prevent firms from taking on risky investments, even if these are expected to be positive NPV projects. Therefore, as argued by Myers (1977), firms with valuable investment prospects, which are typically

(12)

high-8

growth, risky firms, will be exposed to higher agency costs of debt and the risk of underinvestment. These firms would benefit more from holding enough cash in their treasuries to finance themselves.

In view of Opler et al. (1999), the market-to-book ratio or the Tobin’s Q ratio of a firm is a good indication on the investment opportunities of a firm. As argued earlier, growth prospects and investment opportunities are difficult to value and hence, a source of asymmetric information which increases the cost of external funds. Consequently, we expect that firms with higher Tobin’s Q ratios to hold more cash and liquid assets to safeguard the possibility of financing potentially profitable investments.

So far, the discussion has been centered on the information asymmetries and growth prospects which financially constrain firms and, consequently increase their demand for cash. In addition, various company specific characteristics can have an influence on the firm’s demand for liquidity. The most important such determinants, which have been extensively debated in the literature, will be presented in the following paragraphs.

Firstly, the relationship between a firm’s cash flow and its cash demand is rather ambiguous. On the one hand, large cash flows alleviate the risk of default and financial distress costs and, therefore, lessen the need to hold large cash balances (Miller and Orr, 1966). On the other hand, as argued by Jensen (1986), large cash flows enable the management to excessively stockpile cash, for reasons that benefit their interests in the detriment of the shareholders’. Hence, either a positive or negative relationship is expected between cash levels and cash flow.

Further, the cash flow volatility of a firm is also thought to have an impact on the corporate demand for cash. Theoretically, if the cash flows are unstable over time, it is perceived as a risk and firms will find it beneficial to hold large buffers of cash and other liquid assets to mitigate this risk (Opler et al. 1999). Thus, we expect that cash flow volatility to be a positive driver of cash reserves.

Concerning the substitutes of cash, companies can resort to several measures to inject liquidity on their balance sheets when in shortage of internal cash resources. An obvious substitute for cash is borrowing funds from the capital markets. As argued by the financing hierarchy theory and also by Myers and Majluf (1984), firms will only resort to external financing when they have deficits of cheaper internal funds. In addition, as the amount of debt increases, it will come to be more difficult for a firm to build its internal funds (Baskin, 1987)

(13)

9

and ultimately, the increase in leverage will lead to cash levels diminishing. However, Ozkan and Ozkan (2004) state that in order to obviate the higher costs of financial distress which result from the leverage increase, firms may opt to hold more cash. Hence, a firm’s leverage is expected to be positively related to cash holdings. Following this reasoning, a positive relationship is expected between cash and leverage.

Another highly liquid substitute for cash is the amount of net working capital which can be converted into cash through optimization of the operational activity, for instance. In view of Ozkan and Ozkan (2004) the working capital is considered to be an easily convertible into cash and might, therefore, have a substitution effect for the company’s cash holdings. In line with this reasoning, we expect firms with high levels of net working capital to hold lower amounts of cash.

Regarding a firm’s level of capital investment, Opler et al. (1999) state that, theoretically, this should be negatively associated to the firm’s cash levels since firms spend their cash reserves more rapidly. Surprising or not, the authors document a positive relationship between the cash stockpile and capital expenditures, suggesting that firms undergoing large investments retain cash internally in anticipation of potential cash shortfalls. Hence, we expect a rather opaque relationship between cash and capital expenditures, both a positive or negative association being possible.

Lastly, the firm’s dividend policy is also believed to have an ambiguous effect on corporate cash demands. Opler et al. (1999) claim that dividend paying companies are more likely to hold less amounts of cash, arguing that, when short of internal funds, these firms have the option to increase their liquidity by diminishing or cutting the payouts to shareholders in the form of dividends. Ozkan and Ozkan (2004), on the other hand, argue that dividend paying firms try to avoid the circumstances in which they cannot respect their periodic dividend payments and, consequently hold relatively larger amounts of cash. Therefore, we remain open to both a positive and negative association between a company’s dividend policy and its cash levels.

2.2. Executive characteristics and firm performance

Further, we tap into the corporate governance literature and try to quantify the implications of the top manager’s characteristics on the propensity for a firm to accumulate cash balances in excess of their operational and investing needs. The premise is that managers in general and CEOs in particular can greatly influence corporate performance and behavior. The

(14)

10

literature provides evidence in this direction in the sense that the quality of a firm’s performance varies significantly before and after a CEO turnover (Huson et al. 2001, 2004). Huson et al. (2004) document that firms have higher operating returns in the three years following a CEO turnover and that these returns are increasing more for companies with a greater level of institutional ownership and more independent directors in the board. Greene et al. (2009) and Shleifer and Vishny (1989) found that the corporate capital budgeting decisions are most efficient when the presence of agency and informational asymmetry problems are minimized. CEO turnover may be an indicator of the presence of such agency or informational asymmetry issues as they may cause a CEO to anticipate their departure. Consequently, the CEO would tunnel his/her vision on the short-term firm performance which might cause the quality of capital budgeting decision to decrease as the CEO’s departure approaches.

In line with the free cash flow theory, firms with poor corporate governance systems are expected to hoard larger amounts of cash due to a higher level of managerial discretion. Harford, Mansi and Maxwell (2008) report in a study focused on the US market that firms with poor corporate governance actually hold less cash, however, findings show that the lower documented cash balances result from those firms spending their cash on value destroying acquisitions relatively quickly. In line with this, Dittmar, Mahrt-Smith and Servaes (2003) found that corporate cash reserves are significantly higher in countries with relatively weak shareholder protection. The implications are that in such countries, shareholders are unable to pressure management into redistributing the cash through dividends or share buybacks and, consequently, managers will retain the excess cash within the firm for personal objectives. In addition, firms that exhibit severe forms of management entrenchment and excess cash holdings, in combination with weak shareholder rights are valued lower as opposed to their counterparts (Kalcheva and Lins, 2007). Lastly, in a research by Richardson (2006) there is evidence that firms with more excess cash are more prone to overinvesting, however, certain governance structures affect the amount of investment (e.g. the presence of activist shareholders apparently prevents overinvestment).

2.3. Hypotheses development

In their study, Jensen and Meckling (1979) proved that a shorter tenure causes the managers to underinvest as this increases their hurdle rate for investment projects. Intuitively, myopic executives would make investments with a relatively shorter gestation period, which

(15)

11

offer relatively faster returns in order to develop their reputations in the managerial labor market7

or because they are more concerned with their position security8 (Hirshleifer and Thakor, 1992).

This managerial “short-termism”9

causes long-term investments, which have the potential to increase the firm value, to be seen as suboptimal from the managers’ perspective.

A potential disciplinary mechanism for this “decision horizon problem”, which can diminish the principal – agent conflict, is the managerial labor market (Fama, 1980). Brickley et al. (1999) bring some empirical support to this view. They provide evidence that a CEO’s performance in his final years, before retiring form the position, influences the number of positions he/she will hold in other companies’ boards as an outside director. Moreover, Davidson et al. (2007) state that the labor market is a stronger disciplinary incentive to perform, in the case of younger executives in the debut of their career. Furthermore, Brickley et al. (1999) find that the possibility for a CEO of continued service on his own board is strongly influenced by his performance (abnormal stock returns) in the final two years of his/her tenure. Thus, the prospect of board service can reduce horizon problems as the end of the executive service approaches.

However, towards the end of their careers as executives, these job opportunities may become either irrelevant or weaker incentives as opposed to the incentives offered by the current compensation package (Gibbons and Murphy, 1992). For instance, options on the firm’s equity with a vesting period of two years offered to a CEO closing retirement might be a good performance incentive, but even a two-year span of stock price performance is an arguably short horizon, which could induce myopia. Consequently, as CEOs approach retirement, the possibility of agency conflicts related to the horizon problem increases. In accordance with this view, previous studies have argued that long tenured executives are less likely to engage in new initiatives in the sense that they prefer to favor the status quo (Hambrick and Fukutomi, 1991).

It is important to distinguish between past tenure and expected (future) tenure and to acknowledge that they are inversely related. The literature bringing forward arguments in favor of the intuition that a CEO’s future long-term tenure stands to benefit the firm is complemented

7

See Cambell and Marino (1994)

8_{Chakraborty et al. (2007) found that managers facing high termination risk opt for less risky investments,}

which could provide faster returns, as opposed to managers facing low termination risk.

9

Laverty (1996) studies several determinants of this “short-termism”, i.e. flawed management practice, managerial opportunism, fluid and impatient capital, stock market myopia and information asymmetry. Relevant to the current research is the managerial opportunism.

(16)

12

by the negative arguments related to CEO longevity10 (the number of years a CEO has already

been in office at the firm). Executives with a long tenure grow stale in their roles and disregard their environment, which ultimately diminishes financial performance. Hambrick et al. (1993) bring evidence that long firm and industry tenure are strongly related to commitment to the status quo. It is intuitive that long-tenured CEOs could have developed their own managerial ways and might ignore the competitive threats to a business model that has been successful for a long time. In a contrasting view, long expected (future) tenure should be indispensable for implementing a new strategy that would properly respond, in the long run, to a dynamic environment. An executive, with a short-term horizon might take decisions that are not in the best interest of the firm, in the long-term. For instance, a CEO with a short-term focus could aim for a short-term profitability boost by cutting costs, which is an unsustainable source of growth, rather than investing in positive NPV projects that add value in the long run but do not offer immediate returns. From a cash-flow standpoint, long-term investments are higher consumer of cash and are likely to generate high cash-outflows before they become profitable. In contrast, underinvesting or investing with predominance on the short-term would enable cash to pile up in a firm treasury and accumulate in excess. Following this line of reasoning we formulate our first hypothesis:

H1. The likelihood of holding excess cash is negatively associated with CEO decision horizon.

In a pioneering research, Bertrand and Schoar (2003) bring evidence of a “managerial style” which is determined by personal characteristics and that corporate behavior and performance can vary, depending on the CEO who is in office. The authors argue that CEOs from earlier birth cohorts are more risk averse and conservative than younger CEOs. Consequently, older executives tend to accumulate larger levels of cash in order to mitigate the risk of future, undesirable events. Moreover, conservative CEOs prefer resorting to safer, internal funding rather than more risky (and costly) external financing11.

Arguably, older CEOs have shorter career horizons which limit the time to develop and execute long term strategies which might generate returns after their departure from the firm.

10

Bhagat and Bolton (2008) use the CEO age to tenure ratio as a measure of CEO quality, reasoning that a CEO who has a five year tenure at the age of sixty-five is likely to be of different quality than a CEO with five years of tenure at the age of fifty.

11_{The CEO conservatorism is argued in the Upper Echelons Theory (UET), which is an emerging body of finance}

literature which investigates managerial attributes (e.g. cognitive ability), behaviors (e.g. overconfidence) and perceptions (e.g. with regard to valuation) and link them to corporate financial decisions related to capital structures and corporate investments.

(17)

13

Investments spanning over several years take funds from the current profitability and only pay off in the long run, which causes older CEOs to lose interest in such projects, even if they are valued as positive NPV, and rather focus on the short term. Therefore, we state that older CEOs will be less concerned with the opportunity cost of cash, thus resulting higher cash holdings12. This is in line with Bertrand and Schoar (2003) who also find evidence in favor of a direct relationship between age and firm cash holdings. In line with these arguments we hypothesize that:

H2. The likelihood of holding excess cash is positively associated with the CEO’s age.

There are two fundamental reasons for a long CEO tenure in office. First, a CEO might have had a successful tenure, with a good history of decisions that increased firm value in the past, even though the recent performance has been suboptimal. Second, a CEO might have taken actions that made replacement more difficult, but did not increase firm value, e.g. investing in projects that he had superior know-how in managing, compared to potential replacements (Shleifer and Vishny, 1989). Moreover, as CEO tenure grows so does his/her negotiating power, as the CEO can influence the board selection and build personal relationships with the board members. In both scenarios, the risk of termination for the CEO drops as his tenure increases (Brookman and Thistle, 2009), resulting in less diligent monitoring by the board (i.e. more discretion). It is thus more likely that such CEOs are allowed control over higher cash levels at their own discretion. In order to avoid the risk of losing this “position of power”, which may enable them to secure a future board position or negotiate a better compensation package (Bebchuk and Fried, 2004), we affirm that long tenured CEOs will be less preoccupied with the opportunity cost of cash and more focused on the precautionary role of cash, resulting in excess cash accumulation. Based on this reasoning, we bring forward the following hypothesis:

H3. The likelihood of holding excess cash is positively associated with CEO tenure.

Tapping on the Upper Echelons Theory (see footnote 11), we find that managerial decision making is also influenced by the extent to which executives have a diverse management experience. In view of Herrmann and Datta (2006), past experiences of executive managers define how they design their organizations and how they establish and execute their strategies. In our study, we will focus on the observable other-industry experience from the range of possible past experiences and try to determine whether there is a relationship between the excess cash

12_{Orens and Reheul (2013) find that CEO age is positively related to cash holdings in a study on a sample of}

(18)

14

holdings and this CEO characteristic. Intuitively, CEOs with experience in other industries than the one in which they currently operate, possess broader modes of thinking compared to CEOs who have spent their entire career in the same industry. In addition, the choice to gain exposure to different types of industry experiences signals the intrinsic propensity towards change of such individuals (Silva, 2007). Moreover, other-industry experience helps a CEO expand his/her professional network (Richard et al., 2009), which provides a knowledge resource into opportunities to invest in productive projects instead of amounting large cash reserves. In line with this argumentation, we believe that executives with experience in other industries are able to identify more opportunities to deploy cash reserves. Thus, we hypothesize that:

H4. The likelihood of holding excess cash is negatively associated with the CEO’s experience in other industries.

Gao, Harford and Li (2012) argue that large pay cuts are used as a substitute for forced turnover by boards of directors in the attempt to discipline underperforming CEOs, instead of ousting them. They find that, after either a pay cut or forced turnover, firm performance improves on average and CEOs reduce investments and leverage. Intuitively, in order to avoid the pay cut or, in extreme cases, the termination of their contract, a CEO who has been subject to a pay cut would want to avoid engaging in new investments unless their success is highly probable. Therefore a more cautious behavior is induced by such disciplinary measures, thus leading to greater accumulation of cash. Another benefit of holding on to the cash for the CEO would be to avoid future underperformance by mitigating the risk of shortfall in case of difficult times (in line with the precautionary motives of holding cash).

Contrary to this intuition, excess cash levels may be seen by the shareholders as potential for investment underperformance in the future and the board may enforce a pay cut, or at the extreme, forcing the CEO out, as disciplinary measures for this occurrence. Prior literature on cash reserves provides mixed evidence on whether shareholders should be concerned about large cash reserves. Harford (1999) argue that there is reason for concern. He finds that cash-rich firms are more likely to engage in acquisitions and their acquisitions to be value-decreasing. However, Mikkelson and Partch (2003) document that excessive cash stockpiles do not lead to reduced performance and do not represent conflicts of interest between shareholders and managers.

(19)

15

From the board’s perspective however, it is assessed whether the poor performance is

caused by the CEO’s lack of skill or lack of effort. If the issue is about effort13, and the current

CEO’s estimated ability is relatively higher than that of the available alternative candidates, the board will rationally retain the current CEO, exerting different measures to stimulate his/her commitment. However, in the case of a lack of skill, relative to the average CEO skills available in the labor market, the board will rationally terminate the CEO’s contract and bring in a new manager from the labor pool (Gao et al. 2012). From the perspective of a self-interested manager however, the dilemma of whether to engage in investments or whether to hold on to the cash is an issue of the benefits of investing today or additional flexibility in the future versus the costs associated with the potential for disciplinary measures (Harford et al., 2008). Solving this dilemma is expected to have an impact on the propensity for a firm to hold excess cash.

Considering these arguments, we expect either a negative or a positive relationship between disciplinary measures and the likelihood of holding excess cash. As such we posit the following hypotheses:

H5. The likelihood of holding excess cash is associated with the occurrence of pay cuts. H6. The likelihood of holding excess cash is associated with the forced turnover events.

Corporate capital budgeting should be most efficient when agency and informational asymmetry issues are minimized (Jensen, 1986; Shleifer and Vishny, 1989; Greene et al., 2009) which is a likely scenario when a firm has strong internal communication channels (Hornstein and Zhao, 2011). CEO turnover may be a signal of the presence of such agency and information inequality problems, and therefore be associated with discrete changes in the functioning of internal communication channels. In conditions of intensified agency and informational problems, CEOs may anticipate their contracts termination is imminent, and consequently adopt a myopic perspective with a higher focus on the short-term performance of the firm. This might determine the quality of corporate budgeting process to be less effective as the CEO’s departure approaches. Hornstein (2009) finds that the impact of CEO turnover is asymmetric between under- and overinvesting firms (prior to CEO departure) and these types of firms should be studied separately. Moreover, the author documents in the subsample of under-investing firms an increase in the average liquidity position prior the CEO turnover followed by a decrease in the

13

Most CEOs inarguably work hard, however the type of effort problems we refer to are more related to tasks that CEOs may have avoided or postponed, but which might have driven improved performance. For instance, a CEO might be reluctant in firing a member of his executive team, abort a pet project or divest a division.

(20)

16

average liquidity position after the CEO turnover. Our approach on CEO turnover is in line with this view, as we focus on firms holding excess cash, thus implying, on average an under-investing sample of firms. In line with this view, we hypothesize that:

H7. The likelihood of holding excess cash is positively associated with the number of CEO turnovers at the firm level.

3. Sample selection and data description

3.1. Financial Data 3.1.1. Sample selection

The financial data used in estimating the cash-rich and cash-normal companies has been collected by consulting the Compustat Fundamentals Annual database, which contains company financial information for US public companies, for the 1950-present period. We begin with the full sample of US firms over the period of 1950-2013, which we restrict by applying several filters. First, we eliminate the companies that might be subject to regulatory constraints. These are companies from the financial sector (sic codes 6000-6999), utilities sector (sic codes 4900-4949) and public services institutions and government organizations (sic codes 9100-9995). Second, we eliminate all firm year observations that show missing data for the financial information. In this step, we also eliminate firm year observations that show negative balance sheet data, except for the working capital data (which can take negative values in a balance sheet). Third, we eliminate the firms that show gaps in the time series data after we have dropped the observations described at the previous step and restrict the minimum number of consecutive time observations for a company at 10 years with data up to at least 2010. The minimum number of years set to 10 is essential for computing the cash flow volatility, which is specified as the moving standard deviation for a 5 year window,

ending in year t0. Lastly, when merging with the CEO data, which is available from 1992, we

restrict our firm year observations for the minimum year of 1987 (1992 minus 5 years used in the cash flow volatility derivation). The final sample consists of 1,518 unique firms with 26,479 firm-year observations, spanning over the period of 1992-2013.

(21)

17 3.1.2. Variable construction

The financial data that we collected is then used to fit a regression model that estimates the relationship between a firm’s cash holdings and the specific firm characteristics, with the ultimate purpose of predicting the expected level of cash a firm should hold at a given moment in time. We then employ these predictions to observe which firms hold more cash than expected and, thus identify cash-rich and cash-normal firms for every year. We will use the cash ratio as the dependent variable, for which the existing literature provides alternative specifications. The most widely used definition in the press and media is the cash deflated by total assets ratio. However, Opler et al. (1999) affirm that the more relevant specification for the cash ratio is to deflate the cash holdings by total assets net of cash14. For increased robustness, this study will employ both measures of the cash ratio.

The independent variables used in the explanatory model for the cash ratio are employed in the existing literature, mainly the studies by Opler et al. (1999), Harford (1999) and Bates et al. (2006). The variables (and their corresponding Compustat data items reported in brackets) are described below:

1. Size – due to information asymmetries, smaller firms are believed to be subject to higher financial distress risk, and implicitly, higher outside financing cost. Such firms benefit more from having enough cash on hand. Thus, a negative relationship between size and cash ratio is expected. Firm size is estimated as the logarithm of the Book Value of Total Assets (Data item 6)

2. TobinQ (Tobin’s Q) – used as a proxy for growth opportunities and asymmetric information (widely used in the literature with this purpose). Growth firms, which are surrounded by relatively larger information asymmetries, are expected to have a larger Tobin Q ratio. Such firms, with more value locked in intangible assets (e.g. RD, Goodwill) will find it harder to raise external capital and, would therefore, prefer to hold larger reserves of cash. The Tobin Q ratio is calculated as (Market Value of Equity + Book Value of Debt)/(Book Value of Total Assets (Data item 6)), where Market Value of Equity = number of common shares outstanding (Data item 25) x share price at fiscal year-end (Data item 199) and Book Value of Debt = Book Value of Total Assets (Data item 6) – Common Shareholder Equity (Data item 60)

14_{The Compustat data items used in calculating the cash ratio are: cash and short-term investments(Data item}

(22)

18

3. NWC (Net Working Capital) – consists of non-cash liquid assets which act as a substitute for cash holdings. We expect that firms with large net working capital levels to be less in favor for stockpiling cash. The measure is computed as (Working Capital (Data item 179) – Cash and Short-term Investments (Data item 1))/Total Assets Net of Cash (Data item 6 – Data item 1)

4. Leverage – the effect of borrowing from the external markets on the cash holdings can be double sided. On the one hand, when firms can borrow easily they are not so concerned with holding large amounts of cash for precautionary reasons. On the other hand, as leverage increase, the financial distress cost also increases which can be mitigated by accumulation of sufficient amounts of cash. Therefore, either a positive or negative relationship is plausible. Leverage is measured as Total Debt (Long-term debt (Data item 9) + Short-term Debt (Data item 34)) scaled by Total Assets Net of Cash (Data item 6 – Data item 1)

5. ShTD (Short-term Debt) – this variable is absent in most prior models for estimating cash levels, however, we chose to include it to control for the effects of highly liquid borrowings. The variable is defined as the ratio of Short-term Debt (Data 34) over Total Debt.

6. Capex (Capital Expenditures) – firms with more capital expenditures are expected to either hold larger amounts of cash to mitigate cash shortfalls or, in contrast, hold smaller amounts of cash due to the fact that they invest more. The variable is calculated as the ratio of Capital Expenditures (Data item 128) over Total Assets Net of Cash (Data item 6 – Data item 1).

7. RD (Research and Development) – also a proxy for growth opportunities and information asymmetries. Firms with large R&D expenses have a larger probability of default and the riskiness of these investments raises the difficulty of attracting outside capital. Therefore, firms with high R&D expenses are expected to larger amounts of cash. The variable is defined as Total R&D Expenses (Data item 46) scaled by Net Sales (Data item 12).

8. AQC (Acquisitions) – investments in mergers and acquisitions are naturally expected to diminish the cash reserves of an acquiring company. The acquisitions investments measure is calculated as the level of Acquisitions (Data item 129) in a given year scaled by Total Assets Net of Cash (Data item 6 – Data item 1).

9. CF (Total cash flow) – this variable controls for the impact of cash flows on the firm’s cash reserves. Naturally, a positive relationship is expected between cash flow

(23)

19

and levels of cash. The variable is computed as total cash flow (from operations (Data item 308)), investment activity (Data item 311) and financing (Data item 313)) scaled by Total Assets Net of Cash (Data item 6 – Data item 1).

10. Volatility (of operating cash flow) – seen as a signal of uncertainty regarding future operational activity of the firm against which firms should protect themselves by holding higher cash amounts in reserve. Cash flow volatility is measured as the standard deviation of operational cash flow (Data item 308) over the previous 5 year window in any given year.

11. Dividend – The relationship between the amounts of dividends paid is also expected to be either positive or negative. On the one side, dividend paying firms would rather avoid the situation of being unable to pay promised dividends thus opting to accumulate cash to prevent a cash shortfall. However, in a contrasting perspective, dividend paying firms can scale back on their dividend payments when in need of liquidity, thus being less likely to hold large amounts of cash. The dividend variable is defined as a dummy variable, taking the value of one when a firm pays a common dividend (Data item 21) and zero in the year it does not.

3.1.3 Descriptive statistics

All the variables discussed so far are summarized in Table 1, with data from the 26,479 firm-year observations. To deal with the distorting effects of outliers, these variables have been winsorized at the 5% level. Both measures for the cash ratio discussed earlier are included: Cash/net assets is defined as cash scaled by total assets net of cash while Cash/assets is defined as cash scaled by total assets, including cash. The cash to net assets ratio exhibits a mean of 39.7% and a median of 11.2% and the cash to total assets ratio exhibits a mean of 17% and a median of 10%, indicating that both definitions of cash are skewed to the right. There is an extremely wide variation within the sample as indicated by the standard deviations of the cash ratios which, respectively, take the values of 166% and 18.6% for cash to net assets and cash to total assets. Looking at the predictor variables, we can observe a wide dispersion as well. For example, the Tobin’s Q ratio has a mean of 1.65 and a standard deviation of 1.58 and the Size variable shows a mean of 6.31 and a standard deviation of 2.26.

The findings from our sample, regarding the cash ratios show relatively higher mean and median values when compared to existing studies. Opler et al. (1999) who also uses an

(24)

20

US sample of firms documents for the cash to net assets ratio a mean of 17% and median of 6.5%. However, the sample concentration to the right is also observable in his sample. The difference in magnitude of the cash ratios might be attributed to the fact that they study a

different time period (i.e. 1971-1994 in their study, 1992-2013 in the present study).

Table 1: Descriptive Statistics Financial Data

This table reports the sample characteristics for the 1,518 publicly traded US companies, for their respective 26,479 firm-year observations, over the period 1992-2013. Cash/net assets represents the ratio of cash and cash equivalents scaled by total assets minus cash. Cash/assets is the ratio of cash and equivalents divided by total assets. Size is defined as the natural logarithm of total assets. TobinQ represents the market-to-book ratio, derived as market value of equity (shares outstanding x share price close at fiscal year-end) plus book value of debt (book value of total assets - common shareholder equity) all divided by book value of total assets. NWC is calculated as the ratio of net working capital over total net assets. Leverage is the ratio of total debt over total assets. ShTD is the proportion of short term debt from total debt. Capex is the ratio of capital expenditures over net assets. RD is the ratio of research and development expenses over total net sales. AQC represents the ratio of acquisition expenses over total net assets. CF equals to the ratio of total cash flow over total net assets.

Volatility is measured as the standard deviation of the operational cash flows over the previous 5 year window,

scaled by total net assets. Dividend in this table represents the amount of dividends paid scaled by total net assets, instead of the dummy variable used in regressions. Firms operating in the financial services, utilities and public services have been excluded from the sample to avoid the effects of regulation. To offset the distorting effects of the outliers the variables have been winsorized at the 5% level.

Variable Mean 25th % Median 75% SD

Cash/ net assets 0.3973 0.0322 0.1116 0.3252 1.6617

Cash/ assets 0.1702 0.0312 0.1004 0.2454 0.1862 Size 6.3086 4.7287 6.3046 7.8482 2.2568 TobinQ 1.6511 0.8464 1.2226 1.9081 1.5766 NWC 0.1069 -0.0080 0.1088 0.2490 0.3705 Leverage 0.2085 0.0260 0.1821 0.3172 0.2242 Short-Term Debt 0.2183 0.0001 0.0835 0.3106 0.2956 Capex 0.0626 0.0248 0.0461 0.0842 0.0514 RD 0.0420 0.0000 0.0033 0.0534 0.0697 AQC 0.0265 0.0000 0.0000 0.0149 0.0780 CF 0.0145 -0.0175 0.0034 0.0404 0.0874 Volatility 0.0940 0.0274 0.0476 0.0868 0.3129 Dividend 0.0230 0.0000 0.0000 0.0197 0.7510

(25)

21 3.2. CEO Data

3.2.1. Sample selection

In constructing the CEO data sample, the Compustat Execucomp database is consulted. This database contains executive specific information for the US, from 1992 to 2013. Starting from the full available period in Execucomp, our sample is built as follows. We extract data on CEO age and tenure, in order to compute the “expected tenure” measure, which will proxy for the decision horizon. Therefore, we require CEO-year observations to have data available for the age, commencement date and termination date. The compensation data required to

flag the pay cuts are sourced as “total compensation” which is the sum of15

salary (1), bonus (2), options granted (3), restricted stocks granted (4), long-term incentive plan (5), other annual compensation (6) and all other compensation (7). Further, several restrictions are applied. First the same industry restriction as in the financial data sample are applied by excluding financial, utilities and government and public services institutions. Secondly, CEO turnovers within the same year, when the same CEO returns the next year are dropped, justifying that the interim CEOs are irrelevant to our analysis. Lastly, the database is merged with the financial database and firm-year observations without CEO or financial data are excluded. The final CEO population used in our analysis consists of 1,731 individuals. The merged dataset is also the final version of the data we use to conduct the hypotheses testing and consists of 631 unique firms and 11,186 firm-year observations.

3.2.2 Variable construction

In order to assess the relationship between CEO characteristics and the likelihood of a firm to hold excess cash, several studies are consulted as a theoretical background for constructing the CEO specific variables. The descriptive statistics for the CEO sample that will be used in the logistic regression analysis are presented in Table 2.

We conduct our empirical investigation for testing the CEO Decision Horizon hypothesis starting from the premise that the manager’s decision horizon can be adequately estimated by his/her expected tenure with the firm. We follow closely the procedure in Antia et al. (2010). In this set-up we assume that the comparison with other CEOs in their industry signals to a particular CEO their own estimated tenure. This comparison is done in two

(26)

22

dimensions, namely the length of current tenure and age. Thereby, our measure is computed as follows:

[ ] (1)

where is the number of years the CEO has been in office and is the

age of the CEO working for firm in year . and are the industry

medians of Tenure and Age, respectively. We define an industry using the Fama-French 12 industry classification16.

Considering that the Decision Horizon (DH) measure is an industry-adjusted measure, it can take either negative or positive values. A negative value means that the CEO’s expected tenure is shorter than the industry median, because he/she is older and/or has held office for a longer period than the same industry’s median firm’s CEO. A positive value should be interpreted similarly, that is the CEO’s expected tenure is longer than the industry median because the CEO is younger and/or has not held the office as long as the industry median.

CEO Age is defined as a categorical variable with three values: one for CEOs younger than 35 years of age, two for CEO age between 35 and 50, and three for CEO’s older than 50 years. CEO Tenure is specified as a categorical variable as well, which distinguishes between tenures below 5 years (value of one), tenures between 5 and 9 years(value of two) and tenures of 10 years or more (value of three). CEO experience in other industries is defined as a dummy variable, taking the value of one if at a given year the CEO has previously held executive positions in other industries (defined by Fama-French 12 industry classification) and zero otherwise.

To define the CEO pay cuts variable we follow the Gao et al. (2012) methodology. We first identify CEOs who were subject to an extreme reduction of 25% or more of their total

compensation17. It is noteworthy to affirm that a CEO’s total compensation can fluctuate over

time if stock options and grants, which represent the largest component of an executive’s

16

The 12 industries are: (1) Consumer Nondurables, (2) Consumer Durables, (3) Manufacturing, (4) Energy, Oil, Gas and Coal Extraction and Products, (5) Chemicals and Allied Products, (6) Business Equipment, (7) Telecom and Televesion Transmission, (8) Utilities, (9) Wholesale, Retail and Some Services (e.g. Repair Shops), (10) Healthcare, Medical Equipment and Drugs, (11) Financial Services and (12) Other (e.g. Transports, Building Materials, Mines, Hotels etc.)

17_{Execucomp Data item Total compensation is the sum of: (1) Salary, (2) Bonus, (3) Restricted Stock Grants, (4)}

(27)

23

total compensation, are not awarded every year. Assuming that a CEO is compensated in stock and options benefits once every two years, we will focus on pay cuts instances every second year. Further we apply several filters to isolate the pay cuts. First, we require that the

same CEO is in office from year to pay cut year . Second, his/her total compensation in

year is no more than 75% of his/her compensation in year . Third, his/her total

compensation in year is no higher than 125% of the compensation from year . Using

this procedure, we find 626 instances of pay cuts over the period 1994-201318.

We obtained our data for forced CEO turnover from Peters and Wagner (forthcoming) and Jenter and Kanaan (forthcoming), which contain all forced turnover instances between 1993 and 2010. Peters and Wagner (forthcoming) and Jenter and Kanaan (forthcoming) follow Parrino (1997) and classify a CEO departure as being forced if they find supporting evidence in the press and media, i.e. the press reports specifically if the CEO is fired or forced out, retires or resigns due to pressure. This procedure follows closely the study by Parrino (1997) who describe it in detail. The number of CEO turnovers is defined as the sum of CEO turnovers that occurred into a firm in a given year up to that point. We exclude from these the instances of CEO deaths as reported by the Execucomp database record for reason of departure.

The Outside Industry Experience variable takes the form of a dummy, taking the value of one for every firm year observation in which the CEO has previous experiences as executive in other industries. To be more clear, assume for example, a CEO has transitioned from one company to another that operates in another industry as indicated by its SIC code. For every firm year observation in which the CEO is the new position with the second company, the variable Outside Industry Experience will take the value of one. The variable is meant to indicate the number of years the company has been run by the more generalist CEO (i.e. with a vast experience in more than one industry) and enable us to observe the effects on the likelihood of the company holding excess cash.

18

Considering that the Execucomp database starts in 1992, we cannot identify any pay cuts any sooner than 1994 due to our restriction.

(28)

24 Table 2. Descriptive Statistics CEO Sample

The table reports descriptive statistics for the CEO sample (1,731 unique individuals with 11,186 firm year observations, for the 1992-2013 period) used in the logistic regressions. Short DH shows the number of the firm year observations when the CEO has a shorter expected tenure than the industry median. Long

DH represents the number of firm year observations in which the CEO has a longer expected tenure than

the industry median. Age represents the mean age of the population sample, measured in years. Age <35,

Age between 35 and 50, Age > 50 represents the number of firm year observations when the CEO is in

either category of age. Tenure represents the mean of the population sample, measured in years. Tenure <

5, Tenure between 5 and 9, Tenure >10 represent the number of firm year observations when the CEO

fits in either one of the categories. Pay cuts represent the instances of pay cuts that occurred within the CEO population throughout the whole period. Forced Turnovers represent the number of instances of forced turnovers that occurred within the CEO population throughout the period 1993-2010. Other

Industry Experience represents the firm years when the CEO had at least one year of executive

experience in other industry. The percentages of the total number of observations are reported on the right-most column.

^ - The Forced Turnover sample covers the years 1993-2010, with 10,555 firm year observations, and the percentage is adjusted accordingly.

Variable Value %

Short DH 5,798 51.8%

Long DH 5,388 48.2%

Age (Mean) 56.5

Age < 35 3 0.03%

Age between 35 and 50 2,222 19.9%

Age > 50 8,961 80.1%

Tenure (Mean) 8.7

Tenure < 5 3,921 35.1%

Tenure between 5 and 9 3,546 31.7%

Tenure > 10 3,278 29.3%

Pay cuts 626 5.6%

Forced Turnovers 189^ 1.8% ^

Other Industry Experience 51 0.46%

N 11,186

4. Methodology

In the preceding section we have documented the sample selection, variable construction and descriptive statistics for the financial data and the CEO sample. In this section, we will focus on the methodology employed for testing the hypotheses. This section will follow the logical steps we employed in conducting our analysis which are as follows. In the first stage, the expected levels of cash holdings for the firms in the financial data sample are estimated. For this purpose, studies such as Kim et al. (1998), Opler et al. (1999) and