Do dividends signal information about future earnings?
Khaled Hussainey
Department of Accounting and Finance
FK9 4LAScotland
Abstract
This paper aims to examine the extent to which dividends signal information about future earnings. It examines the effects of dividends payment status and the change in dividends payment on the relation between current stock returns and future earnings.
The results show that dividends contain valuerelevant information about future earnings and that this information is incorporated into current stock returns. In particular, we find that firms that pay dividends at the current year exhibit higher levels of share price anticipation of earnings than nondividend paying firms. We also find that the change in dividends is positively related to changes in the importance of future earnings news for current stock returns. Finally, we find that the relevance of dividends for forecasting future earnings is greater for lossmaking firms than for profitable firms.
Introduction
In a recent paper, Hanlon et al (2008) examine the impact of dividend payment status on the stock market’s ability to anticipate future earnings for a sample of US firms. They test the association between current year stock returns and future earnings for dividend paying and nondividend paying firms. They find that dividend paying firms exhibit significantly higher levels of share price anticipation of earnings than nondividend paying firms. The present paper contributes to the literature in three crucial issues. First, it examines the association between dividends and share price anticipation of earnings for UK firms. Second, it examines the extent to which the change in dividends signals relevant information for the stock market about future earnings. Finally, it tests the extent to which the associations between dividends and share price anticipation of earnings differ between lossmaking and profitable firms.
Methodology
We use the Collins et al. (1994) returnsfuture earnings regression model. However, we include only two future earnings growth variables in our regression (N = 2 and k = 1, 2) rather than three future years. In defining the earnings growth variable, we deflate earnings change by price and not by lagged earnings. The latter adjustment is made to preserve a maximum number of observations for our tests (Hussainey et al., 2003). These adjustments yields the following modified model:
Where:
b_{0} : Intercept;
b_{1}–b_{8}: Coefficient of slope parameters;
µ : Error term.
R_{t}_{+1} : Stock return for period t+1
R_{t}_{+2} : Stock return for period t+2
X_{t} : Earnings change per share in period t deflated by the share price four months after the end of the financial year t1
X_{t}_{+1} : Earnings change per share in period t+1 deflated by the share price four months after the end of the financial year t1
X_{t}_{+2} : Earnings change per share in period t+2 deflated by the share price four months after the end of the financial year t1
EP_{t}_{–1 }: Earnings yield is defined as Earnings per share for period t1divided by share price four months after the end of the financial year t1
AG_{t} : Total Assets growth for period t
Further, we expand the above model by including a dividends dummy variable (Div) to examine its potential value to investors. We interact all righthand side variables with this dummy (1= firms that pay dividends or firms that increase dividends; 0 otherwise). Interacting these variables with dividends dummy produces Equation 2 (our main regression model):
Data
Schleicher et al. (2007) examine the association between disclosure and share price anticipation of earnings. Their sample size was 4568 firmsyears for the period 1996 – 2002. The present study uses the same sample used by Schleicher et al. (2007) to examine the effect of dividends on share price anticipation of earnings. However, the number of firms is reduced further due to missing dividends data.
Accounting and return data are collected from Datastream. Earnings measure is operating income before all exceptional items (Worldscope item 01250). Earnings per share is calculated by dividing item 01250 by the outstanding number of shares. Returns are calculated as buyandhold returns (inclusive of dividends) over a 12month period from eight months before the financial yearend to four months after the financial yearend. Earnings yield, , is defined as period t–1’s earnings over share price four months after the financial yearend of period t–1. is the growth rate of book value of total assets (Worldscope item 02999) for period t. Dividends measure is dividends per share (Worldscope item 05101).
Empirical Results
We examine the effect of dividend payment status/ the change in dividends on share price anticipation of earnings. Tables 1 and 2 report the results. Column 2 (3) reports the results of lossmaking (profitable) firms. Column 4 reports the results for the whole sample.
For the whole sample, consistent with prior studies, Table 1 shows that the coefficient on is positive and significant at the 1 percent level. Additionally, there is evidence of share price anticipation of earnings one year ahead for firms that did not pay dividends at the current year. The coefficient on is positive with a pvalue of 0.030. The incremental predictive value of dividends for anticipating future earnings is given by the estimate coefficients on and . These coefficients are positive and significant at the 1 and 10 percent levels, respectively. These results indicate that current stock returns incorporate future earnings information much more strongly for dividend paying firms than nondividend paying firms.
Table 1 reveals a number of significant differences between lossmaking and profitable firms:
First: the current earnings variable exhibits a higher earnings response coefficient for profitable firms than lossmaking firms. The coefficient on is 1.95 with a pvalue of 0.001 for profitable firms, while it is insignificantly negative for lossmaking firms. This is consistent with Hayn (1995) who finds that the strength of the relation between annual stock returns and sameperiod earnings changes is lower for lossmaking firms than for profitable firms.
Second: there is no evidence of share price anticipation of earnings for lossmaking firms that pay no dividends. For these firms, the coefficients on and are negative. This indicates that the stock market is unable to anticipate future earnings changes for lossmaking firms that pay no dividends at the current year. In contrast there is strong evidence that profitable firms that do not pay dividends do exhibit share price anticipation for one year ahead. The coefficient on is positive and significant at the one percent level.
Third: for the effect of dividends payment status on prices leading earnings, we find that the coefficient on for lossmaking firms is 1.03 with a pvalue of 0.024. This coefficient indicates that paying dividends at the current years improve the stock market’s ability to anticipate future earnings one year ahead for lossmaking firms. In contrast there is no significant effect of dividend payment status on share price anticipation of earnings for profitable firms. The coefficient on , for profitable firms is positive, but insignificant.
Overall our evidence for profitable firms suggests that the market is able to anticipate future earnings changes one year ahead, but this ability is neither related to nor enhanced by paying dividends at the current year. The evidence for lossmaking firms supports the view that the market has particular difficulties in predicting such firms’ future earnings changes, but that this difficulty is partially overcome by the paying dividends at the current year.
We also statistically test the extent to which the association between share price anticipation of earnings and dividends payment status is significantly stronger for lossmaking firms than for profitable firms. We undertake this test by including all firms in one dataset and creating a dummy variable to be equal 1 for lossmaking firms and zero otherwise. We interact the loss dummy variable throughout the model. This analysis shows that the coefficient estimate on is positive and significant at the 1 percent level (not reported in the table). This suggests that that the strength of the degree of association between share price anticipation of earnings and dividends payment status is stronger for lossmaking firms than for profitable firms.
Insert Table 1 here
Similarly we test the effect of the change in dividends on share price anticipation of earnings. We also examine the extent to which the association between the change in dividends and share price anticipation of earnings varies between lossmaking firms and profitable firms. Table 2 reports the results.
The results in Table 2 are quite similar to those reported in Table 1. In particular, we find evidence of share price anticipation of earnings two year ahead for the whole sample. We also find that increasing dividend levels over time is positively association with high levels of share price anticipation of earnings.
For lossmaking firms, we find that the stock market has difficulties in predicting such firms’ future earnings changes, but that this difficulty is overcome by the increasing dividends levels over time. The coefficient on for lossmaking firms is 1.53 and statistically significant. For profitable firms, on the other hand, we find that the market is able to anticipate future earnings changes two years ahead, but this ability is neither related to nor improved by increasing dividends over time. The coefficient on for profitable firms is positive, but insignificant. Finally, we statistically test the actual differences between profitable firms and lossmaking firms. We find that the strength of the degree of association between share price anticipation of earnings and dividends payment status is statistically stronger for lossmaking firms than for profitable firms at the 1 percent level (not reported in the table).
Insert Table 2 here
Conclusion
The paper has examined two important research issues. The first of these issues is that the effects of dividend payment status and the increase in dividends over time on the relation between current stock returns and future earnings. The second of these issues is that the extent to which the consequences of dividend payment status and the change in dividends levels on share price anticipation of earnings varies between profitable and lossmaking firms. The paper adds to the literature by providing evidence that firms that pay dividends or that increase dividends levels exhibit higher levels of share price anticipation of earnings than nondividend paying (nonincreasers) firms. It also provides evidence that the association of dividend with share price anticipation of earnings is statistically significant for lossmaking firms and insignificant for profitable firms. Our findings suggest that lossmaking firms use dividends as to signal their future prospects to the stock market participants.
References
Collins, D. W., Kothari, S. P., Shanken, J. and Sloan, R. G. (1994) Lack of timeliness and noise as explanations for the low contemporaneous returnearnings association, Journal of Accounting and Economics, 18 (3): 289–324.
Hanlon M., Myers, J., and Shevlin, T. (2008) Are dividends informative about future earnings?, Working paper, University of Michigan . http://papers.ssrn.com/sol3/papers.cfm?abstract_id=964279; 18 September 2008 .
Hayn, C. (1995) The information content of losses, Journal of Accounting and Economics, 20 (2): 125–53.
Hussainey, K., Schleicher, T. and Walker , M. (2003) Undertaking largescale disclosure studies when AIMRFAF ratings are not available: the case of prices leading earnings, Accounting and Business Research, 33 (4): 275–294.
Schleicher, T., Hussainey, K. and Walker , M. (2007) Loss firms' annual report narratives and share price anticipation of earnings, The British Accounting Review, 39 (2), 153–171.
Table 1: Dividend Payers vs. Non Dividend Payers
Independent
Variable

Lossmaking firms

Profitable firms

Full Sample

Intercept

–0.12***
(0.002)

–0.02
(0.666)

–0.03
(0.184)

X_{t}

–0.30
(0.273)

1.95***
(0.001)

1.44***
(0.000)

X_{t}_{+1}

–0.57*
(0.070)

1.32***
(0.003)

0.48**
(0.030)

X_{t}_{+2}

–0.90**
(0.011)

0.30
(0.319)

0.12
(0.620)

R_{t}_{+1}

0.01
(0.895)

–0.05
(0.445)

0.03
(0.413)

R_{t}_{+2}

–0.15***
(0.001)

–0.15***
(0.009)

–0.16
(0.001)

AG_{t}

0.07**
(0.045)

0.15*
(0.092)

0.10***
(0.005)

EP_{t}_{–1}

–0.69***
(0.005)

0.99**
(0.018)

0.86***
(0.001)

Div

–0.03
(0.573)

–0.03
(0.652)

0.00
(0.968)

Div*X_{t}

0.23
(0.589)

1.03**
(0.024)

1.10***
(0.001)

Div*X_{t}_{+1}

1.03**
(0.024)

0.37
(0.418)

1.08***
(0.001)

Div*X_{t}_{+2}

0.70
(0.120)

0.39
(0.224)

0.45*
(0.084)

Div*R_{t}_{+1}

–0.01
(0.848)

–0.02
(0.798)

–0.08*
(0.066)

Div*R_{t}_{+2}

0.21***
(0.000)

0.05
(0.407)

0.07**
(0.018)

Div*AG_{t}

0.07
(0.297)

0.02
(0.835)

0.03
(0.408)

Div* EP_{t}_{–1}

–0.16
(0.703)

0.13
(0.756)

0.17
(0.431)

Observations

761

3766

4550

Full model Adj. R^{2}

0.0698

0.1971

0.1799

The significance levels (twotail test) are: * = 10 percent, ** = 5 percent, *** = 1 percent.
Table 2. Dividend Increasers vs. Dividend NonIncreasers
Independent
Variable

Lossmaking firms

Profitable firms

Full Sample

Intercept

–0.13***
(0.001)

–0.14***
(0.001)

–0.08***
(0.001)

X_{t}

–0.21
(0.377)

2.43***
(0.001)

1.57***
(0.001)

X_{t}_{+1}

–0.38
(0.146)

1.49***
(0.001)

0.84***
(0.001)

X_{t}_{+2}

–0.68**
(0.025)

0.46***
(0.005)

0.34**
(0.030)

R_{t}_{+1}

0.02
(0.694)

–0.06*
(0.062)

0.01
(0.836)

R_{t}_{+2}

–0.13***
(0.001)

–0.13***
(0.001)

–0.14***
(0.001)

AG_{t}

0.12***
(0.008)

0.18***
(0.001)

0.13***
(0.001)

EP_{t}_{–1}

–0.59***
(0.006)

1.31***
(0.001)

0.82***
(0.001)

Div

0.03
(0.749)

0.14***
(0.001)

0.10***
(0.001)

Div*X_{t}

–0.24
(0.599)

0.55*
(0.097)

0.97***
(0.001)

Div*X_{t}_{+1}

1.35**
(0.013)

0.17
(0.542)

0.73***
(0.001)

Div*X_{t}_{+2}

–0.10
(0.876)

0.27
(0.219)

0.17
(0.437)

Div*R_{t}_{+1}

–0.08
(0.760)

0.00
(0.996)

–0.06
(0.136)

Div*R_{t}_{+2}

0.16*
(0.094)

0.04
(0.234)

0.06**
(0.043)

Div*AG_{t}

–0.05
(0.462)

–0.04
(0.500)

–0.02
(0.559)

Div* EP_{t}_{–1}

–0.65
(0.311)

–0.42*
(0.081)

0.08
(0.662)

Observations

745

3728

4490

Full model Adj. R^{2}

0.068

0.203

0.184

The significance levels (twotail test) are: * = 10 percent, ** = 5 percent, *** = 1 percent.
ليست هناك تعليقات:
إرسال تعليق