Corporate social responsibility and the wealth gains from dividend increases

Published date01 April 2018
Date01 April 2018
DOIhttp://doi.org/10.1016/j.rfe.2017.07.002
ORIGINAL ARTICLE
Corporate social responsibility and the wealth gains from
dividend increases
Charmaine Glegg
|
Oneil Harris
|
Thanh Ngo
East Carolina University, Department of
Finance, Greenville, NC 27858, United
States
Correspondence
Thanh Ngo, East Carolina University,
Department of Finance, Greenville, NC
27858, United States.
Email: Gleggc@ecu.edu, Harriso@ecu.edu,
Ngot@ecu.edu
Abstract
This study examines whether corporate social responsibility (CSR) influences the
stock price response to dividend increase announcements and changes in subse-
quent operating performance. We find that dividend increasing firms with lower
CSR scores elicit higher abnormal announcement returns and greater improve-
ments in industry-adjusted operating performance. These findings support the
argument in the literature that socially responsible firms are more transparent and
commit to higher ethical standards than other firms, suggesting that they suffer
fewer agency and informational problems (Kim, Park, & Wier, 2012). Conse-
quently, larger dividend payouts reduce agency costs in firms with lower CSR
commitments, thereby generating higher wealth gains for shareholders.
1
|
INTRODUCTION
Over the past few years, corporate social responsibility (hereafter CSR) has becom e a notable topic in financial research
owing to the drastic increase in the amount of resources that corporations have allocated to socially responsible initiatives
(Deng, Kang, & Low, 2013; Kim, Li, & Li, 2014). El Ghoul, Guedhami, Kwok, and Mishra (2011) note that more than
half of the Fortune 1000 companies in the US regularly issue CSR reports.
1
Bernea and Rubin (2010) underscore that
CSR-related expenses account for a substantial component of the operations of many US companies. As an example, they
emphasize that the annual CSR expenditure at General Electric accounts for about 15% of the firms profits.
Yet, the question of whether firm-level investments in socially responsible activities represent a misallocation of produc-
tive resources is still an open and debatable issue. Under stakeholder theory, CSR investments have a positive impact on
firm value because focusing on the interests of other stakeholders increases their willingness to support the firm s operation,
which increases shareholderswealth (Deng et al., 2013). In line with this view, Jiao (2010) and Benson and Davidson
(2010) suggest that more socially responsible firms exhibit higher values.
Bae, El Ghoul, Guedhami, Kwok, and Zheng (2017) find that CSR mitigates unfavorable actions by customers and com-
petitors among highly-leveraged firms, thereby reducing the cost of leverage. Consequently, CSR engagement may bring
the interests of shareholders and other stakeholders into greater alignment (see Freeman, Wicks, & Parmar, 2004).
Evidence of value-enhancing effect of CSR investment extends beyond domestic U.S. firms. Boubakri, El Ghoul, Wang,
Guedhami, and Kwok (2016) report that firms cross-listed in the U.S. with more CSR activities have higher values and
their non-cross-listed peers. Home country characteristics also affect the value form enjoy from CSR investment (see Bou-
bakri et al., 2016; El Ghoul, Guedhami, & Kim, 2017). For instance, El Ghoul et al. (2017) examine 2445 firms from 53
countries and find evidence that CSR is associated with (i) improved access to financing in countries with weaker equity
and credit markets, (ii) greater investment and lower default risk in countries with more limited business freedom, and (iii)
longer trade credit period and higher future sales growth in countries with weaker legal institutions.
First published online by Elsevier on behalf of The University of New Orleans, 13 July, 2017, https://doi.org/10.1016/j.rfe.2017.07.002
Received: 11 January 2017
|
Revised: 4 July 2017
|
Accepted: 11 July 2017
DOI: 10.1016/j.rfe.2017.07.002
Rev Financ Econ. 2018;36:149166. wileyonlinelibrary.com/journal/rfe ©2017 The University of New Orleans
|
149
Then again, Bernea and Rubin (2010) contend that CSR spending is a manifestation of agency costs. They argue that
managers overinvest in CSR activities to burnish their reputations as responsible industry leaders at the expense of share-
holders. Some managers also allocate resources to CSR projects in order to cover up corporate misconduct (Hemingway &
Maclagan, 2004) and to reinforce an entrenchment strategy (Surroca & Tribo, 2008). In accord, Br ammer, Brooks, and
Pavelin (2006) find that firms with higher CSR ratings realize lower returns. Nelling and Webb (2009) also find little evi-
dence that CSR spending improves performance. Moreover, Petrovits (2006) reports that firms use their charitable founda-
tions to manipulate their reported earnings, which is consistent with the agency view.
In a recent study, Benlemlih (2014) purports that socially responsible firms may use dividend policy to manage the
agency problems related to overinvestment in CSR activities. He finds that high-CSR firms tend to pay more regular cash
dividends than low-CSR firms, and that dividend payout is more stable in high-CSR firms. Rakotomavo (2012) examines
whether investments in social responsibility take away from expected dividends. He finds that CSR spending does not sub-
tract from dividends because CSR investments tend to be made by mature firms that can afford it. Firms investing highly
in CSR activities are typically larger, more profitable, and have more earned (rather than contributed) equity, which suggest
that established firms tend to invest more in CSR initiatives (Rakotomavo, 2012).
Our study contributes to this scant, but interesting literature, by relating CSR activities to the wealth gains to sharehold-
ers from dividend increases. Thus, while earlier studies relate CSR investments to the amount of dividends that firms pay
out, we take a different approach by testing whether CSR investments affect the abnormal announcement returns around
dividend increases and subsequent changes in operating performance. To the best of our knowledge, no research has, to
date, examined the information content of dividends in conjunction with CSR engagement.
Prior research suggests that socially responsible firms commit to greater transparency and higher ethical standards than
other firms, implying that they suffer fewer informational and agency related problems. For instance, Kim, Park, and Wier
(2012) argue that firms with higher CSR activities provide more transparent and reliable financial information to investors.
Likewise, Gelb and Strawser (2001) find that firms with more CSR investments provide more financial disclosure. Bec-
chetti, Ciciretti, and Giovannelli (2013) also report that higher CSR quality contributes to maki ng earnings forecasts unbi-
ased. In addition, socially responsible firms engage in less bad news hoarding, which reduces price crash risk (Kim et al.,
2014). Therefore, it comes as no surprise that firms with lower CSR have higher equity and debt financing costs because
investors perceive them as having higher levels of risk (El Ghoul et al., 2011; Goss & Roberts , 2011).
Agency theory suggests that firms are more likely to experience higher agency costs when the quality and quantity of
financial disclosure is poor. In the agency framework, a higher dividend payout mitigates agency costs because it reduces
the discretionary cash flows available to managers (Easterbrook, 1984; Jensen, 1986). Consequently, we posit that dividend
increases by high-CSR and low-CSR firms elicit disproportionate wealth effects. We anticipate findi ng higher abnormal
announcement period returns and greater subsequent improvements in operating performance for dividend increasing firms
with lower CSR engagement owing to the reduction in agency costs.
On the other hand, Bernea and Rubin (2010) suggest that managers overinvest in CSR activities for their own private
benefit, rather than to maximize shareholder wealth. Cheng, Hong, and Shue (2014) echo this sentiment; they find strong
support for the agency motive for CSR. Hemingway and Maclagan (2004) underscore that managers may use CSR activi-
ties to cover up corporate misconduct. They spotlight Enron Corporation (and a few other firms) as anecdotal evidence that
managers use CSR to conceal their unethical behavior.
2
By the same token, Petrovits (2006) finds that managers use corpo-
rate foundations to manipulate reported earnings. There is also empirical evidence suggesting that CSR investments are
made in part to enhance managerspolitical connections (Borghesi, Houston, & Naranjo, 2014; Lin, Tan, Zhao, & Karim,
2015).
Hence, under the overinvestment hypothesis, higher CSR investments represent wasteful spending, implying that firms
with higher CSR scores have lower values. Consistent with the agency motive for CSR, some studies find that increases in
CSR ratings are associated with negative stock returns and declines in operating performance (Brammer et al., 2006; Giuli
& Kostovetsky, 2014). Given this backdrop, a dividend increase may reduce investorsestimates of the amount of resources
that will be unproductively allocated to CSR activities, thus increasing firm value. Accordingly, the overinvest ment hypoth-
esis suggests that high-CSR firms may elicit more pronounced stock price responses to dividend increase announcements
and exhibit greater improvements in operating performance than low-CSR firms.
Using a sample of 2800 dividend increases from 1995 to 2012, we document a strong n egative relationship between
CSR engagement and the announcement returns around dividend increases. The results show that on average, the cumula-
tive abnormal returns (CAR) to low- CSR dividend increasing firms (CSR score
CAR to high-CSR dividend increasing firms (CSR score median) are only about 0.5%. Consequently, low-CSR dividend
increasing firms earn nearly twice as much abnormal announcement period returns as high-CSR dividend increasing firms.
150
|
GLEGG ET AL.
This inverse relation continues to hold even after we control for factors known to affect the price reaction to announcements
of dividend increases. The marginal effect of CSR ratings on CAR is approximately 0.038, which is highly significant. In
economic terms, this suggests that all things being equal, a higher CSR score reduces the wealth gains to shareholders of the
median dividend increasing firm in our sample by roughly $127 million.
3
As a result, our results have important implications
for investors. This strong negative correlation between the CAR around dividend increase announcements and CSR engage-
ments withstands various robustness checks, including the Heckman (1979) two-step self-selectivity method.
In line with our expectation, we also find a significantly negative and economically meaningful relation between CSR
investments and the change in industry-adjusted operating performance following dividend increases. Specifically, dividend
increasing firms with lower CSR activities experience greater subsequent improvements in profitability. In fact, our univari-
ate tests show that the mean change in industry-adjusted operating performance is significantly positive only for dividend
increasing firms with low CSR scores, indicating that the improvements in profitability is concentrated in low-CSR divi-
dend increases. The mean change in industry-adjusted performance is significantly negative among high-CSR dividend
increasing firms, suggesting that they suffer a subsequent decline in operating performance.
4
Overall, these results are consistent with the view that more socially responsible firms have fewer agency and informa-
tional problems; they show that dividend increases mitigate the agency conflicts between managers and shareholders in
socially irresponsible firms. As a result, our study has important implications for shareholders. By documenting how CSR
engagement influences shareholderswealth gains from dividend increases, our paper makes notewort hy contributions to
both the CSR and the dividend payout literature. First, our study adds to the growing literature on CSR and its economic
consequences; as such, our findings do not support the agency view of CSR. Second, we extend prior research on the valu-
ation effects of dividend changes by showing that dividend increases by high-CSR and low-CSR firm s relay different infor-
mation to the market about future profitability, thereby eliciting disparate wealth effects.
Although our paper focuses on dividend increases, we also examine a sample of 938 dividend decreases during the sam-
ple period. However, we find no evidence that CSR activities influence either the CAR around dividend decreases or the
ensuing change in industry-adjusted operating performance. So for brevity, we do not formally report these results in our
tables, but they can be made available upon request. Fracassi (2008) argues that the rationale and underlying dynamics of
dividend increases are distinct from those of dividend decreases; these distinctions perhaps explai n the contrasting results.
The remainder of this paper is organized as follows. We discuss the relevant literature and develop testable hypotheses
in Section 2. We then describe our sample selection procedure in Section 3 and discuss the measurement of our main vari-
ables in Section 4. We present our univariate and multivariate empirical findings and discuss their implications in Section 5
and Section 6, respectively. Finally, we offer our closing remarks in Section 7.
2
|
BACKGROUND AND HYPOTHESIS DEVELOPMENT
It is well established that announcements of dividend increases (decreases) elicit positive (negative) stock price returns
(Benartzi, Michaely, & Thaler, 1997; Grullon, Michaely, & Swaminathan, 2002; Nissim & Ziv, 2001). However, the expla-
nation for the price reactions is subject to some debate.
5
Notwithstanding, there is a number of reasons why the degree of
CSR engagement may cause variations in the valuation effects from dividend changes, and the competing views of CSR in
the literature yield different predictions. In this section, we outline these different theoretical argum ents and develop several
testable hypotheses.
2.1
|
The ethical view of CSR
A notable strand of the CSR literature suggests that socially responsible firms commit to higher ethical and financial report-
ing standards than other firms, indicating that they have less informational asymmetries. For example, Jones (1995) purports
that CSR firms have an incentive to be truthful and ethical because such behavior is beneficial to the firm. Kim et al.
(2012) find that firms with higher CSR engage in less earnings management, thereby delivering more transparent and reli-
able financial information to investors. Socially responsible firms also provide more financial disclosure (Gelb & Strawser,
2001) and engage in less bad news hoarding to reduce price crash risk (Kim et al., 2014). Addi tionally, Becchetti et al.
(2013) find that higher CSR activi ties play a role in making earnings forecasts unbiased.
Collectively, these findings suggest that CSR reduces information risk. Hence, it comes as no surprise that firms with
higher CSR enjoy lower debt and equity financing costs because they are perceived as having lower risk (El Ghoul et al.,
2011; Goss & Roberts, 2011). Since a lower cost of capital enhances firm value, this view of CSR complements
GLEGG ET AL.
|
151
stakeholder theory, which implies that CSR is a means to create wealth for shareholders (Benson & Davidson, 2010; Deng
et al., 2013; Jiao, 2010).
If socially responsible firms commit to greater transparency and higher ethical standards, they may suffe r fewer agency
conflicts. The rationale is that managers have more discretionary power in opaque information environments, which
increases agency costs. In the agency theory framework, higher dividend payouts lower agency costs by reducing the free
cash flows available to managers (Easterbrook, 1984; Jensen, 1986). Therefore, agency theory predicts that dividend
increasing firms with lower CSR activities elicit higher abnormal announcement returns and greater improvements in oper-
ating performance due to a reduction in agency costs.
H1a. Low-CSR firms have larger abnormal returns surrounding dividend increase announcements than high-
CSR firms.
H2a. Low-CSR firms experience greater improvements in operating performance subsequent to increasing div-
idends than high-CSR firms.
2.2
|
The overinvestment view of CSR
Alternatively, Bernea and Rubin (2010) argue that CSR spending is a manifestation of agency costs; they argue that man-
agers overinvest in CSR activities to burnish their own reputations as responsible industry leaders at shareholdersexpense.
Borghesi et al. (2014) and Lin et al. (2015) suggest that managers engage in CSR in part to enhance political connections,
while Surroca and Tribo (2008) argue that CSR investments serve to further entrench managers. Cheng, Ioannou, and Ser-
afeim (2014) also find evidence supporting this agency motive for CSR. Petrovits (2006) cast doubt on the ethical view of
CSR by showing that managers use charitable contributions to manipulate corporate earnings, thereby misleading investors.
In addition, Hemingway and Maclagan (2004) suggest that managers invest in CSR to cover up unethical behavior.
6
Brammer et al. (2006) and Giuli and Kostovetsky (2014) report that increases in CSR ratings are associated with nega-
tive stock returns and declines in operating performance, which supports the agency motive for CSR.
7
Therefore, higher
CSR spending may represent perquisite consumption that lowers firm value. In this case, a dividend increase may reduce
investorsestimate of the amount of resources that will be wastefully invested to CSR activities, thereby increasing firm
value.
8
For this reason, the overinvestment hypothesis predicts that dividend increasing firms wi th higher CSR activities
may elicit higher abnormal announcement returns and greater subsequent improvements in operating performance.
H1b. High-CSR firms have larger abnormal returns surrounding dividend increase announcements than low-
CSR firms.
H2b. High-CSR firms experience greater improvements in operating performance subsequent to increasing
dividends than low-CSR firms.
3
|
DATA DESCRIPTION
We obtain quarterly dividend increases from the Center for Research in Security Prices (CRSP) database from 1995 to 2012. As
in earlier studies, we restrict our sample to firms with share codes of 10 or 11 that pay an ordinary quarterly cash dividend (U.S.
dollars) in the current quarter and in the previous quarter (Nissim & Ziv, 2001).
9
Each announcement also satisfies the following
criteria. First, the percent change in dividends is between 12.5% and 500%. The 12.5% lower bound ensures that only economi-
cally meaningful dividend changes are included, while the 500% upper bound eliminates outliers (Grullonet al., 2002).
Second, no other distributions or corporate events are announced between the declaration of the previous dividend and
15 days after the declaration of the current dividend (Denis et al., 1994; Grullon, Michaely, Benartzi, & Thaler, 2005; Lie
& Li, 2006; Nissim & Ziv, 2001). Third, there are no ex-distribution dates between the ex-distribution dates of the previous
and the current dividends (Nissim & Ziv, 2001).
We match the resulting sample to the Compustat quarterly files and exclude firms with standard industrial classification
(SIC) code 60006999. Nissim and Ziv (2001) also eliminate firms in the financial sector. Firms are further required to have
152
|
GLEGG ET AL.
data available in MSCI ESG so that we can compute CSR scores.
10
The MSCI ESG dataset provides a broad spectrum of
socially responsible indicators for the 3000 largest U.S. firms (by market capitalization) and is widely used in the literature to
gauge CSR activities (see Bernea & Rubin, 2010; Deng et al., 2013; El Ghoul et al., 2011; Jha & Cox, 2015; Jiraporn, Jira-
porn, Boeprasert, & Chang, 2014; Kim et al., 2012, 2014; Ng & Rezaee, 2015; Oikonomou, Brooks, & Pavelin, 2014).
This sample selection process identifies 2800 dividend increases and 938 dividend decreases, which totals 3738 events
over the sample period. These sub-samples are consistent with earlier studies reporting that events of dividend increases are
more frequent than events of dividend decreases. However, our analyses indicate that CSR has no noteworthy effect on
either the announcement returns or the subsequent profitability of dividend decreasing firms.
11
So for brevity, we hence-
forth focus our attention on the 2800 dividend increasing events.
We report sample summary statistics in Table 1. Panel A shows the sample distribution by year. The results indicate
that there is a substantial upturn in the number of dividend increasing firms starting in 2003, which coincides with the
Bush tax cutsthat lowered the tax rates on dividends and capital gains and the expanded coverage of firms in the KLD
database that year. By contrast, there is a noticeable fall in the number of firms increasing dividends in 2008 and 2009
owing presumably to the drastic downturn in economic activities during that period.
In Panel B, we group firms based on 1-digit SIC codes to ascertain whether there is industry clustering in the data.
Roughly 29.47% of the observations are from the 30003999 SIC code range, which represents manufacturers of durable
goods. This represents the largest sector. Manufacturers of household goods (SIC 20002999) represent 20.39% of the sam-
ple, whereas wholesale and retail trades (SIC 50005999) represent another 18.25%. Firms in the 40004999 SIC code
range (i.e., transportation, communication, and utilities) account for about 10.96% of the observations and those in the
70007999 range (information technology-related fields) account for about 10.14%. Consequently, our sample is well suited
for this study since it includes dividend increasers of various industries covering nearly two decades.
Panel C of Table 1 reports descriptive statistics on firm characteristics. The mean market capitalization of the dividend
increasers in our sample is $14,407.44 million. MSCI ESG covers the 3000 largest U.S. firms by market capitalization;
therefore, it should come as no surprise that our sample consists of very large firms. We use the natural log of market
TABLE 1 Frequency distribution by year and deal characteristics
Panel A Sample distribution by year
Year NPercent Year NPercent Year NPercent
1995 85 3.04% 2001 36 1.29% 2007 272 9.71%
1996 52 2.21% 2002 39 1.39% 2008 187 6.68%
1997 51 1.82% 2003 176 6.29% 2009 72 2.57%
1998 41 1.46% 2004 251 8.96% 2010 209 7.46%
1999 43 1.54% 2005 303 10.82% 2011 297 10.61%
2000 26 0.93% 2006 269 9.61% 2012 381 13.61%
Panel B Distribution by sector Panel C Descriptive statistics on firms
SIC range NPercent Variables Mean Median
00000999 19 0.68% Market capitalization ($ million) 14,407.44 3344.47
10001999 208 7.43% Dividend change (DIVCHG) 0.578 0.240
20002999 571 20.39% Dividend premium (DIVPREM) 3.264 2.944
30003999 825 29.47% Institutional ownership (INSTOWN) 0.536 0.655
40004999 307 10.96% Cash-to-assets ratio (CTA) 0.130 0.081
50005999 511 18.25% Retained earnings-to-assets (RETA) 0.352 0.343
70007999 284 10.14% Operating income-to-assets (ROA) 0.082 0.075
80008999 75 2.68% Debt-to-assets ratio (DTA) 0.520 0.519
This table provides frequency distributions and descriptive statistics for the sample of 2800 dividend increases obtained from CRSP covering the eighteen-year period
19952012 inclusive. Each dividend announcement satisfies the following criteria: The percent change in dividends is between 12.5% and 500%. No stock split,
stock dividend, other cash distribution, or other corporate events occur within 15 trading days around the dividend announcement. There are no ex-distribution dates
between the ex-distribution dates of the previous and the current dividend payments. We further require that firms are not in the financial sector (SIC 60006999)
and have data available in both Compustat and the MSCI ESG databases at the time of the dividend announcement. We report the sample distribution by year in
Panel A and by sector in Panel B. Panel C provides summary statistics on firm characteristics.
GLEGG ET AL.
|
153
capitalization to gauge firm size. The conditional mean (median) percentage increase in cash dividend per share relative to
the previous quarter is about 57.80% (24.00%), which is similar to prior studies (see Benartzi et al., 1997; Grullon et al.,
2002; Nissim & Ziv, 2001). The mean (median) dividend premium
12
at the time of the announcement is 3.264 (2.944).
Mean institutional holdings is 53.60%, implying that institutional investors prefer dividend paying stock; although they
do not have a preference for firms that pay high dividends (Grinstein & Michaely, 2005). Cash relative to assets averages
about 13.00% at the end of the prior quarter, suggesting that dividend increasing firms have a build-up of excess funds.
Earned equity as a percent of assets averages 35.20%, which is comparable to the estimat e reported by DeAngelo et al.
(2006). Dividend increasers also display favorable prior performance; the mean (median) operating income to asset ratio is
8.20% (7.50%). Moreover, their debt-to-assets ratio averages 52.00%; the median is 51.90%.
4
|
METHODOLOGY
4.1
|
Corporate social responsibility
Pursuant to Kim et al. (2014), we compute a standardized CSR score that takes into account industry-year fixed effects
using the following five MSCI ESG dimensions: community, environment, employee relations, product quality and safety,
and diversity. Each dimension is associated with indicators denoting strengths and concerns (i.e., positive and negative
markers); MSCI ESG assigns a point to each indicator. Following Kim et al. (2014), we compute CSR net counts as total
strengths minus total concerns in the aforementioned five dimensions at the end of the year prior to the dividend change.
CSR net counts are then transformed to create a CSR score that ranges from zero to one to facilitate comparison of CSR
scores across time.
Kim et al. (2014) suggest a transformation that preserves the relative distance between CSR net counts for firms within
the same FamaFrench 48 industry for each year using the following formula, where CSR refers to CSR net counts defined
above:
CSRSCORE5¼CSR for firm i in year t-Min.CSR for firm i0s industry in year tðÞ
Max. CSR for firm i0s industry in year t-Min. CSR for firm i0s industry in year t (1)
Benson and Davidson (2010), Benson, Davidson, Wang, and Worrell (2011), and Coombs and Gilley (2005) also esti-
mate CSR engagement using the five dimensions accentuated by Kim et al. (2014).
13
Consequently, CSR_SCORE
5
is in
line with earlier studies.
In addition, we compute an alternative CSR score. For robustness, we also include the MSCI ESG corporate governance
dimension and its human rights dimension into the Kim et al. (2014) methodology. Thus, our alternate CSR score is based
on the seven MSCI ESG dimensions; we denote this alternative measure as CSR_SCORE
7
.
4.2
|
Abnormal announcement period returns
We measure the wealth effects around dividend increase announcements as the residual from the market model. The CRSP
total value weighted index is used as the benchmark portfolio and the model parameters are estimated over the (350, 31)
event window relative to the announcement date. We compute cumulative abnormal returns (CAR) over the 3-day event
window (1, +1). Similarly, we calculate the run-up in stock price prior to the announcement over the (30, 2) event
window. We use the bootstrapped Patell t-statistic to test whether the mean CAR is statistically significant (Lyon, Barber,
& Tsai, 1999; Patell, 1976).
4.3
|
Change in operating performance
Pursuant to earlier studies, we use the change in industry-adjusted operating performance to gauge the change in profitabil-
ity following dividend increases. Profitability is measured as the ratio of operating income before depreciation and amorti-
zation scaled by the book value of total assets (i.e., ROA). Nissim and Ziv (2001) emphasize that dividend payouts are
typically set in response to annual, rather than quarterly earnings; therefore, we follow their suggestion and use annual data
to measure the change in profitability. Since most firms change their dividends only once per fiscal year (Benartzi et al.,
1997; Nissim & Ziv, 2001), this data adjustment has only a marginal impact on our sample. The revised sample consi sts of
2479 dividend increases.
14
154
|
GLEGG ET AL.
As in Grullon et al. (2002), industry-adjusted operating performance (IAOP) is the ROA of the dividend increasing firm
minus that of the matching firm in the same 2-digit SIC code with the closest average change in ROA during the 3 years prior
to the dividend change (i.e., year 0).
15
Benartzi et al. (1997) use a similar method. Grullo n et al. (2002) advise that this mea-
sure of firm-specific profitability controls for systematic changes in ROA across similar firms. As a result, it is a reliable proxy
for abnormal operating performance. They also argue that subtracting the past average change in IAOP from the future average
change provides a reliable way to control for any firm-specific drift in operating performance that is unrelated to dividend
changes. Consequently, we proxy the change in profitability as the average change in IAOP in the 3 years after the dividend
change minus the average change in IAOP in the 3 years prior to the dividend change (denoted DIAOP).
4.4
|
Descriptive statistics and correlation matrix
Table 2 provides summary information and correlation matrix on the CSR scores, CAR, and DIAOP. Panel A shows that
the mean CSR_SCORE
5
is 0.428 for our sample of dividend increasers (mean CSR_SCORE
7
is 0.446). Kim et al. (2014)
report an average CSR score of 0.404; hence, our estimate is comparable to that of earlier studies. The mean 3-day CAR
around dividend increase announcements is 0.71% (t-statistic 7.87), implying that investors deem these events to be news
worthy. Numerous studies also document a favorable market reaction to dividend increases.
However, we find that on average industry-adjusted operating performance does not change after dividend increases.
Hence, for the average firm in our sample, dividend increases contain no information about future earnings. Studies by
DeAngelo et al. (2006), Benartzi et al. (1997), and Grullon et al. (2005) document similar results.
Panel B reports the Pearson correlation coefficients and their p-values. Collectively, these preliminary results support the
idea that dividend increases by high- and low-CSR firms convey different information to the market, thereby eliciting dis-
parate wealth effects.
16
For instance, both our CSR measures are negatively related to the announcement period CAR. The
correlation between CSR_SCORE
5
and CAR is 0.046 and that between CSR_SCORE
7
and CAR is 0.043; both test statis-
tics are significant at the 5% level or better. Thus, low-CSR dividend increasing firms appear to elicit higher abnormal
announcement returns than their high-CSR counterparts.
In addition, the correlations between the two CSR scores and the DIAOP variable are also negative (a correlation of about
0.04) and significant at the 5% level. This implies that dividend increasing firms with lower CSR investments also exhibit
greater subsequent improvements in profitability. Overall, the results in Panel B of Table 2 support Hypotheses H1a and H2a.
TABLE 2 Descriptive statistics on CSR score, abnormal returns and change in industry-adjusted operating performance
Panel A Summary statistics
Variables NMean Median Std. dev. Skewness t-Statistic
CSR_SCORE
5
2800 0.428 0.385 0.269 0.465 84.21***
CSR_SCORE
7
2800 0.446 0.417 0.263 0.344 89.82***
CAR [1, 1] 2800 0.71% 0.48% 4.692% 0.442 7.87***
DIAOP
3, 3
2285 0.03% 0.04% 6.548% 0.109 0.23
Panel B Pearson correlation matrix
CSR_SCORE
5
CSR_SCORE
7
CAR [1,1] DIAOP
3, 3
CSR_SCORE
5
1.000
CSR_SCORE
7
0.925*** (0.000) 1.000
CAR [1, 1] 0.046*** 0.043** 1.000
(0.014) (0.025)
DIAOP
3, 3
0.045** (0.033) 0.041** 0.019 1.000
(0.044) (0.351)
This table provides summary statistics (inPanel A) and the Pearson correlation matrix (in Panel B) for the key variables in the paper. The sample consists of 2800 dividend
increases obtained from CRSP covering the eighteen-year period 19952012, inclusive. CSR_SCORE
5
is the industry adjusted standardized social responsibility rating
based on the following five MSCI ESG categories at the end of the prior year: community, diversity, employeerelations, environment, and product. CSR_SCORE
7
is the
alternative measure that also includes the corporategovernance and human rights categories. CAR [1, 1] is the 3-day market model abnormal returns benchmarked against
the CRSP total value weighted index; the model parametersare estimated over the [350, 31] event window. Pursuant to Grullon et al. (2002), DIAOP is computed as
the average change in industry-adjusted operating performanceover the 3 years after the dividend change minus the average change over the 3 years prior to the change.
Panel B reports the Pearson correlation coefficients and their p-values.The symbols **, and *** indicate the statistical significance at the 5% and 1% levels, respectively.
GLEGG ET AL.
|
155
5
|
UNIVARIATE ANALYSES
5.1
|
Dividend increasers with high versus low CSR scores
Table 3 provides univariate test results. In Pa nel A, we divide our sample into two portfolios based on the median
CSR_SCORE
5
; firms with CSR scores greater than the median are classified high-CSR firms, while those with CSR scores
below the median are classified low-CSR firms. The mean abnormal returns to high-CSR dividend increasers is 0.459% (t-
statistic =4.01), while that to low-CSR dividend increasers is 0.948% (t-statistic =6.11). The difference is significant at the
1% level (t-statistic =2.72), implying that the market reacts more favorably to dividend increases when CSR engagement is
lower.
The results further show that low-CSR firms elicit significant increases in profitability after increasing dividend, while
high-CSR firms experience substantial decreases in profitability after increasing dividend. On average, low-CSR divi dend
increasers exhibit a 0.277% increase in industry-adjusted operating performance over the 3-year period following the divi-
dend increase, which is highly significant at the 5% level (t-statistic =2.49). In contrast, high-CSR dividend increasers elicit
a0.291% change in industry-adjusted operating performance over the 3-year period after the dividend increase, which is
statistically significant (t-statistic =1.91). The difference in the change in profitability between high- and low-CSR divi-
dend increasing firms is statistically significant (t-statistic =2.10).
Generally speaking, we confirm most of these results using the alternative CSR activities measure. As eviden t from the
table, the results obtained from using CSR_SCORE
7
to proxy CSR are quantitatively similar to those of CSR_SCORE
5
.
These results are consistent with the Pearson correlation results reported in Table 2, and support the argument in the litera-
ture that socially responsible firms are more transparent and commit to higher ethical standards than other firms (Kim et al.,
2012).
TABLE 3 Univariate results
Panel A Comparison between highCSR and lowCSR dividend increasers
CSR_SCORE
5
LowCSR increasers (1) HighCSR increasers (2) Difference (1)(2)
Mean Median t-Statistic Mean Median t-Statistic Mean Median t-Statistic
CAR [1, 1] 0.948% 0.633% 6.11*** 0.459% 0.248% 4.01*** 0.490% 0.385% 2.72***
DIAOP
3, 3
0.277% 0.309% 2.49** 0.291% 0.151% 1.91* 0.567% 0.460% 2.10**
CSR_SCORE
7
CAR [1, 1] 0.903% 0.599% 6.27*** 0.416% 0.237% 3.53*** 0.487% 0.362% 2.60***
DIAOP
3, 3
0.020% 0.000% 0.94 1.057% 0.547% 5.35*** 1.077% 0.547% 3.62***
Panel B Comparison by CSR score quartiles
CSR_SCORE
5
Quartiles Difference
Low-Q1 Q2 Q3 High-Q4 Q1Q4 t-Statistic
CAR [1, 1] 0.970% 0.905% 0.491% 0.417% 0.553% 2.08**
DIAOP
3, 3
0.641% 0.102% 0.147% 0.173% 0.814% 2.47**
CSR_SCORE
7
CAR [1, 1] 0.974% 0.833% 0.559% 0.292% 0.682% 2.14**
DIAOP
3, 3
0.460% 0.509% 1.197% 0.931% 1.391% 2.07**
This table reports univariate results. In Panel A, we partition the sample of dividend increasing firms based on the median value of the CSR score; low-CSR firms
have CSR score median and high-CSR firms have CSR score median. Pursuant to Kim et al. (2014), CSR_SCORE
5
is the industry adjusted standardized social
responsibility rating based on the following five MSCI ESG categories at the end of the prior year: community, diversity, employee relations, environment, and prod-
uct. CSR_SCORE
7
is the alternative measure that also includes the corporate governance and human rights categories. CAR [1, 1] is the 3-day abnormal returns
based on the market model. The market model parameters are estimated over the period from t =350 to t=31 days before the announcement; the CRSP total value
weighted index is used as the benchmark portfolio. Pursuant to Grullon et al. (2002), DIAOP is computed as the average change in industry-adjusted operating per-
formance over the 3 years after the dividend change minus the average change over the 3 years prior to the change. DSRISK
3, 3
is the change in systematic risk,
computed as the average change in systematic risk in the 3 years after the dividend change minus the average change in the 3 years prior to the change. Panel B pro-
vides a comparison of these key variables across quartiles of the CSR score and a test of means between the lowest and highest quartiles. The symbols *,**, and
*** indicate the statistical significance at the 10%, 5%, and 1% levels, respectively.
156
|
GLEGG ET AL.
5.2
|
Dividend increasers in the top versus bottom quartile of the CSR score
Panel B of Table 3 provides comparisons of CAR and DIAOP across quartiles of the CSR score (i.e., CSR_SCORE
5
). The
mean 3-day abnormal returns is largest in the lowest quartile (about 0.97%) and smallest in the highest quart ile (about
0.42%); the difference between the lowest and highest quartiles is significant at the 5% level (t-statistic =2.08). This is con-
sistent with our earlier findings suggesting that low-CSR firms earn higher wealth effects around dividend increase
announcements than high-CSR firms.
We also find that the change in industry-adjusted operating performance after dividend increases is highest in the bottom
quartile of the CSR score (about 0.64%) and lowest in the top quartile (about 0.17%); the difference between the two
quartiles is significant at the 5% level (t-statistic =2.47). Accordingly, low-CSR dividend increasing firms experience nota-
ble growth in profitability after increasing dividend, whereas high-CSR dividend increasing firms suffer a decline in prof-
itability after increasing dividend.
We document quantitatively similar results using the alternative CSR activity measure (CSR_SCORE
7
). For instance, the
mean CAR (DIAOP) is highest in the bottom quartile and lowest in the top quartile; the test statistics show that the differ-
ence between the top and bottom quartiles is significant at the 5% level. Consequently, these findings are robust to the
CSR measure used.
6
|
MULTIVARIATE REGRESSION ANALYSES
6.1
|
Regression results for abnormal returns
In this section, we report cross-sectional regression analyses on the abnormal returns surrounding dividend increase
announcements. We estimate the following model:
CARð1;þ1Þ¼aþb1CSRSCOREiþb2DIVCHGiþb3DIVPREMiþb4RUNUPi
þb5FSIZEiþb6CTAiþb7RETAiþb8INSTOWNiþb9DAiþþb8IRISKi
(2)
Our primary variable of interest is the proxy for corporate social responsibility. Regression (1) specifies results using
our main CSR variable CSR_SCORE
5
, while Regression (2) specifies results using the alternate CSR_SCORE
7
. Since some
firms increase dividends several times over the sample period, we use standard errors clustered by firm.
We control for a number of factors that have been shown in the literature to affect the excess returns around dividend
increases. For instance, Brickley (1983) finds that the magnitude of the dividend change influences the stock price
response. So, we control for the percent change in dividend (DIVCHG). Lie and Li (2006) argue that the price reaction also
depends on the dividend premium. We measure the dividend premium (denoted DIVPREM) as the difference in log book-
value-weighted average market-to-book ratio for dividend payers and for nonpayers at the end of the quarter prior to the
dividend change (Baker & Wurgler, 2004; Lie & Li, 2006).
In addition, we control for the run-up in price before the announcement since large price run-ups may result in more
excess valuations; RUNUP is the CAR over the event window (30, 2) based on the market model. We control for firm
size because size is known to be inversely related to stock returns; FSIZE is the natural log of the market value of equity
10 days prior to the divided increase. We also include the cash-to-asset ratio at the end of the quarter prior (CTA) because
cash-rich firms have more resources to increase dividend.
DeAngelo et al. (2006) suggest that the mix of earned and contributed capital is a logical proxy for a firm s life-cycle
stage because it measures the extent to which the firm is self-financing or reliant on external capital. So, we control for the
retained earnings-to-assets ratio (RETA) at the end of the quarter prior to the dividend change announcement.
17
We also
include the percent of shares owned by institutional investors at the end of the quarter prior to the dividend increase (INS-
TOWN) since institutions prefer dividend paying firms to non-dividend paying firms (Grinstein & Michaely, 2005). Institu-
tional ownership data are collected from the Thomson-Reuters institutional holdings (13F) database.
Mikhail, Walther, and Willis (2003) find that firms with higher earnings quality have lower market reaction to dividend
increases. Therefore, we control for earnings quality using discretionary accruals at the end of the quarter prior (DA), as
measured in Zang (2012). In addition, Hoberg and Prabhala (2009) show that risk affects a firms propensity to increase
dividends. So, we control for idiosyncratic risk (IRISK), which we proxy as the standard deviation of the residuals from the
market model on the daily return in the quarter prior.
GLEGG ET AL.
|
157
Table 4 reports the results from the estimat ion of Eq. (2). Regression (1) shows that excess announcement returns are inver-
sely and significantly related to CSR_SCORE
5
(t-statistic =2.01). Thus, even after controlling for other factors, we still find
more favorable CAR around dividend increases when CSR engagement is lower. The marginal effect of CSR_SCORE
5
on
abnormal announcement returns is approximately 0.038. Consequently, a unit increase in the CSR score reduces the wealth
gains to shareholder of the median dividend increasing firm in our sample by about $127 million.
18
Several control variables are also significant. All else being equal, larger dividend increases elicit more favorable
announcement effects; a larger price run-up to the announcement also leads to higher abnormal returns (t-statistic =2.16).
Yet, larger firms appear to earn lower excess returns around dividend increase announcements. Dividend increasers with
higher cash-to-asset ratios elicit more favorable announcement abnormal returns (t-statistic =3.51). The coefficient for the
variable RETA is negative and highly significant (t-statistic =2.48). Thus, the market reacts less favorably to dividend
increases when retained earnings account for a larger portion of total assets, which is consistent with dividend life-cycle
theory (see DeAngelo et al., 2006).
We find similar results when we use CSR_SCORE
7
to gauge firmsCSR activities, as reported in Regression (2). Conse-
quently, our results are robust to different measures of CSR engagement. The inverse relation between CSR activities and
CAR is also robust to different model specifications.
6.1.1
|
Heckman two-step approach for selection bias
Given that firms enter into our sample only when they increase dividend, there may be self-selection bias in the aforemen-
tioned results. Therefore, we employ the Heckman (1976) two-step method to correct for potential selection bias. In the first
TABLE 4 Cross-sectional regression results on abnormal announcement returns
Regression (1) Regression (2)
Coef. t-Stat Coef. t-Stat
INTERCEPT 0.0209 2.60** 0.0165 2.52**
CSR_SCORE
5
0.038 2.01** ––
CSR_SCORE
7
––0.037 2.15**
DIVCHG 0.031 1.62* 0.032 1.71*
DIVPREM 0.001 0.11 0.000 0.02
RUNUP 0.031 2.16** 0.031 2.09**
FSIZE 0.039 1.77* 0.031 1.57
CTA 0.036 3.51*** 0.042 3.26***
RETA 0.049 2.48** 0.050 2.36**
INSTOWN 0.009 0.40 0.010 0.45
DA 0.006 0.29 0.004 0.20
IRISK 0.004 0.15 0.009 0.28
F-statistic 4.584*** 4.499***
Adjusted R
2
0.02 0.01
Number of Obs. 2780 2780
Quarter fixed effects YES YES
This table reports regression results on the abnormal announcement period returns. The dependent variable is the market model 3-day CAR; model parameters are
estimated over the period from t =350 to t =31 days before the announcement. The variable of interest is the CSR score. Pursuant to Kim et al. (2014), CSR score
(CSR_SCORE
5
) is defined as the total strengths minus total concerns in the following five MSCI ESG categories at the end of the prior year: community, diversity,
employee relations, environment, and product, standardized in each industry for each year. For robustness, we consider an alternative CSR measure that includes all
seven MSCI ESG categories (i.e., we include corporate governance and human rights in the Kim et al. methodology), which we denote CSR_SCORE
7
.DIVCHG is
the absolute value of the percent changes in dividends. DIVPREM is the difference in log weighted average market-to-book ratio for dividend payers and for nonpay-
ers the year prior to the dividend change. RUNUP is the stock price run-up over the event window (30, 2) before the announcement of the dividend change.
FSIZE is the log of market capitalization the quarter prior to the dividend change. CTA is the cash-to-assets ratio and RETA is the retained earnings-to-assets ratio
the quarter prior to the dividend change. INSTOWN is the percent institutional ownership at the end of the quarter prior to the dividend change. DA denotes the dis-
cretionary accruals at the end of the quarter prior, as measured in Zang (2012). IRISK is standard deviation of the residuals from the market model on the daily
return in the quarter prior. The symbols *,**, and *** indicate the significance at the 10%, 5%, and 1% levels, respectively.
158
|
GLEGG ET AL.
stage, we estimatea logit regression of the propensity to payout dividends. Our propensity to payout model is based on
DeAngelo et al. (2006). The dependent variable equals 1 if the firm is a dividend payer in quarter t; zero otherwise. The
model controls for retained earnings-to-total equity (RETE), net income-to-assets (ROA), sales growth rate (SGR), market-
to-book (MTB), and the natural logarithm of market capitalization (SIZE).
We run separate logit regressions for each quarter in the sample period (19952012) for all firms with CRSP share
codes 10 or 11 and non-missing Compustat data items. Panel A of Table 5 reports the mean value of the quarterly fitted
coefficients and the Fama and MacBeth (1973) t-statistics, computed using the time series of the fitted logit coefficients
across the 72 quarters in our sample period. In the second stage of the Heckman approach, we compute the inverse Mills
ratio as a correction factor and include it as an explanatory variable in our model (INVMILLS).
The Heckman regression results are reported in Panel B of Table 5. Model 1 (Model 2) shows that CSR_SCORE
5
(CSR_SCORE
7
) is still negative and highly significant. Furthermore, INVMILLS is not significantly different from zero, sug-
gesting that self-selection is not a problem in our sample. The significance and implication of key control variables are also
unchanged. As a result, our finding of an inverse relation between CSR activities and the abnormal returns around dividend
increase announcements is not only robust, but also free of self-selection biases.
6.2
|
Regression results for change in profitability
In this section, we use multivariate regression analyses to examine the relation between CSR activities and the change in
operating performance following dividend increases. As in earlier studies, we annualize the quarterly dividend ch anges in
order to match the dividend announcements during a fiscal year to the earnings of that year (see Benartzi et al., 1997; Grul-
lon et al., 2002). Pursuant to Nissim and Ziv (2001), we calculate the geometric sum of the quarterly dividend changes
across observations with the same firm and fiscal year values.
We estimate the following model:
DIAOPi¼aþb1CSRSCOREiþb2PIAOPiþb3PDIAOPiþb4ADIVCHGiþb5CMPTIVi
þb6UNIQUEiþb7DDAiþb7DIRISKiþb7DFSIZEiþb8DINSTOWNiþb9INVMILLSi
(3)
The dependent variable is the change in industry-adjusted operating performance (IAOP), gauged as the difference of
the average IAOP in the 3 years after the dividend change minus the average IAOP in the 3 years prior to the change. As
before, our primary variable of interest is the proxy for corporate social responsibility. Regression (1) provides results using
CSR_SCORE
5
, whereas Regression (2) provides results using CSR_SCORE
7
.
Lie and Li (2006) report that past performance patterns influence expected futur e changes in performance. So, we
control for past industry-adjusted performance (PIAOP), as well as the past change in industry-adjusted performance
(PDIAOP). We also include the annualized change in dividends (ADIVCHG) since the signaling theory suggests that divi-
dend changes are reliable indicators of future earnings prospects (Nissim & Ziv, 2001).
Product market competition may discipline managers, thereby improving profitability. However, a counter argument is
that competition may reduce market share, which may hurt profitability. Therefore, we control for whether the firms indus-
try is competitive (CMPTIV) and for whether the firms industry is unique (UNIQUE). As in Chemmanur, Jordan, Liu, and
W (2010), CMPTIV is a dummy variable that takes the value 1 if the industrys Herfindahl index is in the bottom quartile
of all 48 FamaFrench industries, and 0 otherwise. Likewise, UNIQUE is a dummy variable that takes the value 1 if the
industrys median ratio of selling expenses to sales is in the top quartile of all 48 industries, and 0 otherwise.
We include the change in discretionary accruals (DDA) and the change in idiosyncratic risk (DIRISK) as instruments to
control for changes in earnings quality and valuation uncertainty over the period, respectively; likewise, we contr ol for the
change in firm size (DFSIZE). Monitoring by institutional investors can improve profitability (Shleifer & Vishny, 1986),
but they prefer firms that pay fewer dividends (Grinstein & Michaely, 2005). Therefore, we control for the change in insti-
tutional ownership over the period (DINSTOWN). We also include the Heckman inverse Mills ratio (INVMILLS) to control
for the firms propensity to pay dividends.
We report the results from the estimation of Eq. (3) in Table 6. The coefficients for CSR_SCORE
5
(in Regression (1))
and CSR_SCORE
7
(in Regression (2)) are both negative and highly significant.
19
Consequently, these findings suggest that
even after controlling for other factors dividend increasers with lower CSR scores experience great er improvements in
industry-adjusted operating performance than dividend increasers with higher CSR scores. For example, over the 3 years
after the dividend increases, profitability decreases by about 1.24% for a one stand ard deviation increase in CSR_SCORE
5
(0.269 90.046 =0.0124). The decline is notably larger when based on CSR_SCORE
7
(i.e., about 1.55% estimated as
GLEGG ET AL.
|
159
0.263 90.059 =0.0155). These findings are consistent with Hypothesis 2, which suggests that CSR engagement is
inversely related to the change in profitability following dividend increases.
In addition, CMPTIV is negative and highly significant, implying that dividend increasing firms that are subject to more
product market competition experience a subsequent decline in profitability. The change in industry-adjusted operating
TABLE 5 Heckman two-step regression for selection bias
Panel A: Logistic regression
Coef. t-Stat
INTERCEPT 2.258 54.35***
RETE 1.330 40.93***
ROA 0.950 12.62***
SGR 1.162 16.60***
MKBK 0.002 2.62***
FSIZE 0.283 32.87***
Chi X
2
statistic 82.60***
Pseudo R
2
0.28
% correct 76.05
Number of Qtr. 72
Panel B: Heckman correction for dividend increases
Regression (1) Regression (2)
Coef. t-Stat Coef. t-Stat
INTERCEPT 0.0365 2.46** 0.0318 2.19**
CSR_SCORE
5
0.037 1.99** ––
CSR_SCORE
7
––0.039 1.95*
DIVCHG 0.032 1.42 0.034 1.48
DIVPREM 0.002 0.09 0.000 0.01
RUNUP 0.032 1.43 0.032 1.43
FSIZE 0.063 2.22** 0.055 1.91*
CTA 0.061 2.08** 0.067 2.24**
RETA 0.066 2.73*** 0.067 2.75***
INSTOWN 0.008 0.40 0.009 0.47
DA 0.002 0.08 0.000 0.01
IRISK 0.010 0.37 0.014 0.54
INVMILLS 0.040 1.07 0.040 1.07
F-statistic 3.55*** 3.302***
Adjusted R
2
0.02 0.01
Number of Obs. 2594 2594
Quarter fixed YES YES
This table presents regression results for the Heckman (1979) two-step approach. In the first stage, we estimate the propensity to pay dividends. The dependent vari-
able equals 1 if the firm is a dividend payer in quarter t; zero otherwise. The independent variables include retained ear nings-to-total equity (RETE), net income-to-
assets (ROA), sales growth rate (SGR), market-to-book (MTB), and the natural logarithm of market capitalization (FSIZE). We run separate logit regressions each
quarter (totaling 72 quarters) for all firms with CRSP share codes 10 or 11 and non-missing Compustat data. In Panel A, we report the average value of the fitted
logit coefficients and the calculated Fama and MacBeth (1973) t-statistics. In the second stage, we compute the inverse Mills ratio as a correction factor and include
it as an explanatory variable in our model, denoted INVMILLS. The Heckman correction regression results are reported in Panel B. The dependent variable is the
market model 3-day CAR; model parameters are estimated over the period from t=250 to t=30 days before the announcement. The variable of interest is the CSR
score. Pursuant to Kim et al. (2014), CSR score (CSR_SCORE
5
) is defined as the total strengths minus total concerns in the following five MSCI ESG categories at
the end of the prior year: community, diversity, employee relations, environment, and product, standardized in each industry for each year. For robustness, we con-
sider an alternate CSR measure that includes all seven MSCI ESG categories (i.e., we include corporate governance and human rights into the Kim et al. methodol-
ogy), which we denote CSR_SCORE
7
. All control variables that were previously specified in Table 5 are as defined earlier. *,**, and *** indicate the significance
at the 10%, 5%, and 1% levels, respectively.
160
|
GLEGG ET AL.
performance also correlates negatively with the change in discretionary accruals. Thus, performance is low er for dividend
increasers that subsequently exhibit poorer earnings quality. Yet, dividend increasing firms that grow in size over the ensu-
ing 3 years exhibit higher profitability. INVMILLS is positive and significant, implying that firms pay dividends when they
anticipate improvements in profitability.
7
|
ENDOGENEITY
Firms with better CSR may enjoy better firm performance, and so are likely to increase dividends. Then again, better per-
forming firms can afford to concurrently invest more in CSR and pay out more to shareholders. Therefore, we correct for
potential endogeneity biases using the instrumental variable technique.
Pursuant to earlier studies, we use the initial firm-level CSR score (denoted First_CSR) and the industry average CSR
value (denoted Ind_CSR) as instruments (see Attig, Cleary, El Ghoul, & Guedhami, 2014; Attig, El Ghoul, Guedhami, &
Suh, 2013; El Ghoul, Guedhami, Kim, & Park, 2016; El Ghoul et al., 2011). The rationale is that these instruments are
highly correlated with contemporaneous CSR investment because CSR is path-dependent, and are likely to be exogenous
given that they are predetermined. Moreover, these are appropriate instruments since lagged values of firm- and industry-
TABLE 6 Cross-sectional regression results on change in operating performance
Regression (1) Regression (2)
Coef. t-Stat Coef. t-Stat
INTERCEPT 0.046 2.45** 0.04672.51**
CSR_SCORE
5
0.0456 2.14** ––
CSR_SCORE
7
––0.059 2.70***
PIAOP 0.053 1.62 0.052 1.61
PDIAOP 0.031 1.11 0.030 1.09
ADIVCHG 0.012 0.54 0.012 0.54
CMPTIV 0.061 2.62*** 0.060 2.62***
UNIQUE 0.030 1.24 0.033 1.37
DDA 0.045 1.80* 0.045 1.83*
DIRISK 0.041 0.93 0.041 0.93
DFSIZE 0.099 3.41*** 0.107 3.66***
DINSTOWN 0.016 0.66 0.016 0.66
INVMILLS 0.076 2.21** 0.075 2.18**
F-statistic 2.141*** 2.298***
Adjusted R
2
0.031 0.032
Number of Obs. 2097 2097
Year fixed effects Yes Yes
This table reports regression results on the change in profitability following dividend increases. The dependent variable is the change in industry-adjusted operatin g
performance (DIAOP
3, 3
), computed as the average change in industry-adjusted ROA over the 3 years after the dividend change minus the average change over the
3 years prior to the change (see Grullon et al., 2002). The variable of primary interest is the CSR score. Pursuant to Kim et al. (2014), CSR score (CSR_SCORE
5
)
is defined as the total strengths minus total concerns in the following five MSCI ESG categories at the end of the prior year: community, diversity, employee rela-
tions, environment, and product, standardized in each industry for each year. For robustness, we consider an alternative CSR measure that includes all seven MSCI
ESG categories (i.e., we include corporate governance and human rights into the Kim et al. methodology), which we denote CSR_SCORE
7
.ADIVCHG is the annual-
ized percent changes in dividends. Following Lie and Li (2006), we use PIAOP and PDIAOP to control for past industry-adjusted performance and the past change
in industry-adjusted performance, respectively. CMPTIV is a dummy variable that takes the value 1 if the industrys Herfindahl index is in the bottom quartile of all
48 FamaFrench industries, and 0 otherwise. UNIQUE is a dummy variable that takes the value 1 if the industrys median ratio of selling expenses to sales is in the
top quartile of all 48 industries, and 0 otherwise. DA denotes the discretionary accruals at the end of the quarter prior, as measured in Zang (2012). IRISK is standard
deviation of the residuals from the market model on the daily return in the quarter prior. We also control for firm size (FSIZE), measured as the log value of market
capitalization. DINSTOWN is the change in institutional ownership over the period. INVMILLS is the Heckman inverse Mills ratio used to control for the propensity
to pay. The symbols *,**, and *** indicate the significance at the 10%, 5%, and 1% levels, respectively.
GLEGG ET AL.
|
161
TABLE 7 Two-stage least squares (2SLS) results on abnormal announcement returns
Panel A: CSR_SCORE
5
First stage Second stage
Coef. t-stat Coef. t-stat
INTERCEPT 0.067 1.217 0.0315 1.044
First.CSR
5
0.433 20.326*** ––
First. Ind.CSR
5
0.135 2.926*** ––
Pred_CSR
5
–– 0.029 2.134**
DIVCHG 0.024 1.721* 0.037 1.760*
DIVPREM –– 0.144 0.512
RUNUP –– 0.023 0.952
FSIZE 0.220 9.939*** 0.011 0.398
CTA 0.011 0.716 0.023 2.014**
RETA 0.010 0.573 0.042 2.149**
INSTOWN 0.007 0.384 0.016 0.770
DA 0.019 1.141 0.017 0.851
IRISK 0.006 0.314 0.036 1.093
F-statistic 17.19*** 4.436***
Adjusted R
2
0.315 0.037
Number of Obs. 2780 2780
Quarter fixed YES YES
Panel B: CSR_SCORE
7
First stage Second stage
Coef. t-stat Coef. t-stat
INTERCEPT 0.1216 2.013 ** 0.0327 1.330
First.CSR
7
0.395 17.937*** ––
First.Ind.CSR
7
0.113 3.647*** ––
Pred_CSR
7
–– 0.045 2.785***
DIVCHG 0.031 2.032 ** 0.035 1.785*
DIVPREM –– 0.201 0.712
RUNUP –– 0.023 0.949
FSIZE 0.185 8.207 *** 0.008 0.302
CTA 0.007 0.445 0.024 2.039**
RETA 0.012 0.658 0.040 2.073 **
INSTOWN 0.013 0.734 0.017 0.813
DA 0.007 0.390 0.018 0.914
IRISK 0.020 0.911 0.036 1.101
F-statistic 12.40*** 4.457***
Adjusted R
2
0.245 0.038
Number of Obs. 2780 2780
Quarter fixed YES YES
This table presents 2SLS regression results on abnormal announcement returns. In the first stage, we estimate a firms CSR score using the CSR score recorded when the
firm enters the sample (First.CSR
5
) and the industry average CSR score in the first year of data (First.Ind.CSR
5
) as instruments. The other control variables are previ-
ously specified in Eq. (2) (and are as defined earlier). The instrumented CSR score from the first stage is used in the second stage regression (denoted Pred_CSR) to esti-
mate the impact of CSR on the stock price response to dividend increase announcements. The dependent variable in the second stage is the market model 3-day CAR;
model parameters are estimated over the period from t =250 to t =30 days before the announcement. Panel A reports the 2SLS results where CSR engagement is esti-
mated using the five dimensions accentuated by Kim et al. (2014), which is denoted CSR_SCORE
5
. Panel B reports the 2SLS results when we include the MSCI ESG
corporate governance dimension and its human rights dimension into the Kim et al. (2014) methodology, which is denoted CSR_SCORE
7
. All control variables that were
previously specified in Table 5 are as defined earlier. *,**, and *** indicate the significance at the 10%, 5%, and 1% levels, respectively.
162
|
GLEGG ET AL.
level CSR are unlikely to directly affect the current stock price reaction to dividend increases or the subsequent change in
operating performance.
Table 7 presents the instrumented CSR results on the wealth effects of dividend increase announcements. Using a two-
stage least squares (2SLS) method, we first estimate a firms CSR score based on its initial CSR score and the industry
average value. The instrumented CSR score from this first stage regression is then used in the second stage regression to
estimate the impact of CSR on the stock price response to dividend increases. The 2SLS results where CSR score is esti-
mated using the five dimensions accentuated by Kim et al. (2014) are reported in Panel A. Panel B reports the 2SLS results
when we include the MSCI ESG corporate governance dimension and its human rights dimension into the Kim et al.
(2014) methodology.
In both of these cases, the first stage regression shows that the coefficients on First_CSR and Ind_CSR are positive and
highly significant, confirming that they are suitable instruments for the contemporaneous CSR of the firm. We take the
instrumented value of the CSR score from the first stage and use it as our main independent variable in the second stage
(denoted Pred_CSR). As expected, the second stage results in Panel A show that Pred_CSR
5
is negative and statistically
significant. By the same token, Pred_CSR
7
is also negative and highly significant in Panel B. Thus, the inverse relation
between CSR activities and the 3-day CAR around dividend increase announcements persist even after accounting for
endogeneity bias, which supports Hypothesis 1. Therefore, we find very robust evidence that low -CSR dividend increasing
firms elicit higher abnormal returns than their high-CSR counterparts. Several of the control variables are still significant
and their implications are also unchanged.
The instrumented CSR score from the first stage regression is also used to assess the subsequent change in operating
performance and these 2SLS results are presented in Table 8. Regression (1) reports the second stage results based on the
instrumented variable Pred_CSR
5
, while Regression (2) reports the second stage results for the instrumented variable
Pred_CSR
7
. Here again we find that the coefficients for Pred_CSR
5
(in Regression (1)) and Pred_CSR
7
(in Regression (2))
are both still negative and highly significant,
20
implying that endogeneity is not a concern in our findings. Generally
TABLE 8 Two-stage least squares (2SLS) results on change in operating performance
Regression (1) Regression (2)
Coef. t-stat Coef. t-stat
INTERCEPT 0.005 0.387 0.008 0.554
Pred_CSR
5
0.065 2.563**
Pred_CSR
7
0.040 2.242**
PIAOP 0.023 0.586 0.024 0.592
PDIAOP 0.024 0.752 0.025 0.773
ADIVCHG 0.019 0.601 0.019 0.596
CMPTIV 0.071 3.233*** 0.070 3.200***
UNIQUE 0.050 2.192** 0.049 2.146**
DDA 0.015 0.496 0.017 0.542
DIRISK 0.086 1.980** 0.086 1.998**
DFSIZE 0.068 2.169** 0.049 1.684*
DINSTOWN 0.008 0.256 0.008 0.262
F-statistic 2.274*** 2.18***
Adjusted R
2
0.032 0.030
Number of Obs. 2097 2097
Year fixed effects Yes Yes
This table presents 2SLS regression results on the change in profitability following dividend increases. In the first stage, we estimate a firm s CSR score using the
CSR score recorded when the firm enters the sample (First. CSR
5
) and the industry average CSR score in the first year of data (First. Ind. CSR5) as instruments.
The other control variables are previously specified in Eq. (2) (and are as defined earlier). The instrumented CSR score from the first stage is used in the second
stage regression (denoted Pred_CSR) to estimate the impact of CSR on the change in profitability following dividend increases. The dependent variable is the change
in industry-adjusted operating performance (DIAOP
3, 3
), computed as the average change in industry-adjusted ROA over the 3 years after the dividend change
minus the average change over the 3 years prior to the change (see Grullon et al., 2002). Regression (1) reports the second stage regression results based on the
instrumented variable Pred_CSR
5
, while Regression (2) reports the second stage regression results based on the instrumented variable Pred_CSR
7
. All control vari-
ables that were previously specified in Table 6 are as defined earlier.*,**, and *** indicate the significance at the 10%, 5%, and 1% levels, respectively.
GLEGG ET AL.
|
163
speaking, the results in Table 8 support Hypothesis 2, which suggests that CSR engagement is inversely related to the
change in profitability following dividend increases.
8
|
CONCLUSION
This paper relates CSR investments to shareholder wealth gains from dividend changes. While there is earlier work that
looks at associations between CSR and the amount of dividends that a firm pays out (see Benlemlih, 2014; Rakotomavo,
2012), this paper takes a different approach by looking at the announcement returns around dividend increases and the sub-
sequent change in operating performance. We posit that dividend increases by high-CSR and low-CSR firms convey asym -
metrical information to investors about firm profitability, thereby eliciting disparate wealth effects.
Using a sample of 2800 dividend increases from 1995 to 2012, we show that dividend increasing firms with lower CSR
ratings elicit significantly higher abnormal announcement returns compared to dividend increasing firms with higher CSR
ratings. Our empirical tests also show that dividend increasers with lower CSR activities also experience greater subsequent
improvements in industry-adjusted operating performance. Together, these findings indicate that dividend increases help to
curtail the agency costs in socially irresponsible firms. Therefore, our collective results do not support the agency view of
CSR. Rather, we find support for the notion that socially responsible firms commit to high ethical and financial reporting
standards, which reduces agency problems.
Under agency theory, firms are more likely to experience high agency costs when the quality and quantity of financial
disclosure is poor. Since socially responsible firms commit to greater transparency and higher ethical standards than other
firms (Becchetti et al., 2013; Gelb & Strawser, 2001; Kim et al., 2014), they suffer fewer agency and informational prob-
lems. By documenting how CSR ratings influence shareholderswealth gains from dividend increases, we show that divi-
dend increases mitigate the agency conflicts between managers and shareholders in socially irresponsible firms.
ENDNOTES
1
CSR activities have also attracted the attention of professional money managers; about 10% of US investments are screened to ensure that they
meet CSR-related criteria (El Ghoul et al., 2011; Galema, Plantinga, & Scholtens, 2008).
2
Kim et al. (2014) also notes that at the time of the 20002001 accounting scandal, the Enron Corporation was considered to be a strong social
performer.
3
The $127 million was estimated as the product of the coefficient (0.038) and the median market capitalization of the dividend increasing firms
in the sample given in Table 1 ($3344.47 million).
4
These findings are consistent with our results on the wealth effects around announcements of dividend increases.
5
See studies such as those by Bajaj and Vijh (1990), Baker and Wurgler (2004), Brav, Graham, Harvey, and Michaely (2005), Benartzi et al.
(1997), DeAngelo, DeAngelo, and Skinner (2004), DeAngelo, DeAngelo, and Stulz (2006), Denis, Denis, and Sarin (1994), Fama and French
(2001), Grullon et al. (2002), Hoberg and Prabhala (2009), Lang and Litzenberger (1989), Lie and Li (2006), Nissim and Ziv (2001), and Yoon
and Starks (1995).
6
Enron provides ample anecdotal evidence that managers use CSR to hide corporate misconduct.
7
Likewise, Nelling and Webb (2009) find little evidence that CSR improves performance.
8
Jensen (1986) suggests that higher dividend payouts curb the agency costs associated with free cash flow.
9
CRSP distribution code 1232: first digit =1 (ordinary dividend); second digit =2 (cash, United States dollars); third digit =3 (quarterly divi-
dend); fourth digit =2 (taxable).
10
Formerly known as the Kinder, Lyndenberg, and Domini Research and Analytics Inc. (KLD) database.
11
Fracassi (2008) argues that the rationale and underlying dynamics of dividend increases contrasts with those of dividend decreases; this differ-
ence may help to explain the disparate results for dividend increases and decreases.
12
Given as the log difference in the value-weighted average market-to-book value of dividend-payers and the value-weighted market-to-book
value of non-dividend-payers (see Baker & Wurgler, 2004; Lie & Li, 2006).
13
Kim et al. (2014) recommend excluding the corporate governance and human rights categories from the CSR score because these two dimen-
sions are inherently different from the activities represented by the other categories.
14
In cases where a firm has multiple dividend increases in a given fiscal year, we only consider the first dividend increase in that year.
15
Thus, industry-adjusted operating performance (IAOP) is given as: ROA
it
ROA
mt
, where ROA
it
is the profitability of the sample firm and
ROA
mt
is the profitability of the matching firm.
16
In the case of dividend decreases, we find no meaningful correlation between the CSR scores, CAR, and DIAOP. For brevity, we do not pre-
sent these results in a table.
164
|
GLEGG ET AL.
17
The retained earnings-to-assets ratio also reflects firmsgrowth options.
18
The $127 million was estimated as the product of the coefficient (0.007) and the median market capitalization of the dividend increasing
firms in the sample given in Table 2 ($3344.47 million).
19
The point estimates are 0.046 (t-statistic =2.14) and 0.059 (t-statistic =2.70), respectively.
20
The point estimates are 0.065 (t-statistic =2.536) and 0.049 (t-statistic =2.242), respectively.
REFERENCE
Attig, N., Cleary, S., El Ghoul, S., & Guedhami, O. (2014). Corporate legitimacy and investment-cash flow sensitivity. Journal of Business
Ethics,121, 297314.
Attig, N., El Ghoul, S., Guedhami, O., & Suh, J. (2013). Corporate social responsibility and credit ratings. Journal of Business Ethics,117, 679
694.
Bae, K. -H., El Ghoul, S., Guedhami, O., Kwok, C., & Zheng, Y. (2017). Does corporate social responsibility add value? Evidence from capital
structure and product markets interactions. Working paper.
Bajaj, M., & Vijh, A. (1990). Dividend clienteles and the information content of dividend changes. Journal of Financial Economics,26, 193
219.
Baker, M., & Wurgler, J. (2004). A catering theory of dividends. Journal of Finance,59, 11251165.
Becchetti, L., Ciciretti, R., & Giovannelli, A. (2013). Corporate social responsibility and earnings forecasting unbiasedness. Journal of Banking
& Finance,37, 36543668.
Benartzi, S., Michaely, R., & Thaler, R. (1997). Do changes in dividends signal the future or the past? Journal of Finance,52, 10071034.
Benlemlih, M. (2014). Why do socially responsible firms pay more dividends? Working paper. Available at SSRN http://ssrn.com/abstract=
2566876.
Benson, B., & Davidson, W. (2010). The relation between stakeholder management, Firm value, and CEO compensation: A test of enlightened
value maximization. Financial Management,39, 926964.
Benson, W., Davidson, W., III, Wang, H., & Worrell, D. (2011). Deviations from expected stakeholder management, firm value, and corporate
governance. Financial Management,40,3981.
Bernea, A., & Rubin, A. (2010). Corporate social responsibility as a conflict between shareholders. Journal of Business Ethics,97,7186.
Borghesi, R., Houston, J., & Naranjo, A. (2014). Corporate socially responsible investments: CEO altruism, reputation, and shareholder interests.
Journal of Corporate Finance,26, 164181.
Boubakri, N., El Ghoul, S., Wang, H., Guedhami, O., & Kwok, C. (2016). Cross-listing and corporate social responsibility. Journal of Corporate
Finance,41, 123138.
Brammer, S., Brooks, S., & Pavelin, S. (2006). Corporate social performance and stock returns: UK evidence from disaggregate measures. Finan-
cial Management,35,97116.
Brav, A., Graham, J., Harvey, C., & Michaely, R. (2005). Payout policy in the 21st century. Journal of Financial Economics,77, 483527.
Brickley, J. (1983). Shareholder wealth, information signaling and the specially designated dividend. Journal of Financial Economics,12, 187
209.
Chemmanur, T., Jordan, B., Liu, M., & W, Q. (2010). Antitakeover provisions in corporate spin-offs. Journal of Banking & Finance,34, 813
824.
Cheng, I., Hong, H., & Shue, K. (2014). Do managers do good with other peoples money? Working paper. Available at http://papers.ssrn.c
om/sol3/papers.cfm?abstract_id=1962120.
Cheng, B., Ioannou, I., & Serafeim, G. (2014). Corporate social responsibility and access to finance. Strategic Management Journal,35,123.
Coombs, J., & Gilley, K. (2005). Stakeholder management as a predictor of CEO compensation: Main effects and interactions with financial per-
formance. Strategic Management Journal,26, 827840.
DeAngelo, H., DeAngelo, L., & Skinner, D. J. (2004). Are dividends disappearing? Dividend concentration and the consolidation of earnings.
Journal of Financial Economics,72, 425456.
DeAngelo, H., DeAngelo, L., & Stulz, R. M. (2006). Dividend policy and the earned/contributed capital mix: a test of the life-cycle theory. Jour-
nal of Financial Economics,81, 227254.
Deng, X., Kang, J., & Low, B. (2013). Corporate social responsibility and stakeholder value maximization: Evidence from mergers. Journal of
Financial Economics,110,87109.
Denis, D., Denis, D., & Sarin, A. (1994). The information content of dividend changes: Cash flow signaling, overinvestment, and dividend clien-
teles. Journal of Financial and Quantitative Analysis,29, 567587.
Easterbrook, F. (1984). Two agency cost explanations of dividends. American Economic Review,74, 650659.
El Ghoul, S., Guedhami, O., & Kim, Y. (2017). Country-level institutions, firm value, and the role of corporate social responsibility initiatives.
Journal of International Business Studies,48, 360385.
El Ghoul, S., Guedhami, O., Kim, H., & Park, K. (2016). Corporate environmental responsibility and the cost of capital: International evidence.
Journal of Business Ethics,48, 360385.
El Ghoul, S., Guedhami, O., Kwok, C., & Mishra, D. (2011). Does corporate social responsibility affect the cost of capital? Journal of Banking
& Finance,35, 23882406.
GLEGG ET AL.
|
165
Fama, E., & French, K. (2001). Disappearing dividends: changing firm characteristics or lower propensity to pay? Journal of Financial Eco-
nomics,60,344.
Fama, E., & MacBeth, J. (1973). Risk, return, and equilibrium: Empirical tests. Journal of Political Economy,71, 607636.
Fracassi, C. (2008). Stock price sensitivity to dividend changes. Working paper. Available at http://faculty.mccombs.utexas.edu/cesare.fracassi/
Price%20Sensitivity%20to%20Dividend%20Change%20-%20,072,808.pdf.
Freeman, R., Wicks, A., & Parmar, B. (2004). Stakeholder theory and the corporate objective revisited.Organization Science,15, 364369.
Galema, R., Plantinga, A., & Scholtens, B. (2008). The stocks at stake: Return and risk in socially responsible investment. Journal of Banking &
Finance,32, 26462654.
Gelb, D., & Strawser, J. (2001). Corporate social responsibility and financial disclosures: An alternative explanation for increased disclosure.
Journal of Business Ethics,33(1), 113.
Giuli, A., & Kostovetsky, L. (2014). Are red or blue companies more likely to go green? Politics and corporate social responsibility. Journal of
Financial Economics,111, 158180.
Goss, A., & Roberts, G. (2011). The impact of corporate social responsibility on the cost of bank loans. Journal of Banking & Finance,35,
17941810.
Grinstein, Y., & Michaely, R. (2005). Institutional holdings and payout policy. Journal of Finance,60, 13891426.
Grullon, G., Michaely, R., Benartzi, S., & Thaler, R. (2005). Dividend changes do not signal changes in future profitability. Journal of Business,
78, 16591682.
Grullon, G., Michaely, R., & Swaminathan, B. (2002). Are dividend changes a sign of firm maturity? Journal of Business,75, 387424.
Heckman, J. (1979). Sample selection bias as a specification error. Econometrica,47, 153161.
Hemingway,C., & Maclagan, P. (2004).Managerspersonalvalues as drivers of corporate socialresponsibility. Journal of BusinessEthics,50,3344.
Hoberg, G., & Prabhala, N. (2009). Disappearing dividends, catering, and risk. Review of Financial Studies,22,79119.
Jensen, M. (1986). Agency costs of free-cash flow, corporate finance and the market for takeovers. American Economic Review,76, 323329.
Jha, A., & Cox, J. (2015). Corporate social responsibility and social capital. Journal of Banking & Finance,34, 25492561.
Jiao, Y. (2010). Stakeholder welfare and firm value. Journal of Banking & Finance,34, 25492561.
Jiraporn, P., Jiraporn, N., Boeprasert, A., & Chang, K. (2014). Does corporate social responsibility (CSR) improve credit ratings? Evidence from
geographic identification. Financial Management,43, 505531.
Jones, T. (1995). Instrumental stakeholder theory: A synthesis of ethics and economics. The Academy of Management Review,20, 404437.
Kim, Y., Li, H., & Li, S. (2014). Corporate social responsibility and stock price crash risk. Journal of Banking & Finance,43,113.
Kim, Y., Park, M., & Wier, B. (2012). Is earnings quality associated with corporate social responsibility? The Accounting Review,8, 761796.
Lang, L., & Litzenberger, R. (1989). Dividend announcements, cash flow signaling vs. free cash flow hypothesis. Journal of Financial Eco-
nomics,24, 181192.
Lie, E., & Li, W. (2006). Dividend changes and catering incentives. Journal of Financial Economics,80, 293308.
Lin, K. J., Tan, J., Zhao, L., & Karim, K. (2015). In the name of charity: Political connections and strategic corporate social responsibility in a
transition economy. Journal of Corporate Finance,32, 327346.
Lyon, J., Barber, B., & Tsai, C. (1999). Improved methods for tests of long-run abnormal returns. Journal of Finance,54, 165201.
Mikhail, M. B., Walther, B. R., & Willis, R. H. (2003). Reactions to dividend changes conditional on earnings quality. Journal of Accounting,
Auditing and Finance,18, 121152.
Nelling, E., & Webb, E. (2009). Corporate social responsibility and financial performance: The virtuous circlerevisited. Review of Quantitative
Finance and Accounting,32, 197209.
Ng, A., & Rezaee, Z. (2015). Business sustainability performance and cost of equity capital. Journal of Corporate Finance,56, 21112113.
Nissim, D., & Ziv, A. (2001). Dividend changes and future profitability. Journal of Finance,56, 21112133.
Oikonomou, I., Brooks, C., & Pavelin, S. (2014). The effects of corporate social performance on the cost of corporate debt and credit Ratings.
Financial Review,49,4975.
Patell, J. (1976). Corporate forecasts of earnings per share and stock price behavior: Empirical tests. Journal of Accounting Research,14, 246276.
Petrovits, C. (2006). Corporate-sponsored foundations and earnings management. Journal of Accounting and Economics,41, 335361.
Rakotomavo, M. (2012). Corporate investment in social responsibility versus dividends? Social Responsibility Journal,8, 199207.
Shleifer, A., & Vishny, R. (1986). Large shareholders and corporate control. Journal of Political Economy,94, 461488.
Surroca, J., & Tribo, J. (2008). Managerial entrenchment and corporate social performance. Journal of Business Finance and Accounting,35,
748789.
Yoon, P., & Starks, L. (1995). Signaling, investment opportunities, and dividend announcements. Review of Financial Studies,8, 9951018.
Zang, A. (2012). Evidence on the trade-off between real activities manipulation and accrual-based earnings management. Accounting Review,87,
675703.
How to cite this article: Glegg C, Harris O, Ngo T. Corporate social responsibility and the wealth gains from
dividend increases. Rev Financ Econ. 2018;36:149166. https://doi.org/10.1016/j.rfe.2017.07.002
166
|
GLEGG ET AL.

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT