Do Dividends Convey Information About Future Earnings? Charles Ham Assistant Professor Washington University in St. Louis

Transcription

1 Do Dividends Convey Information About Future Earnings? Charles Ham Assistant Professor Washington University in St. Louis Zachary Kaplan Assistant Professor Washington University in St. Louis Mark Leary Associate Professor Washington University in St. Louis We appreciate helpful comments from Alon Kalay (discussant), Roni Michaely, and participants at the University of Utah, University of Kansas, Washington University in St. Louis, and the 2017 FARS Midyear Meeting. We thank James Bosnick for excellent research assistance. Any remaining errors or omissions are ours.

2 Do Dividends Convey Information About Future Earnings? Abstract Yes. In contrast to the current consensus in the literature, we find that dividend changes predict future unexpected earnings changes for at least three years after the dividend change. This result is robust to various measures of expected earnings, including matching, analyst forecasts, and a flexible function of past earnings and returns. We find dividend changes convey more information when information uncertainty is higher. The information content of dividend changes is larger and more persistent than that of share buybacks, suggesting dividend paying firms match their information about future earnings to the duration of their payout choices. Our findings suggest managers make payout decisions using private information about future earnings, which has implications for our understanding of managerial incentives to return cash to investors.

3 1. Introduction Economists have long debated why dividend decisions matter for firms valuations. Managers and investors both seem to believe that dividend announcements contain valuerelevant information. 1 But the nature of the information conveyed by these announcements remains unclear. Miller and Modigliani (1961) recognized the tension between the sharp market reactions to dividend announcements and their dividend irrelevance theorem. They propose that investors interpret dividend changes as reflecting a change in management s views of future profit prospects for the firm (p. 430). A substantial theoretical literature has built on this information content view (Bhattacharya 1979; Miller and Rock 1985). However, the empirical evidence has largely challenged one of its central predictions, that dividend changes should provide incremental information about future realized earnings. For example, in recent prominent review articles, DeAngelo, DeAngelo and Skinner (2009) state Researchers have struggled to find evidence that dividend increases are reliable signals of future earnings increases (p. 185), while Allen and Michaely (2003) conclude that the overall accumulated evidence does not support the assertion that dividend changes convey information about future earnings (p. 73). In this paper, we re-examine the earnings information content of dividends and find strong support. We depart from the prior literature by defining more precisely the timing of dividend declarations relative to subsequent earnings announcements. In particular, we use an event window approach where we compare earnings announced prior to the dividend declaration to those announced after the declaration. This approach assumes that any earnings announced after 1 Brav et al. (2005) report that 80% of CFOs agree that dividends convey information to investors. Numerous studies document significant stock price reactions to announcements of dividend changes (e.g., Pettit 1972; Aharony and Swary 1980). 1

4 the dividend change were at least partially unknown to investors at the time of the dividend declaration. We find that dividend changes predict changes in unexpected earnings in the same direction for each quarter of the first year, as well as the first, second, and third year after the dividend change. These findings are robust to controlling for expected earnings using a flexible function of past earnings levels, earnings changes and stock returns as well as propensity score matching. We find the quarters immediately after the dividend change have larger increases in earnings than more distant quarters. This highlights the importance of defining precisely the timing of dividend changes relative to earnings reports. It also suggests that dividend changes are associated with both permanent and transitory future earnings changes. Since dividend changes are generally thought to represent a long-horizon commitment to pay out cash, it is surprising that they are associated with transitory earnings changes. We show that this is largely due to the accounting timing of certain types of expenses. Under the accounting system, many costs are matched to revenues and therefore expensed when the firm recognizes revenue. Other expenses, such as research and development, advertising, and administrative costs are expensed as incurred, even though the firm will often realize the benefit from these expenses in other accounting periods (Banker et al. 2011; Enache and Srinivasta 2016). We decompose earnings changes into two components: i) gross profit (revenue minus matched expenses (Novy-Marx 2013)), and ii) other period expenses. We find that the information content of dividend changes for gross profit is highly persistent. However, dividend changes also have a positive association with other period expenses, which increases with horizon. Firms alter period expenses in response to the same earnings news that prompted the dividend change, but do so with a delay. This provides a potential explanation for why dividends have a greater 2

5 association with short-horizon than long-horizon earnings changes. 2 As an additional robustness test, we use analyst forecasts as a proxy for earnings expectations and find dividend changes predict errors in pre-dividend forecasts. We also provide evidence investors update their expectations of earnings in response to dividend announcements, so that market reactions reflect information about future earnings. First, we document a significant positive association between abnormal returns in the three-day window surrounding a dividend change announcement and earnings changes over the subsequent four quarters. Dividend announcement returns have a significantly stronger association with future earnings relative to counter-factual return windows, consistent with investors impounding information about future earnings into price in response to dividend changes. Second, we document post-dividend analyst revisions have a strong association with dividend changes. Third, we document using both realized earnings and analyst forecasts, that negative dividend changes have larger information content. As a result, the information content theory can explain the asymmetric market reactions to dividend decreases and increases (Aharony and Swary 1980). Having established that dividends have information content, we address two questions to further our understanding of how payout policy shapes the information environment. First, we examine if dividends convey more information when investors have less precise information about future cash flows. Using several proxies for information asymmetry, we find that the magnitude of market reactions to dividend changes increases with information asymmetry. Given the lack of prior empirical support for the information content hypothesis, the literature has often focused on the agency costs of free cash flow (Jensen 1986) to explain the market reaction to dividend changes. Interestingly, we find no evidence that the market reaction to 2 Dividend changes have information content for three years, so the change in period expenses is not sufficiently large to consume all the change in gross profit. 3

6 dividend announcements is positively related to proxies for the severity of the free cash flow problem. While these explanations are not mutually exclusive, our evidence suggests that market reactions are more consistent with the earnings information content of dividends. We also find dividends are more informative about future earnings when there is more information asymmetry, consistent with either less precise earnings expectations or a stronger correlation between payout choices and managers earnings information in the context of greater information asymmetry. Second, we explore the interaction between the firm s choice of the form of payout and its information content, to understand whether managers substitute between dividends and buybacks using their information about future earnings. Prior literature has argued that firms use repurchases to pay out temporary cash flows, while dividends are perceived to be a longer term commitment (Brav et al. 2005; Skinner 2008). If dividends convey information about future earnings, we would expect this information content to be both larger and more persistent for dividends than for share repurchases. We find several important contrasts between repurchases and dividends that support this conjecture. First, while repurchases contain information about future earnings, this predictability does not persist beyond the first year. Second, among firms that pay out using both dividends and repurchases, the information content of repurchases is much smaller than that of dividends and only marginally significant. Our findings contrast with the current consensus that there is little empirical support for the information content hypothesis for dividends or buybacks (see, for example, reviews by Allen and Michaely (2003) and DeAngelo et al. (2009)). Most of these studies use a fiscal year approach, which groups earnings and dividend changes into fiscal years and then examines changes across fiscal years. This approach has a tendency to categorize earnings announced 4

7 after the dividend change but before the end of the fiscal year as current or past earnings realizations (Watts 1973; Benartzi et al. 1997). We agree with this literature in that we find the strongest correlation between dividend changes and earnings changes right around the time of the dividend change. However, on the pivotal question as to whether dividend changes convey information about future earnings, we obtain substantially different results when utilizing an event window approach, which ensures that all earnings announcements made after a dividend announcement are considered future earnings. Because investors will update their expectations of any earnings realization that has not yet been announced, we argue researchers should use an empirical measure of future which includes all unannounced earnings realizations. Our study also relates to the literature that seeks to understand the market s reaction to dividend announcements and managers incentives to pay dividends or repurchase shares (e.g., Pettit 1972; Lang and Litzenberger 1989; Howe et al. 1992; Grullon et al. 2002; Grullon and Michaely 2002). We do not provide direct evidence that managers consciously bear dead-weight costs to signal information to investors (Miller and Rock 1985; Bhattacharya 1979). However, we do show that dividends contain significant predictive content for future earnings up to three years out, that investors and analysts correctly interpret the information content of dividend announcements, and that this information differs across forms of payout. Because managers likely take the informational effects of payout into account when setting payout policies, our findings have implications for theories of payout. We conclude by discussing these implications. 2. Literature review and hypothesis development 5

8 In this section, we first briefly review the theoretical literature on dividend signaling and the information content view of dividends. We then review the prior empirical literature testing whether dividend declarations predict future earnings changes Theory and hypotheses The seminal dividend irrelevance model of Miller and Modigliani (1961) assumes, among other things, that managers and outside investors have the same information about future investment and cash flows. However, differences in information between insiders and investors are likely to be prevalent in financial markets in practice. Indeed, Miller and Modigliani (1961) recognize that such informational differences are likely to influence investors reactions to dividend changes. 3 As discussed by Miller and Rock (1985), this information content can arise simply through the sources and uses of cash identity. That is, if the firm s investment policy is known or inferred by investors, then the dividend, net of any capital raised, allows investors to back out the (unobserved) earnings. However, Miller and Rock also point out that this version of the information content hypothesis is only sustainable in equilibrium under restrictive assumptions. In particular, if managers have any incentive to boost the current stock price, rather than solely maximizing fundamental value, they might lower investment or take other measures to increase the current dividend. This possibility for manipulation, in turn, would undermine the information content of dividends if investors are rational. 3 We note that other (non-mutually exclusive) explanations have been offered for the price reaction to dividend changes. Following Easterbrook (1984) and Jensen (1986), higher dividends may reduce the free cash flow subject to managerial discretion, thereby increasing the fraction of future earnings captured by investors. Alternatively, Grullon et al. (2002) suggest that dividend increases reflect a reduction in risk, and therefore a lower discount rate, as firms mature. Given our focus on the earnings information content of dividends, we refer the reader to excellent reviews by Allen and Michaely (2003), DeAngelo et al. (2009), and Kalay and Lemmon (2011) for fuller treatments of these alternate views. 6

9 One way in which information content can be restored in equilibrium is if increasing the dividend is costly enough to discourage manipulation by firms whose future earnings prospects don t warrant the increase. This idea has been formalized in a number of dividend signaling models, which differ primarily in the assumed cost of paying a high dividend. 4 However, in all of these models, the dividend announcement allows investors to infer managers private information about current and future profitability. Dividend signaling models, or the information content hypothesis more generally, have several testable implications. First, if dividend decisions are a function of managers private information about current and future earnings, dividend increases (decreases) should be associated with subsequent increases (decreases) in earnings realizations. Second, if investors recognize the earnings news reflected in dividend announcements, dividend changes should be greeted by price changes in the same direction. Related, investors should update their expectations about future earnings following announced dividend changes Empirical tests of the information content of dividends A lengthy literature tests whether dividends contain information about unexpected future earnings changes. While a few studies support the information content view of dividends (Ofer and Siegel 1987; Aharony and Dotan 1994; Yoon and Starks 1995; Nissim and Ziv 2001), most large sample empirical studies argue dividend changes contain little or no information about future earnings (Watts 1973; Gonedes 1978; Penman 1983; Lang and Litzenberg 1989; DeAngelo et al. 1996; Benartzi et al. 1997; Grullon et al. 2002; Grullon et al. 2005). 5 Recent 4 Potential costs include financing costs (Bhattacharya 1979), reduced investment (Miller and Rock 1985), and taxes (John and Williams 1985) 5 We exclude studies from our review that use a small subset of dividend paying stocks, such as Brickley (1983), which studies earnings changes for thirty-five firms that change their dividend. We also exclude studies examining dividend omissions and dividend initiations. 7

10 review papers (DeAngelo et al. 2009; Allen and Michaely 2003) characterize this latter view as the current consensus. Our review of the empirical literature suggests one research design choice has a dramatic influence on the probability a study will confirm or contradict the information content hypothesis. The pivotal choice is whether the study computes earnings changes using an event window approach or over fiscal years. In the event window methodology, earnings (or earnings expectations) immediately following the dividend declaration are compared to earnings prior to the dividend declaration. In the fiscal year methodology, dividend changes are aggregated over a fiscal year. These studies then compare earnings changes in the fiscal year following the dividend declaration to earnings changes in the year in which the firm declared the dividend change. Almost all of the studies employing the fiscal year approach do not support the information content hypothesis. 6 Perhaps the most compehensive of these studies is Benartzi et al. (1997). Using both regression and matched-sample approaches, the authors show that dividend changes are highly correlated with earnings in the current or past fiscal years. However, dividend increases are uncorrelated with earnings growth in the subsequent fiscal years, while dividend cuts are actually followed by earnings increases. Nissim and Ziv (2001) argue that controlling for mean reversion in earnings produces results more consistent with the information content of dividends. They perform similar tests as Benartzi et al. (1997), but add lagged return on equity and lagged earnings changes to control for changes in earnings predicted by financial statement variables. However, Grullon et al. (2005) argue these findings are highly sensitive to the manner of 6 Specifically, seven studies find no information content (Watts 1973; Gonedes 1978; Penman 1983; DeAngelo et al. 1996; Benartzi et al. 1997; Grullon et al. 2002; Grullon et al. 2005), while only Nissim and Ziv (2001) finds supportive evidence. Several of these studies consider dividend changes in the first quarter of the subsequent fiscal year as part of the prior fiscal year s earnings. 8

11 controlling for mean reversion and demonstrate that controlling for non-linearity in mean reversion restores the conclusions of Benartzi et al. (1997). The only study in this set, of which we are aware, with results that support the information content hypothesis and have not been challenged is Aharony and Dotan (1994). Not coincidentally, this study uses an event window methodology. However, this study shows positive information content for only two quarters after the dividend change and negative information content in the fourth quarter. Such short-term information content is hard to reconcile with the perceived long-term commitment of a dividend change (Lintner 1956; Brav et al. 2005). We expand on the methodology in Aharony and Dotan (1994) by including extensive controls for pre-dividend declaration earnings and returns, which allows us to isolate the unexpected information content in the dividend change and delivers consistent results across the earnings horizon. We also make several additional contributions. First, we highlight the source of discrepancy between their findings and the bulk of the related literature. Second, we demonstrate how dividend announcements affect investor expectations by linking the information content of dividends to analyst forecast revisions and announcement period returns. Third, we use cross-sectional tests to document that the information content of dividends is more pronounced in settings of greater information asymmetry and also contrast the information content of dividend and repurchase announcements. In contrast to studies examining information content using actual earnings changes where the fiscal year approach is the norm, the three studies of which we are aware using analyst forecasts all use an event study methodology. Two of the three studies find significant information content (Ofer and Siegel 1987; Yoon and Starks 1995) while one does not (Lang and 9

12 Litzenberg 1989). However, all these studies use summary files, which offer only approximate information about the timing of forecast revisions. As a result, these studies cannot rule out the possibility that the revision was driven by (i) a concurrent earnings release, or (ii) information released before the dividend declaration (Allen and Michaely 2003). By using the I/B/E/S detail file, we are able to ensure that we compare only forecasts made after the previous earnings release but before the dividend change to forecasts made between the dividend change and the next earnings release. Further, we remove the impact of analyst-specific biases by only including forecasts made by the same analyst both before and after the dividend change. 3. Sample selection and descriptive statistics We obtain data on dividend declarations from the CRSP events database. We first select all ordinary quarterly dividend declarations (distribution code 1232) over the period for which the firm made a previous quarterly dividend declaration in the past 180 days. 7 This allows us to compute the percentage dividend change. We limit the sample to: (i) firms listed on the NYSE, AMEX, or Nasdaq exchanges, (ii) ordinary common stocks (i.e., those with share code 10 or 11), and (iii) non-financial firms (we exclude firms with a four digit SIC beginning with six). We also exclude: (i) dividend declarations for which the firm declared a distribution other than a quarterly dividend between the declaration dates of the current and prior quarterly dividends, to focus our analysis on the information content of quarterly dividends (Benartzi et al. 1997; Nissim and Ziv 2001), and (ii) firms that split their shares between the month of the prior dividend declaration and the month of the current dividend declaration, as splits affect the information content of dividend changes (Nayak and Prabhala 2001). We require data on CRSP to compute dividend declaration announcement returns and past returns. We obtain earnings 7 The first year earnings announcements were available on Compustat is

13 data from the CRSP/Compustat Merged database and require twelve consecutive quarters of earnings to calculate seasonal earnings changes around the dividend declaration (i.e., four earnings changes before and after the declaration). We winsorize all continuous variables at the top and bottom one percent to mitigate the influence of outliers, except the percentage dividend change for which we set all dividend increases larger than 200% to 200%. 8 Table 1 presents descriptive statistics for our sample. As shown in Panel A, 85% of dividend declarations maintain the prior dividend level, while 14% (1%) increase (decrease) the dividend. Although dividend decreases are less frequent, they tend to be larger. The average decrease reduces the dividend by 49.8% (Panel C) while the average increase raises the dividend by 19.6% (Panel D). The average decrease has an announcement window return of -2.6%, compared to 0.8% for the average increase, suggesting a greater reaction to dividend decreases. Declarations that change the dividend tend to be preceded by returns of the same sign as the dividend change, suggesting at least some of the information affecting the decision to change the dividend was released to the market before the dividend declaration. Examining earnings changes, we find positive (negative) earnings changes for firms that increase (decrease) the dividend in the four quarters before and after the dividend declaration (except for the fourth quarter ahead for dividend decreases). The goal of our first set of empirical tests is to identify the portion of the post-dividend declaration earnings change that is unexpected. 4. Do dividend changes predict future earnings changes? 8 Several dividend increase observations are extremely large in percentage terms. To mitigate their influence we winsorize the percentage dividend change at +200%. We do not winsorize dividend decreases because they are bounded at -100% and dividend decrease observations comprise just over 1% of the sample. We winsorize all variables involving earnings at the top and bottom one percent for two reasons: (i) the distribution of changes in earnings values is highly kurtotic and skewed so extreme values account for much of the variance in earnings changes (Grammery and Gerakos 2014), and (ii) large changes in accounting income have little relation with economic income (Freemen and Tse 1992). All standard errors are clustered by year of the dividend declaration. 11

14 In this section, we test whether dividend changes have information content about future earnings via three distinct approaches. First, we estimate the relation between dividend changes and future earnings changes in a regression approach where we include variables suggested by the extant literature to control for expected earnings changes in the absence of any dividend change. We also use a matched sample approach that compares earnings changes for dividend changers to non-dividend changers with similar characteristics. Second, we use analyst forecasts of earnings to examine whether dividend changes predict forecast errors and forecast revisions. Third, we infer changes in earnings expectations from market prices and examine whether stock returns on dividend declaration days contain information about future earnings A regression approach to testing for information content In our first approach, we regress earnings changes on the percentage dividend change (ΔDIV) and a series of control variables. E it+n = β 0 + β 1 DIV it + β j Controls + ε (1) E is the change in earnings using income before extraordinary items (IBQ) from the CRSP/Compustat merged quarterly file. All earnings changes are computed as the difference between earnings announced after the dividend change and earnings for the same period in the prior year (before the dividend change) and scaled by the market value of equity the quarter before the dividend announcement, similar to Benartzi et al. (1997). We calculate earnings changes over five different horizons: one, two, three, and four quarters ahead, as well as one year ahead which is the sum of the four quarterly changes after the dividend announcement. Refer to Figure 1 for a visual depiction of the earnings change calculations. If a dividend declaration occurs the day of an earnings announcement, we classify the earnings announced at the time of 12

15 the dividend change as the prior quarter s earnings. We cluster all standard errors by the year of the dividend declaration unless otherwise noted. We present the results from estimating equation (1) in Table 2. The dependent variable in column (1) is the annual change in earnings ( E (y+1) ), calculated as the sum of the four quarterly earnings values following the dividend change minus the sum of the four quarterly earnings values before the dividend change. We find a highly significant coefficient on the dividend change (β=0.028; t=4.2), which suggests a 50% increase in the dividend corresponds to an increase in earnings equal to 1.4% of the market value of equity of the firm. In column (2), we control for the earnings changes that would have been expected in the absence of the dividend change by including the four past quarterly earnings changes and four past earnings levels as independent variables. The magnitude and significance of the estimated coefficient on dividend change is unaffected. In column (3), we add non-linear functions of the annual earnings change and level (Fama and French 2000; Grullon et al. 2005), as well as five variables capturing returns over the 240 trading days before the dividend announcement (Ball and Brown 1968), to more fully control for variation in expected earnings changes. 9 We continue to find a highly significant coefficient on the dividend change (β=0.024; t=4.1). Prior literature suggests some results may be sensitive to the choice of deflator (Nissim and Ziv 2001). To examine whether our results are invariant to the choice of deflator, in columns (4) (6), we use three different deflators: the market value of equity five quarters before the dividend declaration in column (4), the book value of common equity five quarters 9 The past earnings level (earnings change) is the sum of the four quarterly earnings levels (earnings changes) before the dividend announcement. Specifically, we include a total of six variables, three each for the earnings change and level: (i) an interaction between the variable and an indicator equal to one if the variable is negative, (ii) an interaction between a positive indicator and the variable squared, and (iii) an interaction between a negative indicator and the variable squared. We exclude the main effect because it will be multi-colinear with our four quarterly earnings change and levels variables. In unreported analysis, we find our coefficient estimates and significance levels are unchanged by including non-linear controls for each quarterly change and level. 13

16 before the dividend declaration in column (5), and the book value of common equity one quarter before the dividend declaration in column (6). Our results are unaffected. In columns (7) (10), we examine the horizon over which dividend changes correlate with earnings changes. In each column, the dependent variable is earnings from a post-dividend declaration quarter minus earnings from the same quarter in the prior fiscal year, scaled by the market value of equity the quarter before the dividend declaration. Column (7) shows a significant relation between dividend changes and one quarter ahead earnings changes. The magnitude (β=0.008; t=4.6) is about a third as large as when the dependent variable is the annual earnings change. In columns (8), (9), and (10) we examine the association between the dividend change and the earnings change two, three, and four quarters ahead, respectively. The coefficient remains significant at all horizons, but decreases monotonically with horizon: β=0.006 at two quarters ahead, β=0.004 at three quarters ahead, and β=0.003 at four quarters ahead. Our finding of a significant, positive association between dividend changes and unexpected earnings at horizons of three and four quarters contrasts with Aharony and Dotan (1994), who do not control for expected earnings changes and find negative or insignificant information content at these longer horizons Comparison with prior literature The prior literature has typically computed earnings changes over fiscal years, which can result in earnings that occur after the dividend declaration but before the fiscal year end being included in the prior year s earnings (i.e., the year of the dividend change). In other words, in some cases, earnings that are announced well after the dividend change are not considered future earnings. 14

17 To examine whether the disagreement between our findings and those of the prior literature are attributable to computing earnings changes over the fiscal year, in Table 3, we calculate earnings changes as in the prior literature earnings in the fiscal year after the dividend declaration less earnings in the fiscal year of the dividend declaration. Columns (2) and (3) control for non-linear functions of earnings levels and changes over the past fiscal year, following Grullon et al. (2005). In column (4) we include earnings levels and changes for each of the past four quarters. In all cases, we find no evidence of a positive relation between dividend changes and future earnings changes. The only significant coefficient (column (1) with no controls) has the wrong sign. Overall, these results are consistent with the empirical findings of the prior literature that when computing earnings changes over fiscal years, dividend changes appear to have minimal or no information content Information content of dividends beyond the subsequent year The results in columns (7) (10) of Table 2 appear to suggest the information content of dividends declines with horizon. This result is somewhat puzzling given that dividend changes tend to be persistent (Lintner 1956). We expect commitments to change cash outflows over a multi-year period should be associated with cash inflows over a similar duration. In Panel A of Table 4, we examine whether the relation between dividend changes and future earnings changes extends beyond the four quarters after the dividend declaration. In column (1), we examine the relation between the dividend change and the earnings change two years ahead, calculated as the difference between the sum of the earnings announced five to eight quarters after the dividend declaration less the sum of quarterly earnings from the four quarters before the dividend declaration. We use pre-dividend declaration control variables, so we do not control for any of the earnings changes realized in the first year after the dividend change. As a 15

18 result, the coefficient on ΔDIV allows us to examine the persistence of the earnings changes associated with the dividend change. We find a positive and statistically significant coefficient on the dividend change (β=0.013; t=2.6), suggesting dividends have information content for earnings for at least two years. The magnitude is slightly more than half that of the coefficient in column (3) of Table 2, suggesting some mean reversion in the earnings change. In column (2), we specifically test for mean reversion by computing the dependent variable as the difference between quarterly earnings recognized five to eight quarters after the dividend change and the first four quarters announced after the dividend change. The significantly negative coefficient estimate is consistent with mean reversion in the earnings changes associated with the dividend changes. In columns (3) (4), we conduct similar analyses, but extend the horizon to three years after the dividend change. We continue to find significant information content (column (3)), with a coefficient estimate of similar magnitude as that in column (1). In column (4), we take the difference between earnings announced nine through twelve quarters after the dividend declaration and earnings announced five through eight quarters after the dividend declaration. Unlike in column (2), we find no economically or statistically significant mean reversion as we go from the two-year to three-year horizon. Overall, our results suggest that dividend changes have a positive association with future earnings over a long horizon, though the magnitude is somewhat diminished beyond the first year. The strong association with earnings changes shortly after the dividend change can further explain why research designs which do not classify all earnings changes after the dividend change as future earnings find evidence inconsistent with information content Information content of dividends for gross profit and period expenses 16

19 Our results thus far suggest that dividends are associated with future earnings changes, but the earnings changes are both permanent and transitory. The association between transitory earnings and dividend changes is unintuitive and warrants further investigation. To better understand the cause of the association between dividend changes and transitory earnings changes, we decompose accounting income into gross profit (revenues minus cost of goods sold) and period expenses. A challenge with using accounting income to measure changes in the amount of economic income the firm generates each period is that accounting standards accelerate expenses into earnings. 10 These accelerated expenses often constitute investments, and these investments could be positively correlated with the dividend change (or the shock to economic income that prompted the manager to change the dividend). Because cost of goods sold are matched explicitly to the revenues they generate, gross profit should be largely unaffected by inter-temporal variation in investment. 11 Separately examining the mapping between dividend changes and the two components of income, gross margin and period expenses, allows us to better understand the cause of the transitory shock to earnings. In the first five columns of Panel B of Table 4, we estimate equation (1) replacing all earnings amounts in both dependent and independent variables with gross profit amounts. Columns (1) and (2) indicate that over the first year and first quarter, respectively, dividend changes have similar information content for gross profit as they do for earnings (see columns (3) and (7) of Table 2). Columns (2)-(5), though, show much less attenuation in the coefficient on dividend change over the subsequent quarters. The coefficient in the fourth quarter is 75% of 10 Two well documented reasons why accounting income differs from economic income are: (i) the accounting system requires immediate expensing of investments such as advertising and research and development, even though the firm realizes the benefits of these expenses over a period of years (Enache and Srinivasta 2016), and (ii) the accounting system requires assets to be written down when impaired (Basu 1997). 11 Novy-Marx (2013) states that gross profits is the cleanest accounting measure of true economic profitability. 17

20 the coefficient in the first quarter, suggesting the information content of dividends for gross profit is substantially more persistent than for income before extraordinary items. In columns (6)-(10), we replace all earnings variables with period expenses, computed as the difference between gross profit and income before extraordinary items (we multiply period expenses by negative one, so positive values indicate more expense). In column (6), we find at the annual horizon a statistically significant positive coefficient estimate on the dividend change (β=0.023, t=2.5). In other words, shocks that cause managers to increase payout also cause an increase in period expenses. In columns (7)-(10), we find the association between period expenses and dividend changes increases with horizon as we go from the first to the fourth quarter after the dividend change. The increase in period expenses over the horizon contributes to the attenuation of earnings information content within the fiscal year, and could plausibly be caused by increased investment rather than increases in operating expenses. 12 The positive covariance between period expenses and dividend changes is opposite to the sign predicted by the free cash flow hypothesis (Jensen 1986), which predicts payout constrains managers ability to consume perquisites (because perquisites, such as building a fancy new headquarters, higher executive salaries or private jet flights, cannot be matched to revenues, they are treated as period expenses) Matched sample analysis The regression results presented thus far use a series of lagged earnings realizations and return realizations to control for expected earnings changes. However, if the effect of past earnings changes on future earnings changes varies with the size of the firm, its industry, and/or over time, these interactions could lead us to find dividend changes predict earnings changes 12 In untabulated analysis we find a similar increase across the horizon when examining the association between dividend changes and research and development, an expense line item with an intuitive link to firm investment. 18

21 when the predictability arises because of these heterogeneous effects. To address this possibility, we match dividend change firms to similar dividend paying firms that do not change their dividend. Specifically, we estimate a propensity score model of the probability the firm will change the dividend as a function of the past four quarterly earnings changes and levels. We estimate the model separately for dividend increases and decreases. We then match each dividend increase or decrease firm to a firm that did not change the dividend. We choose the firm with the closest propensity score within the same dividend declaration year and industry (two digit SIC). Matching is performed with replacement and we impose a caliper distance of 0.01 (Shipman et al. 2016). Panel A of Table 5 and Figure 2 report the level and change in earnings from four quarters before, to four quarters after, the dividend declaration for dividend decrease firms and the matched firms that did not change the dividend. Both dividend decreasing firms and matched non-changers exhibit declining performance in the four quarters prior to the dividend declaration, though the differences between the dividend changers and non-changers are insignificant. However, the firms that decrease their dividend exhibit significantly worse performance in each of the first four quarters after the dividend declaration. This leads to lower earnings levels that persist for at least three years. In Panel B of Table 5 and Figure 2, we repeat the analysis for firms that increase the dividend. Opposite to Panel A, both the dividend increasing firms and the matched non-changers exhibit improving performance in the four quarters before the dividend declaration. The differences in earnings changes during the pre-period are again insignificant. In the four quarters after the dividend declaration, the dividend increase firms report earnings changes that are significantly more positive than those reported by the matched non-changers, again generating 19

22 persistent differences in earnings levels up to three years out. Collectively, these results support the notion that dividend changes contain information about future earnings changes, and that our prior results are not driven by size, industry, and/or time effects. 5. Do announcement returns reflect the information content of dividends? Our previous tests suggest dividend changes reflect changes in the expected value of future earnings. In our next set of tests, we examine whether investors update their expectations of future earnings in response to the dividend and whether this updating is related to the market reaction to dividend announcements. We note that many of our analyses in this section test the joint hypothesis that (i) dividend changes have information about future earnings and (ii) investors understand and react to this information Dividend declaration returns and future earnings As a first test of whether returns reflect the earnings information in dividends, we examine the relation between returns around dividend declarations and future changes in earnings. The decision to change the dividend will be affected by both previously disclosed and potentially undisclosed information about future cash flows. While earnings realizations will be a function of both types of information, if markets are reasonably efficient, market prices at the time of the dividend declaration will already reflect the information released prior to that point. In this case, the announcement return should only reflect previously undisclosed information and we can regress future earnings on declaration date returns to test whether dividends convey information about future earnings. Essentially, this regression approach attempts to isolate the new information about expected earnings changes from the dividend change by using announcement window returns. 20

23 Theories that dividends convey information about discount rate news (Grullon et al. 2002), or information about wealth transfers from debtholders to equityholders (Handjinicolaou and Kalay 1984), would predict little or no relation between announcement returns and future earnings. 13 In contrast, if dividends convey information about future earnings, we would predict a significantly positive relation. One challenge to this interpretation, however, is the empirical evidence that returns lead earnings (Ball and Brown 1968). Thus, returns over any three-day window are likely to contain some future earnings news. To test whether dividend changes convey incremental information, we compare the information content of returns over the dividend declaration window to two proxies for the (counter-factual) information content of returns in the absence of a dividend change announcement: (i) announcement returns for dividend declarations that do not change the dividend, and (ii) returns from three-day periods centered five trading days before and after the dividend change. The second approach has the advantage of holding constant the future earnings change, so we test whether dividends affect the time at which the market impounds unexpected earnings. We also include earnings controls in all of our analyses to remove expected earnings changes from the dependent variable and increase the power of the tests. E it+n = β 0 + β 1 Ret i(-1,+1) + β 2 Ret i(-1,+1) *I[ DIV 0] it + β j Controls + ε E it+n = β 0 + β 1 Ret i(-1,+1) + β 2 Ret i(-6,-4) + β 3 Ret i(+4,+6) + β j Controls + ε (2a) (2b) We report the results from estimating equation (2a) in columns (1)-(5) of Table 6. Ret (-1,+1) is the three-day return centered on the dividend declaration. We remove dividend declarations that are bundled with earnings announcements by requiring that none of the days from the three-trading-day windows centered on the dividend declaration date coincide with any 13 Or perhaps even a negative relation in the case of the maturity hypothesis (Grullon et al. 2002). 21

24 day of a three-day earnings announcement window. We interact the three-day return with an indicator variable equal to one if the firm changes the dividend (I[ DIV 0]) to determine whether announcement returns have more information content about future earnings when the firm changes the dividend. In column (1), we measure earnings changes over the four quarters after the dividend declaration minus the four quarters before the dividend declaration ( E (y+1) ). The coefficient on the announcement returns is significantly positive (β=0.084, t=7.5), suggesting dividend declaration returns predict future earnings changes. We also find a significantly positive coefficient estimate on the interaction Ret (-1,+1) *I[ DIV 0] (β=0.179, t=6.3), suggesting dividend declaration returns have a stronger association with future earnings when the dividend is changed. The magnitude of the sensitivity of earnings to returns for a dividend change is over three times that of a no-change announcement. In columns (2)-(5), we examine how the information content of returns varies with horizon. We find significantly positive coefficients on the interaction Ret (-1,+1) *I[ DIV 0] for earnings one, two, three and four quarters ahead. At each of the four quarters, announcement returns predict future earnings by at least twice as much for declarations that change the dividend, compared to declarations that leave the dividend unchanged. We report the results from estimating equation (2b) in columns (6)-(11) of Table 6. Here we restrict the sample to dividend changes only and compare the coefficient on the three-day declaration returns to the three-day returns one week before (Ret (-6,-4) ) and one week after (Ret (+4,+6) ) the dividend declaration. We again require that none of the three-trading-day windows coincide with any day of a three-day earnings announcement window. In column (6), when examining the yearly income change over the following four quarters, we find a positive statistically significant coefficient estimate on the declaration date returns (β=0.197, t=7.9). The 22

25 information content of announcement window returns is more than double that of the return window one week before the dividend declaration date and more than five times larger than returns one week after. 14 In column (7), we estimate the same model on the sample of non-dividend changers to evaluate whether dividend declarations that do not change the dividend provide information to the market. The coefficient on each of the three return variables is positive and significant. However, we note that (i) the coefficient on Ret (-1,+1) is less than half that of the coefficient in column (6), and (ii) the coefficient on Ret (-1,+1) is slightly larger, but statistically indistinguishable from the coefficients on returns during the two counter-factual windows. Overall, our tests suggest dividend declarations which do not change the dividend do not have significant incremental information content. In columns (8)-(11), we examine how the information content of dividends varies with horizon. Similar to the results described above, we find the information content of declaration date returns exceeds the information content of returns both before and after the declaration date for each of the first four quarters after the dividend change. Seven of the eight differences are statistically significant. The only insignificant difference is information content for returns before the declaration, for the four quarter ahead earnings horizon Information content of dividend decreases vs. increases Empirically, dividend cuts have a larger effect on returns than dividend increases. If the association between dividend changes and returns reflects earnings information content, we 14 One surprising feature of our results is that the information content of returns is higher before the dividend declaration than after. Prior studies findings suggest the information content of returns increases monotonically with the length of time until the next earnings announcement (Kaplan and Milian 2016). The larger coefficient on returns before the declaration is thus atypical and suggests that the information disclosed through the dividend reduces information content of subsequent returns, consistent with theory that disclosure substitutes for private information production (Diamnond 1985). 23

26 would expect dividend decreases to be more informative about future earnings than dividend increases. In Table 7, we examine whether the relation between future earnings and dividend changes varies with the sign of the dividend change via a modified version of equation (1). E it+n = β 0 + β 1 DIV it + β 2 DIV it *I[ DIV<0] it + β j Controls + ε, (3) where I[ DIV<0] is an indicator variable equal to one if DIV is negative, zero otherwise. We include both the level and change of the past four quarterly earnings changes and controls for past returns in all models. The positive estimated coefficient on the interaction between this indicator and the dividend change in column (1) confirms that there is a stronger relation between future earnings changes and dividend cuts than for dividend increases. In columns (2)-(5), we examine how the information content of positive and negative dividend changes varies with the horizon over which we compute earnings changes. Consistent with the results in Table 2, the earnings information content of both dividend decreases and increases decline with the horizon. However, the dividend decrease has an association with earnings changes at least double that of dividend increases at every horizon, with the difference statistically significant for all but quarter four. Overall, while both positive and negative dividend changes predict future earnings, the relation is stronger for negative dividend changes, consistent with the asymmetric reaction to dividend news Dividend changes and analyst forecasts We next use analyst forecasts of earnings as a benchmark for expected earnings. First, we test whether analysts respond to dividend changes by examining forecast revisions around dividend declarations. Evidence analysts update their earnings expectations in response to dividend changes suggests investors infer information content from dividend changes. Second, 24