CEO Entrenchment and the Information in Dividend Decreases

Abstract

We use unique hand-collected data to conduct an initial examination of the relationship between the information in dividend decreases and proxies of chief executive officer (CEO) entrenchment. The evidence suggests that CEO entrenchment weakens the negative stock market reaction to dividend decreases. However, the evidence relating CEO entrenchment to long-term firm outcomes is mixed. Following dividend cuts, CEO entrenchment is associated with better short-term profitability, but bankruptcy is more likely. Following dividend suspensions, long-term profitability is worse, but bankruptcy is less likely. Overall, the evidence is consistent with the notion that entrenched CEOs obscure the bad news in dividend announcements, which is later revealed in the long run.

1. Introduction

Dividend decreases are typically associated with negative reactions in the stock market, which suggests that dividend decreases contain information about firm value (Miller & Modigliani, 1961). We investigate three potential mechanisms whereby CEO entrenchment may moderate the relationship between dividend decreases and changes in firm value. First, agency theory suggests that dividends remove cash from the firm that would otherwise be expropriated by entrenched managers (Jensen, 1986) or that dividend payments increase monitoring by capital markets Easterbrook (1984). Thus, a dividend decrease by a firm with an entrenched CEO should reduce firm value more than a firm without an entrenched CEO. Second, modern dynamic agency models such as Wu (2018) suggest that CEO entrenchment may reduce the information in dividends about the firm’s prospects.1 In this model, managers face a greater risk of job loss if they reduce dividends and therefore will only do so when earnings will not recover fast enough to maintain the dividend. Because turnover risk is less of a concern for entrenched CEOs, dividend decreases will contain less information if the CEO is entrenched. Third, CEO entrenchment is correlated with corporate information production (Anderson et al., 2009; Baik et al., 2007; Lin et al., 2020) and optimism in financial reporting (Hendricks et al., 2022; Huang & Kang, 2019). Thus, an entrenched CEO may also influence the information contained in dividend announcements.

Using a hand-collected sample of proxies for CEO entrenchment, we conduct an initial examination of the relationships between our proxies of CEO entrenchment and information contained in dividend decreases.2 As far as we are aware, this is the first paper to examine these relationships. Overall, the evidence suggests that entrenched CEOs obscure the bad news in dividend announcements, consistent with entrenched CEOs providing optimistic disclosures (Hendricks et al., 2022; Huang & Kang, 2019).

We begin by examining the relationship between measures of CEO entrenchment and cumulative abnormal returns (CARs) surrounding dividend cuts and suspensions. We find evidence of a positive relationship between founder CEOs and CARs around dividend cuts, although this is only statistically significant in one of the three examined event windows. We also see a negative relationship between CARs and firms with greater institutional ownership. Moving to the dividend suspension sample, we find evidence of a positive relationship between CARs and CEO stock ownership and a negative relationship between CARs and whether the CEO is a new hire. This evidence suggests that firms with an entrenched CEO tend to exhibit higher CARs surrounding dividend decreases compared to those without an entrenched CEO. This finding supports the notion that CEO entrenchment reduces the bad news provided by dividend cuts and suspensions.

We then investigate the relationship between changes in earnings surrounding dividend decreases and our proxies of CEO entrenchment. We find little evidence that changes in earnings surrounding dividend cuts are associated with measures of CEO entrenchment, which suggests that perhaps CEO entrenchment does not moderate the information about earnings in dividend cuts. Moving to the sample of dividend suspensions, we find evidence of a negative association between the change in earnings three years after a suspension and founder CEOs, which suggests that CEO entrenchment is associated with relatively worse long-run earnings following dividend suspensions.

Ham et al. (2020) point out that gross profit is better suited to measuring persistent economic income because it is less affected by asset write-downs and investments that coincide with dividend changes. In the first year following the dividend cut, there is a positive association between changes in gross profit and whether the CEO is also the chairperson of the board and a negative association if the CEO is a new hire. These results suggest that entrenched CEOs are associated with relatively better profitability in the short run following dividend cuts. However, we also find evidence of a negative relation between CEO stock ownership and gross profit two years after a dividend suspension, which suggests that firms with an entrenched CEO experience relatively worse profitability two years following a dividend suspension.

We then investigate whether the likelihood of a CEO being replaced following a dividend decrease is related to CEO entrenchment. We also examine whether collective actions, such as the firm being acquired or filing for bankruptcy following dividend decreases, are related to CEO entrenchment. Most of the evidence from these tests suggests that shareholders are less likely to intervene following dividend decreases when the CEO is entrenched. CEOs are less likely to be replaced following dividend cuts if they are an insider. They are less likely to be replaced after a suspension if they are founders or family members of the controlling family. In the dividend cut sample, the likelihood of acquisition decreases with CEO tenure. In contrast, the likelihood of bankruptcy increases if the CEO is a founder. Turning to the dividend suspension sample, the likelihood of an acquisition diminishes with CEO stock ownership, and the likelihood of bankruptcy decreases if the CEO is either a founder or an inside hire.

To summarize, the stock market evidence suggests that CEO entrenchment reduces the bad news contained in dividend cuts and suspensions. However, dividend cuts of firms with an entrenched CEO appear to contain bad news in the long run, as evidenced by a greater likelihood of bankruptcy. Additionally, dividend suspensions of firms with an entrenched CEO appear to have relatively worse long-run profitability. Thus, the combined evidence appears to support the notion that entrenched CEOs obscure bad news in the short run, which is later revealed in the long run. The conflicting evidence that firms with an entrenched CEO are more (less) likely to file for bankruptcy after a dividend cut (suspension) indicates that suspensions provide firms with an entrenched CEO more resources to stave off creditors.

Our evidence contributes to the literature examining the relationship between CEO entrenchment and dividend policy. For example, Hu and Kumar (2004) present evidence that firms with entrenched managers make larger dividend payments to avoid the cost of monitoring by shareholders. Onali et al. (2016) present evidence from a sample of European banks that entrenchment is associated with lower dividend payments, balancing shareholder demands and regulatory requirements for liquidity. Adams et al. (2024) present evidence from the UK insurance sector that reinsurance allows firms with an entrenched CEO to pay more dividends to avoid monitoring because reinsurance reduces liquidity risk. Despite the significant interest in the relationship between CEO entrenchment and the level of dividend payments or, more generally, agency costs and dividend policy (Officer, 2011), to our knowledge, there is no research examining the relationship between our measures of CEO entrenchment and the market reaction to dividend cuts or suspensions, nor is there any exploration of the relationship between these measures and changes in profitability following dividend cuts or suspensions.3 We suspect that this gap is due to data constraints. We overcome this challenge by collecting data on CEO characteristics for a large sample of CEOs who have cut or suspended dividends.

Several papers examine the relationship between dividend cuts and changes in earnings, the most recent being by Ham et al. (2020), which shows that the inconsistent evidence supporting the signaling theory in previous work is due to the methodologies employed in earlier studies. More directly related to our paper, Deng et al. (2024) presents cross-country evidence of a negative association between investor protections and the signal in dividend changes regarding future earnings. They posit that the removal of cash from the firm to reduce managerial rents is more critical when investor protections are low, and dividend changes are therefore more sensitive to temporary shocks to earnings, conveying a weaker signal about permanent changes to earnings.

There are several differences between our analyses and those in Deng et al. (2024). First, we examine stock market reactions, changes in profitability, and the collective action of shareholders and stakeholders, while they focus solely on changes in earnings. Second, our analyses rely on variations in measures of CEO entrenchment in U.S. companies, whereas they depend on cross-country variation in investor protections. Our results are therefore not subject to cross-country differences that could potentially be correlated with investor protections.4 That is, holding investor protections fixed, we present evidence that proxies of CEO entrenchment are related to the information in dividend cuts or suspensions regarding firm value.

Finally, we contribute to the literature documenting that corporate information production is related to CEO entrenchment (Anderson et al., 2009; Hendricks et al., 2022; Lin et al., 2020). The evidence in our paper indicates that entrenched CEOs obscure the negative information in dividends, which is later revealed in lower long-term profitability in the case of suspensions and bankruptcy filings in the case of dividend cuts.

The paper proceeds as follows. In Section 2, we review the related literature and develop our main hypotheses. In Section 3, we describe the data sources used in this study. In Section 4, we report the empirical results. In Section 5, we conclude.

2. Related Literature and Motivation

Financial economists have invested considerable resources into understanding why firms pay dividends.5 The work of Miller and Modigliani (1961) suggests that dividend policy is irrelevant to firm value; however, it also suggests that dividend changes may contain information about the value of the firm. Indeed, empirical studies show that dividend changes are correlated with stock returns: announcements of dividend increases are associated with positive stock returns, and announcements of dividend decreases are associated with negative stock returns (Aharony & Swary, 1980; Benartzi et al., 1997; Eades, 1982; Pettit, 1972; Woolridge, 1983).

Tests of the relationship between stock returns surrounding dividend cut announcements and agency costs are usually conducted by examining the relationship between CARs and Tobin’s Q. The rationale is that firms with Q less than one should distribute cash, because they are likely to be overinvesting. Consistent with this view, Lang and Litzenberger (1989) present evidence that CARs of dividend-cutting firms with Q less than one are more negative than those with Q greater than one. However, Denis et al. (1994) and Yoon and Starks (1995) report that this evidence is due to inadequate controls for firm size and the pre-cut dividend yield. Officer (2011) suggests that the conflicting results are due to differences in operating cash flow. Firms with Q less than one also need high cash flows to overinvest. Indeed, Officer (2011) demonstrates that CARs surrounding dividend initiations are greater for firms with Q less than one and high operating cash flows compared to other firms.6 Compared to this line of research, we proxy for agency costs using measures of CEO entrenchment. That is, we hold Tobin’s Q and other determinants fixed and directly examine the relationship between CARs and measures of CEO entrenchment.7

Economists have also developed signaling models to explain the observed stock price reactions to dividend changes. For example, Bhattacharya (1979) suggests that because dividends are taxed at a higher rate than capital gains, firms with long-lived productive assets can signal their profitability to long-term investors who have a relatively lower dividend tax rate than short-term investors. Miller and Rock (1985) posit that dividends can signal future earnings because increasing payouts comes at the expense of foregoing valuable investment projects; thus, only firms with strong earnings can afford to pay high dividends.

The empirical evidence regarding the information in dividend changes related to future profitability is mixed. Some studies (e.g., Aharony & Dotan, 1994; Brickley, 1983; Carroll, 1995; Ham et al., 2020) document that dividend changes predict future earnings, with higher (lower) earnings typically following a dividend increase (cut). In contrast, other studies (e.g., Brickley, 1983; DeAngelo et al., 1996; Grullon et al., 2005; Penman, 1983) find that dividend changes do not provide incremental information about future earnings. Ham et al. (2020) suggest that the mixed evidence is due to the empirical methodologies used in the previous studies. They propose a new methodology to overcome some of the issues in the prior work and show that dividend changes predict changes in future earnings. Kaplan and Pérez-Cavazos (2022) demonstrate that the information in dividend changes about earnings is moderated by investment opportunities.8 We contribute to this line of research by examining whether CEO entrenchment moderates the information about earnings in dividend decreases.

Several studies present evidence that firms with an entrenched CEO release different information than those without an entrenched CEO. For example, Anderson et al. (2009) suggests that founder and family CEOs release less information to facilitate rent-taking. However, using a regression discontinuity design to identify managerial entrenchment, Lin et al. (2020) present evidence that CEO entrenchment reduces corporate information production. Baik et al. (2007) present evidence that firms with high managerial ownership are less likely to issue earnings forecasts, particularly in anticipation of bad news. Hendricks et al. (2022) present evidence that founder CEOs present more optimistic information. Additionally, Huang and Kang (2019) present evidence that family firms report more optimistic information, while Liao et al. (2025) report that family firms provide more readable annual reports. Building on these studies, we posit that information in dividend announcements is also likely to be different for firms with an entrenched CEO compared to firms without an entrenched CEO. We complement this literature by investigating the information contained in the dividend decreases of firms with an entrenched CEO.

2.1. Motivation

In this section, we draw on the extant literature to motivate our main empirical tests. Our empirical tests investigate three hypotheses that link CEO entrenchment to the information contained in dividend decreases.

The first hypothesis is derived from agency theories of payout policy, which suggest that dividend cuts will leave more cash in the firm for managers to expropriate from shareholders Jensen (1986). Because we expect agency costs to increase with CEO entrenchment, we anticipate that the market reaction to dividend cuts by firms with an entrenched CEO will be more negative compared to those of firms without an entrenched CEO. Furthermore, assuming that entrenched CEOs are more likely to overinvest than non-entrenched CEOs, we expect that dividend decreases convey more negative information about future profitability if the CEO is entrenched compared to a firm where the CEO is not entrenched.9

Hypothesis 1.

If entrenched CEOs decrease dividends to expropriate cash that would otherwise be invested in positive net-present value projects, then stock returns and changes in profitability surrounding dividend decreases will be more negative, relative to a firm without an entrenched CEO.

The second hypothesis is derived from Wu (2018). In this setting, managers view earnings and dividends as informational substitutes; low earnings accompanied by a dividend decrease increase the manager’s turnover risk because shareholders are more likely to infer a poor match between the manager and the firm from this action. Therefore, managers have incentives to smooth dividends relative to earnings because it reduces turnover risk. If low earnings are likely to persist, managers are forced to cut the dividend. Thus, dividend decreases indicate that the firm’s poor performance is likely to continue. However, if managers do not fear intervention because they are entrenched, dividend changes may not convey information about firm value. Indeed, Wu (2018) shows that their model has a slightly better fit for low-reputation and short-tenure executives who have higher turnover risk. This discussion leads to our second hypothesis:

Hypothesis 2.

If the information in dividend decreases is linked to CEO turnover risk, then stock returns and changes in profitability surrounding dividend decreases will be less negative for a firm with an entrenched CEO, relative to a firm without an entrenched CEO.

The final hypothesis is derived from the corporate information production literature. If firms with an entrenched CEO produce more (less) information, then we expect a muted (stronger) market response to a dividend cut because the bad news is more (less) likely to have already been partially revealed. However, if communications from firms with an entrenched CEO tend to be optimistic, then we also expect a muted market reaction to dividend cuts of firms with an entrenched CEO.

We do not expect a relationship among CEO entrenchment, information production, and reported long-run profitability except in the case where entrenched CEOs are excessively optimistic, that is, cases where future profitability is likely to be more negative than that of firms without entrenched CEOs, but this information is obscured by entrenched CEOs.10 This discussion leads to the following hypothesis:

Hypothesis 3.

If information in dividend announcements by entrenched CEOs is excessively optimistic, then stock returns will be less negative and changes in profitability will be more negative for a firm with an entrenched CEO, relative to a firm without an entrenched CEO.

3. Data

This section describes the various data sources used in this study.

3.1. Dividend Cuts and Suspensions

We obtain our sample of dividend cuts from Alderson et al. (2021). They show that traditional methods for identifying dividend cuts in CRSP result in samples where most observations are not ordinary dividend cuts. We obtain their clean dividend cut data and summarize their data collection process.

They begin with all GVKEY-PERMNO pairs in the CRSP/Compustat merged database, where share codes 10 or 11 indicate common stock. The sample is restricted to quarterly ordinary common dividends (CRSP distribution code 1232) in the CRSP distribution file. The amounts of dividends per share reported in the CRSP distribution file are adjusted using the CRSP cumulative adjustment factor on the ex-dividend date. The dividends are then sorted by payment date within each GVKEY-PERMNO combination. Observations with a successive change in dividend per share (DPS) less than −10% are flagged as dividend cuts. Observations where another type of distribution, a merger, or a stock split occurred between payment dates are removed from the sample. The remaining observations are then manually reconciled with each firm’s SEC filings in the EDGAR database to ensure that an ordinary dividend cut occurs. The sample contains 1131 dividend cuts from 1994 to 2019.

We also use a sample of dividend suspensions in our analysis. Dividend suspensions are also obtained from Alderson, Betker, and Halford. These data are not used in Alderson et al. (2021) because they focus on dividend cuts, but they were collected using a similar process. Specifically, they begin with the complete list of quarterly ordinary common dividends sorted by payment date, as described in the previous section. Then, for each GVKEY-PERMNO pair, they identify gaps in each firm’s time series that are greater than 128 days (the 95th percentile in the sample) or instances where there are more than 128 days between the firm’s last dividend payment and the date of the last fiscal year-end reported in Compustat.

These observations are then reconciled with the EDGAR data to eliminate the cases where the dividend suspension is confounded with a merger, spin-off, bankruptcy, change in dividend timing, or accelerated dividend payments. We then manually collect dividend suspension announcement dates using EDGAR and Proquest U.S. Newsstream, which includes major and local newspapers and the Wall Street Journal. We remove 10 observations because they have the same announcement date as a dividend cut. The sample consists of 862 dividend suspensions from 1994 to 2019. Finally, we exclude firms in the financial and utility industries from both samples because their payout policies are heavily regulated in these sectors. This screen reduces the sample to 568 dividend cuts and 615 dividend suspensions.

3.2. Dividend Cutting CEOs

In this section, we discuss the process of identifying CEOs who cut or suspend dividends. For each of the 1183 dividend cuts or suspensions in our list, we identify the CEO in office at the time of the cut using EDGAR or Proquest news searches. We collect the CEO’s start date, end date, turnover announcement date, gender, approximate age at the time of the cut, and if the CEO is a co-CEO, an insider (an employee for more than two years before becoming CEO), an outsider (the CEO is not an employee in the two years before becoming CEO), a family member of the controlling family, a founder or co-founder of the company, the chairperson of the board at the time of the cut, an interim CEO, or a lame duck CEO (i.e., the CEO has a turnover announcement date before the dividend announcement but does not leave office until after the dividend cut). After eliminating interim CEOs (55 obs.), lame duck CEOs (26 obs.), co-CEOs (17 obs.), and those without complete data (14 obs.), our sample contains 520 dividend cuts and 615 suspensions.11

3.3. Measures of CEO Entrenchment

We employ several variables to measure CEO entrenchment. Based on the theories discussed in Section 2.1, we take a broad view of entrenchment and use proxies that are likely to measure the ease of shareholder intervention.

Specifically, CEOs with longer tenure are more likely to be entrenched as they have had more time to install a favorable board of directors. In our regressions, we measure CEO tenure using the variable $ln\left(Tenure\right)$. The variable $NewHire$ indicates whether the CEO has been on the job for less than one year. CEOs who are also chairpersons of the board are likely to be harder to remove from office. The variable $Chairperson$ indicates if the CEO is also the chairperson of the board at the time of the dividend decrease. CEOs who are founders are likely to have more goodwill established with the company and be harder to replace. The variable $Founder$ indicates if the CEO is a founder or co-founder. CEOs who are family members of the firm’s controlling family are also less likely to be replaced. CEOs with high stock ownership are less likely to be voted out of office, and according to Benmelech et al. (2010), they have incentives to hide the truth about the firm’s earnings growth. The variable $Ownership$ measures the CEO’s proportional stock ownership. More sophisticated investors are likely to wield greater power and therefore have an easier time intervening or removing the CEO. The variable $Inst.Holdings$ measures the proportional stock ownership of institutional investors. We include all of these variables in our regressions to capture the various potential dimensions of CEO entrenchment.

3.4. Firm Level Data

We collect firm-level data from Compustat and CRSP. We also supplement this data by hand-collecting the CEOs’ stock ownership reported in the Proxy Statements (DEF 14A) filed with the SEC. Institutional ownership is obtained from Thompson 13-F filings. After removing observations with incomplete data, our samples comprise 426 dividend cuts and 469 dividend suspensions.

4. Results

4.1. Summary Statistics

We present summary statistics for the dividend cut sample in

Table 1. Summary statistics: dividend cuts.

	Obs.	Mean	Std	25%	50%	75%
CAR $(-1,+1)$	426	−0.038	0.093	−0.081	−0.024	0.009
CAR $(-3,+3)$	426	−0.044	0.116	−0.097	−0.032	0.020
CAR $(-5,+5)$	426	−0.042	0.137	−0.113	−0.035	0.035
Age	426	57.22	8.557	51.00	57.00	62.00
Tenure	426	7.472	8.549	2.000	4.000	10.00
Female	426	0.023	0.152	0.000	0.000	0.000
Chairperson	426	0.533	0.500	0.000	1.000	1.000
Founder	426	0.099	0.298	0.000	0.000	0.000
Family	426	0.120	0.325	0.000	0.000	0.000
Insider	426	0.500	0.501	0.000	0.500	1.000
New Hire	426	0.160	0.367	0.000	0.000	0.000
Ownership	426	0.049	0.132	0.001	0.005	0.020
Inst. Holdings	426	0.595	0.266	0.403	0.615	0.805
% Change in DPS	426	−56.97	18.69	−72.00	−50.00	−47.37
Dividend Yield	426	0.053	0.034	0.029	0.046	0.066
OCF	426	0.088	0.080	0.039	0.078	0.122
Size	426	2879	7888	180.7	555.3	2028
MTB	426	1.488	1.052	0.987	1.162	1.545
Leverage	426	0.273	0.193	0.119	0.270	0.403
Repurchases	426	0.017	0.031	0.000	0.001	0.019
Stock Return	426	−0.202	0.306	−0.430	−0.228	−0.019
DD	382	4.396	3.985	1.479	3.519	6.415

This table presents summary statistics of the firm and CEO variables in our sample of dividend cuts. Variables are described in the text and Appendix A. Non-binary variables are winsorized at the 1st and 99th percentiles.

. As reported, the average cumulative abnormal return (CAR) surrounding the announcement of the dividend cut is between −3.8% and −4.4%, depending on the window used.12 The CARs are computed using the market model with the CRSP value-weighted index as the market portfolio. On average, CEOs in the sample are 57 years old, have been in office for seven years, are female in 2.3% of the cases, serve as chairperson of the board in 53% of the cases, are founders in 10% of the cases, are family members of the controlling family in 12% of the cases, are inside hires (defined as being an employee for at least three years before becoming CEO) in 50% of the cases, are new hires (defined as being hired within one year of the dividend cut announcement) in 16% of the cases, and, on average, own 4.9% of the company’s outstanding stock.

Moving to the firm characteristics, on average, firms have 60% of their shares held by institutions, cut the dividend by 57%, and have an annual dividend yield before the cut of 5.3%, operating cash flows of 8.8% relative to total assets, a market value of equity of $2.879 billion, a market-to-book ratio of about 1.5, and financial leverage of 0.27, repurchased 1.7% of their stock relative to the market value of equity at the end of the prior fiscal year, a stock return of −20% over the prior fiscal year, and a distance to default measure of 4.4. These firm-level variables are measured at the end of the fiscal year before the announcement of the dividend cut.

The summary statistics for the dividend suspension sample are presented in

Table 2. Summary statistics: dividend suspensions.

	Obs.	Mean	Std	25%	50%	75%
CAR $(-1,+1)$	469	−0.060	0.147	−0.129	−0.039	0.013
CAR $(-3,+3)$	469	−0.072	0.180	−0.150	−0.061	0.018
CAR $(-5,+5)$	469	−0.069	0.199	−0.171	−0.059	0.024
Age	469	56.10	8.372	51.00	55.00	61.00
Tenure	469	7.060	8.647	1.000	4.000	10.00
Female	469	0.036	0.187	0.000	0.000	0.000
Chairperson	469	0.441	0.497	0.000	0.000	1.000
Founder	469	0.122	0.327	0.000	0.000	0.000
Family	469	0.104	0.306	0.000	0.000	0.000
Insider	469	0.412	0.493	0.000	0.000	1.000
New Hire	469	0.224	0.417	0.000	0.000	0.000
Ownership	469	0.053	0.112	0.001	0.006	0.030
Inst. Holdings	469	0.547	0.291	0.307	0.552	0.793
Dividend Yield	469	0.041	0.037	0.017	0.030	0.050
OCF	469	0.065	0.095	0.018	0.063	0.104
Size	469	877.7	2043	82.03	227.7	698.3
MTB	469	1.301	0.810	0.918	1.082	1.344
Leverage	469	0.310	0.226	0.140	0.307	0.450
Repurchases	469	0.014	0.030	0.000	0.000	0.012
Stock Return	469	−0.220	0.379	−0.483	−0.266	−0.035
DD	416	3.038	3.225	0.945	2.507	4.453

This table presents summary statistics of the firm and CEO variables in our sample of dividend suspensions. Variables are described in the text and Appendix A. Non-binary variables are winsorized at the 1st and 99th percentiles.

. In most aspects, the sample is similar to the dividend cut sample. The average CAR surrounding the dividend suspension announcement is lower than that in the dividend cut sample, ranging from −6.0% to −7.2%, depending on the window used.13 The average dividend yield is also slightly lower at 4.1% compared to 5.3% in the dividend cut sample. The suspending companies appear to have lower operating cash flows relative to total assets at 6.5% compared to 8.8% in the dividend cut sample and a lower market value of equity, at $0.877 billion, compared to $2.9 billion in the dividend cut sample. The average distance to default is also lower, 3.7 compared to 4.4 in the dividend cut sample. In general, the summary statistics of key variables are comparable to those used in other studies.

4.2. Market Reaction

In this section, we examine the relationship between CARs surrounding dividend cuts or suspensions and CEO entrenchment. We compute cumulative abnormal announcement returns relative to the market model over three-, seven-, and eleven-day windows surrounding the dividend cut or suspension announcement date. We then regress these returns on our various measures of CEO entrenchment, along with several firm-level control variables, including the ratio of market value of assets to book value, a proxy for Tobin’s Q (Lang & Litzenberger, 1989), firm size, pre-cut dividend yield (Denis et al., 1994; Yoon & Starks, 1995), and operating cash flow (Officer, 2011). The regressions also include industry dummy variables at the two-digit SIC code level and year of announcement fixed effects, and the standard errors are clustered by the year of the announcement.

The regression results for the dividend cut sample are reported in

Table 3. Announcement returns: dividend cuts.

	$\mathbf{CAR} \mathbf{(}\mathbf{-}\mathbf{1}\mathbf{,}\mathbf{+}\mathbf{1}\mathbf{)}$	$\mathbf{CAR} \mathbf{(}\mathbf{-}\mathbf{3}\mathbf{,}\mathbf{+}\mathbf{3}\mathbf{)}$	$\mathbf{CAR} \mathbf{(}\mathbf{-}\mathbf{5}\mathbf{,}\mathbf{+}\mathbf{5}\mathbf{)}$
ln(Tenure)	−0.003	−0.007	−0.014
	(−0.318)	(−0.680)	(−1.320)
Female	−0.031	−0.038	−0.012
	(−1.552)	(−1.420)	(−0.439)
Chairperson	−0.007	0.000	0.014
	(−0.745)	(0.019)	(0.609)
New Hire	−0.020	−0.025	−0.040
	(−1.182)	(−1.107)	(−1.256)
Founder	0.018	0.028	0.058 **
	(0.626)	(1.214)	(2.096)
Family	−0.011	−0.033	−0.026
	(−0.584)	(−1.396)	(−1.026)
Insider	−0.001	−0.023	−0.016
	(−0.090)	(−1.434)	(−0.949)
Ownership	−0.013	−0.066	−0.103
	(−0.305)	(−1.513)	(−1.073)
Inst. Holdings	−0.043 ***	−0.058 **	−0.065 **
	(−2.863)	(−2.727)	(−2.384)
% Change in DPS	0.000	0.000	0.000
	(0.076)	(1.108)	(0.931)
Dividend Yield	−0.187	−0.348	−0.596 ***
	(−0.969)	(−1.647)	(−3.372)
No. of Obs.	426	426	426
Adj. R-Squared	−0.005	0.002	0.036

This table presents regressions of cumulative abnormal returns (CARs) surrounding dividend announcements on CEO characteristics and control variables. The regressions control for the following firm-level variables reported in Table 1: $ln\left(Size\right)$, $OCF$, $MTB$, $Leverage$, $Repurchases$, and $StockReturn$. The regressions also include industry controls at the two-digit SIC code level and fixed effects for dividend announcement years. Standard errors robust to dividend announcement year clustering are reported in parentheses. The symbols *** and ** indicate statistical significance at the 1%, and 5% levels. All variables are described in the text and Appendix A. Non-binary variables are winsorized at the 1st and 99th percentiles.

. As reported, a positive relationship exists between eleven-day CARs surrounding dividend cuts and founder CEOs (statistically significant at the 5% level), and a negative relationship exists between CARs and the proportion of the firm’s shares held by institutions in every event window examined (statistically significant at the 5% level or less).

The positive coefficient on founder CEOs provides some support for the notion that CEO entrenchment is associated with a more positive market reaction to dividend cuts. However, we recognize that this relationship only exists in one of the three event-windows we examine. The consistently negative relationship between institutional holdings and CARs is consistent with the idea that the threat of shareholder intervention strengthens the signal in dividend cuts.14

Moving to the suspension sample in

Table 4. Announcement returns: dividend suspensions.

	$\mathbf{CAR} \mathbf{(}\mathbf{-}\mathbf{1}\mathbf{,}\mathbf{+}\mathbf{1}\mathbf{)}$	$\mathbf{CAR} \mathbf{(}\mathbf{-}\mathbf{3}\mathbf{,}\mathbf{+}\mathbf{3}\mathbf{)}$	$\mathbf{CAR} \mathbf{(}\mathbf{-}\mathbf{5}\mathbf{,}\mathbf{+}\mathbf{5}\mathbf{)}$
ln(Tenure)	−0.001	−0.012	−0.026
	(−0.076)	(−0.669)	(−1.058)
Female	−0.093 **	−0.092 *	−0.035
	(−2.326)	(−1.972)	(−0.479)
Chairperson	0.005	0.011	0.010
	(0.203)	(0.460)	(0.331)
New Hire	−0.008	−0.029	−0.041 *
	(−0.355)	(−1.048)	(−1.719)
Founder	−0.037	−0.030	−0.022
	(−1.186)	(−0.752)	(−0.518)
Family	0.003	−0.001	0.024
	(0.086)	(−0.033)	(0.895)
Insider	0.015	0.007	−0.013
	(0.802)	(0.283)	(−0.533)
Ownership	0.248 ***	0.246 *	0.229
	(3.154)	(1.999)	(1.620)
Inst. Holdings	0.016	0.031	0.033
	(0.377)	(0.637)	(0.573)
Dividend Yield	−0.579 *	−0.840 **	−0.963 **
	(−1.807)	(−2.597)	(−2.646)
No. of Obs.	469	469	469
Adj. R-Squared	0.027	0.054	0.045

This table presents regressions of cumulative abnormal returns (CARs) surrounding dividend announcements on CEO characteristics and control variables. The regressions control for the following firm-level variables reported in Table 2: $ln\left(Size\right)$, $OCF$, $MTB$, $Leverage$, $Repurchases$, and $StockReturn$. The regressions also include industry controls at the two-digit SIC code level and fixed effects for dividend announcement years. Standard errors robust to dividend announcement year clustering are reported in parentheses. The symbols ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels. All variables are described in the text and Appendix A. Non-binary variables are winsorized at the 1st and 99th percentiles.

, we find evidence of a negative relationship between eleven-day CARs and newly hired CEOs (statistically significant at the 10% level), and a positive relationship between three-day CARs and CEO stock ownership (statistically significant at the 1% level). Newly hired CEOs are less likely to be entrenched, whereas greater stock ownership indicates that the CEO is more likely to be entrenched. Thus, the regression results suggest that CEO entrenchment reduces the bad news in dividend suspensions. Hence, overall, the evidence in this section regarding dividend cuts and suspensions is consistent with Hypotheses 2 and 3: CEO entrenchment reduces the bad news in dividend decreases.

4.3. Changes in Profitability

In this section, we investigate whether CEO entrenchment is associated with changes in profitability surrounding dividend cuts. We follow the advice in the work of Ham et al. (2020) and run the regressions in event time. We form annual profitability variables by summing the quarterly earnings announcements relative to the date of the dividend announcement. For example, the variable $E_{y-1}$ is the sum of the four consecutive quarterly earnings announcements that occur before the dividend announcement date. The variables $\Delta E_{y+1}$, $\Delta E_{y+2}$, and $\Delta E_{y+3}$ are formed as the sum of four consecutive earnings announcements beginning one, five, and nine quarters after the dividend announcement date, respectively, minus $E_{y-1}$. All earnings measures are scaled by the market value of equity one year before the dividend announcement date.

We follow the specifications in the work of Ham et al. (2020) by regressing the change in the profitability measures on the CEO entrenchment variables, the four prior quarterly earnings and their quarter-over-quarter changes, and six nonlinear earnings measures: $E_{y-1}$ and $\Delta E_{y-1}$ multiplied by a negative indicator if the measure is negative, as well as their squares multiplied by a negative indicator if the measure is negative and a positive indicator if the measure is positive. We also include stock returns measured during the $(-2,-20)$, $(-21,-40)$, $(-41,-60)$, $(-61,-120)$, and $(-121,-240)$ trading day windows relative to the dividend announcement.15

We examine two profitability measures: changes in earnings and changes in gross profit. Ham et al. (2020) demonstrate that changes in gross profit are less likely to be affected by changes in capital expenditures and asset sales; thus, they are more likely to represent permanent changes in earnings. The results for the dividend cut sample are reported in

Table 5. Earnings: dividend cuts.

	$\mathbf{\Delta }{\mathit{E}}_{\mathit{y}\mathbf{+}\mathbf{1}}$	$\mathbf{\Delta }{\mathit{E}}_{\mathit{y}\mathbf{+}\mathbf{2}}$	$\mathbf{\Delta }{\mathit{E}}_{\mathit{y}\mathbf{+}\mathbf{3}}$
ln(Tenure)	−0.005	−0.014	−0.009
	(−0.527)	(−1.312)	(−0.801)
Female	0.000	−0.058	−0.048
	(0.000)	(−0.914)	(−0.567)
Chairperson	−0.008	−0.006	−0.002
	(−0.714)	(−0.387)	(−0.091)
New Hire	0.003	0.032	0.000
	(0.110)	(1.027)	(0.012)
Founder	−0.001	0.016	0.049
	(−0.054)	(0.481)	(1.256)
Family	−0.037	0.050	0.033
	(−1.319)	(1.432)	(1.235)
Insider	0.017	0.010	0.022
	(0.554)	(0.471)	(0.801)
Ownership	−0.026	0.006	0.019
	(−0.284)	(0.066)	(0.267)
Inst. Holdings	0.030	−0.004	−0.053
	(1.189)	(−0.085)	(−1.682)
% Change in DPS	0.000	0.000	0.001
	(1.113)	(1.248)	(1.088)
Dividend Yield	−0.273	−0.243	0.108
	(−1.447)	(−1.147)	(0.310)
No. of Obs.	433	415	396
Adj. R-Squared	0.495	0.562	0.507

This table presents regressions of changes in earnings following dividend cuts on CEO characteristics and controls. The dependent variables are of the form $\Delta E_{y+x}=E_{y+x}-E_{y-1}$, where $E_{y-1}$, $E_{y+1}$, $E_{y+2}$, and $E_{y+3}$ are formed by summing the four consecutive quarterly earnings announcements beginning in quarters minus four, one, five, and nine relative to the dividend announcement, respectively. The regression controls also include four prior quarterly earnings, their quarter-over-quarter changes, and six nonlinear earnings measures: $E_{y-1}$ and $\Delta E_{y-1}$ multiplied by a negative indicator if the measure is negative, their squares multiplied by a negative indicator if the measure is negative, and a positive indicator if the measure is positive. All earnings measures are scaled by the market value of equity one year before the dividend announcement. We also include five lagged stock returns measured over $(-2,-20)$, $(-21,-40)$, $(-41,-60)$, $(-61,-120)$, and $(-121,-240)$ trading day windows relative to the dividend announcement. The regressions also include industry controls at the two-digit SIC code level and fixed effects for dividend announcement years. Standard errors robust to dividend announcement year clustering are reported in parentheses. All other variables are described in the text and Appendix A. Non-binary variables are winsorized at the 1st and 99th percentiles.

and

Table 6. Gross profit: dividend cuts.

	$\mathbf{\Delta }{\mathit{GP}}_{\mathit{y}\mathbf{+}\mathbf{1}}$	$\mathbf{\Delta }{\mathit{GP}}_{\mathit{y}\mathbf{+}\mathbf{2}}$	$\mathbf{\Delta }{\mathit{GP}}_{\mathit{y}\mathbf{+}\mathbf{3}}$
ln(Tenure)	−0.007	−0.019	−0.010
	(−0.635)	(−1.142)	(−0.697)
Female	0.085	0.067	0.068
	(1.411)	(0.781)	(0.516)
Chairperson	0.025 **	0.032 *	0.029
	(2.384)	(1.830)	(1.464)
New Hire	−0.064 **	−0.083	−0.080
	(−2.413)	(−1.507)	(−1.485)
Founder	−0.031	−0.020	0.006
	(−1.161)	(−0.499)	(0.133)
Family	−0.043 *	0.018	−0.013
	(−1.776)	(0.395)	(−0.417)
Insider	−0.037 *	−0.042 *	−0.030
	(−1.964)	(−2.044)	(−1.235)
Ownership	−0.085	−0.125	−0.102
	(−1.393)	(−1.051)	(−0.882)
Inst. Holdings	0.006	−0.001	0.014
	(0.186)	(−0.037)	(0.440)
% Change in DPS	−0.000	0.001	−0.000
	(−0.013)	(1.501)	(−0.392)
Dividend Yield	0.408 *	0.515 *	0.281
	(2.016)	(1.781)	(0.710)
No. of Obs.	414	396	383
Adj. R-Squared	0.378	0.356	0.262

This table presents regressions of changes in gross profit following dividend cuts on CEO characteristics and controls. The models have the same specifications as those in Table 5, but gross profit is used as the measure of earnings. Standard errors robust to dividend announcement year clustering are reported in parentheses. The symbols ** and * indicate statistical significance at the 5% and 10% levels. All other variables are described in the text and Appendix A. Non-binary variables are winsorized at the 1st and 99th percentiles.

. The results for the suspension sample are reported in

Table 7. Future earnings: dividend suspensions.

	$\mathbf{\Delta }{\mathit{E}}_{\mathit{y}\mathbf{+}\mathbf{1}}$	$\mathbf{\Delta }{\mathit{E}}_{\mathit{y}\mathbf{+}\mathbf{2}}$	$\mathbf{\Delta }{\mathit{E}}_{\mathit{y}\mathbf{+}\mathbf{3}}$
ln(Tenure)	−0.010	0.024	0.012
	(−0.429)	(1.194)	(0.644)
Female	−0.040	0.039	0.093
	(−0.719)	(0.785)	(0.611)
Chairperson	−0.043	−0.019	0.010
	(−1.254)	(−0.647)	(0.374)
New Hire	0.019	0.025	−0.005
	(0.486)	(0.462)	(−0.126)
Founder	0.053	0.035	−0.112 **
	(0.795)	(0.768)	(−2.106)
Family	0.002	0.063	−0.030
	(0.028)	(1.104)	(−0.891)
Insider	0.035	0.004	−0.024
	(1.260)	(0.106)	(−0.732)
Ownership	−0.158	0.022	0.223
	(−1.120)	(0.218)	(1.639)
Inst. Holdings	−0.028	0.014	0.031
	(−0.464)	(0.346)	(0.713)
Dividend Yield	−1.538 **	−0.765	0.350
	(−2.486)	(−1.024)	(0.608)
No. of Obs.	469	440	410
Adj. R-Squared	0.442	0.523	0.565

This table presents regressions of changes in earnings following dividend announcement on CEO characteristics and controls. The dependent variables are of the form $\Delta E_{y+x}=E_{y+x}-E_{y-1}$, where $E_{y-1}$, $E_{y+1}$, $E_{y+2}$, and $E_{y+3}$ are formed by summing the four consecutive quarterly earnings announcements beginning in quarters minus four, one, five, and nine relative to the dividend announcement, respectively. The regression controls also include four prior quarterly earnings, their quarter-over-quarter changes, and six nonlinear earnings measures: $E_{y-1}$ and $\Delta E_{y-1}$ multiplied by a negative indicator if the measure is negative, their squares multiplied by a negative indicator if the measure is negative, and a positive indicator if the measure is positive. All earnings measures are scaled by the market value of equity one year before the dividend announcement. We also include five lagged stock returns measured over $(-2,-20)$, $(-21,-40)$, $(-41,-60)$, $(-61,-120)$, and $(-121,-240)$ trading day windows relative to the dividend announcement. The regressions also include industry controls at the two-digit SIC code level and fixed effects for dividend announcement years. Standard errors robust to dividend announcement year clustering are reported in parentheses. The symbol ** indicates statistical significance at the 5% level. All other variables are described in the text and Appendix A. Non-binary variables are winsorized at the 1st and 99th percentiles.

and

Table 8. Future gross profit: dividend suspensions.

	$\mathbf{\Delta }{\mathit{GP}}_{\mathit{y}\mathbf{+}\mathbf{1}}$	$\mathbf{\Delta }{\mathit{GP}}_{\mathit{y}\mathbf{+}\mathbf{2}}$	$\mathbf{\Delta }{\mathit{GP}}_{\mathit{y}\mathbf{+}\mathbf{3}}$
ln(Tenure)	−0.016	0.008	−0.017
	(−0.680)	(0.343)	(−0.608)
Female	−0.023	−0.077	−0.035
	(−0.698)	(−1.557)	(−0.651)
Chairperson	0.008	0.009	0.028
	(0.331)	(0.292)	(0.776)
New Hire	−0.028	−0.018	−0.064
	(−1.077)	(−0.478)	(−1.082)
Founder	0.056	0.076	0.078
	(1.374)	(1.597)	(1.281)
Family	0.011	−0.039	−0.048
	(0.236)	(−0.664)	(−0.741)
Insider	−0.005	−0.020	−0.008
	(−0.306)	(−0.631)	(−0.233)
Ownership	−0.196	−0.439 **	−0.124
	(−1.007)	(−2.071)	(−0.645)
Inst. Holdings	−0.032	0.012	−0.011
	(−0.771)	(0.289)	(−0.209)
Dividend Yield	−0.144	0.167	−0.488
	(−0.512)	(0.536)	(−0.998)
No. of Obs.	438	411	386
Adj. R-Squared	0.344	0.352	0.308

This table presents regressions of changes in gross profit following dividend suspensions on CEO characteristics and controls. The models have the same specifications as those in Table 7, but gross profit is used as the measure of earnings. Standard errors robust to dividend announcement year clustering are reported in parentheses. The symbol ** indicates statistical significance at the 5% level. All other variables are described in the text and Appendix A. Non-binary variables are winsorized at the 1st and 99th percentiles.

.

As reported in Table 5 and Table 6, we do not find evidence of a strong relationship between long-term earnings measures following dividend cuts and measures of CEO entrenchment. This is consistent with the conjecture and evidence of Wu (2018) that dividend cuts signal a continuation of low current earnings. However, when examining changes in gross profit, the positive coefficient on $Chairperson$ and the negative coefficient on $NewHire$ (both statistically significant at the 5% level) suggest that dividend cuts provide a weaker signal of short-term earnings when the CEO is potentially entrenched. However, there are also negative coefficients on $Family$ and $Insider$ (both statistically significant at the 10% level) that suggest the signal in earnings is stronger when the CEO is entrenched. The evidence is therefore mixed, supporting all three hypotheses that we examine, depending on the proxy of CEO entrenchment. However, the evidence supporting the notion that entrenchment strengthens the signal in dividend cuts about earnings is only statistically significant at the 10% level.

Turning to the suspension sample results in Table 7 and Table 8, the negative coefficient on $Founder$ (statistically significant at the 5% level) in Table 7 indicates that CEO entrenchment strengthens the signal in dividend suspensions about earnings. When examining changes in gross profit in Table 8, the negative coefficient on $Ownership$ (statistically significant at the 5% level) also suggests that CEO entrenchment strengthens the signal in dividend suspensions related to earnings.

The evidence in this section does not conclusively support any of the hypotheses we examine. In the case of dividend cuts, some proxies of CEO entrenchment suggest that dividend cuts provide a weaker signal of earnings, while others support the opposite conclusion. The evidence surrounding dividend suspensions supports the notion that CEO entrenchment strengthens the signal about earnings in dividend suspensions, which is consistent with entrenched CEOs being excessively optimistic (Hypothesis 3).

4.4. CEO Turnover

In this section, we examine whether entrenched CEOs are less likely to be replaced following dividend cuts and suspensions. Finding that they are indicates that shareholders have difficulty intervening to replace entrenched CEOs. About 38% of CEOs in our sample are replaced after a dividend cut. This figure increases to 47% following dividend suspensions.

In

Table 9. CEO turnover.

	Dividend Cuts	Dividend Suspensions
ln(Tenure)	0.292	0.072
	(1.310)	(0.419)
Female	1.304	0.196
	(1.363)	(0.283)
Chairperson	0.545 *	0.191
	(1.839)	(0.741)
New Hire	−0.572	0.128
	(−1.309)	(0.264)
Founder	0.012	−0.878 **
	(0.020)	(−2.051)
Family	−0.583	−0.738 **
	(−1.101)	(−2.104)
Insider	−0.580 **	−0.288
	(−2.288)	(−1.381)
Ownership	−1.257	−0.184
	(−1.273)	(−0.122)
Inst. Holdings	−0.480	0.136
	(−1.028)	(0.248)
% Change in DPS	−0.003
	(−0.289)
Dividend Yield	0.340	−4.654
	(0.113)	(−1.161)
No. of Obs.	395	452
Pseudo R-Squared	0.138	0.105

This table presents logistic regressions of indicator variables equal to 1 if the firm filed for bankruptcy, was acquired, or either within three years following the dividend announcement on CEO characteristics and controls ($ln\left(Size\right)$, $OCF$, $MTB$, $Leverage$, $Repurchases$, $StockReturn$). The regressions also include industry effects at the two-digit SIC code level. Standard errors robust to dividend announcement year clustering are reported in parentheses. The symbols **, and * indicate statistical significance at the 5%, and 10% levels. All variables are described in the text and Appendix A. Non-binary variables are winsorized at the 1st and 99th percentiles.

, we run logistic regressions of an indicator variable equal to one if the CEO is replaced within three years following a dividend cut or suspension on our measures of CEO entrenchment and control variables. These regressions do not include industry or year of dividend announcement controls because there is little variation in the dependent variable, which then requires us to drop several observations. As shown, following dividend cuts, CEOs are less likely to be replaced if they are an insider (statistically significant at the 5% level) and are more likely to be replaced if they are chairperson of the board (statistically significant at the 10% level). Turning to the suspension sample, founder CEOs and those who are related to the controlling family are less likely to be replaced following dividend suspensions. Most of the evidence from these analyses suggests that entrenched CEOs are harder to replace following dividend decreases.

4.5. Bankruptcy or Acquisition

In this section, we examine whether firms with an entrenched CEO that cut or suspend dividends are more likely to experience an acquisition or bankruptcy in subsequent years. We create three separate indicator variables equal to one if the firm files for bankruptcy, is acquired, or experiences either of these events in the three years after a dividend cut or suspension. Approximately 2.5% of firms file for bankruptcy, while 7.2% of firms are acquired in the dividend cut sample. These figures are 7.1% and 7.2% for the sample of suspensions. We then regress these indicator variables on measures of CEO entrenchment and control variables. These regressions do not include industry or year of dividend announcement controls because there is little variation in the dependent variables, which then requires us to drop most of the observations. The standard errors are clustered by the year of the dividend announcement.

As reported in

Table 10. Bankruptcy or acquisition: dividend cuts.

	Acquisition	Bankruptcy	Either
ln(Tenure)	−0.643 **	−0.587	−0.535 **
	(−2.192)	(−1.125)	(−2.259)
Female	2.303 **		1.698 *
	(2.086)		(1.671)
Chairperson	0.816	−0.184	0.662
	(1.188)	(−0.246)	(1.316)
New Hire	−0.425	0.586	−0.203
	(−0.960)	(0.411)	(−0.410)
Founder	0.169	2.677 **	0.819
	(0.256)	(2.089)	(1.508)
Family	0.665		0.091
	(1.215)		(0.125)
Insider	−0.028	−1.022	−0.324
	(−0.044)	(−1.618)	(−0.540)
Ownership	1.521	9.512 *	1.694
	(0.935)	(1.941)	(0.888)
Inst. Holdings	−0.993	−3.277	−1.306
	(−1.160)	(−1.633)	(−1.601)
% Change in DPS	−0.018	−0.057 ***	−0.026 *
	(−1.024)	(−3.287)	(−1.659)
Dividend Yield	−12.888	8.749	−4.553
	(−1.395)	(0.886)	(−0.575)
No. of Obs.	382	329	382
Pseudo R-Squared	0.122	0.477	0.128

This table presents logistic regressions of indicator variables equal to 1 if the firm filed for bankruptcy, was acquired, or either within three years following the dividend announcement on CEO characteristics and controls ($ln\left(Size\right)$, $OCF$, $MTB$, $Leverage$, $Repurchases$, $StockReturn$, and $DistancetoDefault$). Standard errors robust to dividend announcement year clustering are reported in parentheses. The symbols ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels. All variables are described in the text and Appendix A. Non-binary variables are winsorized at the 1st and 99th percentiles.

, the entrenchment variables are not statistically significant in the acquisition regressions; however, founder CEOs and potentially those with greater stock ownership are more likely to file for bankruptcy following a dividend cut. These relationships suggest that the long-run prospects of firms with an entrenched CEO are worse than those without an entrenched CEO.16

Moving to the dividend suspension sample in

Table 11. Bankruptcy or acquisition: dividend suspensions.

	Acquisition	Bankruptcy	Either
ln(Tenure)	−0.129	−0.176	−0.195
	(−0.558)	(−0.392)	(−0.685)
Female	1.714 ***	0.530	1.179 *
	(2.667)	(0.562)	(1.914)
Chairperson	0.156	0.292	0.282
	(0.397)	(0.640)	(0.955)
New Hire	−0.510	−0.289	−0.465
	(−0.859)	(−0.292)	(−0.793)
Founder	0.953	−1.531 **	−0.648
	(1.077)	(−2.280)	(−1.110)
Family	−0.651	−0.680	−0.658
	(−0.867)	(−1.240)	(−1.587)
Insider	−0.545	−1.272 **	−0.967 **
	(−1.539)	(−2.085)	(−2.272)
Ownership	−6.745 **	2.194	0.125
	(−2.520)	(1.373)	(0.079)
Inst. Holdings	−0.110	−1.408 *	−0.855
	(−0.137)	(−1.657)	(−1.304)
Dividend Yield	−17.291	−3.038	−7.562
	(−1.080)	(−0.384)	(−1.063)
No. of Obs.	416	416	416
Pseudo R-Squared	0.135	0.115	0.086

This table presents logistic regressions of indicator variables equal to 1 if the firm filed for bankruptcy, was acquired, or either within three years following the dividend announcement on CEO characteristics and controls ($ln\left(Size\right)$, $OCF$, $MTB$, $Leverage$, $Repurchases$, $StockReturn$, and $DistancetoDefault$). Standard errors robust to dividend announcement year clustering are reported in parentheses. The symbols ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels. All variables are described in the text and Appendix A. Non-binary variables are winsorized at the 1st and 99th percentiles.

, we find evidence that CEO stock ownership is negatively correlated with an acquisition. We also find evidence that founder CEOs, inside-hire CEOs, and potentially firms with greater institutional ownership are less likely to file for bankruptcy following a dividend suspension. In general, these relationships indicate that it is more difficult for shareholders to take collective action following dividend suspensions when the CEO is entrenched and that the long-run prospects of dividend-suspending firms are better if the CEO is entrenched.

On its own, the evidence in this section does not conclusively support any of our hypotheses. Additionally, the conflicting bankruptcy results in this section suggest that dividend suspensions, relative to dividend cuts, provide firms with an entrenched CEO more resources to hold off creditors compared to firms without an entrenched CEO.

5. Conclusions

The results from our analyses are summarized in

Table 12. Summary of results.

	CARs		Earnings		Gross Profit
	Cut	Suspension	Cut	Suspension	Cut	Suspension
ln(Tenure)
Chairperson					+ **
New Hire		− *			− **
Founder	+ **			− **
Family					− *
Insider					− *
Ownership		+ ***				− **
Inst. Holdings	− ***
	CEO Turnover		Acquisition		Bankruptcy
	Cut	Suspension	Cut	Suspension	Cut	Suspension
ln(Tenure)			− **
Chairperson	+ *
New Hire
Founder		− **			+ **	− **
Family		− **
Insider	− **					− **
Ownership				− **	+ *
Inst. Holdings						− *

This table presents a summary of the statistically significant relationships in Table 3, Table 4, Table 5, Table 6, Table 7, Table 8, Table 9, Table 10 and Table 11. We report whether the sign of the relationship is positive (+) or negative (−), and the symbols ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively.

. Focusing on results that are statistically significant at the 5% level or less, the stock market evidence (CARs) in the dividend cut sample is consistent with the notion that CEO entrenchment reduces the bad news associated with dividend decreases. However, the other dividend cut regressions tell a more nuanced story; CEO entrenchment results in a weaker signal of gross profit (in the short run), a lower likelihood of CEO turnover and acquisition, but a higher probability of bankruptcy in the long run.

The market reaction and bankruptcy results appear to be at odds. Why would investors react more favorably to a dividend cut by founder CEOs if, in the long run, the firm is more likely to file for bankruptcy? One potential explanation is that founder CEOs are better at hiding the truth in the short run, which eventually manifests in the long term. This is consistent with evidence of founder CEOs providing excessively optimistic disclosures (Hendricks et al., 2022).

Turning to the sample of dividend suspensions, the stock market evidence (CARs) is consistent with the notion that CEO entrenchment reduces the bad news associated with dividend suspensions. However, dividend suspensions signal that future long-run earnings and gross profits of firms with an entrenched CEO are worse than those of firms without an entrenched CEO. This combination is consistent with entrenched CEOs being excessively optimistic. However, in contrast to the dividend cut sample, bankruptcy is less likely for firms with a founder CEO. This potentially indicates that, relative to dividend cuts, suspensions provide additional cash flows that enable entrenched CEOs to stave off interventions by creditors.

Finally, we leave for future work the examination of why certain of our proxies are related to CARs and other firm outcomes while others are not, or whether we document causal relationships. Finally, we focus on industrial firms in this paper, but future work may wish to examine these relationships in the financial sector. On the one hand, payout policy is more heavily regulated in that sector, which is likely to affect the information in dividend decreases. But, on the other hand, there is less cross-sectional variation in the products produced by financial firms, which may reduce the impact of unobserved variables in such an analysis.