Timely vs. delayed CEO turnover

Abstract

This paper investigates changes in company performance following timely versus delayed CEO resignations due to financial wrongdoings. A timely resignation is proactively pushed by the company, and a delayed resignation is driven by investigations initiated by the SEC or other regulatory authorities. Our results show significant negative abnormal returns following the announcement of CEO resignations. In addition, compared with timely resignations, delayed resignations experience a larger and longer lasting negative stock market reaction. This suggests that CEO resignations due to financial wrongdoings are not perceived as good news by investors, and the delayed resignations could make investors lose more confidence, possibly because of worries about the ineffective corporate governance and supervision mechanism. We have found a significant negative relationship between CEO-chairman duality and the timeliness of CEO resignations. Our results have important implications for investors and policy makers.

Appendices

Appendix A

A CEO’s compensation is often considered an important reflection of a CEO’s power within the company. Scholars have argued that, as the CEO’s power relative to the board increases, the efficiency of the board will decline and that the executives’ power significantly increases their compensation amount (Hermalin and Weisbach 1998; Fahlenbrach 2009). Therefore, it can be assumed that higher compensation means higher power for a CEO and a lower likelihood of turnover.

Two broadly accepted hypotheses explain the relationship between tenure and the probability of a turnover: The managerial entrenchment hypothesis suggests that an executive’s social network grows broader over time and provides some resistance against outside pressures, thereby reducing the probability of executives being replaced (Morck et al. 1988). The learning hypothesis suggests that when a new CEO first takes office, the board has little information about the CEO’s true capability; therefore, the board has a relatively high degree of tolerance for the expected performance of the CEO (Gibbons and Murphy 1991). As the board learns more about the CEO, this tolerance decreases and performance that was previously acceptable may no longer be acceptable.

The majority of scholars agree that it is easier for smaller boards to dismiss a CEO who has inferior performance: when a company encounters poor performance, small boards are more inclined to dismiss their CEO (Yermack 1996). The threat of dismissal declines when board size increases. Chakraborty and Sheikh (2008) also find a positive correlation between smaller boards and the probability of a CEO turnover.

Prior research generally suggests a positive relationship between independent directors and the probability of misbehaved or incapable CEOs being replaced. One study shows that outside independent directors are more likely to replace underperforming CEOs and others also confirm the positive relationship between the proportion of independent directors on the board and the probability of CEO turnovers (Weisbach 1988; Kaplan 1994; Kang and Shivdasani 1995; Denis et al. 1997).

Prior research indicates that the larger the company size, the more dispersed its ownership structure; thus, the more difficult it is to obtain sufficient votes on the board to dismiss the CEO. Also, the larger the company, the higher the requirements for the new CEO successor’s knowledge and experience to run the company; thus, the more difficult it is to find a suitable candidate to replace the incumbent CEO. Therefore, it is often argued that company size is negatively correlated with the probability of CEO turnover, and this has been empirically confirmed (Finkelstein and Hambrick 1990). Company performance is an important factor that may influence the probability of CEO turnover.

As a matter of fact, the negative correlation between the likelihood of CEO turnover and company performance has been documented by a large body of empirical research (Defond and Hung 2004; Kaplan 1994; Puffer and Weintrop 1991).

Appendix B

We manually collect data on company characteristics, as well as accounting-related data. Specifically, to investigate the influence of board and CEO characteristics on a firm’s pre- and post-resignation performance, we collect information from Execucomp and SEC proxy statements on CEO-chairman duality, compensation, board size, independent directors, the number of years of experience the CEO has within the firm, and the new CEO’s origin. We collect accounting data from COMPUSTAT and market data from CRSP databases.

To investigate the within-firm variation in performance, we follow a previous study and calculate the operating return on assets (OROAs) and the market-to-book ratio (MTB) for a period from 3 years before the CEO’s resignation to 3 years after the resignation from Compustat and CRSP (Pérez-González 2006). Specifically, we collect information on the firms’ total assets (Compustat item AT), total liabilities (Compustat item LT), net income/loss (Compustat item NI), book value of equity (Compustat item CEQ), market value of equity (Compustat item MKVALT), operating income before depreciation (Compustat item OIBDP), and book value of deferred taxes (Compustat item TXDB). The MTB ratio is defined as the ratio of the sum of the book value of assets, plus the market value of equity, minus the sum of the book value of equity and deferred taxes to the book value of assets (((AT + MKVALT-(CEQ + TXDB))/AT). To control for different industry trends or mean reversion from a firm’s pre-transition performance, we adjust our performance measures by using industry-matched benchmarks (Brad and Lyon 1997). Specifically, we create industry controls by subtracting the median performance of all firms in the same industry from each company’s performance measure. Industries are classified by using the Fama-French industry classification system, which distinguishes between 48 industry sectors and can be found on Kenneth French’s Data Library Web site.

Appendix C

1.1 Buy-and-Hold Abnormal Returns

The buy-and-hold abnormal return (BHAR) is the difference between the buy-and-hold return of a sample firm and its expected buy-and-hold return usually based on a benchmark portfolio. For example, the market-adjusted buy-and-hold abnormal return uses the market return as the benchmark. BHARs are calculated as follows:

$$ {\mathrm{BHAR}}_{\mathrm{i}}=\prod_{\mathrm{t}=0}^{\mathrm{T}}\left(1+{\mathrm{R}}_{\mathrm{i}\mathrm{t}}\right)-\prod_{\mathrm{t}=0}^{\mathrm{T}}\left[1+{\mathrm{R}}_{\mathrm{mt}}\Big)\right] $$

where R_it is the buy-and-hold return for company i in month t, and R_mt is the market buy-and-hold return in month t. The main difference between CARs and BHARs is that CARs do not take the compounding effect into account. One study advocates for using BHARs, which better resemble investors’ investment behaviors, to measure long-term abnormal returns (Brad and Lyon 1997).

Finally, Jensen’s alpha approach, also known as the calendar-time approach, is expressed as.

$$ {\mathrm{R}}_{\mathrm{p}\mathrm{t}}-{\mathrm{R}}_{\mathrm{ft}}={\mathrm{a}}_{\mathrm{p}}+{\mathrm{b}}_{\mathrm{p}}\left({\mathrm{R}}_{\mathrm{m}\mathrm{t}}-{\mathrm{R}}_{\mathrm{ft}}\right)+{\mathrm{S}}_{\mathrm{p}}{\mathrm{S}\mathrm{MB}}_{\mathrm{p}\mathrm{t}}+{\mathrm{h}}_{\mathrm{p}}{\mathrm{HML}}_{\mathrm{p}\mathrm{t}}+{\mathrm{m}}_{\mathrm{p}}{\mathrm{UMD}}_{\mathrm{p}\mathrm{t}}+{\mathrm{e}}_{\mathrm{p}\mathrm{t}} $$

where R_pt is the equally or value-weighted return for calendar month t for the portfolio of event firms that experienced the event within the previous T months, R_ft is the risk-free rate, R_mt is the return on the CRSP value-weight market portfolio, SMB_pt is the difference between the return on the portfolio of small- and large-sized firms stocks, HML_pt is the difference between the return on the portfolio of high and low book-to-market stocks, UMD_pt is the difference between the return on the portfolio of winner and loser stocks during the previous 12 months, and a_p is the average monthly abnormal return, that is, the Jensen’s alpha, on the portfolio of event firms over the T-month post-event period. b_p, S_p, h_p and m_p are sensitivities, that is, betas of the event portfolio for the four factors. It is often thought that Jensen’s alpha approach is more likely to obtain results consistent with market efficiency as the returns are weighted equally by each period rather than by firm.

1.2 Logistic Regression

The dependent variable in our study is a dummy variable that identifies whether CEO resignation is a timely or delayed action. Therefore, the binary logistic regression model is used to analyze the relationship between the dependent binary variable and the independent variables. The logistic regression model is specified as follows:

$$ \begin{array}{l}\kern0.72em \Pr (TIMELY)={\upalpha}_0+{\upbeta}_1 DUAL+{\upbeta}_2 LOGCOMP+{\upbeta}_3 WITHFIRM+{\upbeta}_4 BDSIZE+{\upbeta}_5 INDEPROP+\hfill \\ {}{\upbeta}_6{SIZE}_{t=-1}+{\upbeta}_7{MVEQUITY}_{t=-1}+{\upbeta}_8{OROA}_{t=-1}+\upvarepsilon, \hfill \end{array} $$

where TIMELY is a dummy variable that equals one when the CEO resignation is a timely action and zero when it is a delayed action. As explained in the literature review, we include DUAL, LOGCOMP, WITHFIRM, BDSIZE, and INDEPROP in the logistic regression as the independent variables and SIZE _t=−1 , MVEQUITY _t=−1 and OROA _t=−1 as the controlled variables. DUAL is a dummy variable that equals one when the positions of CEO and chair are held by the same person, that is, duality exists, and zero otherwise. LOGCOMP is the natural logarithm of the resigning CEO’s compensation. WITHFIRM represents the experience (measured in the number of years) the resigning CEO has had with the firm. BDSIZE is the number of board members on the board. INDEPROP is the proportion of the number of independent directors on the board. SIZE _t=−1 , MVEQUITY _t=−1 , and OROA _t=−1 denote the natural logarithm of the firm’s total assets, the market value of equity, and the operating return on assets one year before the CEO’s resignation.

1.3 Within-Firm Variation in Performance

This paper closely follows a previously used methodology that examines the within-firm variation in performance; the original study notes that when concentrating on differences in within-firm performance, one does not need to control for time-invariant company characteristics that may jointly affect a company’s prospects and decision to appoint a new CEO (Pérez-González 2006). The performance measures we use include the operating return on assets (OROA) and the firm’s market-to-book ratio (MTB). Also, we adjust our performance variables using industry-matched benchmarks to control for potential industry trends and mean reversion. This decision is comparable to that of Pérez-González (2006).