Embed Size (px)
Citation preview
Managerial FinanceInvestor sentiment and the market reaction to dividend news: European evidenceElisabete Simões Vieira
Article information:To cite this document:Elisabete Simões Vieira, (2011),"Investor sentiment and the market reaction to dividend news: Europeanevidence", Managerial Finance, Vol. 37 Iss 12 pp. 1213 - 1245Permanent link to this document:http://dx.doi.org/10.1108/03074351111175100
Downloaded on: 20 December 2014, At: 02:09 (PT)References: this document contains references to 54 other documents.To copy this document: [email protected] fulltext of this document has been downloaded 1657 times since 2011*
Users who downloaded this article also downloaded:A.A. Lonie, G. Abeyratna, D.M. Power, C.D. Sinclair, (1996),"The stock market reaction to dividendannouncements: A UK study of complex market signals", Journal of Economic Studies, Vol. 23 Iss 1 pp.32-52 http://dx.doi.org/10.1108/01443589610106534Khaled Hussainey, Chijoke Oscar Mgbame, Aruoriwo M. Chijoke-Mgbame, (2011),"Dividend policyand share price volatility: UK evidence", The Journal of Risk Finance, Vol. 12 Iss 1 pp. 57-68 http://dx.doi.org/10.1108/15265941111100076N. Bhattacharyya, (2007),"Dividend policy: a review", Managerial Finance, Vol. 33 Iss 1 pp. 4-13 http://dx.doi.org/10.1108/03074350710715773
Access to this document was granted through an Emerald subscription provided by 304077 []
For AuthorsIf you would like to write for this, or any other Emerald publication, then please use our Emerald forAuthors service information about how to choose which publication to write for and submission guidelinesare available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.comEmerald is a global publisher linking research and practice to the benefit of society. The companymanages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well asproviding an extensive range of online products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committeeon Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archivepreservation.
*Related content and download information correct at time of download.
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
Investor sentiment and themarket reaction to dividend news:
European evidenceElisabete Simoes Vieira
GOVCOPP Unit Research, ISCA Department,University of Aveiro, Aveiro, Portugal
Abstract
Purpose – The purpose of this paper is to examine the effect of investor sentiment (ISENT) on themarket reaction to dividend change announcements.
Design/methodology/approach – The author used the European Economic Sentiment Indicatordata, from Directorate General for Economic and Financial Affairs, as a proxy for ISENT and focus onthe market reaction to dividend change announcements, using panel data methodology.
Findings – Using data from three European markets, the results indicate that ISENT has someinfluence on the market reaction to dividend change announcements, for two of the three analysedmarkets. Globally, no evidence was found of ISENT influencing the market reaction to dividend changeannouncements for the Portuguese market. However, evidence was found that the positive share pricereaction to dividend increases enlarges with sentiment, in the case of the UK markets, whereas thenegative share price reaction to dividend decreases reduces with sentiment, in the French market.
Research limitations/implications – The author had no access to dividend forecasts, so, thefindings are based on naıve dividend changes and not unexpected change dividends.
Originality/value – This paper offers some insights on the effect of ISENT on the market reaction tofirms’ news, a strand of finance that is scarcely developed and contributes to the analysis of Europeanmarkets that are in need of research. To the best of the author’s knowledge, this is the first study toanalyse the effect of ISENT on the market reaction to dividend news, in the context of European markets.
Keywords Europe, Financial markets, Dividends, Investor sentiment, Dividend news, Market reaction,Behavioural finance
Paper type Research paper
1. IntroductionBhattacharya (1979), John and Williams (1985) and Miller and Rock (1985) developed thesignalling theory classic models, showing that, in a world of asymmetric information,better informed insiders use the dividend policy as a costly signal to convey their firm’sfuture prospect to less-informed outsiders. So, a dividend increase signals an improvementon firm’s performance, while a decrease suggests a worsening of its future profitability.Consequently, a dividend increase (decrease) should be followed by an improvement(reduction) in a firm’s profitability, earnings and growth. Moreover, there should be apositive relationship between dividend changes and subsequent share price reaction.
There has been a significant number of empirical tests showing that dividend changeannouncements are positively related with share returns in the days surrounding thedividend change announcement. Pettit (1972, 1976) found evidence that dividend changeannouncements convey information to the market. Similar results were obtained byseveral authors, such as Aharony and Swary (1980), Benesh et al. (1984) and Dhillon andJohnson (1994) for dividend change announcements, Asquith and Mullins (1983) for
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/0307-4358.htm
Market reactionto dividend news
1213
Managerial FinanceVol. 37 No. 12, 2011
pp. 1213-1245q Emerald Group Publishing Limited
0307-4358DOI 10.1108/03074351111175100
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
dividend initiations, Lee and Ryan (2000, 2002) for dividend initiations and omissionsand Lippert et al. (2000) for dividend increase announcements. Although all these studieswere carried out for the USA, Travlos et al. (2001) analysed Cyprus, Gurgul et al. (2003),Austria, Harada and Nguyen (2005) Japan, McCluskey et al. (2006) Ireland and Yilmazand Gulay (2006) Turkey, finding also support for the dividend information contenthypothesis.
However, some studies have not supported this assumption. Studies by Lang andLitzenberger (1989), Benartzi et al. (1997) and Bernhardt et al. (2005) for the Americanmarket, Conroy et al. (2000) for the Japanese market, Chen et al. (2002) for the Chinesemarket and Abeyratna and Power (2002), for the UK, find no evidence of a significantrelationship between dividend announcements and share returns.
According to Miller (1986), the behavioural finance might help resolvelong-standing anomalies in the financial area of dividend policy. The behaviouralfinance introduces the investor sentiment (ISENT) in the investors’ decision-makingprocess. Some authors developed proxies of sentiment (Lemmon and Portniaguina,2006; Qiu and Welch, 2006; Baker and Wurgler, 2006, 2007) and others have exploredthe role of sentiment in financial markets (Han, 2008 and Yu and Yuan, 2010). Theconsumer confidence might be an indication about the investors’ feelings in whatconcerns the economy and financial markets.
Using a sample of three distinct European markets, Portugal, France and the UK,we try to provide further evidence on the role of the ISENT on the market reaction todividend change announcements, analysing how the market reaction to dividendchange announcements diverges with ISENT.
Our results suggest that the market reaction to dividend change announcements ismore sensitive to dividend increases when sentiment is increasing, for the UK market,and that the market reaction to dividend change announcements is less sensitive todividend decreases when sentiment is increasing, for the French market. For thePortuguese market, we find no evidence of ISENT influencing the market reaction todividend change announcements.
This study contributes to the literature in three ways. First, it offers some insights onthe effect of ISENT on the market reaction to corporate news, a strand of finance that isscarcely developed. Second, the systematic variation in the stock market reaction todividend news in periods of high or low sentiment means that empirical studies couldimprove their reliability by including the effect of market sentiment when examining themarket reaction to corporate news, such as dividends and earnings announcements.Finally, the methodology is applied to three European markets that are in need ofresearch in this domain. To the best of our knowledge, this is the first study to analysethe effect of ISENT on the market reaction to dividend change announcements in thecontext of European markets.
The remainder of this paper is organised as follows. Section 2 derives testablehypotheses and presents the research methodology. The sample selection is described inSection 3. Section 4 presents and discusses the empirical results. Finally, Section 5concludes.
2. Hypotheses and methodologyBecause we are interested in measuring the effect of sentiment on the market reactionto dividend change announcements, we need to calculate dividend changes.
MF37,12
1214
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
Although dividend signalling is about unexpected changes in dividends, we have noaccess to dividends analysts’ forecasts. However, similarly to other studies, we assumethat dividends follow a random walk[1], so the dividend changes were used as the proxyfor the unexpected dividend changes. The annual dividend change corresponding tothe dividend announcement is defined as the difference between the announceddividend in year t and the prior year dividend, scaled by the announcement day shareprice[2]:
DDi;t ¼Di;t 2 Di;t21
Pi;0ð1Þ
where:
D Di,t ¼ change of dividend per share (DPS) i for year t; and
Pi,0 ¼ price of share i in the announcement day.
To measure the market reaction to dividend change announcements, we opt to considerthe buy-and-hold abnormal returns (BHARs)[3]. The abnormal return for a share isdefined as the geometrically compounded return on the share minus the geometricallycompounded return on the market index. Therefore, the BHAR for share i from time21 to þ1 (BHARi(21 to þ 1)( generating model takes the following form (we consider athree-day event window, where t ¼ 0 is the dividend announcement day):
BHARið21toþ1Þ ¼Y1
t¼21
ð1 þ Ri;tÞ2Y1
t¼21
ð1 þ Rm;tÞ ð2Þ
where:
Ri,t ¼ return for share i in day t; and
Rm,t ¼ market return for day t.
In order to measure the ISENT, we consider two approaches. First, we rely on theEuropean Economic Sentiment Indicator (ESI), published by the European Commissionand obtained from Directorate General for Economic and Financial Affairs (DG ECFIN)database. The ESI index is based on sentiment surveys carried out in all Member Statesof the European Union (EU)[4], considering 15 sentiment components[5]. Schmeling(2009) also used the DG ECFIN consumer confidence measure for the European marketsconsidered in its sample.
In addition, we closely follow the methodology of Lemmon and Portniaguina (2006)to obtain a proxy for ISENT. We regress the ESI indicator on a set of macroeconomicvariables, in order to separate the rational and sentimental components of the ESI[6]and obtain a variable that is unrelated to fundamental risk factors. We consider theresidual from this regression as our sentiment measure (optimism or pessimism).Qiu and Welch (2006) document that investors are excessively optimistic or pessimisticbecause of good or bad news, returns or macro developments, thus, sentiment shouldbe related to return and macro variables.
Figure 1 shows the level of ISENT (Figure 1(a)) and ESI (Figure 1(b)) indexes, for theperiod from 1995 to 2002 and from 1989 to 2002, respectively. The first period is limited
Market reactionto dividend news
1215
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
by the availability of some data in what concerns macroeconomic variables used in themethodology of Lemmon and Portniaguina (2006) to obtain the proxy for ISENT. As wecan see, although the indexes have the same type of evolution for the three countries,globally, the sentiment is lower in Portugal and higher in the UK, for both sentimentproxies. The indexes values increase in the boom period of the late 1990s, and decreaseafter the crash of 2000-2001.
Before studying the market reaction to dividend change announcements, we wouldlike to analyse how the ISENT influences the dividend changes. Baker and Wurgler(2007) suggest that a firm with stable dividend shall be less sensitive to the sentiment.Therefore, we formulate the following hypothesis:
H1. The dividend changes are positively related with ISENT.
To analyse the relationship between ISENT and dividend changes, we estimate thefollowing regression:
Figure 1.Time series of theannually sentimentindexes
8
6
4
2
01995 1996 1997 1998 1999 2000 2001 2002
–2
–4
–6
–8
UK FR PT
(a)
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
1990
1989
2002
400.00
300.00
200.00
100.00
150.00
250.00
350.00
50.00
0.00
Notes: (a) ISENT index (1995-2002); (b) ESI index (1989-2002)
UK FR PT
(b)
MF37,12
1216
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
DDi;t ¼ aþ b1 SENT þ b2 ROEi;t þ 1i;t ð3Þ
where:
SENT ¼ measure of ISENT at the end of the year before the dividend change year(considering both ESI and ISENT indexes).
ROEi,t ¼ return on equity (ROE) for share i in year t, computed as the incomebefore extraordinary items at the end of the year before the dividendchange year divided by shareholders equity at the end of the year beforethe dividend change year.
We run two regressions for each country: one for dividend increases and other fordividend decreases. The coefficients b1 allow us to test whether the dividend changesvaries with the ISENT. We use the ROE as a control variable in the regression, in orderto control for the profitability effect.
Employing the panel data methodology, we run the pooled ordinary least squares(OLS), the fixed-effects model (FEM), and the random-effects model (REM).Subsequently, we use an F-statistic and the Hausman (1978) test to choose the mostappropriate model for our samples. We present the standard errors corrected forheteroscedasticity and covariance, based on White’s (1980) heteroscedasticity consistentstandard errors method.
Baker and Wurgler (2004) conclude that the decision to pay dividends is driven byinvestor demand. According to the authors, the catering theory supports the idea thatfirms tend to pay dividends when investors put a share price premium on firms thatdistribute dividends and not pay dividends when investors prefer non-payers.So, dividend payment by firms responds to investor demand for dividend proxies bythe dividend premium, the difference between the market-to-book (MB) ratios ofdividend payers and non-payers in a given year. Baker and Wurgler (2006, 2007) andLemmon and Portniaguina (2006) argue that stocks become underpriced (overpriced)during periods of low (high) sentiment on dividend non-payers, compared to dividendpayers. Based on these results, we formulate the hypothesis concerning therelationship between dividend change announcements and the share price movementsaround dividend announcements, considering the ISENT:
H2. The market reaction to dividend change announcements is higher (lower) fordividend increases (decreases) when sentiment is increasing.
To analyse the market reaction to dividend change announcements, considering theISENT, we estimate the following regression:
BHARið1 toþ1Þ ¼ aþ b1 DI £ DDi;0 þ b2 DD £ DDi;0 þ b3 DI £ DDi;0 £ SENT
þ b4 DD £ DDi;0 £ SENT þ b5 SIZEi þ 1i;tð4Þ
where:
DI ¼ dummy variable that takes value 1 if dividend increases and zerootherwise;
DD ¼ dummy variable that takes value 1 if dividend decreases and zerootherwise; and.
Market reactionto dividend news
1217
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
SIZEi ¼ size for share i, computed as the natural log of total assets at the end ofthe year before the dividend change year.
The coefficients b1 and b2 would capture the market reaction to dividend changes notconsidering the sentiment.
The coefficients on SENT allow us to test whether the market reaction to dividendchanges varies with ISENT. We use the prior years’ sentiment to avoid a look-aheadbias in our tests.
According to H2, we expect that the market reaction to dividend changeannouncements is more (less) sensitive to dividend increases (decreases) whensentiment is increasing. Consequently, we expect b3 to be positive and statisticallysignificant, indicating that the market reacts more to divided increase announcementswhen sentiment is increasing and b4 to be negative and statistically significant,indicating that the market reacts less to divided decrease announcements whensentiment is increasing.
We use the SIZE as a control variable in the regression, in order to control forpotential scale differences (Barth and Kallapur, 1996).
We need to adapt the methodology when analysing the UK sample, as the UK firmsusually announce both dividends and earnings simultaneously, making it difficult toseparate out the dividend announcement effect from that of earnings. However, it givesthe opportunity to incorporate the interaction of the joint signals into the analysis.Therefore, for the UK market, the impact of earnings announcements is examined bydividing the sample of dividend changes into four categories instead of the previoustwo (dividend increases and decreases): dividend increase-earnings increase (DIEI),dividend increase-earnings decrease (DIED), dividend decrease-earnings increase(DDEI), and dividend decrease-earnings decrease (DDED). In the analysis, we split theUK sample into these groups, or consider dummy variables that distinguish thedifferent situations in the regressions, in order to isolate the impact of dividendannouncements and investigate whether dividends provide information beyond thatprovided by earnings announcements.
Afterwards, we consider the following regression, an extension of equation (4), inorder to estimate the model considering some more control variables:
BHARið21 toþ1Þ ¼ aþ b1 DI £ DDi;0 þ b2 DD £ DDi;0 þ b3 DI £ DDi;0
£ SENT þ b4 DD £ DDi;0 £ SENT þ b5 DI £ DDi;0 £ SIZEi;t
þ b6 DD £ DDi;0 £ SIZEi þ b7 DI £ DDi;0 £ VOLATi;t
þ b8 DD £ DDi;0 £ VOLATi;t þ b9 DI £ DDi;0 £ BMi;t
þ b10 DD £ DDi;0 £ BMi;t þ b11 PROFi; t £ DDi;0
þ b12 SIZEi þ 1i;t ð5Þ
where:
VOLATi,t ¼ share return volatility, measured as the standard deviation of dailyprices over the preceding year;
BMi,t ¼ book-to-market ratio for share i, calculated by dividing book valueper share at the end of the year before the dividend change year bythe market price per share at the dividend change announcementdate; and
MF37,12
1218
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
PROFi,t ¼ dummy variable that takes value 1 for profitable firms (ROEpositive) and zero otherwise. Profitability is measured by the ROE.
The variable SIZE allows for the differential response of the market to dividend news,according to the firms size. The variable VOLAT enables to analyse the differentialresponse of the market to dividend news, according to the firms’ volatility. The PROFvariable allows for the differential response of the market to dividend news, accordingto the firms’ profitability and the book-to-market (BM) variable analyses whether themarket reaction to dividend change announcements is influenced by extreme growth ordistressed shares. High values of BM ratio may indicate distress and low values mayindicate high-growth opportunities.
To the extent that informational asymmetry is greater for small firms than for largefirms (Haw and Kim, 1991), the information content of dividend announcements will begreater for small firms. Although large firms have higher media coverage and greaterinstitutional ownership, the smaller firms have less information available in themarket, so, when they announce dividend changes, it will generate greater marketsurprises that induce a larger reaction by the market. Consequently, we expect marketreaction to dividend change announcements to be lower for large firms than for smallfirms. Accordingly, we formulate the following hypothesis:
H3. There is a negative relationship between the market reaction to dividendchange announcements and the firm size.
Eddy and Seifert (1988), Haw and Kim (1991), Mitra and Owers (1995) and Tudor (2008)found a negative relation between firm size and abnormal returns for firms thatincrease or/and initiate dividends.
According to the behaviour literature, the impact of behavioural biases is greater forshares that are harder to value and difficult to arbitrage (Shleifer and Vishny, 1997).Consequently, market reaction of such firms would be more pronounced by sentimentthan are the reaction of the other firms (Qiu and Welch, 2006; Baker and Wurgler, 2006,2007; Sankaraguruswamy and Mian, 2008).
Based on Baker and Wurgler (2006), we formulate the following hypotheses:
H4. The impact of sentiment on the market reaction to dividend changeannouncements is greater for young firms.
H5. The impact of sentiment on the market reaction to dividend changeannouncements is intensified for small firms.
H6. The impact of sentiment on the market reaction to dividend changeannouncements is greater for high volatile firms.
H7. The impact of sentiment on the market reaction to dividend changeannouncements is greater for extreme growth and distressed firms than formedium firms.
Baker and Wurgler (2006) conclude that the impact of sentiment on market reaction tonews is greater for young firms, high volatile shares, extreme growth shares anddistressed shares. The results of Sankaraguruswamy and Mian (2008) corroborate, inglobal terms, these conclusions.
Market reactionto dividend news
1219
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
To test the H4-H7, we closely follow the works of Baker and Wurgler (2006) andSankaraguruswamy and Mian (2008), ranking the firms into quintiles. Firms that fall thetop (bottom) quintiles are those with high (low) values for a particular variable. In whatconcerns the BM ratio, shares in the lowest quintile are identified as the growth sharesand those in the quintile with the highest values are recognised as the distressed shares.
To analyse cross-sectional differences on the impact of ISENT, we estimateequation (4) considering sub-samples according to the bottom and the top quintiles(20 per cent) of the variables as explained before. For instance, to investigate whetherfirm age causes cross-sectional differences on the impact of sentiment, we estimateequation (4) for young firms (bottom 20 per cent) and mature firms (top 20 per cent).If share price sensitivity of young firms is more predisposed to the impact of sentiment,the interaction variables involving sentiment must be more pronounced for theseshares. AGE is computed as the number of years since the firm’s first appearance onDatastream. We follow the same procedure for all variables, except for BM. In the BMcase, we compare the extreme quintiles coefficients with those of the middle threequintiles, because extreme quintiles are likely to be more susceptible to the impact ofISENT. The extreme quintiles are associated with growth and distressed shares,respectively, for the lowest and highest quintiles.
3. Sample selectionWe choose to examine the UK, the French and the Portuguese markets. Although theyare all European markets, they are different from each other for several reasons. First,they differ on size and liquidity. Second, they are different in what concerns theownership of equity. This phenomenon can influence the importance of dividends as asignalling mechanism and consequently share price reaction to dividend changeannouncements would be expected to be lower in countries where ownership is moreconcentrated (Portugal and France). Third, the UK is a market-based system, whereasPortugal and France are bank-based systems, which can influence dividend policy in adifferent way. Fourth, they also differ in what concerns the legal rules coveringprotection of corporate shareholders, which can also influence the dividend policy(La Porta et al., 1998; Aivazian et al., 2002). Finally, the sentiment can influencedifferently the market reaction to dividend change announcements in the threecountries. Globally, the UK presents the higher level of sentiment, and Portugalthe lowest one, being France in a middle position (Figure 1).
Given these characteristics, we expect to find a weaker support to thedividend-signalling hypothesis as well as a weaker influence of ISENT in Portugaland France than in the UK.
The sample is drawn from dividend announcements of firms listed on the EuronextLisbon, Euronext Paris and London Stock Exchange. Announcement dates areavailable on Bloomberg database and all other needed information is available onDatastream database. For the French and the UK markets, we consider the dividendannouncements between 1994 and 2002, and for the Portuguese market we consider thedividend announcements between 1988 and 2002[7].
To be included in the final sample, the dividend announcements must satisfy thefollowing criteria:
. The firm is not a financial institution.
. The company paid an ordinary dividend in the current and previous year.
MF37,12
1220
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
. The firm’s financial data are available on the Datastream database (or the Dathisdatabase in the case of Portugal) and announcement dates are available onBloomberg database.
. For the Portuguese and French markets, we consider that the firms’ earningsannouncements or other contaminate announcements, such as stock splits, stockdividends and mergers, did not occur within five trading days of the dividendannouncement. For the UK market, we exclude all these announcements, exceptthe case of earnings announcements[8].
Our sample events include dividend increases, no changes and decreases from 1995 to2002 for the French and the UK markets and from 1989 to 2002 for the Portuguesemarket. Our sample is an unbalanced panel dataset.
Table I reports the number of dividend events classified by sample selection criteria.The Portuguese sample contains 380 events: 158 increases, 121 decreases and 101 nochange observations. The French sample has 356 events: 235 increases, 62 decreasesand 59 no change observations. Finally, the UK sample contains 3,278 events: 2,662increases, 273 decreases and 343 no change dividend events. The preponderance ofdividend increases over no change and decreases in the three samples is consistentwith prior results that firms are reluctant to cut dividends[9]. The French and the UKpercentage of dividend changes, especially the case of the UK sample, are similar to theones of Abeyratna and Power (2002) for the UK market[10]. Portuguese percentages aresimilar to the ones of some emergent markets, such as Thailand and Korea[11].
4. Empirical resultsTable II provides summary statistics on dividend events and some financial ratios.We consider changes in DPS both in monetary units and in percentage, the payoutratio (the ratio of the DPS to the earnings before extraordinary items per share) and thedividend yield (DPS divided by the share price on the day before the dividendannouncement). We analyse the debt ratio (computed as the total debt divided by thetotal assets), the ROE (calculated as the earnings before extraordinary items divided by
Portugal France UKNumber % Number % Number %
Dividend increases 158 41.6 235 66.0 2,662 81.2No change 101 26.6 59 16.6 343 10.5Dividend decreases 121 31.8 62 17.4 273 8.3Total dividend events 380 100.0 356 100.0 3,278 100.0
Notes: This table reports the number of dividend events for the Portuguese, the French and the UKsamples; to be included in the final sample, a dividend announcement must satisfy the followingcriteria: (1) the firm is not a financial institution (2) the firm paid an annual ordinary dividend inthe current and previous year (3) the firm’s financial data are available on the Datastream or Dhatis(in the Portuguese sample) and announcement dates are available on Bloomberg database (4) for thePortuguese and French samples, the dividend, earnings or other potentially contaminatingannouncements did not occur within five trading days of each other; for the UK firms we consider thesame condition, except for earnings announcements; as they are simultaneous in almost the cases, weexclude dividend announcements which earnings announcements are announced on separate dates
Table I.Sample
Market reactionto dividend news
1221
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
DP
S(e
)D
ivid
end
chan
ges
(e)
Div
iden
dch
ang
es(%
)P
ayou
tra
tio
Div
iden
dy
ield
Deb
tra
tio
RO
EC
urr
ent
rati
o
Por
tug
al:
1989
-200
2Alldividendevents(n
¼380)
Mea
n0.
458
20.
031
2.05
50.
641
0.13
20.
389
0.08
91.
989
Med
ian
0.34
90.
000
0.00
00.
440
0.05
90.
368
0.07
41.
335
SD
0.62
40.
771
46.1
531.
251
0.28
80.
213
0.08
63.
055
Dividendincreases(n
¼158)
Mea
n0.
631
0.19
337
.573
0.45
80.
145
0.36
70.
109
2.26
1M
edia
n0.
449
0.10
020
.000
0.31
80.
073
0.34
30.
091
1.41
0S
D0.
902
0.77
642
.093
0.69
80.
346
0.20
50.
086
4.07
5Nochanges(n
¼101)
Mea
n0.
350
0.00
00.
000
0.53
90.
136
0.43
20.
078
1.92
0M
edia
n0.
324
0.00
00.
000
0.41
40.
050
0.42
60.
057
1.32
8S
D0.
208
0.00
00.
000
0.69
30.
238
0.21
60.
079
2.33
8Dividenddecreases(n
¼121)
Mea
n0.
322
20.
350
242
.197
0.96
50.
111
0.38
20.
071
1.69
1M
edia
n0.
249
20.
175
241
.176
0.88
20.
051
0.37
40.
054
1.25
7S
D0.
246
0.95
923
.613
1.93
60.
240
0.21
80.
087
1.73
4F
ran
ce:
1995
-200
2Alldividendevents(n
¼356)
Mea
n1.
243
0.10
213
.046
0.29
60.
020
0.24
70.
051
1.36
5M
edia
n0.
860
0.05
59.
222
0.18
00.
018
0.24
80.
045
1.17
7S
D1.
267
0.49
832
.848
2.67
20.
016
0.13
60.
040
0.54
1Dividendincreases(n
¼235)
Mea
n1.
319
0.25
026
.367
0.37
10.
021
0.24
60.
052
1.39
2M
edia
n0.
910
0.13
015
.797
0.16
60.
018
0.24
60.
046
1.20
5S
D1.
336
0.41
730
.497
3.24
40.
018
0.13
30.
038
0.53
7Nochanges(n
¼59)
Mea
n1.
148
0.00
00.
000
0.20
20.
020
0.23
70.
054
1.30
1M
edia
n0.
830
0.00
00.
000
0.20
00.
018
0.21
40.
049
1.19
0S
D0.
995
0.00
00.
000
0.33
50.
013
0.14
20.
039
0.50
4
(continued
)
Table II.Summary statistics
MF37,12
1222
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
DP
S(e
)D
ivid
end
chan
ges
(e)
Div
iden
dch
ang
es(%
)P
ayou
tra
tio
Div
iden
dy
ield
Deb
tra
tio
RO
EC
urr
ent
rati
o
Dividenddecreases(n
¼62)
Mea
n1.
042
20.
362
223
.742
0.09
80.
019
0.26
50.
042
1.32
4M
edia
n0.
640
20.
150
218
.768
60.
224
0.01
60.
276
0.03
71.
097
SD
1.21
80.
680
22.1
631.
007
0.01
20.
140
0.04
60.
589
UK
:19
95-2
002
Alldividendevents(n
¼3,278)
Mea
n8.
474
0.66
113
.906
0.50
90.
035
0.20
70.
131
1.47
8M
edia
n6.
355
0.50
09.
655
0.42
90.
030
0.18
60.
133
1.30
2S
D7.
930
2.06
132
.355
0.81
20.
024
0.16
40.
201
0.92
2Dividendincreases(n
¼2,662)
Mea
n8.
757
1.04
719
.941
0.45
30.
032
0.20
80.
145
1.44
6M
edia
n6.
550
0.65
011
.355
0.41
50.
028
0.18
60.
141
1.29
0S
D8.
189
1.78
031
.606
0.27
30.
021
0.16
50.
191
0.82
2N
och
ang
e(n
¼34
3)M
ean
7.43
20.
000
0.00
00.
902
0.04
80.
182
0.06
11.
702
Med
ian
6.00
00.
000
0.00
00.
630
0.04
40.
169
0.07
41.
339
SD
6.11
30.
000
0.00
02.
381
0.02
90.
147
0.20
71.
532
Dividenddecreases(n
¼273)
Mea
n7.
103
22.
272
227
.160
0.62
10.
044
0.22
90.
042
1.48
9M
edia
n5.
165
21.
070
220
.471
0.48
30.
036
0.21
30.
072
1.36
3S
D7.
282
3.08
823
.434
0.62
70.
034
0.17
80.
230
0.71
3
Notes:
Th
ista
ble
rep
orts
som
ed
escr
ipti
ve
stat
isti
csfo
rd
ivid
end
even
tob
serv
atio
ns
du
rin
gth
esa
mp
lep
erio
d;D
PS
isth
ed
ivid
end
per
shar
e;d
ivid
end
chan
ges
are
the
chan
ges
inD
PS
rela
tiv
eto
the
pre
vio
us
yea
r,ca
lcu
late
db
oth
inm
onet
ary
un
its
and
inp
erce
nta
ge;
pay
out
rati
ois
the
DP
Sd
ivid
edb
yth
eea
rnin
gs
bef
ore
extr
aord
inar
yit
ems
per
shar
e;d
ivid
end
yie
ldis
the
DP
Sd
ivid
edb
yth
esh
are
pri
ceon
the
day
bef
ore
the
div
iden
dan
nou
nce
men
t;d
ebt
rati
ois
the
tota
ld
ebt
div
ided
by
the
tota
las
sets
;RO
Eis
the
earn
ing
sb
efor
eex
trao
rdin
ary
item
sd
ivid
edb
yth
eeq
uit
y;c
urr
ent
rati
ois
the
curr
ent
asse
td
ivid
edb
yth
ecu
rren
td
ebt;
all
the
acco
un
tin
gv
aria
ble
sar
eco
nsi
der
edat
the
end
ofth
efi
scal
yea
rb
efor
eth
ed
ivid
end
ann
oun
cem
ent
Table II.
Market reactionto dividend news
1223
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
the equity) and the current ratio (computed as the current asset divided by the currentdebt). All the accounting variables are considered at the end of the fiscal year before thedividend announcement.
Comparing the values of each group of dividend events, the results show that for allthe countries, dividend decrease events are associated with a weaker financial positionthan dividend increases with higher debt ratios and lower ROE. Firms that neither cutnor increase their dividends are in a middle range. Finally, comparing the three samplestatistics, we can see that, for all the events, the UK sample has higher DPS, it is themost profitable sample, and presents the lowest value for the debt ratio, which is inagreement with a developed capital market such as the USA.
Similar to DeAngelo and DeAngelo (1990) and Nissim and Ziv (2001), we observe thatfor all countries dividend increases, although more frequent than dividend decreases, aresmaller in magnitude. In fact, the average decrease in DPS (percentage of change in DPS)is e0.35 (42.20 per cent), compared with an average increase in dividends of nearlye0.19 (37.57 per cent) in Portugal. In France, the average decrease in DPS (percentage ofchange in DPS) is e0.36 (23.74 per cent), compared with an average increase in dividendsof nearly e0.25 (26.37 per cent) and finally, in the UK market, the average decrease in DPS(percentage of change in DPS) is £2.27 (27.16 per cent), compared with an averageincrease in dividends of nearly £1.05 (19.94 per cent).
Overall, the evidence indicates that the UK is the main capital market of our sampleand Portugal is the smallest one, leaving France in a middle position.
Table III shows the estimates of regression model (3) of the most appropriate modelfor each country samples based on the F-statistic and on the Hausman (1978) test, inorder to analyse the relationship between ISENT and dividend changes. We find noevidence of ISENT influencing dividend changes. Consequently, we do not support H1.
However, in the UK market, the results suggest that dividend changes are positivelyrelated with profitability, since the coefficient on ROE is positive and statisticallysignificant.
Table IV reports the estimates of the regression model (4). The first regression results(base model) do not consider the interaction variables involving sentiment. In this model,for all three countries, no coefficient presents a significant value. Consequently, we findno evidence for the dividend-signalling hypothesis, which is in agreement with some ofthe studies carried out before, such as the ones of Lang and Litzenberger (1989), Benartziet al. (1997), Abeyratna and Power (2002) and Vieira and Raposo (2007).
Considering the ISENT effect on the market reaction to dividend changeannouncements, the results are different for the three countries.
For the Portuguese market, we find no evidence that ISENT influences the shareprice response to dividend change news, since no coefficient is statistically differentfrom zero.
In what concerns the French market, the only coefficient that is statisticallysignificant, and only considering the ISENT index, is the coefficient for DD_SENT,which is negative, as expected. This is an indication that the share price sensitivity tobad dividend news is lower when the sentiment is increasing, in agreement withhypothesis one for dividend decrease announcements situations. In addition, theresults suggest that the ISENT proxy for invest sentiment is more robust that the ESIone. Indeed, although they are both low, the adjusted R 2 is slightly higher for theregression considering the ISENT as the ISENT measure.
MF37,12
1224
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
Finally, for the UK market, only the results of the regression considering the ISENTindex present significant values for the coefficients, which reinforce the robustness orthis ISENT measure. The coefficient for DIED_SENT is positive, as expected, andstatistically significant at 5 per cent level, suggesting a stronger market reaction todividend increase announcements when the ISENT is increasing. The variable SIZE isnegative and statistically significant, which is an indication that, during the sampleperiod, the returns of large shares are smaller than those of small shares, in accordancewith the results of Mitra and Owers (1995) and Tudor (2008), among others.
Concluding, we find some evidence for the H2, but only for dividend bad newsannouncements (decreases) in the French market and for dividend good newsannouncements (increases) in the UK market.
Pooled OLSBase model ISENT ESI
Coefficient t Coefficient t Coefficient t
PortugalIntercept 0.0075 0.449 20.0009 20.041 0.0071 0.425DI 0.0113 1.223 0.0210 0.512 20.4281 20.703DD 0.0073 0.587 20.0103 20.396 20.0913 20.517DI_SENT 20.0027 20.227 0.0042 0.721DD_SENT 20.0047 0.0010 0.559
20.498SIZE 20.0005 20.499 20.0001 20.074 20.0004 20.469n 380 125 380Adjusted R 2 0.006 0.015 0.008FranceIntercept 20.0067 20.323 20.0049 20.239 20.0081 20.390DI 20.1000 20.414 20.5299 21.475 20.2915 20.923DD 0.1061 0.669 0.2600 1.444 0.7121 1.618DI_SENT 20.5867 21.577 0.0291 0.893DD_SENT 20.1866 * 21.673 20.0689 21.596SIZE 0.0014 0.430 0.0013 0.407 0.0015 0.484n 356 356 356Adjusted R 2 0.002 0.017 0.012
Notes: Significantly different from zero at the *10 percent level; this table reports the followingregression:
BHARið1 to þ1Þ ¼ aþ b1 DI £ DDi;0 þ b2 DD £ DDi;0 þ b3 DI £ DDi;0 £ SENT þ b4 DD£ DDi;0 £ SENT þ b5 SIZEi þ 1i;t
BHAR3 is the buy-and-hold accumulated abnormal return on the three-day period as calculated byequation (2); DI is a dummy variable that takes value 1 if dividend increases and zero otherwise; DD isa dummy variable that takes value 1 if dividend decreases and zero otherwise; SENT is a measure ofISENT at the end of the year before the dividend change year, considering both the ESI and the ISENTindexes and SIZEi is the size for share i, computed as the natural log of total assets at the end of theyear before the dividend change year; the table presents the best model among pooled OLS, FEM andREM; in order to choose the most appropriate model for each particular sample, we run the F-test, atest for the equality of sets of coefficients, and the Hausman (1978) test, a test with H0: random effectsare consistent and efficient, versus H1: random effects are inconsistent; the numbers in parentheses arethe t-statistics corrected for heteroscedasticity using the White (1980) method
Table III.Regression of market
reaction to dividendchange announcements,
considering the sentiment
Market reactionto dividend news
1225
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
Pooled OLSBase model ISENT ESI
Coefficient t Coefficient t Coefficient t
PortugalIntercept 0.0075 0.449 20.0009 20.041 0.0071 0.425DI 0.0113 1.223 0.0210 0.512 20.4281 20.703DD 0.0073 0.587 20.0103 20.396 20.0913 20.517DI_SENT 20.0027 20.227 0.0042 0.721DD_SENT 20.0047 0.0010 0.559
20.498SIZE 20.0005 20.499 20.0001 20.074 20.0004 20.469n 380 125 380Adjusted R 2 0.006 0.015 0.008FranceIntercept 20.0067 20.323 20.0049 20.239 20.0081 20.390DI 20.1000 20.414 20.5299 21.475 20.2915 20.923DD 0.1061 0.669 0.2600 1.444 0.7121 1.618DI_SENT 20.5867 21.577 0.0291 0.893DD_SENT 20.1866 * 21.673 20.0689 21.596SIZE 0.0014 0.430 0.0013 0.407 0.0015 0.484n 356 356 356Adjusted R 2 0.002 0.017 0.012UK
Base model-FEM ISENT-REM ESI-FEMCoefficient t Coefficient t Coefficient t
Intercept 0.0691 * 1.683 0.0404 * * * 3.615 0.0700 * 1.702DIEI 0.0001 0.575 20.0002 0.741 20.0156 1.125DIED 20.0068 0.923 20.0079 1.120 20.0387 0.165DDEI 20.0197 0.667 20.0265 1.022 20.5647 0.847DDED 20.0190 1.137 20.0139 0.906 0.1914 0.414DIEI_SENT 0.0002 0.644 0.0002 1.132DIED_SENT 0.0149 * * 2.281 0.0003 0.137DDEI_SENT 20.0259 1.548 0.0053 0.818DDED_SENT 0.0110 1.046 20.0021 0.456SIZE 20.0095 1.234 20.0041 * * 1.999 20.0097 1.255n 3,276 3,276 3,276Adjusted R 2 0.193 0.196 0.194
Notes: Significantly differ from zero at the *10, * *5, * * *1 percent levels; this table reports the followingregression:
BHARið1 toþ1Þ ¼ aþ b1 DI £ DDi;0 þ b2 DD £ DDi;0 þ b3 DI £ dDi;0 £ SENT þ b4 DD £ dDi;0
£ SENT þ b5 SIZEi þ 1i;t
BHAR3 is the buy and hold accumulated abnormal return on the three-day period as calculated by equation(2); DI is a dummy variable that takes value 1 if dividend increases and zero otherwise; DD is a dummyvariable that takes value 1 if dividend decreases and zero otherwise; SENT is a measure of ISENT at the endof the year before the dividend change year, considering both the ESI and the ISENT indexes and SIZEi is thesize for share i, computed as the natural log of total assets at the end of the year before the dividend changeyear; the table presents the best model among pooled OLS, FEM and REM; in order to choose the mostappropriate model for each particular sample, we run the F-test, a test for the equality of sets of coefficients,and the Hausman (1978) test, a test with H0: random effects are consistent and efficient, versus H1: randomeffects are inconsistent; the numbers in parentheses are the t-statistics corrected for heteroscedasticity usingthe White (1980) method
Table IV.Regression of marketreaction to dividendchange announcements,considering the sentiment
MF37,12
1226
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
Table V reports the estimates of the regression model (5) for the most appropriatemodel for each country samples based on the F-statistic and the Hausman (1978) test.This regression is useful, namely because it allows the analysis of the regressionrobustness equation (4) results, when we introduce a set of control variables.
We start by estimating the correlations among the independent variables, in orderto detect econometric problems such as multicollinearity, namely between BM andsentiment variables. The higher correlation coefficients are between ESI and BM inPortugal, and ESI and VOLAT in France, but both below 20 per cent[12]. In general,the correlation coefficients do not appear to be sufficiently large to cause concern aboutmulticollinearity problems.
For the Portuguese sample, and considering the ISENT index, we have two controlvariables with significant values, which are the BM and the PROF. In the firstsituation, the DI_BM variable is negative and statistically significant, suggesting thatthe market reaction to dividend increase announcements is higher for the firms thathave a lower BM, which is a proxy for growth firms. This result can be an indicationthat investors believe that firms presenting growth prospects have higher capability tosustain dividends payment in the future, which is somewhat in agreement with thesignalling hypothesis. The D_PROF variable is also negative and statistically differentfrom zero, suggesting that market responds more to dividend change announcementsfor non-profitable firms. When we use the ESI index, only the DD_BM variable isstatistically significant and positive, suggesting that share prices react more to thenegative dividend changes for firms with higher BM ratios, i.e. for distressed firms.
Analysing the French market results, we can see that only the ISENT indexpresents some significant coefficients. Once more, we find evidence of this index to bemore robust that the ESI one. The DD coefficient is positive and significantly differentfrom zero, suggesting that share prices decline in response to the dividend decreaseannouncements. The coefficient for DD_SIZE is negative and significant, indicatingthat the market reacts more to dividend decrease news for small firms, which is inagreement with the H2 and with some other authors who find evidence of a significanteffect of sentiment on returns for small but not for large stocks, such as Schmeling(2009), Brown and Cliff (2005) and Lemmon and Portniaguina (2006).
In what concerns the UK results, we can see that the variables that are useful toexplain the market sensitivity to dividend change announcements are SENT, VOLAT,BM and PROF. The coefficient for the interaction term DIEI_SENT is positive andstatistically significant at 1 per cent level, suggesting that share price changesfollowing good dividend and earnings news is greater when sentiment is higher, whichsupports H1 for the case of dividend increases. However, the coefficient forDIED_SENT, although positive, is statistically insignificant. Comparing thesignificance of DIEI_SENT and DIED_SENT variables, the results suggest thatearnings announcements have information power beyond that of dividendannouncements, which is consistent with the conclusion of DeAngelo et al. (1992)and Conroy et al. (2000), among others.
The coefficient on DIEI_VOLAT is negative and statistically different from zero forthe two ISENT indexes, indicating that the market reacts more to the dividend increaseannouncements for less-volatile firms, and suggesting also that investors reward firmswhich present lower levels of volatility.
Market reactionto dividend news
1227
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
Pooled OLSISENT ESI
Coefficient t Coefficient t
PortugalIntercept 0.0089 0.399 0.0556 * 1.738DI 0.3410 1.476 20.6091 20.815DD 20.1458 21.349 20.4753 21.415DI_SENT 0.0063 0.229 0.0094 1.240DD_SENT 0.0001 0.007 20.0009 21.131DI_SIZE 0.0091 1.050 0.0543 0.554DD_SIZE 20.0001 21.178 0.0029 0.048DI_VOLAT 20.1480 20.509 0.0010 0.513DD_VOLAT 0.0262 0.201 20.0012 21.112DI_BM 20.0287 * 21.726 20.0159 20.604DD_BM 0.0098 1.003 20.0030 * 21.736D_PROF 20.0556 * 22.057 0.0055 0.793SIZE 20.0003 20.286 0.0037 1.271n 125 378Adjusted R 2 0.091 0.255FranceIntercept 0.0077 0.286 0.0052 0.192DI 2.8105 0.617 20.1928 20.029DD 1.0260 * 1.707 1.1734 1.343DI_SENT 20.4513 21.155 20.4027 20.530DD_SENT 20.2601 21.442 21.0684 21.321DI_SIZE 20.5151 20.720 20.0963 21.181DD_SIZE 21.6284 * 21.709 0.0539 0.812DI_VOLAT 20.0660 20.811 0.0174 0.288DD_VOLAT 0.0394 0.580 0.0049 10467DI_BM 0.0142 0.235 20.0578 20.935DD_BM 0.0036 1.054 0.0002 0.045D_PROF 20.0491 21.798 0.0275 0.771SIZE 20.0001 20.035 20.0511 20.916n 356 356Adjusted R 2 0.038 0.033UK
FEMISENT ESI
Coefficient t Coefficient tIntercept 20.0047 20.109 20.0072 20.165DIEI 0.0001 21.121 20.0001 210142DIED 20.0041 20.240 20.0001 21.539DDEI 0.1059 1.540 20.2875 20.415DDED 20.0338 20.965 0.0062 0.013DIEI_SENT 0.0102 * * * 3.430 0.0003 0.531DIED_SENT 0.0113 1.546 20.0002 20.125DDEI_SENT 20.0297 21.635 0.0037 0.568DDED_SENT 0.0152 1.335 20.0005 20.099DIEI_SIZE 0.0001 1.121 0.0001 1.142DIED_SIZE 0.0001 1.580 0.0001 1.539DDEI_SIZE 0.0210 1.562 20.0561 20.413
(continued )
Table V.Regression of marketreaction to dividendchange announcements,considering the sentimentand control variables
MF37,12
1228
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
It is interesting to see that the DIEI_BM coefficient is positive and the DDEI_BMis negative, both statistically significant, indicating that the market reacts more todividend increases for higher BM firms (the distressed ones), and reacts more todividend decreases for lower BM firms (indicating high-growth opportunities), whichis in contrast with the evidence found for the Portuguese sample. Although thePortuguese results are somewhat in agreement with the signalling hypothesis, the UKresults give some support for the free cash-flow hypothesis ( Jensen, 1986).
Pooled OLSISENT ESI
Coefficient t Coefficient t
DDED_SIZE 20.0067 20.983 0.0005 0.006DIEI_VOLAT 20.0001 * * 22.074 20.0001 * 21.725DIED_VOLAT 0.0001 0.618 0.0001 0.634DDEI_VOLAT 20.00080 20.882 20.0007 20.772DDED_VOLAT 0.0004 0.767 0.0004 0.696DIEI_BM 0.0085 * 1.706 20.0044 21.314DIED_BM 20.0102 20.680 20.0138 20.915DDEI_BM 20.0857 * * 22.355 20.0795 * 22.140DDED_BM 20.0028 20.145 0.0010 0.053D_PROF 0.1340 * * * 5.166 0.1292 * * * 4.955SIZE 0.0021 0.257 0.0025 0.309n 3,276 3,276Adjusted R 2 0.210 0.205
Notes: Significantly differ from zero at the *10 per cent leve; this table reports the followingregression:
BHARið1 toþ1Þ ¼ aþ b1 DI £ dDi;0 þ b2 DD £ dDi;0 þ b3 DI £ dDi;0 £ SENT þ b4 DD £ dDi;0
£ SENT þ b5 DI £ dDi;0 £ SIZEi;t þ b6 DD £ dDi;0 £ SIZEi þ b7 DI £ dDi;0
£ VOLATi;t þ b8 DD £ dDi;0 £ VOLATi;t þ b9 DI £ dDi;0 £ BMi;t þ b10 DD£ dDi;0 £ BMi;t þ b11 PROFi £ dDi;0 þ b12 SIZEi þ 1i;t
BHAR3 is the buy-and-hold accumulated abnormal return on the three-day period as calculated byequation (2); DI is a dummy variable that takes value 1 if dividend increases and zero otherwise; DD isa dummy variable that takes value 1 if dividend decreases and zero otherwise; SENT is a measure ofISENT at the end of the year before the dividend change year considering both the ESI and the ISENTindexes; VOLATi is the share return volatility, measured as the standard deviation of daily prices overthe preceding year; BMi,t is the book-to-market ratio for share i, calculated by dividing book valueper share at the end of the year before the dividend change year by the market price per share at thedividend change announcement date; PROF is a dummy variable that takes value 1 for profitable firms(ROE positive) and zero otherwise; profitability is measured by the ROE, computed as the incomebefore extraordinary items at the end of the year before the dividend change year divided byshareholders equity at the end of the year before the dividend change year and SIZEi is the size forshare i, computed as the natural log of total assets at the end of the year before the dividend changeyear; the table presents the best model among pooled OLS, FEM and REM; in order to choose the mostappropriate model for each particular sample, we run the F-test, a test for the equality of sets ofcoefficients, and the Hausman (1978) test, a test with H0: random effects are consistent and efficient,versus H1: random effects are inconsistent; the numbers in parentheses are the t-statistics corrected forheteroscedasticity using the White (1980) method Table V.
Market reactionto dividend news
1229
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
Also in contrast with the Portuguese results, the D_PROF variable is positive andstatistically different from zero, suggesting that market responds more to dividendchange announcements for profitable firms.
Next, we will test H4-H7, associated with the assumption that firms that are moredifficult to arbitrage are more influenced by sentiment (Baker and Wurgler, 2006).Consequently, we estimate equation (4) considering sub-samples of firms arranged byspecific characteristics such as size and age. The results are reported in Table VI.Tables VI-IX present the results of equation (4) for sub-samples sorted on firm age,size, volatility and growth/distressed, respectively.
Table VI presents the results concerning the impact of sentiment on firm AGE(young versus mature firms). We present the results for young and for mature firms,considering the two ISENT indexes. For the Portuguese sample, none coefficient isstatistically significant. Consequently, we find no support for the hypothesis that theimpact of sentiment on the market reaction to dividend change announcements isgreater for young firms (H4).
In what concerns the French sample, the only variable involving sentiment that isstatistically significant is the DI_SENT. Comparing the significance of the variablebetween young and mature firms, we find evidence that the impact of sentiment on themarket reaction to dividend increase announcements is greater for young firms, whichis predicted in H4. However, its signal is negative, contrary to what we expect, but inagreement with the results found in Table IV. The DD_SENT is always insignificant,suggesting that the market reaction to dividend decreases does not diverge across highand low sentiment periods. The SIZE is negative for all the situations, but onlystatistically significant for the young firms, suggesting that, for this group of firms,during the sample period the returns of large firms are smaller than those of smallfirms.
Analysing the results for the UK market, we find some evidence supporting thehypothesis that the impact of sentiment on the market reaction to dividend changeannouncements is greater for young firms. Indeed, for the dividend increaseannouncements, the only interaction variable that is significantly positive is theDIED_SENT (for the ISENT index), which indicates that the market reaction todividend good news is higher for young firms. In what concerns the dividend decreaseannouncements, the coefficient is also only statistically significant for the young firms(for the ESI index). However, it presents a positive signal for the DDEI events and anegative signal for DDED events (the expected signal). Once more, this result suggeststhat earnings announcements have information power beyond that of dividendannouncements. To sum up, we find some evidence supporting H4 for the UK market.
Table VII reports the differential impact of sentiment on small firms. For thePortuguese and the French samples, no coefficient is statistically significant.Consequently, we find no support for the hypothesis that the impact of sentiment onthe market reaction to dividend change announcements is intensified for small firms(H5). For the UK sample, the interaction coefficient for dividend increaseannouncements is always statistically insignificant, suggesting that the marketreaction to dividend good news does not vary across high and low sentiment periods.However, comparing the significance of the DDEI_SENT variable, the results suggestthat the market response to dividend bad news, considering the ISENT, is more
MF37,12
1230
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
You
ng
-IS
EN
TM
atu
re-I
SE
NT
You
ng
-ES
IM
atu
re-E
SI
Coe
ffici
ent
tC
oeffi
cien
tt
Coe
ffici
ent
tC
oeffi
cien
tt
Portugal-pooledOLS
Inte
rcep
t0.
0177
0.75
12
0.04
012
0.38
70.
0137
0.62
90.
0425
0.85
1D
I0.
0874
0.40
00.
0771
0.26
80.
4475
1.19
30.
0841
0.06
5D
D2
0.03
092
0.44
00.
2918
1.31
30.
2941
0.21
32
0.72
182
0.97
9D
I_S
EN
T2
0.00
382
0.06
80.
0155
0.21
22
0.03
952
1.16
32
0.00
062
0.05
1D
D_
SE
NT
20.
0027
20.
015
20.
0067
20.
087
20.
0032
20.
243
0.00
680.
989
SIZ
E2
0.00
092
0.75
30.
0015
0.26
92
0.00
072
0.67
52
0.00
272
0.96
6n
5332
7210
3A
dju
sted
R2
0.02
40.
102
0.03
950.
023
France-FEM
Inte
rcep
t0.
1047
**
2.52
90.
0681
0.82
20.
0986
*1.
961
0.06
530.
792
DI
20.
2804
20.
173
0.41
010.
518
0.36
040.
270
20.
4431
21.
072
DD
20.
4788
**
*2
3.14
60.
0346
0.12
51.
5339
0.53
42
0.55
120.
074
DI_
SE
NT
20.
3678
**
*2
3.31
80.
5439
0.63
52
0.04
232
0.31
90.
0457
1.06
1D
D_
SE
NT
21.
8621
21.
530
0.00
070.
004
20.
2088
20.
727
0.00
570.
079
SIZ
E2
0.01
75*
*2
2.65
42
0.00
982
0.80
62
0.01
68*
*2
2.09
62
0.00
942
0.77
4n
5385
5385
Ad
just
edR
20.
771
0.01
30.
686
0.02
2
(continued
)
Table VI.Regression of market
reaction to dividendchange announcements,
considering the sentimentand firm-specific
characteristics (panel A –age: young versus mature
firms)
Market reactionto dividend news
1231
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
You
ng
-IS
EN
TM
atu
re-I
SE
NT
You
ng
-ES
IM
atu
re-E
SI
Coe
ffici
ent
tC
oeffi
cien
tt
Coe
ffici
ent
tC
oeffi
cien
tt
UK-REM
Inte
rcep
t0.
0487
1.35
40.
0610
**
2.35
00.
0989
0.88
20.
0609
**
2.32
3D
IEI
20.
0007
20.
141
0.02
14*
*2.
071
0.13
110.
696
0.33
881.
053
DIE
D0.
0003
0.01
70.
0177
0.90
82
0.03
102
0.06
62
0.63
032
1.56
3D
DE
I2
0.00
582
0.06
72
0.02
492
0.51
12
0.39
98*
21.
890
1.74
051.
279
DD
ED
20.
0990
**
22.
084
20.
0321
21.
091
0.43
87*
*2.
040
0.81
750.
819
DIE
I_S
EN
T0.
0006
0.13
92
0.00
202
0.19
02
0.00
132
0.69
62
0.00
312
0.98
8D
IED
_S
EN
T0.
0194
**
2.13
02
0.00
552
0.18
90.
0001
0.02
80.
0064
1.60
2D
DE
I_S
EN
T2
0.00
252
0.02
12
0.00
492
0.88
80.
0382
*1.
873
20.
0172
21.
295
DD
ED
_S
EN
T2
0.01
512
0.58
22
0.00
082
0.05
12
0.04
48*
*2
2.08
72
0.00
842
0.85
3S
IZE
20.
0066
20.
968
20.
0076
*2
1.73
52
0.01
622
0.74
82
0.00
78*
21.
744
n43
975
243
975
2A
dju
sted
R2
0.30
40.
186
0.31
50.
192
Notes
:S
ign
ifica
ntl
yd
iffe
rfr
omze
roat
the
* 10,
** 5
,*
** 1
per
cen
tle
vel
s;th
ista
ble
rep
orts
the
foll
owin
gre
gre
ssio
n:
BHARið
1toþ
1Þ¼
aþ
b1
DI£d
Di;
0þ
b2
DD£d
Di;
0þ
b3
DI£d
Di;
0£
SE
NTþ
b4
DD£d
Di;
0£
SE
NTþ
b5
SIZ
Eiþ
1i;t
BH
AR
3is
the
bu
y-a
nd
-hol
dac
cum
ula
ted
abn
orm
alre
turn
onth
eth
ree-
day
per
iod
asca
lcu
late
db
yeq
uat
ion
(2);
DI
isa
du
mm
yv
aria
ble
that
tak
esv
alu
e1
ifd
ivid
end
incr
ease
san
dze
root
her
wis
e;D
Dis
ad
um
my
var
iab
leth
atta
kes
val
ue
1if
div
iden
dd
ecre
ases
and
zero
oth
erw
ise;
SE
NT
isa
mea
sure
ofIS
EN
Tat
the
end
ofth
ey
ear
bef
ore
the
div
iden
dch
ang
ey
ear
con
sid
erin
gb
oth
the
ES
Ian
dth
eIS
EN
Tin
dex
esan
dS
IZE
iis
the
size
for
shar
ei,
com
pu
ted
asth
en
atu
rall
ogof
tota
lass
ets
atth
een
dof
the
yea
rb
efor
eth
ed
ivid
end
chan
ge
yea
r;T
able
sV
I-IX
pre
sen
tth
eeq
uat
ion
resu
lts
for
sub
-sam
ple
sso
rted
onfi
rmag
e,si
ze,
vol
atil
ity
and
gro
wth
/dis
tres
sed
,re
spec
tiv
ely
;ag
eis
com
pu
ted
asth
en
um
ber
ofy
ears
sin
ceth
efi
rm’s
firs
tap
pea
ran
ceon
Datastream
;v
olat
ilit
yis
the
shar
ere
turn
vol
atil
ity
,mea
sure
das
the
stan
dar
dd
evia
tion
ofd
aily
pri
ces
over
the
pre
ced
ing
yea
ran
dth
ep
rox
yfo
rg
row
th/d
istr
esse
dis
the
BM
rati
o,ca
lcu
late
db
yd
ivid
ing
boo
kv
alu
ep
ersh
are
atth
een
dof
the
yea
rb
efor
eth
ed
ivid
end
chan
ge
yea
rb
yth
em
ark
etp
rice
per
shar
eat
the
div
iden
dch
ang
ean
nou
nce
men
td
ate;
the
tab
lep
rese
nts
the
bes
tm
odel
amon
gp
oole
dO
LS
,FE
Man
dR
EM
.In
ord
erto
choo
seth
em
ost
app
rop
riat
em
odel
for
each
par
ticu
lar
sam
ple
,we
run
theF
-tes
t,a
test
for
the
equ
alit
yof
sets
ofco
effi
cien
ts,a
nd
the
Hau
sman
(197
8)te
st,a
test
wit
hH0
:ran
dom
effe
cts
are
con
sist
ent
and
effi
cien
t,v
ersu
sH1
:ran
dom
effe
cts
are
inco
nsi
sten
t;th
en
um
ber
sin
par
enth
eses
are
thet-
stat
isti
csco
rrec
ted
for
het
eros
ced
asti
city
usi
ng
the
Wh
ite
(198
0)m
eth
od
Table VI.
MF37,12
1232
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
Sm
all-
ISE
NT
Lar
ge-
ISE
NT
Sm
all-
ES
IL
arg
e-E
SI
Coe
ffici
ent
tC
oeffi
cien
tt
Coe
ffici
ent
tC
oeffi
cien
tt
Portugal-pooledOLS
Inte
rcep
t2
0.04
632
0.30
32
0.04
262
0.95
60.
0094
0.18
22
0.02
292
0.62
6D
I0.
0258
0.50
30.
0179
0.43
62
0.47
452
0.62
00.
0427
0.04
9D
D0.
0326
0.35
90.
0019
0.07
72
0.30
362
0.87
92
0.07
392
0.39
2D
I_S
EN
T2
0.00
342
0.23
52
0.00
172
0.14
50.
0047
0.63
52
0.00
032
0.03
6D
D_
SE
NT
0.01
350.
394
20.
0002
20.
025
0.00
330.
901
0.00
080.
423
SIZ
E0.
0019
0.22
70.
0022
0.95
02
0.00
062
0.19
80.
0011
0.60
2n
4611
023
523
8A
dju
sted
R2
0.05
30.
023
0.01
20.
009
France-pooledOLS
Inte
rcep
t2
0.06
992
1.22
30.
0851
1.44
72
0.07
832
1.38
70.
0799
1.35
6D
I2
0.51
202
0.65
42
0.74
142
0.87
90.
9632
0.91
62
0.72
522
0.89
8D
D0.
5999
1.13
70.
1799
0.26
31.
5051
1.22
51.
0334
0.67
0D
I_S
EN
T2
0.91
012
0.93
00.
2133
0.33
32
0.09
532
0.89
90.
0629
0.81
5D
D_
SE
NT
20.
3935
21.
159
21.
2290
21.
157
20.
1456
21.
223
20.
0949
20.
656
SIZ
E0.
0124
1.27
52
0.01
122
1.37
00.
0137
1.42
22
0.01
042
1.26
9n
152
152
152
152
Ad
just
edR
20.
032
0.02
50.
034
0.02
2
(continued
)
Table VII.Regression of market
reaction to dividendchange announcements,
considering the sentimentand firm-specific
characteristics (panel B –size: small versus large
firms)
Market reactionto dividend news
1233
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
Sm
all-
ISE
NT
Lar
ge-
ISE
NT
Sm
all-
ES
IL
arg
e-E
SI
Coe
ffici
ent
tC
oeffi
cien
tt
Coe
ffici
ent
tC
oeffi
cien
tt
UK-pooledOLS
Inte
rcep
t0.
0470
1.01
90.
0934
**
*2.
740
0.04
931.
066
0.08
88*
**
2.61
1D
IEI
20.
0039
20.
792
20.
0232
21.
122
20.
0782
20.
658
0.16
730.
455
DIE
D0.
0133
0.72
90.
0002
0.01
32
0.00
082
0.01
80.
2190
0.60
0D
DE
I2
0.06
772
1.58
20.
0262
0.39
42
0.20
78*
*2
2.11
80.
2749
*1.
821
DD
ED
20.
0187
20.
666
20.
0385
21.
037
20.
4666
20.
594
0.76
550.
690
DIE
I_S
EN
T0.
0035
0.77
50.
0010
0.07
60.
0008
0.65
72
0.00
182
0.52
2D
IED
_S
EN
T0.
0260
1.35
42
0.01
482
0.59
90.
0001
0.01
82
0.00
222
0.61
5D
DE
I_S
EN
T2
0.03
342
1.20
40.
0148
0.29
40.
0198
**
2.06
12
0.02
62*
21.
801
DD
ED
_S
EN
T0.
0265
1.24
62
0.03
982
1.46
10.
0044
0.56
52
0.00
792
0.72
2S
IZE
20.
0056
20.
535
20.
0123
**
22.
323
20.
0064
20.
614
20.
0116
**
22.
187
n83
883
883
883
8A
dju
sted
R2
0.01
00.
012
0.01
00.
014
Notes
:S
ign
ifica
ntl
yd
iffe
rfr
omze
roat
the
* 10,
** 5
,*
** 1
per
cen
tle
vel
s;th
ista
ble
rep
orts
the
foll
owin
gre
gre
ssio
n:
BHARið
1toþ
1Þ¼
aþ
b1
DI£d
Di;
0þ
b2
DD£d
Di;
0þ
b3
DI£d
Di;
0£
SE
NTþ
b4
DD£d
Di;
0£
SE
NTþ
b5
SIZ
Eiþ
1i;t
BH
AR
3is
the
bu
y-a
nd
-hol
dac
cum
ula
ted
abn
orm
alre
turn
onth
eth
ree-
day
per
iod
asca
lcu
late
db
yeq
uat
ion
(2);
DI
isa
du
mm
yv
aria
ble
that
tak
esv
alu
e1
ifd
ivid
end
incr
ease
san
dze
root
her
wis
e;D
Dis
ad
um
my
var
iab
leth
atta
kes
val
ue
1if
div
iden
dd
ecre
ases
and
zero
oth
erw
ise;
SE
NT
isa
mea
sure
ofIS
EN
Tat
the
end
ofth
ey
ear
bef
ore
the
div
iden
dch
ang
ey
ear
con
sid
erin
gb
oth
the
ES
Ian
dth
eIS
EN
Tin
dex
esan
dS
IZE
iis
the
size
for
shar
ei,
com
pu
ted
asth
en
atu
rall
ogof
tota
lass
ets
atth
een
dof
the
yea
rb
efor
eth
ed
ivid
end
chan
ge
yea
r;T
able
sV
I-IX
pre
sen
tth
eeq
uat
ion
resu
lts
for
sub
-sam
ple
sso
rted
onfi
rmag
e,si
ze,
vol
atil
ity
and
gro
wth
/dis
tres
sed
,re
spec
tiv
ely
;ag
eis
com
pu
ted
asth
en
um
ber
ofy
ears
sin
ceth
efi
rm’s
firs
tap
pea
ran
ceon
Datastream
;v
olat
ilit
yis
the
shar
ere
turn
vol
atil
ity
,mea
sure
das
the
stan
dar
dd
evia
tion
ofd
aily
pri
ces
over
the
pre
ced
ing
yea
ran
dth
ep
rox
yfo
rg
row
th/d
istr
esse
dis
the
BM
rati
o,ca
lcu
late
db
yd
ivid
ing
boo
kv
alu
ep
ersh
are
atth
een
dof
the
yea
rb
efor
eth
ed
ivid
end
chan
ge
yea
rb
yth
em
ark
etp
rice
per
shar
eat
the
div
iden
dch
ang
ean
nou
nce
men
td
ate;
the
tab
lep
rese
nts
the
bes
tm
odel
amon
gp
oole
dO
LS
,FE
Man
dR
EM
.In
ord
erto
choo
seth
em
ost
app
rop
riat
em
odel
for
each
par
ticu
lar
sam
ple
,we
run
theF
-tes
t,a
test
for
the
equ
alit
yof
sets
ofco
effi
cien
ts,a
nd
the
Hau
sman
(197
8)te
st,a
test
wit
hH0
:ran
dom
effe
cts
are
con
sist
ent
and
effi
cien
t,v
ersu
sH1
:ran
dom
effe
cts
are
inco
nsi
sten
t;th
en
um
ber
sin
par
enth
eses
are
thet-
stat
isti
csco
rrec
ted
for
het
eros
ced
asti
city
usi
ng
the
Wh
ite
(198
0)m
eth
od
Table VII.
MF37,12
1234
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
pronounced for the small firms. Consequently, we find some evidence supporting theH5, but only for the UK market and in the DDEI events.
Table VIII examines whether the relation between sentiment and the market reaction todividend change announcements is exacerbated for volatile firms relative to the otherones, in order to test the hypothesis that the impact of sentiment on the market reaction todividend change announcements is greater for high volatile firms (H6). The Portugueseresults indicate that the market reaction to dividend change announcements does not varyacross high and low sentiments periods as well as between stable or volatile firms. Thus,we find no evidence supporting H6 for the Portuguese sample.
The French results reveal that sentiment plays a greater role for stable firms, beingevident that the effect of sentiment is more pronounced for stable firms, disappearing forthe volatile firms. Consequently, the French evidence does not give support to theH6. Thesame conclusion is obtained for the UK market, as the interaction variables involvingsentiment that are statistically significant occurs for the stable firms sub-sample.However, the coefficient on DDEI_SENT is positive, contrary to the expected signal.Again, it seems that when dividend and earnings are announced together, the earningsgood news have more impact in the market reaction than the dividend bad news,suggesting that earnings have more information content than dividends.
Finally, Table IX shows the results for the separation of firms according to the BMratio, comparing the distressed and growth firms with the others, in order to test thehypothesis that the impact of sentiment on market reaction to dividend changeannouncements is greater for extreme growth and distressed firms than for mediumfirms (H7). We find no evidence supporting H7 for the Portuguese sample, as themarket reaction to dividend change announcements does not vary across high and lowsentiments periods as well as between medium and distressed or growing firms.
In what concerns the French sample, the only coefficient that captures the effect ofISENT that is statistically significant, is the DD_SENT (and negative) for thesub-sample of growth and distressed firms (ESI index), suggesting that the effect ofsentiment on market reaction to dividend decrease news is higher for growth anddistressed firms, relative to the firms with a medium value for the BM ratio, which is inaccordance with H7.
The UK results are not so clear. The results show that, considering the effect ofISENT, the market reaction to DIEI and DDEI events is higher for the medium BMratio firms, which is in contrast with the prediction of H7. However, share pricemovements following the DDED events are stronger for the growth and distressedfirms, in accordance with H7. All the UK interaction variables that have statisticallysignificant values are associated with the ISENT measure, being these regressionresults more robust.
In summary, our results in Tables VI-IX give no support for the hypothesis that theimpact of sentiment on market reaction to dividend change announcements is greaterfor high volatile firms, which contradicts the results of Baker and Wurgler (2006) andSankaraguruswamy and Mian (2008). However, we find some evidence that the role ofISENT is reinforced for young firms, small firms and growth and distressed firms, butonly for the French and the UK markets, and only for some of the dividend events.Consequently, the results are only partially consistent with the assumption that ISENTinfluence is stronger on market reaction for firms that are harder to arbitrage and moredifficult to value.
Market reactionto dividend news
1235
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
Sta
ble
-IS
EN
TV
olat
ile-
ISE
NT
Sta
ble
-ES
IV
olat
ile-
ES
IC
oeffi
cien
tt
Coe
ffici
ent
tC
oeffi
cien
tt
Coe
ffici
ent
t
Portugal-pooledOLS
Inte
rcep
t2
0.00
182
0.07
92
0.03
472
0.59
02
0.00
882
0.46
70.
0228
0.80
3D
I0.
0193
0.45
92
0.00
822
0.12
12
0.31
072
0.48
50.
1384
0.10
4D
D2
0.01
602
0.59
40.
3349
1.58
32
0.09
602
0.52
40.
9979
1.05
3D
I_S
EN
T2
0.00
212
0.17
80.
0058
0.29
50.
0031
0.50
32
0.00
122
0.09
6D
D_
SE
NT
20.
0065
20.
672
0.01
210.
158
0.00
100.
534
20.
0086
20.
958
SIZ
E2
0.00
012
0.04
10.
0017
0.55
30.
0003
0.30
52
0.00
122
0.78
5n
9558
238
238
Ad
just
edR
20.
021
0.08
20.
010
0.01
9France-pooledOLS
Inte
rcep
t0.
0119
0.42
32
0.04
652
1.33
80.
0072
0.25
72
0.04
432
1.25
9D
I2
0.07
832
0.21
00.
2479
0.22
32
0.49
62*
21.
713
0.64
460.
457
DD
0.29
66*
1.91
72
0.75
762
0.47
70.
5719
1.50
83.
1367
1.37
1D
I_S
EN
T0.
0471
0.11
92
0.21
962
0.24
00.
0503
*1.
680
20.
0606
20.
444
DD
_S
EN
T2
0.18
74*
*2
1.98
32
1.20
492
0.69
42
0.05
472
1.47
22
0.31
012
1.39
7S
IZE
20.
0010
20.
228
0.00
071.
330
20.
0004
20.
089
0.00
651.
223
n15
215
215
215
2A
dju
sted
R2
0.03
50.
054
0.42
00.
028
(continued
)
Table VIII.Regression of marketreaction to dividendchange announcements,considering the sentimentand firm-specificcharacteristics (panel C –volatility: stable versusvolatile firms)
MF37,12
1236
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
Sta
ble
-IS
EN
TV
olat
ile-
ISE
NT
Sta
ble
-ES
IV
olat
ile-
ES
IC
oeffi
cien
tt
Coe
ffici
ent
tC
oeffi
cien
tt
Coe
ffici
ent
t
UK-REM
Inte
rcep
t0.
0579
**
*3.
065
0.00
380.
174
0.05
74*
**
3.02
70.
0005
0.02
5D
IEI
0.00
170.
763
20.
0051
20.
806
20.
0408
20.
676
0.00
070.
005
DIE
D2
0.01
54*
21.
803
0.02
011.
706
20.
2560
20.
574
0.05
650.
140
DD
EI
20.
0530
21.
485
20.
0322
20.
442
21.
7329
**
22.
362
0.18
250.
689
DD
ED
20.
0242
20.
884
20.
0645
21.
510
20.
1698
20.
277
0.74
150.
454
DIE
I_S
EN
T0.
0024
0.86
50.
0045
0.79
60.
0004
0.67
32
0.00
012
0.00
6D
IED
_S
EN
T0.
0149
**
2.05
40.
0165
0.80
90.
0024
0.55
12
0.00
032
0.06
9D
DE
I_S
EN
T2
0.03
182
1.39
92
0.03
272
0.91
60.
0163
**
2.30
62
0.01
792
0.69
5D
DE
D_
SE
NT
0.02
420.
556
0.02
070.
863
0.00
140.
247
20.
0081
20.
501
SIZ
E2
0.00
66*
21.
736
0.00
110.
286
20.
0065
*2
1.70
70.
0015
0.38
2n
839
837
838
837
Ad
just
edR
20.
017
0.01
10.
016
0.00
8
Notes:
Sig
nifi
can
tly
dif
fer
from
zero
atth
e* 1
0,*
* 5,
**
* 1p
erce
nt
lev
els;
this
tab
lere
por
tsth
efo
llow
ing
reg
ress
ion
:
BHARið
1toþ
1Þ¼
aþ
b1
DI£d
Di;
0þ
b2
DD£d
Di;
0þ
b3
DI£d
Di;
0£
SE
NTþ
b4
DD£d
Di;
0£
SE
NTþ
b5
SIZ
Eiþ
1i;t
BH
AR
3is
the
bu
y-a
nd
-hol
dac
cum
ula
ted
abn
orm
alre
turn
onth
eth
ree-
day
per
iod
asca
lcu
late
db
yeq
uat
ion
(2);
DI
isa
du
mm
yv
aria
ble
that
tak
esv
alu
e1
ifd
ivid
end
incr
ease
san
dze
root
her
wis
e;D
Dis
ad
um
my
var
iab
leth
atta
kes
val
ue
1if
div
iden
dd
ecre
ases
and
zero
oth
erw
ise;
SE
NT
isa
mea
sure
ofIS
EN
Tat
the
end
ofth
ey
ear
bef
ore
the
div
iden
dch
ang
ey
ear
con
sid
erin
gb
oth
the
ES
Ian
dth
eIS
EN
Tin
dex
esan
dS
IZE
iis
the
size
for
shar
ei,
com
pu
ted
asth
en
atu
rall
ogof
tota
lass
ets
atth
een
dof
the
yea
rb
efor
eth
ed
ivid
end
chan
ge
yea
r;T
able
sV
I-IX
pre
sen
tth
eeq
uat
ion
resu
lts
for
sub
-sam
ple
sso
rted
onfi
rmag
e,si
ze,
vol
atil
ity
and
gro
wth
/dis
tres
sed
,re
spec
tiv
ely
;ag
eis
com
pu
ted
asth
en
um
ber
ofy
ears
sin
ceth
efi
rm’s
firs
tap
pea
ran
ceon
Dat
astr
eam
;v
olat
ilit
yis
the
shar
ere
turn
vol
atil
ity
,mea
sure
das
the
stan
dar
dd
evia
tion
ofd
aily
pri
ces
over
the
pre
ced
ing
yea
ran
dth
ep
rox
yfo
rg
row
th/d
istr
esse
dis
the
BM
rati
o,ca
lcu
late
db
yd
ivid
ing
boo
kv
alu
ep
ersh
are
atth
een
dof
the
yea
rb
efor
eth
ed
ivid
end
chan
ge
yea
rb
yth
em
ark
etp
rice
per
shar
eat
the
div
iden
dch
ang
ean
nou
nce
men
td
ate;
the
tab
lep
rese
nts
the
bes
tm
odel
amon
gp
oole
dO
LS
,FE
Man
dR
EM
.In
ord
erto
choo
seth
em
ost
app
rop
riat
em
odel
for
each
par
ticu
lar
sam
ple
,we
run
theF
-tes
t,a
test
for
the
equ
alit
yof
sets
ofco
effi
cien
ts,a
nd
the
Hau
sman
(197
8)te
st,a
test
wit
hH0
:ran
dom
effe
cts
are
con
sist
ent
and
effi
cien
t,v
ersu
sH1
:ran
dom
effe
cts
are
inco
nsi
sten
t;th
en
um
ber
sin
par
enth
eses
are
thet-
stat
isti
csco
rrec
ted
for
het
eros
ced
asti
city
usi
ng
the
Wh
ite
(198
0)m
eth
od
Table VIII.
Market reactionto dividend news
1237
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
Med
ium
-IS
EN
TD
/G-I
SE
NT
Med
ium
-ES
ID
/G-E
SI
Coe
ffici
ent
tC
oeffi
cien
tt
Coe
ffici
ent
tC
oeffi
cien
tt
Portugal-pooledOLS
Inte
rcep
t2
0.08
52*
21.
788
20.
0009
20.
041
20.
0358
21.
172
0.00
710.
425
DI
0.07
540.
386
0.02
100.
512
1.79
070.
633
20.
4281
20.
703
DD
0.02
570.
147
20.
0103
20.
396
0.45
990.
345
20.
0913
20.
517
DI_
SE
NT
20.
0320
20.
641
20.
0027
20.
227
20.
0150
20.
587
0.00
420.
721
DD
_S
EN
T2
0.00
612
0.12
22
0.00
472
0.49
82
0.00
402
0.31
70.
0010
0.55
9S
IZE
0.00
43*
1.73
42
0.00
012
0.07
40.
0019
1.14
22
0.00
042
0.46
9n
7012
523
638
0A
dju
sted
R2
0.07
50.
015
0.01
80.
008
France-pooledOLS
Inte
rcep
t2
0.00
192
0.04
82
0.01
322
0.62
02
0.00
722
0.18
82
0.01
552
0.73
0D
I2
0.10
78*
21.
893
20.
0180
20.
046
20.
6376
20.
151
20.
4103
21.
304
DD
0.66
950.
588
0.26
061.
461
0.65
870.
571
0.76
71*
1.76
8D
I_S
EN
T2
0.56
402
1.21
40.
0349
0.85
22
0.00
032
0.00
60.
0421
1.29
6D
D_
SE
NT
0.20
970.
349
20.
1907
21.
733
20.
0656
20.
547
20.
0742
*2
1.74
5S
IZE
0.00
150.
258
0.00
240.
743
0.00
200.
349
0.00
270.
819
n15
215
215
230
4A
dju
sted
R2
0.02
60.
014
0.01
80.
019
(continued
)
Table IX.Regression of marketreaction to dividendchange announcements,considering the sentimentand firm-specificcharacteristics (panel D –BM: medium versusdistressed/growth (D/G)firms)
MF37,12
1238
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
Med
ium
-IS
EN
TD
/G-I
SE
NT
Med
ium
-ES
ID
/G-E
SI
Coe
ffici
ent
tC
oeffi
cien
tt
Coe
ffici
ent
tC
oeffi
cien
tt
UK-pooledOLS
Inte
rcep
t0.
0574
**
*3.
536
0.04
05*
**
3.02
20.
0547
**
*3.
346
0.03
94*
**
2.93
8D
IEI
20.
0044
20.
723
20.
0001
20.
508
20.
0737
20.
497
20.
0020
0.14
9D
IED
20.
0098
21.
201
20.
0024
20.
173
20.
3897
20.
916
20.
0554
0.21
6D
DE
I2
0.05
782
0.93
12
0.03
172
1.04
22
0.11
342
0.66
92
0.10
041.
389
DD
ED
20.
0561
*2
1.65
32
0.00
852
0.42
50.
3590
0.29
20.
1929
0.36
9D
IEI_
SE
NT
0.00
500.
748
0.00
010.
216
0.00
070.
488
0.00
010.
142
DIE
D_
SE
NT
0.01
84*
**
2.80
80.
0125
0.71
00.
0037
0.90
50.
0006
0.23
0D
DE
I_S
EN
T2
0.07
74*
*2
2.05
22
0.02
092
1.06
92
0.01
122
0.69
60.
0095
1.35
3D
DE
D_
SE
NT
20.
0171
20.
973
0.03
39*
*2.
318
20.
0040
20.
329
20.
0020
0.39
2S
IZE
20.
0064
**
22.
200
20.
0046
*2
1.83
12
0.00
60*
*2
2.04
62
0.00
44*
*1.
760
n83
61,
679
836
1,67
9A
dju
sted
R2
0.02
50.
007
0.01
10.
004
Notes
:S
ign
ifica
ntl
yd
iffe
rfr
omze
roat
the
* 10,
** 5
,*
** 1
per
cen
tle
vel
s;th
ista
ble
rep
orts
the
foll
owin
gre
gre
ssio
n:
BHARið
1toþ
1Þ¼
aþ
b1
DI£d
Di;
0þ
b2
DD£d
Di;
0þ
b3
DI£d
Di;
0£
SE
NTþ
b4
DD£d
Di;
0£
SE
NTþ
b5
SIZ
Eiþ
1i;t
BH
AR
3is
the
bu
y-a
nd
-hol
dac
cum
ula
ted
abn
orm
alre
turn
onth
eth
ree-
day
per
iod
asca
lcu
late
db
yeq
uat
ion
(2);
DI
isa
du
mm
yv
aria
ble
that
tak
esv
alu
e1
ifd
ivid
end
incr
ease
san
dze
root
her
wis
e;D
Dis
ad
um
my
var
iab
leth
atta
kes
val
ue
1if
div
iden
dd
ecre
ases
and
zero
oth
erw
ise;
SE
NT
isa
mea
sure
ofIS
EN
Tat
the
end
ofth
ey
ear
bef
ore
the
div
iden
dch
ang
ey
ear
con
sid
erin
gb
oth
the
ES
Ian
dth
eIS
EN
Tin
dex
esan
dS
IZE
iis
the
size
for
shar
ei,
com
pu
ted
asth
en
atu
rall
ogof
tota
lass
ets
atth
een
dof
the
yea
rb
efor
eth
ed
ivid
end
chan
ge
yea
r;T
able
sV
I-IX
pre
sen
tth
eeq
uat
ion
resu
lts
for
sub
-sam
ple
sso
rted
onfi
rmag
e,si
ze,
vol
atil
ity
and
gro
wth
/dis
tres
sed
,re
spec
tiv
ely
;ag
eis
com
pu
ted
asth
en
um
ber
ofy
ears
sin
ceth
efi
rm’s
firs
tap
pea
ran
ceon
Dat
astr
eam
;v
olat
ilit
yis
the
shar
ere
turn
vol
atil
ity
,mea
sure
das
the
stan
dar
dd
evia
tion
ofd
aily
pri
ces
over
the
pre
ced
ing
yea
ran
dth
ep
rox
yfo
rg
row
th/d
istr
esse
dis
the
BM
rati
o,ca
lcu
late
db
yd
ivid
ing
boo
kv
alu
ep
ersh
are
atth
een
dof
the
yea
rb
efor
eth
ed
ivid
end
chan
ge
yea
rb
yth
em
ark
etp
rice
per
shar
eat
the
div
iden
dch
ang
ean
nou
nce
men
td
ate;
the
tab
lep
rese
nts
the
bes
tm
odel
amon
gp
oole
dO
LS
,FE
Man
dR
EM
.In
ord
erto
choo
seth
em
ost
app
rop
riat
em
odel
for
each
par
ticu
lar
sam
ple
,we
run
theF
-tes
t,a
test
for
the
equ
alit
yof
sets
ofco
effi
cien
ts,a
nd
the
Hau
sman
(197
8)te
st,a
test
wit
hH0
:ran
dom
effe
cts
are
con
sist
ent
and
effi
cien
t,v
ersu
sH1
:ran
dom
effe
cts
are
inco
nsi
sten
t;th
en
um
ber
sin
par
enth
eses
are
thet-
stat
isti
csco
rrec
ted
for
het
eros
ced
asti
city
usi
ng
the
Wh
ite
(198
0)m
eth
od
Table IX.
Market reactionto dividend news
1239
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
The evidence that the impact of sentiment on market reaction to dividend changeannouncements is greater for young and small firms, but only for the UK market,might be associated with the fact that Portuguese and French firms are smaller firms,with a smaller dispersion in size as well as in firms’ age.
Our results are only partially in accordance with the ones of Sankaraguruswamyand Mian (2008), who found that the positive (negative) market response to good (bad)earnings news increases (decreases) with sentiment. They found that the influence ofsentiment is stronger on the stock price response to earnings news for small youngfirms, volatile firms, and growth and value firms. Perhaps the difference in results canbe associated with the fact that we analyse the ISENT effect on market reaction todividend change news, whether they analyse this effect for earnings news, and, as weconclude, the last events have more information power than dividends.
Robustness of resultsTo evaluate the robustness of the results, we consider the use of the capital asset pricingmodel to compute the abnormal returns, in place of the BHAR. Furthermore, and in orderto reduce the impact of outliers, we consider the elimination of the observations that areless than 1 per cent or greater than 99 per cent of the distribution. In addition, we considermore control variables to regression models to enhance the reliability, such as the assettangibility (Baker and Wurgler, 2006, 2007) and earnings changes.
Finally, based on the fact that Baker and Wurgler (2004, 2006, 2007) points clearly toMB ratio differences between payers and non-payers as a strong indicator of marketsentiment regarding dividends, we consider another proxy for sentiment, which is usedby Baker and Wurgler. It consists on the log difference of the average MB ratio ofdividend payers and non-payers. We compute MB ratio as the inverse of the BM,already considered[12]. However, only one coefficient present significant values at the10 per cent level.
As the results do not differ substantially, our conclusions remain unchanged.
5. ConclusionsThe main relevant issue of our study is to analyse whether the market reaction todividend change announcements vary with the existing ISENT.
We start by analysing the relationship between ISENT and dividend changes,finding no support for the hypothesis that ISENT influences dividend changes.
We find some differences according to the analysed sample. For the Portuguesemarket, we find no evidence of ISENT influencing the market reaction to dividendchange announcements. For the French and the UK markets, we find mixed results.
We find some evidence that the market reaction to dividend change announcementsis more sensitive to dividend increases when sentiment is increasing for the UKmarket, and that market reaction to dividend change announcements is less sensitiveto dividend decreases when sentiment is increasing for the French market.
For the French market, the results suggest that the market reacts more to dividenddecrease news for small firms, which is in agreement with the hypothesis that the marketreaction to dividend change announcements is lower for large firms than for small firms.However, this evidence is only associated with the dividend decrease events.
We find some evidence supporting the hypothesis that the impact of sentiment onthe market reaction to dividend change announcements is greater for young and small
MF37,12
1240
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
firms, but only for the UK market. In addition, we find some evidence, but only for thedividend decrease events and for the French and the UK markets, that the impact ofsentiment on the market reaction to dividend change announcements is greater forextreme growth and distressed firms than for medium firms. However, our results donot support the hypothesis that the impact of sentiment on the market reaction todividend change announcements is greater for high volatile firms.
In addition, the results suggest that the ISENT proxy for invest sentiment is morerobust that the ESI measure.
Furthermore, the evidence suggests that earnings announcements have informationpower beyond that of dividend announcements, which is consistent with the conclusionof DeAngelo et al. (1992) and Conroy et al. (2000), among others.
Globally, our results are somewhat in agreement with the ones of Brown and Cliff(2004), who find no evidence of fund discounts reflecting ISENT when they use anISENT indicator from the American Association of Individual Investors, and find littleevidence of sentiment having forecasting power for near term returns, using a measureof sentiment constructed by them.
Notes
1. We define the dividend process to be a martingale, having the background in the reluctanceto change dividends evidence, which assumes that managers are averse to change dividendsunless they perceive substantial changes in the future economic situation of their firm. Thisproxy (naıve dividend changes) has been used in other studies, such as in Nissim and Ziv(2001), Benartzi et al. (2005) and Sankaraguruswamy and Mian (2008).
2. Although deflating the dividend change by the prior dividend is not unusual, deflating byprice is more prevalent in the literature and is likely to be a better measure. See Nissim (2003)for an extensive discussion of the merits of normalizing the change in dividends by price pershare.
3. Barber and Lyon (1997) investigated the bias sources in abnormal returns. They suggest thatcumulative abnormal returns (CARs) are subject to a measurement, a new listing and askewness bias, which all lead to positively biased test statistics. BHARs are subject to a newlisting, a skewness (which is worse than that for CARs) and a rebalancing bias, which leadsto negatively biased test statistics. However, in assessing these different biases, Barber andLyon (1997, p. 347) states that “we favor the use of BHARs on conceptual grounds”.
4. Gelper and Croux (2007) conclude that, although constructed in a rather ad hoc way, the ESIcan compete with other indicators constructed according to statistical principles.
5. The DG ECFIN conducts regular harmonized surveys for different sectors of the economiesin the EU to provide information for economic surveillance, short-term forecasting andeconomic research. The surveys provide information on a wide range of variables (forexample, production, business activity, consumer financial situation, unemployment,savings, among others) that are useful to monitor cyclical developments. The ESI is madewith a range of individual components of the industry, services, consumers, construction andretail trade confidence indicators. The economic sentiment data were collected in DG ECFINwebsite: http://ec.europa.eu/economy_finance/db_indicators/surveys/time_series/index_en.htm
6. Our variable set includes short- and long-term interest rates, consumption, inflation,exportations and importations, as well the lags of these variables.
7. The year of 1994 is conditioned by the availability of announcement dates on Bloombergdatabase. For the Portuguese sample we consider a longer period, in order to maximize the
Market reactionto dividend news
1241
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
number of observations, since this is a small market, with a small number of dividendevents. Because Bloomberg and Datastream lack information on the Portuguese market, weobtain data from Dhatis, an EL database and we also needed to collect some financialstatements directly from the companies.
8. For the UK market, dividends and earnings are usually announced in the same date. We,therefore, exclude the dividend events for which dividends and earnings information wereannounced on separate dates, which is a small number (six events). In addition, we adapt themethodology in order to separate the two effects (dividends and earnings).
9. We emphasise, for the Portuguese sample, the significant number of dividend decreases(about 32 per cent of sample events). One possible explanation for these sample statisticsmay be the exposure of emerging and Portuguese markets to more economic risks.
10. They found the following percentages for dividend increases, no change dividends anddividend decreases, respectively: 75, 15.7 and 9.3 per cent, for the period between 1989 and1993.
11. Aivazian et al. (2003) found the following percentages for dividend increases, no changedividends and dividend decreases, respectively, for the Thailand market: 47, 22.6 and 30.4per cent, and for the Korea market: 42, 14.6 and 43.4 per cent, both for the period between1981 and 1990.
12. Results are not reported here for the sake of brevity, but are available upon request.
References
Abeyratna, G. and Power, D.M. (2002), “The post-announcement performance ofdividend-changing companies: the dividend-signalling hypothesis revisited”, Accountingand Finance, Vol. 42, pp. 131-51.
Aharony, J. and Swary, I. (1980), “Quarterly dividend and earnings announcements andstockholders’ returns: an empirical analysis”, Journal of Finance, Vol. 35 No. 1, pp. 1-12.
Aivazian, V., Booth, L. and Cleary, S. (2002), “Why some firms smooth their dividends and whyothers don’t?”, working paper, Rotman School of Management, University of Toronto,Toronto.
Aivazian, V., Booth, L. and Cleary, S. (2003), “Do emerging market firms follow differentdividend policies from US firms?”, Journal of Financial Research, Vol. 26 No. 3, pp. 371-88.
Asquith, P. and Mullins, D.W. (1983), “The impact of initiating dividend payments onshareholder wealth”, Journal of Business, Vol. 56 No. 1, pp. 77-96.
Baker, M. and Wurgler, J. (2006), “Investor sentiment and the cross-section of stock returns”,Journal of Finance, Vol. 61 No. 4, pp. 1645-80.
Baker, M. and Wurgler, J. (2007), “Investor sentiment in the stock market”, Journal of EconomicPerspectives, Vol. 21, pp. 129-51.
Barber, B.M. and Lyon, J.D. (1997), “Detecting long-run abnormal stock returns: the empiricalpower and specification of test statistics?”, Journal of Financial Economics, Vol. 43 No. 3,pp. 341-72.
Barth, M.E. and Kallapur, S. (1996), “Effects of cross section scale differences on regressionresults in empirical accounting research”, Contemporary Accounting Research, Vol. 13,pp. 527-67.
Benartzi, S., Michaely, R. and Thaler, R. (1997), “Do changes in dividends signal the future or thepast?”, Journal of Finance, Vol. 52 No. 3, pp. 1007-34.
MF37,12
1242
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
Benartzi, S., Grullon, G., Michaely, R. and Thaler, R. (2005), “Dividend changes do not signalchanges in future profitability”, Journal of Business, Vol. 78 No. 5, pp. 1659-82.
Benesh, G.A., Keown, A.J. and Pinkerton, J.M. (1984), “An examination of market reaction tosubstantial shifts in dividend policy”, Journal of Financial Research, Vol. 7 No. 2, pp. 131-42.
Bernhardt, D., Douglas, A. and Robertson, F. (2005), “Testing dividend signaling models”,Journal of Empirical Finance, Vol. 12 No. 1, pp. 77-98.
Bhattacharya, S. (1979), “Imperfect information, dividend policy, and ‘the bird in the handfallacy’”, Bell Journal of Economics, Vol. 10 No. 1, pp. 259-70.
Brown, G.M. and Cliff, M.T. (2004), “Investor sentiment and the near-term stock market”, Journalof Empirical Finance, Vol. 11, pp. 1-27.
Brown, G.M. and Cliff, M.T. (2005), “Investor sentiment and asset valuation”, Journal of Business,Vol. 78, pp. 405-40.
Chen, G., Firth, M. and Gao, N. (2002), “The information content of concurrently announcedearnings, cash dividends, and stock dividends: an investigation of the Chinese stockmarket”, Journal of International Financial Management & Accounting, Vol. 13 No. 2,pp. 101-24.
Conroy, R.M., Eades, K.M. and Harris, R.S. (2000), “A test of the relative pricing effects ofdividends and earnings: evidence from simultaneous announcements in Japan”, Journal ofFinance, Vol. 55 No. 3, pp. 1199-227.
DeAngelo, H. and DeAngelo, L. (1990), “Dividend policy and financial distress: an empiricalinvestigation of troubled NYSE firms”, Journal of Finance, Vol. 45 No. 5, pp. 1415-31.
DeAngelo, H., DeAngelo, L. and Skinner, D.J. (1992), “Dividends and losses”, Journal of Finance,Vol. 47 No. 5, pp. 1837-63.
Dhillon, U.S. and Johnson, H. (1994), “The effect of dividend changes on stock and bond prices”,Journal of Finance, Vol. 49 No. 1, pp. 281-9.
Eddy, A. and Seifert, B. (1988), “Firm size and dividend announcements”, Journal of FinancialResearch, Vol. 11 No. 4, pp. 295-302.
Gelper, S. and Croux, C. (2007), “The predictive power of the European economic sentimentindicator”, working paper, available at: http://ssrn.com/abstract¼1093637
Gurgul, H., Madjosz, P. and Mestel, R. (2003), “GARCH modelling of Austrian stock marketreactions on dividend announcements”, Working Paper, VII Ogolnopolskie SeminariumNauukowe.
Han, B. (2008), “Investor sentiment and option prices”, Review of Financial Studies, Vol. 21 No. 1,pp. 387-414.
Harada, K. and Nguyen, P. (2005), “Dividend changes context and signalling efficiency in Japan”,Pacific-Basin Finance Journal, Vol. 13 No. 5, pp. 504-22.
Hausman, J.A. (1978), “Specification tests in econometrics”, Econometrica, Vol. 46 No. 6,pp. 1251-71.
Haw, I.H. and Kim, W. (1991), “Firm size and dividend announcement effect”, Journal ofAccounting, Auditing and Finance, Vol. 6 No. 3, pp. 325-47.
Jensen, M. (1986), “Agency costs of free cash flow, corporate finance, and takeover”, AmericanEconomic Review, Vol. 76 No. 2, pp. 323-9.
John, K. and Williams, J. (1985), “Dividends, dilution, and taxes: a signalling equilibrium”,Journal of Finance, Vol. 40 No. 4, pp. 1053-70.
La Porta, R., Lopes-de-Silanes, F., Shleifer, A. and Vishny, R.W. (1998), “Law and finance”,Journal of Political Economy, Vol. 106 No. 6, pp. 1113-55.
Market reactionto dividend news
1243
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
Lang, L.H.P. and Litzenberger, R.H. (1989), “Dividend announcements: cash flow signallingversus free cash flow hypothesis”, Journal of Financial Economics, Vol. 24 No. 1, pp. 181-91.
Lee, H.W. and Ryan, P.A. (2000), “The information content of dividend initiations and omissions:the free cash flow and dividend signaling hypotheses”, Journal of Research in Finance,Vol. 3 No. 2, pp. 196-277.
Lee, H.W. and Ryan, P.A. (2002), “Dividends and earnings revisited: cause or effect?”, AmericanBusiness Review, Vol. 20 No. 1, pp. 117-22.
Lemmon, M. and Portniaguina, E. (2006), “Consumer confidence and asset prices: some empiricalevidence”, Review of Financial Studies, Vol. 19, pp. 1499-529.
Lippert, R.L., Nixon, T.D. and Pilotte, E.A. (2000), “Incentive compensation and the stock priceresponse to dividend increase announcements”, Financial Review, Vol. 35 No. 4, pp. 69-94.
McCluskey, T., Burton, B.M., Power, D.M. and Sinclair, C.D. (2006), “Evidence on the Irish stockmarkets reaction to dividend announcements”, Applied Financial Economics, Vol. 16 No. 8,pp. 617-28.
Miller, M.H. (1986), “Behavioral rationality in finance: the case of dividends”, Journal of Business,Vol. 59 No. 4, pp. 451-68.
Miller, M.H. and Rock, K. (1985), “Dividend policy under asymmetric information”, Journal ofFinance, Vol. 40 No. 4, pp. 1031-51.
Mitra, D. and Owers, J.E. (1995), “Dividend initiation announcement effects and the firm’sinformation environment”, Journal of Business Finance & Accounting, Vol. 22 No. 4,pp. 551-73.
Nissim, D. (2003), “Discussion – reactions to dividend changes conditional on earnings quality”,Journal of Accounting, Auditing & Finance, Vol. 18 No. 1, pp. 153-61.
Nissim, D. and Ziv, A. (2001), “Dividend changes and future profitability”, Journal of Finance,Vol. 56 No. 6, pp. 2111-33.
Pettit, R. (1972), “Dividend announcements, security performance, and capital market efficiency”,Journal of Finance, Vol. 27 No. 5, pp. 993-1007.
Pettit, R. (1976), “The impact of dividends and earnings announcements: a reconciliation”,Journal of Business, Vol. 49 No. 1, pp. 86-96.
Qiu, L. and Welch, I. (2006), “Investor sentiment measures”, working paper, available at: http://ssrn.com/abstract¼589641
Sankaraguruswamy, S. and Mian, G.M. (2008), “Investor sentiment and stock market response tocorporate news”, working paper, available at: http://ssrn.com/abstract¼1107619
Schmeling, M. (2009), “Investor sentiment and stock returns: some international evidence”,Journal of Empirical Finance, Vol. 16, pp. 394-408.
Shleifer, A. and Vishny, R.W. (1997), “The limits to arbitrage”, Journal of Finance, Vol. 52 No. 1,pp. 35-55.
Travlos, N., Trigeorgis, L. and Vafeas, N. (2001), “Share-holder wealth effects of dividend policychanges in an emerging stock market: the case of Cyprus”, Multinational Finance Journal,Vol. 5 No. 2, pp. 87-112.
Tudor, C. (2008), “Firm-specific factors as predictors of future returns for Romanian commonstocks: empirical evidence”, working paper, available at: http://ssrn.com/abstract¼1277213
Vieira, E. and Raposo, C.C. (2007), “Signalling with dividends? The signalling effects of dividendchange announcements: new evidence from Europe”, working paper, available at: http://ssrn.com/abstract¼955768
MF37,12
1244
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
White, H. (1980), “A heteroscedasticity-consistent covariance matrix estimator and a direct testfor heteroscedasticity”, Econometrica, Vol. 48 No. 4, pp. 149-70.
Yilmaz, M.K. and Gulay, G. (2006), “Dividend policies and price-volume reactions to cashdividends on the stock market: evidence from the Istanbul stock exchange”, EmergingMarkets, Finance & Trade, Vol. 42 No. 4, pp. 19-49.
Yu, J. and Yuan, Y. (2010), “Investor sentiment and the mean-variance relation”, Journal ofFinancial Economics (in press).
About the authorProfessor Elisabete Simoes Vieira is a Teacher of ISCA and DEGEI at the University of Aveiro.She is a member of the GOVCOOP investigation unit and is teaching and researching in financesubjects. She is a member of the Financial Analysts Association and member of a ScientificCommittee of a Portuguese Review. Elisabete Simoes Vieira can be contacted at: [email protected]
To purchase reprints of this article please e-mail: [email protected] visit our web site for further details: www.emeraldinsight.com/reprints
Market reactionto dividend news
1245
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
This article has been cited by:
1. Elisabete F. Simões Vieira, Márcia S. Valente Pereira. 2014. Herding behaviour and sentiment: Evidencein a small European market. Revista de Contabilidad . [CrossRef]
Dow
nloa
ded
by I
STA
NB
UL
UN
IVE
RSI
TY
At 0
2:09
20
Dec
embe
r 20
14 (
PT)
