03.vvvvv

8/12/2019 03.vvvvv

1/29

O R I G I N A L R E S E A R C H

Market uncertainty, market sentiment,

and the post-earnings announcement drift

Ron Bird Daniel F. S. Choi Danny Yeung

Springer Science+Business Media New York 2013

Abstract Post-earnings announcement drift (PEAD) which was first identified over

40 years ago seems to be as much alive today as it ever was. Numerous attempts have been

made to explain its continued existence. In this paper we provide evidence to support a new

explanation: that the PEAD is a reflection of the level of market uncertainty and sentiment

that prevails during the post-announcement period. The overriding conclusion from our

analysis is that both uncertainty and sentiment play a central role in determining investor

behaviour and it is this behaviour that ultimately determines the pricing that is observed infinancial markets.

Keywords Anomalies Post-earrings announcement drift Uncertainty Sentiment

JEL Classification G12 G14 D81

1 Introduction

Post-earnings announcement drift (PEAD) is the oldest continuing market anomaly, dating

back to the first event study published over 40 years ago (Ball and Brown 1968). Ball

(1978) notes that at least 20 papers in the decade post-Ball and Brown found evidence of a

PEAD. If we move forward almost 35 years, empirical studies are still finding evidence of

a PEAD with stocks reporting that good news continues to be associated with positive

excess returns for an extended period beyond the earnings announcement date and bad

R. Bird (&) D. YeungPaul Woolley Centre for the Study of Capital Market Dysfunctionality, University of Technology

Sydney, Sydney, Australia

e-mail: [email protected]

D. Yeung


R. Bird D. F. S. ChoiWaikato Management School, The University of Waikato, Hamilton, New Zealand


1 3

Rev Quant Finan Acc

DOI 10.1007/s11156-013-0364-x

8/12/2019 03.vvvvv

2/29

news continues to realise negative excess returns for an extended period (Ali et al. 2008;

Konchitchki et al. 2010; Forner et al. 2009). A number of authors have suggested that

PEAD (along with momentum), having proved robust across time, markets and method-

ologies (e.g., Fama 1998; Kothari 2001), represents the most serious challenge to the

Efficient Market Hypothesis (EMH).Since Ball (1978), a continuing stream of authors has attempted to explain PEAD, either

within a rational expectations framework or by appealing to behavioural explanations.

Explanations proposed include arbitrage risk (Mendenhall 2004), liquidity risk (Sadka

2006), and unsophisticated investors (Bartov et al. 2000). Two explanations that have

gained much attention in recent years have been information uncertainty (Francis et al.

2007) and investor sentiment (Livnat and Petrovitis2009). The underlying proposition in

both cases is that the uncertainty or sentiment at the time of the information release impacts

on the market reaction at that time with the PEAD representing the subsequent market

correction to the initial mispricing. For example, high uncertainty or low sentiment will

result in an initial underreaction to good news with the subsequent upward drift (i.e.

PEAD) representing the market correction to this initial underreaction.

Recent studies find evidence of an asymmetric response at the time of information

release, with the market reacting more to bad news than to good news. One line of research

attributes the asymmetry to the pessimism induced by the level of uncertainty at the time

the information becomes available (Williams 2009; Kim et al. 2010; Bird and Yeung

2012). A second line of research finds that the market sentiment at the time of the

information release also results in an initial asymmetric response to information. Mian and

Sankaraguruswamy (2008) and Jiang (2011) both find that markets are more responsive to

good news at the time of its release when sentiment is high and more responsive to badnews when sentiment is low.

Bird and Yeung (2012) examine the combined impact of market uncertainty and market

sentiment on how investors react to information. The usual presumption is that investors

react positively to good news and negatively to bad news. However, Bird and Yeung find

that investors largely ignore bad news if it is released at a time when market uncertainty is

low and market sentiment is high, while investors actually react negatively to good news

when it is released at a time when market uncertainty is high and sentiment is low. In the

spirit of Francis et al. (2007) and Livnat and Petrovitis (2009), this would suggest the

potential for significant PEAD depends on the levels of uncertainty and sentiment at the

time of the information release.Previous empirical studies have highlighted that the PEAD is impacted by the level of

uncertainty and/orby the sentiment at the time of the information release. In this paper we

extend these findings by evaluating the combined effect on the PEAD of the prevailing

uncertainty and sentimentover the post-announcement period. Therefore, our study differs

from those that come before in two important ways: (i) it is the first to examine the

combined effect of uncertainty and sentiment on the PEAD, and (ii) it is the first to

examine the impact of uncertainty and sentiment over the post-announcement period rather

than just at the time of the information release.

We find significant evidence that during the post-announcement period: (i) the strongestdownward drift after a bad earnings announcement occurs when uncertainty is high and

sentiment is low, and (ii) the strongest upward drift following good news announcements

occurs during periods when uncertainty is low and sentiment high. The best example of the

combined impact of prevailing uncertainty and sentiment on the PEAD is our finding that

there is a strong downward drift after the release of good news during periods when

uncertainty is high and sentiment is low.

R. Bird et al.

1 3

8/12/2019 03.vvvvv

3/29

Our results also contribute to our understanding of what influences the speed with which

the market adjusts to information releases. By confirming the existence of a PEAD, our

findings support previous evidence that the market typically underreacts to most infor-

mation sources including both good news and bad news earnings announcements

(Kadiyala and Rau 2005). Such a conclusion is at variance with those of other studieswhich find evidence that there is a greater initial response to bad news than good news at

times of high uncertainty, and suggest that this is because investors take a pessimistic view

at such times and so underreact to good news but overreact to bad news (Williams2009;

Kim et al.2010). This pessimism is being driven by investors following maxmin expected

utility (MEU) and so they base their decisions on the worst case outcomes at times of high

uncertainty (Gilboa and Schmeidler 1989).

We do find evidence that, consistent with MEU, there is a larger downward drift during

the announcement period after bad news than an upward drift after good news and that this

asymmetry increases as uncertainty increases. However, we also find that the initial

reaction at the time of the earnings announcement represents a higher proportion of the

total adjustment over the 60 trading days post the announcement in the case of good news

announcements as compared to bad news announcements. In other words we find evidence

of both a greater response to bad news as compared to good news at the time of the

information release but also evidence to suggest that the initial underreaction to bad news

is greater than what it is for good news. Both of these seemingly inconsistent findings are

strengthened when the information is released at a time of high market uncertainty and are

explained by the fact that overall there is a much greater adjustment to bad news than there

is to good news.

The remainder of the paper is structured as follows: Section 2 reviews the PEADliterature, concentrating on the explanations postulated to explain its existence. Section 3

sets out the data and methodology employed in the study. Section 4reports and discusses

the findings. Section5presents the concluding remarks and discusses possible future work

in the area.

2 Literature review on post-earnings announcement drift

Information efficiency implies thatmarkets quickly impound information intoprices. However,

much of the empirical evidence indicates that the adjustment process can be quite slow,extending over several months or in some cases, several years. Ball and Brown (1968) were the

first to identify a continuing drift in returns subsequent to an announcement (PEAD) and Ball

(1978) notes that PEAD was apparent in at least 20 studies conducted over the next decade.

Today, the continuing existence of PEAD remains an anomaly, as evidenced by the recent

findings of Ali et al. (2008), Konchitchki et al. (2010) and Forner et al. (2009). Therefore, it is

not surprising that Kothari (2001) concluded his survey paper by saying the PEAD anomaly

poses a serious challenge to the efficient markets hypothesis. It has survived a battery of tests

and many attempts to explain it away.

Over the last 35 years we have seen a continuing stream of papers that attempt toexplain the PEAD, with many such as Ball (1978) drawing upon model misspecification

and other reasons to argue that it does not necessarily represent a departure from the EMH.

In contrast, Barnard and Seyhun (1997) using a stochastic dominance approach put aside

the model misspecification argument and suggest that it is difficult to present a case that

PEAD does not represent a departure from the EMF. One explanation for the PEAD

phenomenon is that good news companies are inherently more risky than bad news

The post-earnings announcement drift

1 3

8/12/2019 03.vvvvv

4/29

companies (Bernard and Thomas 1989). Bernard and Thomas conclude that at best, risk

could explain only a small proportion of the PEAD and this conclusion has gone largely

unchallenged. Another line of explanation dating back to Ball (1978) argues that the PEAD

is just an artefact of the methodology and/or the data employed to calculate abnormal

returns rather than an indication of any market inefficiency (Jones and Litzenberger 1970;Jacob et al.2000). However, the results have remained robust despite the use of numerous

alternative methodologies and the data problems are no longer a concern. Another pos-

sibility is that difficulties in implementation and/or transaction costs could mean that it is

impossible to profit from the perceived opportunities to exploit the PEAD. This seems

unlikely as Bernard and Thomas (1989) demonstrate how the PEAD can be exploited

following a very low turnover strategy. However, more recent studies have suggested that

the PEAD may at least be partially explained by the high costs of arbitrage (Mendenhall

2004), and liquidity-related risk (Sadka2006).

All explanations for the PEAD discussed to date are attempts to reconcile the evidence

on PEAD with the efficient markets hypothesis. One proposal suggestive of inefficiencies

in markets is that investors just get it wrong and consistently underreact to both good and

bad earnings news. Bernard and Thomas (1990) suggest that investors adopt a very nave

approach when evaluating new earnings numbers and fail to recognise their full implica-

tions for future earnings. This explanation is consistent with the possibility that less

sophisticated investors might be driving the PEAD. Bartov et al. ( 2000) provide some

support for this premise when they find a negative relationship between the level of

institutional holdings (sophisticated investment) and the level of the PEAD. Other authors

have taken a behavioural approach and attempted to explain the PEAD as being due to one

or more of the cognitive biases attributed to investors. Examples of this include Frazzini(2006) who demonstrates a link between PEAD and the disposition effect and Barberis

et al. (1998) who explain it in terms of conservative and representativeness biases.

Clearly we are far from achieving closure as to the factors that drive the continued

existence of the PEAD. One recent explanation that we are yet to consider is that the PEAD

is driven by investor uncertainty as to how to interpret information. The argument is that

investors uncertainty about the quality of an information signal causes them to underreact

at the time of the release of the information, and as uncertainty is resolved, we will begin to

see the full reaction to the information and so the PEAD is created.1 A number of studies

provide empirical support for this proposition by confirming that information uncertainty is

positively related to the magnitude of the PEAD (Zhang 2006; Francis et al. 2007;Anderson et al. 2007; Angelini and Guazzarotti 2010).

Another line of research has examined the impact of sentiment on the PEAD (Livnat

and Petrovitis 2009) with similar reasoning that the initial reaction to the release of

information will be conditioned by the sentiment in the market and the PEAD reflects a

subsequent price adjustment. For example, Mian and Sankaraguruswamy (2008) and Jiang

(2011) both find that there is a stronger reaction to good news when the information is

released at a time when sentiment is strong. Consistent with these results, Livnat and

Petrovitis (2009) find a greater positive PEAD following a good news announcement made

when market sentiment is weak than when market sentiment is strong.One unique feature of this study is that we measure the impact of market uncertainty

(and sentiment) by how investors react to information over the post-announcement period.

Previous studies have considered uncertainty at the firm level, basing their measure of

uncertainty on factors such as the companys use of accruals, its size, its return volatility

1 The conceptual argument for this proposition can be found in Caskey 2009.

R. Bird et al.

1 3

8/12/2019 03.vvvvv

5/29

and the dispersion of analysts earnings forecasts. The value of a company at any time is

contingent on the influence of thousands of factors which impact on the future profitability

and risk characteristics of the firm. The ability of investors to cope with all of these factors

varies significantly over time and becomes extremely difficult in the aftermath of certain

events such as the 9/11 disaster, the collapse of Lehman Brothers, and the threats to theviability to the European Economic Union. The thesis of this paper is that the interpretation

that the market places on any information is conditioned by the level of market uncertainty

at the time of, and subsequent to, the information release.

Another unique feature of this paper is that it looks at the combined effect of market

uncertainty and market sentiment on the price behaviour of a stock during the post-

announcement period. Bird and Yeung (2012) is the only paper that has considered the

combined effect of market uncertainty and sentiment on how investors react to information

at the time of its release. In this paper we extend this analysis to examine how uncertainty

and sentiment work together to explain the PEAD, and whether they have separate effects

or whether one just proxies for the other.

The third unique feature of this paper is that it examines the importance of the pre-

vailing market uncertainty and market sentiment during the post-announcement period in

explaining the PEAD. Previous studies have only considered the impact that uncertainty or

sentiment at the time of the information release has on initial price adjustment to the

information and so gives rise to the mispricings that are the basis for the subsequent price

adjustments. This might be a satisfactory way to address the general question of why there

is a PEAD but fails to adequately explain the post-announcement price behaviour of

particular stocks. We propose that the behaviour of particular stocks will also be influenced

by the prevailing level of market uncertainty and sentiment during the post-announcementperiod.

3 Data and methodology

The research question in this paper is: Does the level of both market uncertainty and

market sentiment prevailing over the post-announcement period impact on the price

behaviour of the stock during this period (i.e., the PEAD)? To answer this question we will

test the following hypotheses:

Hypothesis 1 High market uncertainty during the post-announcement period will

increase the downward drift associated with a bad news announcement but mitigate the

upward drift associated with a good news announcement.

Hypothesis 2 High market sentiment during the post-announcement period will tends to

nullify any PEAD associated with bad news stocks and magnify any PEAD associated with

good news stocks.

In other words we are proposing the prevailing market uncertainty and sentiment over

the post-announcement period will play a major role in explaining the PEAD. Studies have

shown that different market conditions can elicit different investor reactions to identicalevents (Klein and Rosenfeld 1987; Docking and Koch 2005) and that the timing of the

information release will also affect the reaction because of uncertainty resolution

(Sharathchandra and Thompson 1995; Eden and Loewenstein 1999). High sentiment

creates a climate of optimism (Baker and Wurgler 2006), while uncertainty breeds pes-

simism. So the twin forces of sentiment and uncertainty will tend to counteract each other.


1 3

8/12/2019 03.vvvvv

6/29

In our analysis we will also provide insights into the relative importance of the levels of

uncertainty and sentiment at the time of the announcement as compared with the levels of

uncertainty and sentiment prevailing during the post-announcement period.

By examining the relationship between market uncertainty and market sentiment over

sixty trading days post-announcement,2

we will provide useful insights into a number ofother important questions:

Is the initial response of the market to new information consistent with market

efficiency?

Is the initial market response to new information an underreaction or an overreaction,

and to what extent does this response vary depending on whether news is good or bad?

Is there an asymmetric response to bad and good news during the post-announcement

period?

Is the market better at quickly incorporating one type (bad or good) of information into

prices than it is the other type?

3.1 Data

The sample period used in this study extends from January 1986 to September 2009. We

use three types of data: data from the equity market, data from the options market, and

accounting data. The return data from the equity market were obtained from CRSP through

WRDS. Our measure of market uncertainty is the Implied Volatility Index (VIX) from

CBOE.3 The accounting data which includes reported earnings were obtained from the

CRSP/COMPUTSTAT merged database which is sourced through WRDS. Finally,information on actual earnings and financial analysts earnings forecasts were sourced from

the IBES summary.

To be included in the final sample, we required the firms to have issued earnings

announcements in each of the previous five quarters. We also required information on firm

characteristics (such as book-to-market and firm size), VIX and firm returns at the time of

the earnings announcements. Finally, in order to reduce the impact of outliers, firm

characteristics have been trimmed at the 1st and 99th percentiles.

In the following section, we provide a brief discussion of the calculation of the three

major variables used in our study: (i) unexpected earnings, (ii) market uncertainty, and

(iii) market sentiment:

3.1.1 Unexpected earnings (UE)

The study revolves around evaluating stocks returns in the period after the release of an

earnings announcement. More specifically, we study how uncertainty and sentiment play

vital roles in determining the PEAD. Central to our analysis is the unexpected component

of the earnings announcement which we measure as the difference between the actual EPS

and the consensus earnings estimate in the month immediately prior to the announcement

(Han and Wild1990; Francis et al.2007; Kaestner2006). So the unexpected portion of the

earnings announcement of firm i can be expressed as:

2 We also examined the same relationship over the 30 trading days post announcement period and the

results were basically unchanged.3 For a detailed explanation of the calculation of the PEAD, see Williams (2009).

R. Bird et al.

1 3

8/12/2019 03.vvvvv

7/29

Unexpected earningsi Actual EPSiExpected earningsi

where Expected Earnings is defined as the Consensus EPS Estimate for firm i.

Consistent with the literature (Kaestner2006), we scaled the unexpected earningsby the

absolute value of actual EPS to arrive at our final measure of unexpected earnings.4 The

scaled unexpected earnings measure is therefore:

UEi Actual EPSiExpected EPSi

Actual EPSi

The scaling of the unexpected earnings standardises earnings surprises across our

sample and thus allows us to examine the influence of news on the returns of the firms.5 A

positive unexpected earnings event (PUE) occurs when the earnings just announced exceed

expected earnings. Similarly, a negative unexpected earnings event (NUE) occurs when the

earnings just announced fall short of expected earnings.

3.1.2 Market uncertainty

Using a suitable proxy for market uncertainty is critical in this paper. Francis et al. (2007)

have used the quality of information emanating from a firm as indicated by its use of

accruals to proxy for uncertainty, while others have used disagreement among experts such

as analysts as a measure of the difficulty market participants had in interpreting the

implications of the information (Barron et al. 1998; Zhang 2006). However, all these

proxies are designed to measure uncertainty at the firm level whereas we required a

market-wide measure of uncertainty. Anderson et al. (2009) obtain such a measure by

aggregating the analysts earnings forecasts for all firms and using the dispersions in these

aggregated forecasts as a quarterly macro-measure of uncertainty. However, Anderson

et al.s measure cannot be calculated on the daily basis required in this study.6 In this study,

we have measured uncertainty by the implied volatility from the options market (i.e. VIX)

which is used by Williams (2009), Drechsler (2009), Bird and Yeung (2012), and Kim

et al. (2010), as this is available on a daily basis.7 Although some critics have suggested

that VIX provides an estimate of risk rather than uncertainty, we contend that recent

studies suggest otherwise. Drechsler (2009) provides support for VIX through a general

equilibrium model that incorporates time-varying Knightian uncertainty. The model

explains that a large hedging/variance premium is evidenced in stock markets. Drechslerargues that the large time-varying option premium (reflected in the implied volatility) is

consistent with investors using options for protection against uncertainty (and time-

variation in uncertainty). He shows through calibration that fluctuations in the variance

premium reflect changes in the level of uncertainty.

4 We also tried several other measures of unexpected earnings, including the unscaled unexpected earnings

and SUE the unexpected earnings standardised by the standard deviation of analysts estimates with similarfindings.5 Both Williams (2009) and Bird and Yeung (2012) adopted a similar methodology to standardise earnings

surprises prior to analysing the impact of uncertainty on investors behaviour.6 Another problem with the uncertainty proxy used by Anderson et al. (2009) is that it can be affected by a

number of other factors such as the heterogeneous beliefs of the analysts.7 VIX is calculated continually through the day but we use the level of VIX as at the end of each day.


1 3

8/12/2019 03.vvvvv

8/29

3.1.3 Market sentiment

High investor sentiment has the potential to mitigate some of the negative effects of

uncertainty. Baker and Wurgler (2007) have developed a model for measuring the overall

level of investor sentiment and used this measure to establish that investors take an overlyoptimistic stance to pricing stocks when sentiment is high but they take a much more

subdued stance when sentiment is low. The problem in using the Baker and Wurgler

sentiment index is that it cannot be calculated with sufficient frequency to capture short-

term variations in sentiment through time. As we wish to capture sentiment at the market

level we use the S&P 500 index returns realised over the post-announcement period to

proxy for sentiment.

3.2 Methodology

The basic model used in our analysis to establish the association between the PEAD and

unexpected earnings is:

Rit bob1NUEitb2PUEitb3log MVit b4BTMVitYear Effects eit 1

where Rit denotes the accumulated excess return8 over the post-announcement period

which commences on the second day after the announcement and ends on the 60th trading

day after the announcement (i.e., t ? 2 to t ? 60).9

PUE is calculated by multiplying the unexpected earning by a dummy variable which

takes the value of 1 if there are positive earnings surprises and 0 otherwise. Similarly, a

negative unexpected earnings (NUE) event occurs when the earnings just announced fallshort of expected earnings. So if UE[ 0, PUE = UE, otherwise PUE = 0. MV represents

the market capitalisation at the time of the announcement and is measured in millions.

BTMV measures the book-to-market value of the firm at the time of the announcement.

With no drift, the coefficients b1 and b2 are expected to be not significantly different

from zero.

We next test the extent to which the level of PEAD is affected by the level of uncertainty

(VIX) at the time of the announcement and the extent to which uncertainty changes (DVIX)

over the post-announcement period. In order to do this, we determine the level of VIX at the

time of each announcement and the change in VIX over the post-period. We expand Eq. (1) to

incorporate these two additional variables into the following regression equation:

Rit bo b1NUEit b2PUEitb3X1NUEitb4X1PUEitb5X2NUEit b6X2PUEit

b7log MVit b8BTMVitYear Effects eit

2

where X1 = 1 when firm i makes an earnings announcement at time t and the level of

uncertainty (level of VIX) at time t is above the median level of VIX when all levels of

VIX are ranked from low to high; otherwise, X1 = 0; X2 = 1 where there is an increase in

the level of VIX over the post-announcement period (as measured by the difference

8 The excess return is calculated on a daily basis as the difference between the daily return on a particular

stock and the daily return on the S&P500 index.9 We also undertook the same analysis using a post-announcement period extending from the second day

after the announcement to the 30th trading day after the announcement. As the findings were the same we

only report our findings for the longer post-announcement period.

R. Bird et al.

1 3

8/12/2019 03.vvvvv

9/29

between the level of VIX at the time of announcement, VIX t, and the level of VIX 60 days

post announcement, VIXt?60); otherwise X2 = 0.

We will define high uncertainty as being where the level of VIX is above the median

level at the time of the announcement and increases over the post-announcement period.

Similarly we define low uncertainty being where the level of VIX is below the medianlevel at the time of the announcement and decreases over the post-announcement period.

We next expand our analysis to incorporate market sentiment into the analysis. We do

this by introducing as an additional variable the momentum over the post-announcement

period. Our expanded regression equation is set out below:

Rit bob1NUEitb2PUEitb3X1NUEitb4X1PUEitb5X2NUEitb6X2PUEit

b7X3NUEitb8X3PUEitb9X4NUEit b10X4PUEitb11log MVit

b12BTMVitYear Effects eit

3where X3 = 1 if the return on the S&P 500 Index from t ? 2 to t ? 60 is in the second

tercile when all S&P 500 Index returns {t ? 2, t ? 60} are ranked from low to high;

otherwise X3 = 0; X4 = 1 if the return on S&P 500 Index returns from t ? 2 to t ? 60

ranks in the third tercilewhere all S&P 500 Index returns {t ? 2, t ? 60} are ranked from

low to high; otherwise X4 = 0.

The final part of our analysis involves separating two sub-samples from our total

sample. One sub-sample comprises those announcements made at a time when the market

uncertainty was in the lowest tercile, and the market sentiment was in the highest tercile,

across our whole sample. These are the conditions where we would expect to see thegreatest initial market response to good news and the lowest market response to bad news.

Based on the findings of previous research, we would expect for this sub-sample to see a

much larger PEAD associated with bad news than with good news.

The other sub-sample comprises those earnings announcements made at a time when the

market uncertainty was in the highest tercile, and the market sentiment was in the lowest

tercile, across our whole sample. These are the conditions where we would expect the

greatest initial market response to bad news but the most muted market response to good

news. Based on the initial price adjustment, we would expect to see for this sub-sample a

much larger PEAD associated with good news than with bad news.

We complete the analysis by analysing the impact of the market uncertainty and sen-timent prevailing over the whole post-announcement periodon the PEAD experienced by

the stocks in each of the two sub-systems. By so doing we have a great opportunity to

examine how these post-announcement conditions impact on the PEAD experienced by

individual stocks (both in absolute terms and relative to the impact of the level of

uncertainty and sentiment at the time of the announcement). The regression equation that

we investigate for each sub-sample is:

Rit bo b1NUEit b2PUEitb5X2NUEitb6X2PUEitb7X3NUEitb8X3PUEit

b9X4NUEitb10X4PUEit b11log MVit b12BTMVitYear Effects eit

4

where: X2 = 1 where there is an increase in the level of VIX over the post-announcement

period (as measured by the difference between the level of VIX at the time of

announcement (VIXt), and the level of VIX 60 days post announcement (VIXt?60));

otherwise, X2 = 0; X3 = 1 if the return on the S&P 500 Index from t ? 2 to t ? 60ranks


1 3

8/12/2019 03.vvvvv

10/29

in the second tercilewhen all S&P 500 Index returns {t ? 2, t ? 60} are ranked from low

to high; otherwise X3 = 0; X4 = 1 if the return on S&P 500 Index from t ? 2 to t ? 60

ranks in the third tercilewhere all S&P 500 Index returns {t ? 2, t ? 60} are ranked from

low to high; otherwise X4 = 0.

3.2.1 Summary statistics

Our final sample set comprises 325,888 observations of quarterly earnings announcements.

Summary statistics for our final sample are reported in Table 1. It can be seen that the

magnitude of bad news is approximately twice as large as it is for good news with this

proportion remaining fairly constant across all levels of uncertainty (VIX) and sentiment

(MOM). There is only a slight variation in the size of the firms making announcements

across the various sub-samples, the major departure being the preponderance of smaller

firms that make earnings announcements at times when uncertainty is low. Finally, the

greatest variation highlighted in Table1is that growth stocks are far more likely than value

stocks to release their earnings figures during periods when markets are experiencing low

uncertainty.

4 Empirical results

In this section, we show that our empirical results affirm the existence of a PEAD in our

sample data, that the PEAD is influenced by the level of market uncertainty and market

sentiment prevailing over the post-announcement period, and that the PEAD for small capand growth stocks is both greater and more impacted by the joint effects of uncertainty and

sentiment than large cap and value stocks.

4.1 PEAD

The first question that we address is whether there is a post-earnings announcement drift in

our data. In order to evaluate this we applied our data to Eq 1. The coefficients reported for

NUE and PUE in Table2are both significant and positive. Our findings thus confirm the

existence of a PEAD andof the signs expected to be associated with the release of good

and bad earnings news.10 The most interesting finding is that the magnitude of the coef-ficient attached to the bad news announcements is significantly larger to that attached to the

good news announcements. The implication of this finding is that the market underreacts to

both bad and good news earnings announcements but there is a greater underreaction to

bad news than to good news. In order to investigate this further we ran the same regression

as for Table2but this time with excessive return over the 3-day announcement variables

(i.e. t - 1 to t ? 1) as the dependent variable. We found the coefficient on both NUE and

PUE to be positive and highly significant with little difference in their magnitude. By

comparing these coefficients to those reported in Table 2, we conclude that about half of

10 As can be seen from Table2, the coefficients attached to both the control variables are significant. Indeed

in all of our egressions there is a positive coefficient attached to the size variable and a significant negative

coefficient attached to the value/growth variable. In the interest of clear exposition, we do not report the

coefficient for these variables in future tables.

R. Bird et al.

1 3

8/12/2019 03.vvvvv

11/29

8/12/2019 03.vvvvv

12/29

the reaction of the market to bad news over the 62 trading days inclusive of the

announcement period occurred during the announcement period, whereas in the case of

good news two-thirds of the reaction took place during the announcement period. This

finding is also consistent with a slightly larger underreaction to bad news than to good

news.

4.2 Market uncertainty at the time of the announcement

Although previous studies conclude that the market response to the release of informationis impacted by the level of uncertainty at the time, there is disagreement as to the nature of

this impact. Some studies claim that uncertainty causes investors to underreact to both bad

and good news, with the PEAD reflecting a subsequent adjustment to the information (e.g.,

Francis et al. 2007) whereas other studies claim that uncertainty causes investors to take a

pessimistic stance and so overreact to bad news and underreact to good news (e.g., Williams

2009). Although these two explanations both suggest a subsequent upward adjustment to

good news, the former suggests a further downward adjustment following a bad news

announcement while the latter suggests a correction with a subsequent drift upwards in price

(e.g., Francis et al. 2007; Williams 2009; Bird et al. 2011). We evaluate these propositions by

running the following regression which is a reduced form of Eq.2:

Rit bob1NUEitb2PUEitb3X1NUEitb4X1PUEitb5X2NUEitb6X2PUEit

b7log MVit b8BTMVitYear Effects eit

Table 2 Analysis of post-earnings announcement drift (PEAD)

Variable Coefficient

NUE 0.0151***

PUE 0.0091***Ln(MV) -0.0022***

BTMV 0.0079***

Test of difference NUE[PUE***

The above table reported the basic results for the basic regression (or Eq.1): Rit bo b1NUEitb2PUEit b3log MVit b4BTMVit Year Effects eit:

The dependent variable, Rit, is the accumulated excess return over the post-announcement period which

commences on the second day after the announcement and ends on the 60th trading day after the

announcement (i.e., t ? 2 to t ? 60). The unexpected portion of an earnings announcement is defined as the

difference between the actual earnings and the consensus earnings estimate in the month immediately prior

to announcement. We scaled the unexpected portion of the earnings announcement by the actual earningsannounced to arrive at our final measure of unexpected earnings. PUE are events where the announced

earnings are greater than the consensus analyst forecast earnings. PUE is calculated by multiplying the

unexpected earning by a dummyvariable which takes the value of 1 if there are positive earnings surprises

and 0 otherwise. Similarly, a negative unexpected earnings (NUE) event occurs when the earnings just

announced fall short of the consensus analyst forecast earnings. PUE are events where the announced

earnings are greater than the expected earnings where median analysts forecast earnings. PUE is calculated

by multiplying the unexpected earning by a dummy variable which takes the value of 1 if there are positive

earnings surprises and 0 otherwise. Similarly, a negative unexpected earnings (NUE) event occurs when the

earnings just announced fall short of expected earnings. MV represents the market capitalisation at the time

of the announcement and is measured in millions. BTMV measures the book-to-market value of the firm

making the announcement. Yearly fixed effects are included but not reported in the results. The notations

***, ** and * denote statistical significance at the 1, 5 and 10 % levels respectively

R. Bird et al.

1 3

8/12/2019 03.vvvvv

13/29

Our findings reported in Table 3indicate that there is a significant and positive PEAD

associated with both bad and good news irrespective of whether uncertainty is low or high

at the time of the announcement. Consistent with expectations, bad news has a greater

PEAD when the announcement is made at a time of high market uncertainty, whereas there

is a greater positive PEAD associated with good news when the announcement is made at atime of low uncertainty. Further, the PEAD associated with a bad news announcement is

always larger than that associated with a good news announcement but this difference is

only significant in those cases when uncertainty is high at the time of the announcement.

Two important insights can be drawn from our empirical findings. First, the evidence

confirms that the PEAD is higher, particularly for bad news, when the news is released at a

time when market uncertainty is high. Second, we find the same asymmetric response to

bad news and good news in the post-announcement period when market uncertainty is

high. This is the same phenomenon that others have noted with respect to the market

reaction at the time of the release of the information. To conclude, the evidence confirms

the proposition that the market underreacts to both good and bad earnings announcements

and that the PEAD over the subsequent 60 trading days is greater in the case of bad news,

particularly when high uncertainty prevails at the time of the information release. These

findings thus challenge the validity of market efficiency and suggest that investors faced

with high uncertainty as to how factors will evolve in the future will fail (even more) to

realise the importance of new information as is evidenced by the trend that the market

follows in the post-announcement period.

4.3 Changes in uncertainty over the post-announcement period

It is evident from the previous analysis that the level of uncertainty at the time of an

earnings announcementimpacts on the magnitude of subsequent PEAD. We now examine

the extent to which the level of uncertainty prevailing over the post-announcement period

impacts on the magnitude of PEAD. In order to evaluate this issue, we apply the sample

data to Eq. 2. In Table2, the coefficient for NUE is 0.0151, indicating that on average

there is a significant downward drift after bad news announcements. From Table 4, when

uncertainty (VIX) starts low and decreases during the post-announcement period, the

coefficient for NUE reduces to 0.0081, indicating a less than average downward drift. In

contrast, when uncertainty starts high and increases during the post-announcement period,

the coefficient for NUE is 0.0226. So once again there is a significant downward drift butthis time one that is well above the average. It is interesting to note that in those instances

when uncertainty is low at the time of the announcement but subsequently rises, and in

those instances when it is high at the time of the announcement but subsequently falls, the

coefficients are almost identical to the average of 0.0151. In addition to confirming a

general underreaction to information at the time of its release, these results underline the

importance of the level of uncertainty over the post-announcement period in determining

the magnitude of the PEAD after a bad news announcement.

In Table2, the coefficient for PUE is 0.0091, indicating a significant upward drift after

good news announcements. From Table 4, it is apparent that the level of uncertainty overthe post-announcement period has an even greater impact on PEAD after a good news

announcement than it does after a bad news announcement. When uncertainty (VIX) starts

low and decreases, there is a much larger than average upward drift over the post-

announcement period (i.e., the coefficient of PUE = 0.0215) with the correction to an

initial underreaction unmitigated by the negative impact that high market uncertainty can

have on investor behaviour. When we examine the PEAD after a good news announcement


1 3

8/12/2019 03.vvvvv

14/29

Table3

Theimpact

ofuncertaintylevelsonthepost-

earningsannouncementdrift

Coefficient

Sig

nificanceTest

NUE

PUE

Testofdifference

VIXLo

0.0122***

0.0112***

NU

E[

PUE

VIXHi

0.0177***

0.0075***

NU

E[

PUE***

TestofDifference

VIX

Hi[VIXLo*

VIXLo[

VIXHi

Theabovetablereportedtheresultsfortheregression:Rit

bo

b1NUEit

b2PUEit

b3X1NUEit

b4X1PUEit

b7logMVit

b8BTMVit

YearEffects

eit

Thedependentvariable,Rit

istheaccumulatedexcessre

turnoverthepost-announcementperiodwhichcommencesonthese

conddayaftertheannouncementandendsonthe

60thtradingdayafter

theannouncement(i.e.,t?

2tot

?

60).Theunexpectedportionofanearningsannouncementisdefi

nedasthedifferencebetweenthe

actualearnings

andtheconsensusearningsestimateinthemonthimmediatelypriortoannouncement.We

scaledtheunexpectedportionoftheearningsannouncementbythe

actualearnings

announcedtoarriveatourfinalmeasureofunexpectedearnings.PUEareeventswheretheannouncedearningsaregreatert

hantheconsensusanalystforecastearnings.PUE

iscalculatedbymultiplyingtheunexpectedearningby

adummyvariablewhichtakesthevalueof1iftherearepositive

earningssurprisesand0otherwise.Similarly,a

negativeunexpectedearnings(NUE)eventoccurswhen

theearningsjustannouncedfallshortoftheconsensusanalystforecastearnings.X1

isanindicatorva

riablewhichis

equalto1wherefirm

imakesanearningsannouncementattimetandthelevelofVIXattimeisabovemedianwhenallle

velofmarketuncertainty(VIX)a

rerankedfrom

lowtohigh;otherwiseX1

=

0.Foreaseofinterpretation,wehaveformattedtheabovetab

letoshowonlytheresultsdirectly

relatedtoVIXlevels.Forexample,thedisplayed

coefficientforVIXHiandnegativeearningssurprise(N

UE)is0.0177(i.e.sumofb1

and

b3).TotestforasymmetryinresponsestoNUEandPUE,weconductWaldteston

thecoefficients,theresultsarereportedinthelast2co

lumnsofthetable.YearlyControlvariableincludingbook-to-marketvalues,marketcapitalisationandyearlyfixed

effectsareincludedb

utnotreportedintheresults.Thenotations***,**and*denotes

statisticalsignificanceatthe1,5

and10%levelsrespectively

R. Bird et al.

1 3

8/12/2019 03.vvvvv

15/29

over a period when market uncertainty (VIX) starts high and increases, we now see a

reversal of the drift which is now downward and significant (the coefficient of PUE =

-0.0055). In other words, the negative impact that high uncertainty can have on investor

behaviour is sufficient to offset the normal upward drift associated with a correction to an

initial underreaction to the good news announcement. Indeed, the importance of the levelof uncertainty prevailing during the post-announcement period for determining the nature

of the subsequent drift is highlighted by the fact that there is almost no drift associated with

a good news announcement made at a time when uncertainty is low if it subsequently rises

during the post-announcement period (the coefficient of PUE = 0.0015). Our findings

highlight that there is a difference between the path that a stock price follows after an

initial underreaction to a news announcement depending on whether the news is good or

bad: overall, the path of the stock price is impacted even more by the prevailing uncertainty

during this post-announcement period if the news is good.

We previously found that investors underreact to both good and bad news announce-

ments, especially when the announcement is made at a time of high market uncertainty.

We have now found that the PEAD associated with a correction to the initial underreaction is

significantly affected by the level of uncertainty that prevails over the post-announcement

period. In the case of a bad news announcement, the effect of high uncertainty is to heighten

the typical downward drift. However, in the case of a good news announcement, the effect of

high uncertainty is to negate and even reverse the more typical upward drift during the post-

announcement period.

4.4 Market sentiment

Our second proposition is that in addition to uncertainty, market sentiment will have an

impact on the PEAD. We divided our sample on the basis of the level of sentiment

prevailing over the post-announcement period, with high sentiment being when market

momentum is strong and low sentiment being when it is weak. In order to evaluate this

proposition, we apply our sample data to Eq. 3. Based on the information contained in

Table5, it can be seen that there is always a downward drift after a bad news earnings

announcement irrespective of the level of market uncertainty and the level of market

sentiment prevailing over the post-announcement period (i.e. all of the coefficients are

positive and significant). As already noted, the greatest downward drift occurs when high

uncertainty prevails over the post-announcement period and this can be seen from Table 5.The downward drift is stronger after bad news announcements when low sentiment pre-

vails (with the difference between high and low sentiment for any given level of uncer-

tainty being significant at the 10 % level). The combined effect of uncertainty and

sentiment can be seen when we compare the coefficient attached to NUE when uncertainty

is high and sentiment is low (0.0237) with the coefficient attached to NUE when uncer-

tainty is low and sentiment is high (0.0066). The difference is significant at the 1 % level

and highlights the extent to which both uncertainty and sentiment impact on the PEAD

after bad news, with the negative drift when uncertainty is high and sentiment is low being

four times greater than when uncertainty is low and sentiment is high.The findings with respect to PEAD after a good news announcement are similar, though

more complicated, than those reported above for bad news announcements. As is the case

with bad news, high uncertainty during the post-announcement period is shown to have a

negative impact on investors and this translates into a lower, indeed negative, PEAD (e.g.

the coefficient for high uncertainty and low sentiment is -0.0140). We can now see from

the information presented in Table5 that sentiment has a larger impact on the markets


1 3

8/12/2019 03.vvvvv

16/29

Table4

Theimpact

ofuncertaintylevelsandchangesinuncertaintyonPEAD

Trend

N

UE

PUE

TestofDifference

V

IXLo

VIXHi

Trend

V

IXLo

VIXHi

VIXLo

V

IXHi

;DVIX

0.0081***

0.0151***

;DVIX

0

.0215***

0.0145**

*

PUE[

NUE***N

UE[

PUE

:DVIX

0.0156***

0.0226***

:DVIX

0

.0015

-0.0055**

NUE[

PUE***N

UE[

PUE***

Testof

difference

:

DVIX[

;DVIX***

:DVIX[

;DVIX***

Testof

difference

;

DVIX[

:DVIX***

;DVIX[

:DVIX***

Weexaminethecom

binedimpactofuncertaintylevelandchangesinuncertaintyonPEAD.Changesinuncertaintyaredefinedasthedifferencebetweenthelevelof

uncertaintyontheday

oftheannouncementandtheleve

lofuncertainty60dayspostannouncement.Werankthechangesin

uncertaintyacrossthesampleofa

nnouncements.

;DVIXrepresentsthe

sampleofannouncementsthatfallwithinthefirsttercileintermsofchangesinuncertainty(i.e.announ

cementsthatarefollowedbydecreasesinmarket

uncertaintyinthepost-announcementperiod).:DVIXrepresentsthesampleofannouncementsthatrankinthethirdtercile(i.e.announcementsfollowby

theincreasein

uncertaintylevel).Thenwerunthefollowingregression(orEq.2):Rit

bo

b1NUEit

b2PUEit

b3X1NU

Eit

b4X1PUEit

b5X2NUEit

b6X2PUEit

b7logMVit

b8BT

MVit

YearEffects

eit

Thedependentvariab

le,Rit

istheaccumulatedexcessreturnoverthepost-announcemen

tperiod(i.e.,t?

2tot?

60).PUEareeventswheretheannouncedearningsare

greaterthantheexpec

tedearningswhereexpectedearnings.PUEiscalculatedbymultiply

ingtheunexpectedearningbyadummyvariablewhichtakesthevalueof1ifthere

ispositiveearningssurpriseand0otherwise.Similarly

,anegativeunexpectedearnings

(NUE)eventoccurswhentheearningsjustannouncedfallshortoftheconsensus

analystforecastearnings.X1

isanindicatorvariablewh

ichisequalto1wherefirmimakesanearningsannouncementattim

etandthelevelofVIXattimeisabovemedian

whenalllevelsofmarketuncertainty(VIX)areranked

fromlowtohigh;otherwiseX1=

0.X2

isequalto1wherethereis

anincreaseinthelevelofVIXin

creaseoverthe

post-announcementperiod(asmeasuredbythedifferencebetweenthelevelofVIXatthetimeofannouncement,VIXtandthelevelofVIX60dayspostannouncement,

VIXt?60);otherwiseX

2

=

0.Foreaseofinterpretation,

wehaveformattedtheabovetabletoshowonlytheresultsdirectlyrelatedtoVIXlevels.Forexample,thedisplayed

coefficientforVIXHi,decreasinguncertaintyandnega

tiveearningssurprise(NUE)isthesumofb1

andb3.Yearlyfixedeffectsareincludedbutnotreportedintheresults.

Thenotations***,**and*denotesstatisticalsignificanceatthe1,5and10%levelsrespectively

R. Bird et al.

1 3

8/12/2019 03.vvvvv

17/29

post-earnings responses to good news than it does to post-earnings responses to bad news.

In fact, there is always an upward drift after a good earnings announcement when market

sentiment is strong over the post-announcement period, irrespective of what level of

market uncertainty prevails. Perhaps even more interesting is that the drift during the post-

announcement period after the release of good news is always negative when marketsentiment is low, again irrespective of the prevailing level of uncertainty. The combined

effect of uncertainty and sentiment can be seen when we compare the extent of the upward

drift following a good news announcement when uncertainty is low and sentiment is high

(coefficient = 0.0301) with the extent of the downward drift when uncertainty is high and

sentiment is low (coefficient = -0.0140). The difference is significant at the 1 % level and

this highlights that the combination of the prevailing market uncertainty and sentiment

over the post-announcement period has a much greater impact on the PEAD after good

news than it does after bad news, although both are highly significant. Further, post-

announcement sentiment is more important in explaining the PEAD than is the level of

uncertainty at the time of the announcement.

That market uncertainty and sentiment over the post-announcement period has a much

greater impact on the PEAD after good news than it does after bad news requires further

comment. Studies have shown that a large difference exists between how individuals react

and update their behaviour following good news and bad news (Akhtar et al.2011; Akhtar

et al.2012; Eil and Rao2011). Both Akhtar et al. (2011) and Akhtar et al. (2012) highlight

the existence of a negativity bias where investors pay greater attention to bad news.

They use this negativity bias to explain the greater reaction to bad news (than good

news) in the Australian and US markets. Consistent with the notion of a negativity bias,

if investors pay greater attention to bad news, then there is likely to be negative drift (forbad news) irrespective of the levels of uncertainty and sentiment. Put simply, investors

place a greater weight on the negative signal than the prevailing market conditions. On the

other hand, investors tend to greet good news with an optimistic bias (Eil and Rao2011).

We propose that this optimistic bias is at its greatest when good news arrives in periods of

high investor sentiment and low levels of uncertainty. At such times, the combination of a

low level of uncertainty and a high sentiment boosts investor confidence. So there is likely

to be a very positive reaction to good news in periods of high sentiment and low uncer-

tainty. However if the market sentiment is low, investors are likely to have a smaller

reaction to good news because the signal is inconsistent with their priors (i.e. of a poorly

performing stock market).

4.5 Announcement uncertainty/sentiment versus PEAD uncertainty/sentiment

The analysis to date has already provided us with some valuable insights into the

importance of market uncertainty and sentiment in explaining the PEAD. We now refine

our analysis in order to provide a clearer picture as to whether it is the level of uncertainty

and sentiment at the time of the announcement, or the levels prevailing over the post-

announcement period, that have the greatest impact on the PEAD.

In order to undertake this analysis we divided our sample up into two sub-samples:1. The first subsample contains those announcements made when market uncertainty is

high (above the median VIX level) and market sentiment is low (where S&P 500

returns are negative leading up to the announcement). These are the conditions where

one might expect the greatest market response to bad news and the least market

response to good news. Hence they are the conditions where one might expect the


1 3

8/12/2019 03.vvvvv

18/29

Table5

Theeffectofuncertaintyandsentimentonth

epost-earningsannouncementdrift

NUE

NUE

Testofdifference

Post-announcement

sentiment

Post-announcement

sentiment

Post-announcements

entiment

Levels

Trend

Lo

Hi

Levels

Trend

Lo

Hi

Lo

H

i

Lo

;DVIX

0.0103***

0.0066***

Lo

;DVIX

-0.0002

0.0

301***

NUE[

PUE**

PUE[

NUE***

Lo

:DVIX

0.0168***

0.0131***

Lo

:DVIX

-0.0089***

0.0

215***

NUE[

PUE***PUE[

NUE**

Uncertainty

:DVIX/Lo[

;DVIX

/Hi***

Uncertainty

;DVIX/Hi[:DVIX

/Lo***

Hi

;DVIX

0.0171***

0.0134***

Hi

;DVIX

-0.0053*

0.0

250***

NUE[

PUE***PUE[

NUE***

Hi

:DVIX

0.0237***

0.0200***

Hi

:DVIX

-0.0140***

0.0

164***

NUE[

PUE***N

UE[

PUE

Testof

difference

:DVIX/Lo[

;DVIX

/Hi***

Testofdifference

;DVIX/Hi[:DVIX

/Lo***

Theimpactofmomen

tumisintroducedintotheanalysis.Momentumreferstomarketmom

entumandismeasuredbytheS&

P500indexreturnsintheperiod260daysafter

theannouncement(i.e.t?

2tot?

60).Theab

ovetablereportstheresultsfortheregression(orEq.3):

Rit

bo

b1NUEit

b2PUEit

b3X1NUEit

b4X1PUEit

b5X2NUEit

b6X2PUEit

b7X3NUEit

b8X3PUEit

b9X4NUEit

b10X

4PUEit

b11logMVit

b12BTM

Vit

YearEffects

eit


istheaccumulatedexcessre

turnoverthepost-announcementp

eriod(i.e.,t?

2tot?

60).PUEandNUErepresentpositiveandnegativeearnings

surprisesrespectively.X1

isanindicatorvariablewhich

isequalto1wherefirmimakes

anearningsannouncementattime

tandthelevelofVIXattimetisabovemedian

whenalllevelofmarketuncertainty(VIX)arerankedfromlowtohigh;otherwiseX1

=

0.X2

isequalto1wherethereis

anincreaseinthelevelofVIXincreaseoverthe

post-announcementperiod(asmeasuredbythedifferencebetweenthelevelofVIXatthetimeofannouncement,VIXtandthelevelofVIX60dayspostannouncement,

VIXt?60);otherwiseX

2

=

0.X3

=

1ifthereturnonS&P500Indexfromt?

2tot?6

0ranks

inthesecondterc

ilewhereallS&P500Indexreturns{t?

2,t?

60}are

rankedfromlowtohigh;otherwiseX3

=

0.X4

=

1ifthereturnonS&P500Indexfrom

t?

2tot?

60ranks

intheth

ird

terc

ilewhereallS&P500Index

returns{t?

2,

t?

60}arerankedfromlowtohigh;otherwiseX4

=0.Foreaseofinterpretation,wehav

eformattedtheabovetabletoshowonlytheresultsdirectlyrelatedtotheimpactof

VIXlevelsandthepo

st-announcementsentimentlevel.Forexample,thetableindicateswherethereisalowlevelofuncertaintyatannouncementfollowedby

highsentiment

butalsoanincreaseinuncertainty(VIX:)inthepost-a

nnouncementperiod,thecoefficie

ntassociatedwithNUEis0.0013

4(i.e.,

b1

?

b5

?

b9).Controlva

riablessuchas

marketcapitalisation,

book-to-marketvalueandyearlyfixedeffectsareincludedbutnotreportedintheresults.Thenotations***,**and*denotesstatisticalsignificanceat

the1,5and10%levelsrespectively

R. Bird et al.

1 3

8/12/2019 03.vvvvv

19/29

lowest PEAD associated with bad news and the greatest PEAD associated with good

news.

2. The second subsample contains those announcements made when market uncertainty

is low (below the median) and market sentiment is high (where the S&P500 returns are

positive leading up to the announcement). These are the conditions where one mightexpect the greatest market response to good news and the least market response to bad

news. Hence they are the conditions where one might expect the lowest PEAD

associated with good news and the greatest PEAD associated with bad news.

The next step is to analyse the PEAD of each of these subsamples in terms of how it is

impacted by the prevailing market uncertainty and sentiment over the post-announcement

period. This is done by regression Eq.4to our two sub-samples. The results of our analysis

are displayed in Table6. The most striking aspect of the results is that it is the prevailing

conditions (of sentiment and uncertainty) during the post-announcement period that is

critical in driving the PEAD. For both subsamples (high VIX/low sentiment and low VIX/high sentiment at the time of the announcement), we observe that high sentiment during the

post-announcement period significantly reduces the negative drift after bad news but

increases positive PEAD following good news. For example, for good news released in the

least favourable investment climate of high uncertainty (i.e. VIX) and low sentiment, there

is a very strong positive drift when there is strong sentiment in the market during the PEAD

period, but when sentiment is weak there is either negative drift or no drift. It is worthwhile

noting that times of low sentiment are the conditions where Livnat and Petrovitis ( 2009)

suggest that there will be the greatest PEAD following good news. As for uncertainty, we

can see that irrespective of the level of uncertainty and sentiment at the time of the

announcement, when uncertainty is rising during the post-announcement period there isalways greater negative drift following bad news and lesser positive drift following good

news.

Having previously established that each of the prevailing uncertainty and prevailing

sentiment has an impact on the PEAD, we now turn our attention to considering their com-

bined impact. What we do in each case is compare the impacts on the PEAD of depressed

times (:DVIX ? LowSent) and favourable times (;DVIX ? HiSent) over the post-

announcement period. These impacts are indicated by the coefficients in bold in Table6. In

all cases, there is a significant impact on the PEAD after both good and bad news irrespective

of the levels of uncertainty and sentiment at the time of the announcement. These findings

confirm the importance of the levels of market uncertainty and sentiment during the post-

announcement period in determining the price behaviour of stocks during this period.

The other important question to address is the relative importance of the uncertainty and

sentiment at the time of the announcement as compared to uncertainty and sentiment over

the post-announcement period, in terms of their impact on the PEAD. If we compare the

coefficient associated with bad news in Panel A (depressed environment) and in Panel B

(favourable environment), we see little difference between them. In other words, in

determining the PEAD the effect of the conditions existing at the time of the announce-

ment are insignificant compared to effect of the prevailing conditions over the post-

announcement period. The situation in relation to the PEAD after good news is not as clearcut. Irrespective of the conditions at the time of the announcement there is a larger positive

PEAD when sentiment is high during the post-announcement period and a larger negative

PEAD when sentiment is low. In other words the level of market sentiment at the time of

the announcement does have some impact on the PEAD but this impact is swamped by the

level of sentiment prevailing over the post-announcement period.


1 3

8/12/2019 03.vvvvv

20/29

Table6

Theimpact

ofuncertaintylevelsandChangesinuncertaintyonPEAD:Extrem

eSample

NUE

PUE

Testofdifference

Post-earningsannouncementsentim

ent

Post-earningsannouncementsentime

nt

Post-earningsannouncementsentiment

Lo

Hi

Lo

Hi

Lo

Hi

PanelA:Subsample

highVIX/lowsentimentatannounc

ement

;DVIX

0.0173***

0.0070**

;DVIX

-0.0071

0.0343***

NUE[

PUE***

PU

E[

NUE***

:DVIX

0.0251***

0.0148***

:DVIX

-0.0182***

0.0232***

NUE[

PUE***

PU

E[

NUE

Testofdifference

:DVIX[

;DVIX

:DVIX[

;DVIX

Testofdifference

:DVIX[

;DVIX

:DVIX[

;

DVIX

Testofextremes

:DVIX?

LowSent[;DVIX?HiSent*

Testofextremes

;DVIX?

HiSent[:DVIX?

LowSent***

PanelB:SubsamplelowVIX/HighSentimentatannouncement

;DVIX

0.0172***

0.0089***

;DVIX

0.0006

0.0219***

NUE[

PUE***

PU

E[

NUE**

:DVIX

0.0207***

0.0123***

:DVIX

-0.0085**

0.0128***

NUE[

PUE***

PU

E[

NUE**

Testofdifference

:DVIX[

;DVIX

:DVIX[

;DVIX

Testofdifference

;DVIX\

:DVIX*

;DVIX\

:

DVIX*

Testofextremes

:DVIX?

LowSent[;DVIX?HiSent***

Testofextremes

;DVIX?

HiSent[:DVIX?

LowSent***

Weevaluatewhetheritisthelevelofuncertaintyandsentim

entatthetimeoftheannouncementorthelevelsprevailingoverthepost-announcementperiodthathavethe

greatestimpact

onthePEAD.Inorder

toundertakethisanalysiswedividedoursampleupintotwosub-samples:(i)thoseannouncementsmadewhenmarketuncertaintyishigh(abov

e0)andmarket

sentimentarelow(below0)and(ii)announcementsmadewhenmarketuncertaintyislow(belowmedian)andmarketsentimentis

high(abovemedian).Wereportthe

coefficientsfor

thefollowingregressio

n:Rit

bo

b1NUEit

b2PUEit

b5X2NUEit

b6X2PUEit

b7X3N

UEit

b8X3PUEit

b9X4NUEit

b10X4PUEit

b11logMVit

b12BTMVitYe

arEffects

eit

4


istheaccumulatedexcessreturnoverthepost-announcementperiod(i.e.,t?

2tot?

60).PUEandNU

Erepresentpositiveandnegativeearningssurprises

respectively.X2

isequalto1wherethereisanincreaseinth

elevelofVIXincreaseoverthepost-announcementperiod(asmeasureb

ythedifferencebetweenthelevelof

VIXatthetime

ofannouncement,VIX

t

andthelevelofVIX60daysposta

nnouncement,VIX

t?60);otherwiseX2

=

0.X3

=

1ifthereturnonS&P

500Indexfromt?

2tot?

60ran

ksinthesecond

terc


2,t?


=

0.X4

=

1ifthereturnonS&

P500Indexfromt?

2tot?

60ran

ksintheth

ird

terc


2,t?


=

0.Foreaseofinterpretation,

wehaveformattedtheabovetabletoshowonlythe

resultsdirectlyrelatedtotheimpactofVIXlevelsandthep

ost-announcementsentimentlevel.Y

earlyfixedeffectsareincludedbutnotreportedintheresults.Thenotations***,**and

*denotesstatisticalsignificanceatthe1,5and10%leve

lsrespectively

R. Bird et al.

1 3

8/12/2019 03.vvvvv

21/29

Previous studies such as Francis et al. (2007) and Livnat and Petrovitis (2009) have

emphasised that it is the level of uncertainty and/or sentiment prior to, or at the time of, the

announcement that determine the PEAD. Our analysis has shown that in the case of bad

news, the conditions at the time of the announcement make little or no contribution to

explaining variations in the PEAD and provide only a minimal contribution in the case ofgood news. Rather, it is the uncertainty and sentiment prevailing during the post-

announcement period that plays a much more important role in explaining the PEAD. In

the case of bad news it is the prevailing post-announcement uncertainty that plays the

major role while in the case of good news it is clear that the greatest role is played by the

post-announcement prevailing sentiment.

4.6 Stock characteristics

The evidence provided to date confirms that the level of prevailing market uncertainty and

market sentiment over the post-announcement period play a critical role in explaining the

PEAD phenomenon. The issue pursued in this sub-section is whether our findings are sensitive

to certain characteristics of the firm making the announcement. The two characteristics eval-

uated are the firms market capitalisation and its book-to-market ratio as both have been found

to have a major influence on a firms market returns (Fama and French1992).

4.6.1 Small cap and large cap

We divided our sample into small cap stocks (defined as firm less than the sample median

market capitalisation) and large cap stocks (defined as stocks greater than the median marketcapitalisation) and then applied Eq.4 to each sub-sample. In Table7, we repeatthe information

provided in Table6but this time for small cap and large cap stocks respectively. The main

result is that the finding that prevailing market uncertainty and sentiment impact on the PEAD

holds for both small and large cap stocks. In all cases the PEAD is much stronger after bad news

when uncertainty is high and sentiment is low and much stronger after good news when

uncertainty is low and sentiment is high. It does appear that the level of market uncertainty and

sentiment at the time of the announcement, particularly when uncertainty is high and sentiment

low, does cause some differences in post-announcement price behaviour of small and large cap

stocks. However the previous finding remains robust: the main driver of PEAD is the post-

announcement uncertainty and sentiment. The other major finding is that in almost all cases thePEAD for small cap firms is more impacted by the post-announcements levels of market

uncertainty and sentiment than is the case for large cap stocks. For example, the negative drift

after a bad news announcement is higher for smaller firms when conditions are depressed (i.e.

high uncertainty and low sentiment) and lower when conditions are favourable (i.e. low

uncertainty and high sentiment).

4.6.2 Value and growth firms

In Table8, we present a summary of our findings where we repeat the analysis reported inTable6, this time dividing the stocks into growth and values stocks as indicated by each

stocks book-to-market ratio. The PEAD of growth stocks follows the typical pattern when

the greatest post-announcement reaction to bad news occurs during periods of high

uncertainty and low sentiment while the greatest reaction to good news occurs during

periods of low uncertainty and high sentiment. While this finding also applies to bad news


1 3

8/12/2019 03.vvvvv

22/29

Table7

Theimpact

ofuncertaintylevelsandChange

sinuncertaintyonPEAD:Extrem

eSample

NUE

PUE

Significancetests

Post-earningsannouncement

sentiment


sentiment


sentiment

Lo

Hi

Lo

Hi

Lo

H

i

A(SmallFirms)

PanelA:Subsample

HighVIXLowsentimentatann

ouncement

;DVIX

0.0224***

0.0064

;DVIX

0.0017

0.0422***

NUE[

PUE***P

UE[

NUE***

:DVIX

0.0255***

0.0095

:DVIX

-0.0130*

0.0274***

NUE[

PUE***P

UE[

NUE

Testofdifference

:DVIX[

;DVIX

:DVIX[

;DVIX

Testofdifference

;DVIX\

:DVIX*

;DVIX\

:DVIX*

Testofextremes

:DVIX?

LowSent[;DVIX?H

iSent***

Testofextremes

;DVIX?

HiSent[:DVIX?Lo

wSent***

PanelB:Subsample

LowVIX/Highsentimentatannouncement

;DVIX

0.0235***

0.0234**

;DVIX

0.0014

0.0238***

NUE[

PUE**P

UE[

NUE

:DVIX

0.0159**

0.0158**

:DVIX

-0.0086

0.0138**

NUE[

PUE***P

UE[

NUE

Testofdifference

:DVIX[

;DVIX

:DVIX[

;DVIX

Testofdifference

;DVIX\

:DVIX

;DVIX\

:DVIX

Testofextremes

:DVIX?

LowSent[;DVIX?H

iSent

Testofextremes

;DVIX?

HiSent[:DVIX?Lo

wSent***

NUE

PUE

Significancetests


sentiment


sentiment

Post-earningsannou

ncement

sentiment

Lo

Hi

Lo

Hi

Lo

Hi

(B)Largefirms

PanelA:Subsample

HighVIXLowSentimentatann

ouncement

;DVIX

-0.0017

0.0009

;DVIX

-0.0285***

0.0246**

*

NUE[

PUE**

PUE[

NUE**

:DVIX

0.0149**

0.0175*

:DVIX

-0.0237***

0.0295**

*

NUE[

PUE***

PUE[

NUE

Testofdifference

:DVIX[

;DVIX*

:DVIX[

;DVIX*

Testofdifference

:DVIX[

;DVIX

:DVIX[

;DVIX

Testofextremes

:DVIX?

LowSent[;DVIX?

HiSent

Testofextremes

;DVIX?

HiSent[:DVIX?LowSent***

R. Bird et al.

1 3

8/12/2019 03.vvvvv

23/29

Table7

continued

NUE

PUE

Significancetests


sentiment


sentiment

Post-earningsannou

ncement

sentiment

Lo

Hi

Lo

Hi

Lo

Hi

PanelB:Subsample

LowVIX/HighSentimentatannouncement

;DVIX

0.0081

-0.0017

;DVIX

-0.0002

0.0155**

NUE[

PUE

PUE[

NUE**

:DVIX

0.0154***

0.0056

:DVIX

-0.0060

0.0097

NUE[

PUE***

PUE[

NUE**

Testofdifference

:DVIX[

;DVIX

:DVIX[

;DVIX

Testofdifference

;DVIX\

:DVIX

;DVIX\

:DVIX

Testofextremes

:DVIX?

LowSent[;DVIX?

HiSent***

Testofextremes

;DVIX?

HiSent[:DVIX?LowSent***

Weevaluatewhetheritisthelevelofuncertaintyandsentim

entatthetimeoftheannouncementorthelevelsprevailingoverthepos

t-announcementperiodthathavethe

greatestimpact

onthePEAD.Inordertoundertakethisanalysiswedividedoursampleintotwosub-samples

:(i)thoseannouncementsmadewh

enmarketuncertaintyishigh(abov

e0)andmarket

sentimentislow(below

0)and(ii)announcementsmadewh

enmarketuncertaintyislowandmarketsentimentishigh.Hereweconfineoursampletosmallfirms(definedasfirmslessthan

thesamplemedianmarketcapitalisation)andlargefirms(definedasfirmsgreaterth

anthesamplemedianmarketc

apitalisation).Weranthefollow

ingregression:

Rit

bo

b1NUEit

b2PUEit

b5X2NUEit

b6X2PUEit

b7X3NUEit

b8X3PUEit

b9X4NUEitb

10X4PUEit

b11logMVit

b12BTMVit

YearEffects

eit

4

Panel7AandPanel7

Bdisplaytheresultsofouranalysisonsmallandlargestocksrespe

ctively.Thedependentvariable,R

it

istheaccumulatedexcessreturn

overthepost-

announcementperiod(i.e.,t?

2tot?

60).PUEandNU

Erepresentpositiveandnegativeearningssurprisesrespectively.X2is

equalto1wherethereisanincreas

einthelevelof

VIXincreaseoverthe

post-announcementperiod(asmeasuredbythedifferencebetweenth

elevelofVIXatthetimeofannouncement,VIXt

andthelevelofVIX60dayspost

announcement,VIXt?

60);otherwiseX2

=

0.X3

=


2tot?

60ranksinthesecondtercilewhereallS&P500Index

returns{t?

2,

t?


=0.

X4

=

1ifthereturnonS&P500I

ndexfromt?

2tot?

60ranksin

thethirdtercilewhereallS&P50

0Indexreturns

{t?

2,t?

60}arerankedfromlowtohigh;otherwiseX

4

=

0.Foreaseofinterpretation,w

ehaveformattedtheabovetableto

showonlytheresultsdirectlyrelatedtotheimpact

ofVIXlevelsandthepost-announcementsentimentlevel.Yearlyfixedeffectsareincludedbu

tnotreportedintheresults.Thenotations***,**and*denotesstatisticalsignificance

atthe1,5and10%levelsrespectively


1 3

8/12/2019 03.vvvvv

24/29

Table8

Theimpact

ofuncertaintylevelsandchangesinuncertaintyonPEAD:extrem

esample

NUE

P

UE

Testofdifference


ent

P

ost-earningsannouncementsentiment

Post-earningsannounce

mentsentiment

Lo

Hi

L

o

Hi

Lo

Hi

(A)Valuefirms

PanelA:Subsample

HighVIXLowsentimentatannouncement

;DVIX

0.0157***

0.0035

;DVIX

-

0.0021

-0.0118**

NUE[

PUE**

NU

E[

PUE**

:DVIX

0.0203***

0.0081*

:DVIX

-

0.0063

-0.0160**

NUE[

PUE***

NU

E[

PUE***

Testofdifference

:DVIX[

;DVIX

:DVIX[

;DVIX

Testofdifference

;

DVIX[

:DVIX

;DVIX[

:

DVIX

Testofextremes

:DVIX?

LowSent[;DVIX?H

iSent

***

Testofextremes

:

DVIX?

LowSent[;DVIX?HiSent

PanelB:Subsample

LowVIX/Highsentimentatannou

ncement

;DVIX

0.0173***

0.0144***

;DVIX

-

0.0091

-0.0087

NUE[

PUE***

NU

E[

PUE***

:DVIX

0.0221***

0.0192***

:DVIX

0.0116*

0.0120

NUE[

PUE***

NU

E[

PUE

Testofdifference

:DVIX[

;DVIX

:DVIX[

;DVIX

Testofdifference:

DVIX\

;DVIX***

:DVIX\

;

DVIX***

Testofextremes

:DVIX?

LowSent[;DVIX?H

iSent

Testofextremes

:

DVIX?

LowSent[;DVIX?HiSent**

NUE

PUE

Testofdifference



ent

Pre-announcementse

ntiment

Lo

Hi

Lo

Hi

Lo

H

i

(B)Growthfirms

PanelA:Subsample

HighVIXLowsentimentatannouncement

;DVIX

0.0073

0.0176**

;DVIX

0.0011

0.0274***

NUE[

PUE

P

UE[

NUE***

:DVIX

0.0274***

0.0376***

:DVIX

-0.0351***

-0.0088

NUE[

PUE***N

UE[

PUE***

Testofdifference

:DVIX[

;DVIX**

:DVIX[

;DVIX**

Testofdifference

;DVIX[

:DVIX***

;DVIX[

:DVIX***

Testofextremes

:DVIX?

LowSent[;DVIX?H

iSent

Testofextremes

;DVIX?

HiSent[:DVIX?

Low

Sent***

PanelB:Subsample

LowVIX/Highsentimentatannou

ncement

R. Bird et al.

1 3

8/12/2019 03.vvvvv

25/29

Table8

continued

NUE

PUE

Testofdifference



ent

Pre-announcementse

ntiment

Lo

Hi

Lo

Hi

Lo

H

i

;DVIX

0.0277***

0.0176***

;DVIX

-0.0116

0.0280***

NUE[

PUE***P

UE[

NUE***

:DVIX

0.0269***

0.0168***

:DVIX

-0.0177***

0.0219***

NUE[

PUE***P

UE[

NUE

Testofdifference

;DVIX[

DVIX

;DVIX[

DVIX

Testofdifference

;DVIX[

DVIX

;DVIX[

DVIX

Testofextremes

:DVIX?

LowSent[;DVIX?H

iSent

Testofextremes

;DVIX?

HiSent[:DVIX?

Low

Sent***

Weevaluatewhetheritisthelevelofuncertaintyandsen

timentatthetimeoftheannouncementorthelevelsprevailingoverthepost-announcementperiodthathavethegreatest

impactonthePEAD.Inordertoundertakethisanalysiswedividedoursampleintotwosub

-samples:(i)thoseannouncements

madewhenmarketuncertaintyish

ighandmarket

sentimentislowand(ii)announcementsmadewhenmark

etuncertaintyislowandmarketsentimentishigh.Hereweconfineou

rsampletogrowthfirms(definedasfirmslessthan

thesamplemedianbook-to-marketratio)andvaluefirms(definedasfirmsaboveth

anthesamplemedianbook-to-m

arketratio).Werunthefollow

ingregression:

Rit

bo

b1NUEit

b2PUEit

b5X2NUEit

b6X2PU

Eit

b7X3NUEit

b8X3PUEit

b9X4NUEitb

10X4PUEit

b11logMVit

b12BTMVit

YearEffects

eit

4

Panel8AandPanel8Bdisplaytheresultsofouranalysisonvalueandgrowthstocksresp

ectively.Thedependentvariable,Rit

istheaccumulatedexcessreturnoverthepost-

announcementperiod(i.e.,t?

2tot?

60).PUEandNU

Erepresentpositiveandnegativeearningssurprisesrespectively.X2is

equalto1wherethereisanincreas

einthelevelof

VIXincreaseoverthe

post-announcementperiod(asmeasurebythedifferencebetweenthelevelofVIXatthetimeofannouncement,VIXt

andthelevelofVIX60dayspost

announcement,VIXt?

60);otherwiseX2

=

0.X3

=


2tot?

60ranksinthesecondtercilewhereallS&P500Index

returns{t?

2,

t?

60}arerankedfro

mlowtohigh;otherwiseX3

=0.

X4

=


2tot?

60ranksin

thethirdtercilewhereallS&P50

0Indexreturns

{t?

2,t?

60}arerankedfromlowtohigh;otherwiseX

4

=

0.Foreaseofinterpretation,w

ehaveformattedtheabovetableto

showonlytheresultsdirectlyrelatedtotheimpact

ofVIXlevelsandthepost-announcementsentimentlevel.

Yearlyfixedeffectsareincludedbu

tnotreportedintheresults.Thenotations***,**and*denotesstatisticalsignificance

atthe1,5and10%levelsrespectively


1 3

8/12/2019 03.vvvvv

26/29

announcements of value companies, their reaction to good news is surprisingly greatest

over periods when uncertainty is high and sentiment is low. The uncertainty and sentiment

levels at the time of the announcement do have an impact on the PEAD for both value and

growth stocks but this relationship is quite complicated. For example, if the initial con-

ditions are favourable (i.e. uncertainty is low and sentiment is high) rather than depressed,then downward drift following bad news for value stocks is greater but only when senti-

ment is high over the post-announcement period. However, if the initial conditions are

depressed then there is a greater upward drift for value stocks in the post-announcement

period after a good news announcement but only during periods when uncertainty is high.

Our overall assessment is that although the conditions at the time of the announcement do

play a role in explaining the PEAD, it is the prevailing uncertainty and sentiment that exist

during this period that play the greater role.

As with large and small cap stocks, there is a clear difference between the PEAD

behaviour of growth stocks and value stocks, with the variation of the drift for growth

stocks over the post-announcement period being much larger than that for value stocks. For

example in the case of growth stocks there is a much greater upward drift during the post-

announcement period after good news when contemporaneous conditions are favourable

(i.e. low uncertainty and high sentiment) and a much greater downward drift when con-

ditions are depressed (i.e. high uncertainty and low sentiment). The bigger reaction to bad

news made by growth firms may be driven by the fact that the valuation of growth stocks is

very much dependent on the maintenance of investor confidence which is likely to be

eroded when a disappointing earnings report is combined with a period of high market

uncertainty and low market confidence (Skinner and Sloan 2002). Conversely, the com-

bination of a favourable earnings report and a positive market environment may fueleuphoria that explains the larger reaction associated with good news announcements for

growth stocks. Thus we conclude that although the prevailing uncertainty and sentiment

during the post-announcement period has a significant impact on the PEAD of both the

value and growth stocks, the influence is much greater for the growth stocks.

5 Concluding remarks

Our findings provide support for the suggestion that the state of mind of investors (i.e.

market sentiment) in concert with the clarity with which they interpret the information (i.e.the level of market uncertainty) work together to determine how investors respond to an

earnings signal in the weeks immediately after that signal is made public. At one extreme,

we have a situation of high market uncertainty and low market sentiment which means

investors have difficulty in interpreting the implications of the earnings announcement for

the value of the firm at a time when a negative tone overlays all of their investing. It is not

surprising that at such times there is a negative post-announcement drift associated with the

release of both bad and good news with the drift being considerably larger for bad news. At

the other extreme, we have a situation of low market uncertainty and high market senti-

ment

03.vvvvv

Documents

Transcript of 03.vvvvv