Comparison of earning surprise effect on stock prices in
developed and emerging economies
Author: Adrian Sajanek
Date: 15.07.2020
This document is written by Student Adrian Sajanek who declares to take full
responsibility for the contents of this document.
I declare that the text and the work presented in this document are original and that
no sources other than those mentioned in the text and its references have been used
in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of
completion of the work, not for the contents.
Abstract:
This paper analyzes the earnings surprise effect and the PEAD in the developed and emerging markets. Focal point of the study is finding the difference in magnitude of the earnings
surprise effect and PEAD between the two types of markets. Developed markets are
represented by the stocks from the US market while the emerging markets are represented by stock from the Indian national stock exchange. The earnings surprise itself is measured by the standardized difference between the average forecast and the actual announced earnings per share on the date of announcement, called SUE. Cumulative abnormal returns are calculated to represent the PEAD through the CAPM model. The hypotheses are drawn on the
assumptions of studies regarding the liquidity and its effect on the earnings surprise effect and PEAD. Study finds that both earnings surprise effect and PEAD are present in both markets and that the emerging markets suffer from higher degree of earnings surprise. Additionally the emerging markets are found to have larger earnings surprise effect and PEAD than developed markets. However the study did not find conclusive evidence on the magnitude of this difference and could not find relationship between the liquidity factor and the earnings surprise in either of the markets
1. Introduction
Long-standing financial literature on the market efficiency, stating that all publicly available information should be reflected in the price of the stock immediately has been challenged by a variety of frictions shown in the day to day trading data as mentioned by Malkiel (2003). Specifically multiple studies like Abarbanell (1992) and Bernard (1989) have been showing existence of a earnings surprise effect or in other words stock price expectations not being correctly adjusted after earnings announcements. These adjustments are not instantaneous as market efficiency states, but rather are found to be underacted to and creating a post-earnings announcement drift. Understanding the stock market and its pricing offers crucial insights for financial institutions as well as financial analysts, however the consensus of the exact reasons of this inefficiency from the standpoint of classic theories has not yet been reached. Financial specialist have been offering multitudes of theories ranging from behavioural finance
(Ekholm, 2006), psychology (Frazzini, 2006) to pure financial stock market analysis
(Bernard, 1990) to try to explain the phenomenon and collectively build more understanding and knowledge on the topic. While the two main ideological flows point to 1.
Misspecification of risk in the underlying market efficiency model and 2. Underraction of the market agents to the earnings announcement, the former one has not been well supported in recent studies, the later is not able to fully explain the magnitude of the earning effect either. This in turn increases the importance and relevance of other theories that combined might bring forth more understanding of the issue as this paper also tries to achieve by engaging with less explored perspective.
The one perspective from which the earnings surprise effect has been less looked into is using the emerging markets in comparison to developed ones to find possible explanations for the existence of this anomaly. Most of the studies on this topic utilize datasets from the
NASDAQ or NYSE only, introducing possible bias or confounding factor. This paper will instead focus on the comparison between the earnings surprise effect on the stocks from the NASDAQ to represent a developed markets and NSE as representation of the developing ones. Several studies on the earnings surprise effect focusing on emerging markets do lead to conclusion of its existence (Li, 2017), however there is a lack of comparative studies that
could lead to new understanding and viewpoints on the topic leading to the research question of this paper:
To what extent is the presence and magnitude of the earnings surprise effect and resultant post-earnings announcement drift, dependent on the development of the studied market?
Another positive contribution to the body of literature on the topic that exploration of given research question hopes to bring is the further examination and clarification of the possible factors that could be causing the effect in question through comparison of markets with differing values of these factors like liquidity. One of the theories by Sadka, 2005 concerns the role of liquidity risk as well as the liquidity of the market and the stock at which its traded. It claims that the stock liquidity does have small impact on the magnitude of the earnings surprise effect. The difference of stock liquidity between the NSE and NASDAQ (Sehgal 2015), could provide further clarity on the importance of this factor when measuring earnings surprise effect and explaining PEAD. The results of this study may be interesting to the academics involved with the topic and investors that consider or have already invested in the stock markets of the emerging countries as the differences between the markets highlight the difference in the presence of the earnings surprise effect. Which can be further used in the development of individual investment strategies or seen as investment opportunity.
The remaining body of this study is structured in a following way: Section 2 describes in detail the existing literature background on the topic and contains hypotheses development. In the section 3 the methodology of the empirical tests of the hypotheses is explained while section 4 describes the data selection process .Consequently the results of the empirical tests are provided in the Section 5 .The Section 6 contains discussion of the results, provides ideas for further research on the topic and offers concluding remarks.
2. Literature review
The prior research on the earnings surprise effect is very extensive and contains large variety of approaches, methodologies and theories trying to explain the anomaly.Therefore only the main subsections of this research that are relevant for the paper and its research question are further presented in this section in the following order: 2.1 Efficient market hypothesis explaining the underlying accepted market theory and reasoning behind the earnings surprise effect being called anomaly, 2.2 Earnings surprise effet and PEAD presents the findings of the previous literature presenting various theories and factors on the anomaly in question. Lastly subsection 2.3 Liquidity focuses on brief summary of literature on the connection of liquidity to the magnitude of the PEAD. Subsequently the rationale behind the hypothesis building is explained.
2.1 Efficient market hypothesis
Financial markets and stock exchanges have became increasingly more crucial for the economic growth and development of many countries and even have great impact on the global trade and business. The stocks have also become a tool to increase one's wealth or maintain savings of an individual investor. As a result many investors, financial managers along many others focus much of their attention on the prediction of the stock price changes and the underlying factors. The academics themselves have been closely monitoring the prices and developing models to describe and predict the prices of stocks as can be seen in example of CAPM models and Fama-French factor models.
While there are several theories on how the markets operate the efficient market hypothesis introduced by Fama (1970) has been prevalent through history. Idea of the efficient market hypothesis is that the markets are able to incorporate the available information into the stock prices. There are 3 distinct tests of this efficiency as suggested by Fama(1970), first being the weak form test where the markets reflect the historical prices of the stocks and second being semi-strong form that tests if the stock prices reflect all available public information. The most strict efficiency test is the strong-form tests that involves the stock prices ability to reflect all, even the private information. The conclusion of the efficient market hypothesis
claims that the uniformed and informed investors will have the same returns, because of the idea of “random walk”. This random walk describes the time series of prices where
subsequent future prices are simply random deviations from the past ones, because the information that becomes available is immediately reflected in the price of the stock
(Malkiel,2003). Therefore the future price of the stock depends on the future information that is not predictable and could be described as random. Given the price changes are depended on random future information the future prices are reflecting all of the available information.
Even though this theory has been relevant and widely defended there have been several phenomenons that are challenging the ability of the stocks to immediately include all the new information that arises (Malkiel,2003), among which also the existence of the earnings surprise is gaining relevance (Abarnabel 1992). The stock prices have also been observed to not be able to fully reflect the incoming earnings news and instead only gradually reach the price point reflecting all of the information, creating phenomenon known as
post-earnings-announcement drift (PEAD) .
2.2 Earnings surprise effect and post-earnings-announcement drift
The existence of the PEAD in connection to the earnings surprise effect was first introduced by Ball and Brown 1968 and since then numerous other studies focused on the subject, such as Bernard 1989 elaborating on the possible explanations for this anomaly.
The earnings surprise effect itself occurs when the firm in question releases an earnings statement that is not in line with what the market predicted and the prices therefore do not reflect the current financial situation of the given firm (Livnat, 2006). The market
expectations can be estimated assuming the seasonal random walk or directly through the forecast analyst predictions of the earnings through measure called standardized unexpected earnings (SUE). SUE compares the expectations of the earnings from the seasonal random walk or forecast analysts and standardizes the measure by dividing it with its standard deviation. From the literature it is observed that the stock prices of most firms adjust steeply in the direction of the earnings surprise (Bernard, 1989), or in other words the price increases after unexpectedly good news and decreases for the opposite. On its own the the stock price
adjusting to information that becomes available is not violation of the efficient market hypothesis, however studies by Bernard (1990), Abarbanell (1992) and others reach the conclusion that the market fails to adjust their expectations of the future earnings based on the past information and therefore challenging even the weak-form of the efficient market hypothesis. The conclusion that Abarbanell 1992 reaches is that the forecast analysts display over reliance on the seasonal random walk while not properly incorporating past earnings announcement into their predictions. While most studies focus on this phenomenon in the settings of the developed market the studies by Li (2017) on the Chinese market and Sen (2009), also observe its existence albeit with slightly different magnitudes.
The mispricing abnormality can be further shown by the occurrence of the PEAD. The post-earnings-announcement drift encapsulates the persisting abnormal returns on the stock which occur after a earnings announcement and drift in the direction of the earnings surprise (Ball, 1968). Abarbanell(1992) and Bernard(1990) utilize the PEAD to reach the above mentioned conclusion on the inefficiency of the market and its inability to properly react to the available information in the past earnings announcement, since there should be no persisting abnormal returns if the available information was immediately reflected in the prices. Instead the literature for both developed and emerging markets show significant abnormal returns on the zero-investment portfolio around the date of the earnings announcement.
While widely documented, there is yet to be a consensus on the explanation of the PEAD in the academic world and there have been many approaches to this issue. Study by Bernard (1989) has found that the misspecification of risk in the CAPM, one of the prevailing
theories, does not explain the presence of the PEAD and found some support for the theory of the PEAD being resultant from the delayed response of the market participants. The later theory is being supported by findings of Abarnabell(1992) and Brown(2009). Several
explanations from the field of psychology and behavioral science also persist. The disposition effect studied by Frazzini (2006) attempts to clarify the reasoning behind the delayed market response, by looking at the behavior of investors and found that the tendency to ride losses and realize gains has significant effect on the magnitude of the PEAD. Another sub-section of studies of PEAD focuses on finding the factors that could be responsible for the anomaly
through market inefficiencies. Brown (2009) studies the relationship between the information asymmetry and the earnings surprise effect and finds that it is significant. On the other hand Sadka (2006) focuses on the liquidity risk as a reason for the existence of PEAD studying the barriers that investors face in case they would want to capitalize on the abnormal returns that form PEAD. The findings of the mentioned the study provide some evidence to the idea that the liquidity of the stock and the stock market may be partially responsible for the magnitude of the PEAD. Further support of this theory was presented by Sen (2009) when the liquidity was found to have significant effect on the abnormal returns in the Indian stock market.
2.3 Liquidity
Liquidity is the measure of cost and availability of transforming a given asset into cash in a short period of time. In terms of stock it entails the ability, cost and ease with which one can buy and sell the stock. The inability to sell or a buy a stock or incurring costs when doing so, prevents trades that could seem profitable when not accounting for these factors and create illusion of abnormal returns (Amihud, 1986). Stocks that are illiquid cannot fully reflect all available information as for example one may not be able to buy the stock because nobody is selling or only sellers are selling at premium high enough to absorb all of the abnormal returns (Malkiel 2003). Therefore accounting for liquidity is important when dealing with stock price changes and even when researching the PEAD as the investment strategies that could exploit and tradeaway the anomaly require high portfolio turnover (Sadka 2006).
While liquidity does not have one standard measure, among the most used types of measures for stocks would be the stock turnover and the transaction costs (Sadka 2006). More
specifically the natural measure of transaction costs and thus liquidity would be the bid-ask spread, as in order to trade stock immediately the investor has to trade at the values of bid and ask, effectively acting as the cost for this immediate transaction (Amihud 1986). Directly relating to the the topic of PEAD and the liquidity measure that has been often examined in the studies of the PEAD is the transaction costs, therefore bid-ask spread, as in Bernard (1989) and Sen (2009), because of the previously mentioned high turnover portfolio strategy needed to exploit PEAD. While Bernard (1989) finds some inconclusive evidence for the transaction cost-based theory, Sen (2009) presents significant PEAD even after controlling for the transaction costs. On the other hand pricing in the liquidity factor was found to have
significant effect on the PEAD by Sadka (2006), who concludes that stock liquidity risk could be further related to the information asymmetry in the market. This would point to the possible differences in outcomes from the papers of Sen (2009) who used data from the Indian market and Bernard (1989) who used the stocks from the developed market, as the liquidity and more specifically transaction costs and information asymmetry are both more prevalent in the emerging markets and India is no exception (Sen 2009).
2.4 Hypothesis building
The efficiency market hypothesis as mentioned is not able to explain the occurence of the PEAD and inspite of the long-standing support for the theory many academics are trying to find alternative theories that would be able to incorporate this anomaly The focus of most studies is on the developed market with just few looking into the earnings surprise effect and PEAD in the emerging markets. Therefore this paper would like to compare the PEAD in the developed and emerging markets. First the markets have to be investigated for the existence of the PEAD. Prior literature such as Bernard(1989), Kaestner (2006), Sen (2009) and Abarnabell (1992) among others show the connection of the PEAD to the standardized unexpected earnings. The SUE measures the difference between what the market expects before the future earnings announcement and the actual realization. From the works of
Bernard(1990) and Frazzini (2006) some of the observed factors that influence the magnitude and presence of the earnings surprise effect are the liquidity and information asymmetry. Both of these are presented to have positive relationship with the SUE. Considering that the emerging markets have higher coefficients of both these factors than the developed ones, the resulting assumption would be that the emerging markets will be observed to have larger magnitude of SUE, from which the first hypothesis follows:
H1:There is a negative relationship between the development of the market and magnitude of standardized unexpected earnings
Presence of the SUE itself does not constitute an anomaly if immediately priced in and therefore the existence of the PEAD resulting from the earnings announcement will be tested
on both markets. Literature on the topic such as studies by Kaestner(2006) and Li(2017) show positive relation between the SUE and the cumulative abnormal returns bringing forth the second hypothesis:
H2:Standardized unexpected earnings are positively related to cumulative abnormal returns in both markets
Given the previously mentioned liquidity concerns and transaction costs based theory on PEAD (Bernard, 1990) the investors in the emerging markets would not be able to react to the earnings announcements as fast as the investors in the developed market. The lower availability of the stock quantities might prevent the investors from buying sufficient amount of stock to have the prices of the stocks correctly reflect the available information or reflect it ot the degree developed markets can. Moreover the higher transaction cost might be enough to deter some of the trading of the stock creating larger PEAD than in the developed markets. Based on the literature the third hypothesis is:
H3:The development factor of a market will have negative impact on the cumulative abnormal returns
The effect of the liquidity and more specifically transaction cost based liquidity is not conclusive (Sadka 2006) and therefore would benefit from further research. The transaction costs are naturally measure by the bid-ask spread of the stock on a given day, as it directly causes less trading to occur (Amihud 1986). The bid-ask spread is narrower for the
economies with more trading and investors active and therefore narrower for more developed markets. This difference could be used as one of the explanations for the different magnitude of the PEAD between the developed and emerging markets. The paper will also focus on understanding the relationship between the bid-ask spread and PEAD and with the knowledge of the prior literature will examine the fourth hypothesis:
3.Methodology
The following section describes in detail the empirical models used to tests the hypotheses, their development process and then provides motivation behind their application
3.1 Earnings surprise effect
An accurate measure of the earnings surprise effect needs to capture the difference between the realization of the actual earnings on the announcement date and the prior expectations of the market, the unexpected earnings (Bernard, 1990). Throughout the academic literature there have been several ways of calculating the unexpected earnings and quantifying the earnings surprise effect. Bernard (1989) utilizes the historical earnings to estimate the earnings forecast and takes difference between those values and the actual ones. Then the unexpected earnings are standardized by being scaled by the standard deviation of the errors of the forecast. Another commonly used approach defines the unexpected earnings as the difference between the actual earnings and the consensus forecast of those earnings as in Sloan (2002) and Ekholm (2006). Sloan also standardizes the unexpected earnings by the standard deviation of the forecasts for given earnings announcement. The standardized unexpected earnings (SUE) can then be used for estimation of the magnitude of the earnings surprise effect and the PEAD. This paper will follow the methodology of Sloan (2002) as the forecast analysis are being increasingly more available and do well reflect the expectations of the investors and through them the market itself. SUE is calculated quarterly for each firm in the dataset and the average forecasted EPS is the average of all the forecasts for given earnings announcement.
UE (Actual EP S Average forecasted EP S)/δ(Actual EP S forecasted EP S) (1)
S = − −
The SUE is then divided into positive and negative earnings surprises through variable Esurprise, which is equal to 1 in case of positive SUE and 0 otherwise. Aforementioned variable is necessary for the testing of the hypothesis including the CAR and PEAD to observe if the PEAD develops in the same direction as earnings surprise effect.
Having established the above mentioned measures the first hypothesis can be tested by regressing the SUE on the Emerging dummy(1 = Emerging market, 0 = Developed) for positive and negative surprises individually in order for the SUE values to not offset themselves:
Hypothesis 1:
There is a negative relationship between the development of the market and magnitude of standardized unexpected earnings
Empirical model:
UE β β merging e (2)
S i,t = 0 + 1* E + i,t
SUE measures the earnings surprise effect, showing by how much did the earnings forecast differ from the actual earnings for the given quarter. Coefficient β0measures the SUE for the developed markets, while the β1shows the difference in the SUE between the emerging and developed markets. The expectation for the value of coefficient β1 is to be significant and have the same sign as the direction of the surprise. based on the literature review, to reflect that emerging markets (India) do experience higher SUE. The model is performed firstly on the sample of negative surprises and then on the sample of positive surprises in order to prevent them from offsetting each other.
3.2 Post-earnings-announcement drift
In order to observe the post-earnings announcement drift the first step is to estimate abnormal returns, which are then cumulated over the duration of the event period and used in the regression analysis to test aforementioned hypotheses. The abnormal returns are the
difference between the expected returns and the actual returns. While the actual returns of the stock can be obtained from the dataset, the expected returns need to be estimated.
An event study of earnings surprise is done to find the expected returns through the estimation window starting 20 days before announcement and ending 5 days before announcement and to capture the magnitude of the PEAD through the persisting abnormal
returns in the event window starting from 3 days before the announcement and ending 60 days after. Bernard (1989) found that most of the PEAD occurs within 60 days after the announcement, however significant abnormal returns were observed 3 days prior as well by Ball(1968). Based on the existing literature such as the study by Bernard (1989) and Ball (1968) the event window was chosen. The estimation window was selected to prevent the event of earnings announcement from influencing the returns within the window as much as possible. As a result of dataset containing multiple subsequent quarters, only days that are more than 60 days after any given announcement and more than 3 days prior can be used for estimation, leaving only small subset suitable for the task. Using the estimation window of [-20,-5] the influence of previous and upcoming announcement are minimized while leaving as large of a subset as possible. However several other studies found that the best window to highlight the PEAD occurs during the short event window of 3 days prior and 3 days after the earnings announcement. Therefore this paper will utilize this event window as well. Having two event windows that are very far apart might bring forth extreme differences in results and to provide a midway step 3rd event window of 3 day prior and 30 days after will be examined as well
The expected returns of stocks for given day are calculated through the capital asset pricing model (CAPM) as sum of market risk premium(different for both emerging and developed markets ) for the given time period multiplied by the beta of the stock and risk free rate for the given quarter as follows:
(r ) f (E[Rm ] f) (3)
E i,t = R t + βi,t t − R
Betas in the equation 2 are calculated during the estimation window through a CAPM model by the following OLS regression:
After the expected returns are estimated the variable cumulative abnormal return (CAR), which is used for most of the regression analyses is created as the sum of abnormal returns over all of the event windows [-3.3], [-3,30] [-3,60].
AR r E(r ) (5) C i,t = ∑n
−3 i,t − i,t
With the cumulative abnormal returns estimated for each quarter and firm in both markets, it is important to test if the earnings effect is significantly related to the CAR or in other words the PEAD in the dataset chosen for both markets as it is necessary for further examination of the PEAD and earnings surprise phenomenon to ultimately answer the research question of this paper.
Hypothesis 2:
Standardized unexpected earnings are positively related to cumulative abnormal returns in both markets
Empirical model:
AR β β UE e (6)
C i,t = 0 + 1* S i,t + i,t
CAR in the equation (6) is a measure of abnormal returns cumulated over the period of 3 days prior to the earnings announcement in the given quarter and 60 days after. SUE in turn is the measure of the earnings surprise effect, showing by how much did the earnings forecast differ from the actual earnings for the given quarter.
The β1coefficient measures the responsiveness of the abnormal returns the SUE and is expected to be positive and significant to reflect the tendency of the abnormal returns to drift in the direction of the earnings announcement. β 0 captures the abnormal returns in case of no unexpected earnings.
After establishing the relationship between the PEAD and SUE in order to come closer to answering the research question the analysis turns to comparison of the emerging markets and developed markets with regards to the earnings surprise coefficient.
Hypothesis 3:
The development factor of a market will have negative impact on the cumulative abnormal returns
Empirical model:
(7)
AR β β UE β merging merging UE e
C i,t = 0 + 1* S i,t + 2* E i + β3* E i * S i,t + i,t
Variables CAR and SUE retain the same definitions as in the previous empirical model and the variable Emerging = 1 for emerging markets and 0 for the developed ones. The
coefficient β1 is examined to find the relationship between the SUE and CAR for developed markets, because in that circumstance the regression equation simplifies to
and as in previous regression should be positive.The
AR β β UE e
C i,t = 0 + 1* S i,t+ i,t
interactive termEmergingi * SUEi,t is utilized to capture the effect of the emerging market on the earnings surprise coefficient. The difference in the emerging markets to the developed ones therefore is measured by β2+β3 SUEi,t and is expected to be positive due to higher barriers to efficiency like higher transaction costs and information asymmetry in the emerging markets. The β0captures the abnormal returns of stock from developed market under no unexpected earnings.
In order to explain the difference in the earnings surprise coefficient this paper will utilize the liquidity as possible factor that influences the CAR based on studies such as Sadka (2006). Liquidity is naturally measured by bid-ask spread(Amihud 1986) and therefore daily bid-ask spread is calculated for every stock and averaged for each company-quarter creating variable BAS. The newly created variable is then added to the regression equation (6) as independent variable along variable Emerging to answer the hypothesis 4
Hypothesis 4:
Bid-ask spread has positive significant effect on the cumulative abnormal returns
Empirical model
AR β β UE β merging AS
The β0captures the abnormal returns of stock from developed market under no unexpected earnings and bid-ask spread = 0, while β1measures the earnings surprise effect of developed market and is expected to be positive. Difference between the earnings surprise effect on the CAR is presented in the β2coefficient, which in literature is shown to be positive. Lastly coefficient β3 captures the effect of the bid-ask spread on the CAR and is predicted to be positive as higher transaction costs ought to prevent trades that could potentially adjust the returns of the stock.
3. Data selection
Earnings per share, earnings announcements dates and the earnings forecasts with the
quarterly frequency for American and Indian firms were gathered from the IBES database for period of 3 years from 2013-2015. Indian markets and the Indian firms are used to represent the emerging markets, due to the availability of information on the stocks while the US firms represent the developed world. The SUE for the Indian firms is calculated based on the 1102 data quarter observations for firms and their respective average forecasts while the SUE of American firms is estimated through 29 398 data quarter observations. The complete dataset contains 1611389 daily observations. While the American dataset is far more extensive than the Indian due to availability of information and lower interest in the forecasts of stocks from emerging markets it does not skew the empirical analysis by overliance on the US data. The data from the both countries are used in comparative way and therefore the number of observation within them only increases the precision of the estimates and their external validity and does not add weight in the regressions. Any firms without 10 consecutive quarters and their respective SUE were excluded from the dataset.
To approximate the market returns the S&P 500 and NIFTY 500 indices were used as they are the broadest indices that most closely mimic their respective markets.
Daily market returns were downloaded from the CRSP database for US firms and from NSE for the Indian firms for the duration of the 3 years mentioned before. Market betas,
3 month Treasury bill rates and 3 month Indian government bond rates are used as the proxy for the risk-free rate for developed emerging market respectively.While the Treasury bill rates were downloaded from the Federal reserve bank of St. Louis the data Indian
government bond rates were obtained from the renown investment portal investment.com
Daily stock prices as well as bid-ask spread is gathered from the CRSP database for the US firms chosen based on availability of the stocks in the IBES database and the number of forecasts they provide. Data on daily stock prices of Indian firms are gathered from the NSE based on the same criterion.
5. Results
In a brief summary, the upcoming section contains the results of the empirical models used to test the previously mentioned hypotheses and provide substantial information towards the goal of answering the research questions. Firstly the outcomes of the prior literature are tested on the dataset by finding the relationship between the standardized unexpected returns and dummy variable of emerging market. The presence of the PEAD resulting from earnings surprise is then confirmed by trying to explain the existing CAR by the SUE. Secondly the regression to find the difference in the magnitude of PEAD in the developed and emerging market is performed. Lastly the effect of stock liquidity on the difference in the CAR between the two types of market is assessed through a linear regression.
5.1 Standardized earnings surprise
To start answering the research question of this paper and begin the comparison of the Indian, market representing the emerging markets, and US market, representing the developed
markets its important to investigate the presence of SUE and PEAD in the markets. Firstly it is important to show the differences in terms of the earnings surprises between the markets in order to validate the further analysis. The results of the empirical model testing the first hypothesis is presented in the Table 1.
The variable Emering is equal to 1 in case of stock and its surprise effect belonging to the Indian stock market and 0 if its from the US market. Esurprises tracks if the earnings surprise is positive or otherwise to prevent the observations from offsetting each other and providing inaccurate results. From Table 1 the SUE can be observed to be on average -1.514 if earnings surprise is negative 2.139 if its positive. Compared to that the emerging market experiences significantly higher level of surprises as seen from the significantly positive value of coefficient Emerging in the subset of positive surprises and significantly negative value in the subset of negative surprises. The value of the coefficient is 0.0917 in positive subset and -0.496 in the negative one. The absolute value of SUE is higher for the emerging markets in both scenarios and therefore the results support the theoretical background and the validity of the Hypothesis 1: There is a negative relationship between the development of the market and magnitude of standardized unexpected earnings.
5.2 Cumulative abnormal returns
After confirming the presence of higher earnings surprises in the emerging market, the attention is turned towards the explanation of the PEAD represented through the abnormal returns cumulated over the event window. The entire dataset is used to first confirm the existence of the PEAD and then subsequently the difference in the magnitude of PEAD between the developed and emerging market is analysed. In order to test the former and
whether the PEAD is influenced and can be explained at least partially by the earnings surprises, the CAR of different estimation windows are regressed on the SUE. Table 2 provides the results from these regressions with the column 1 presenting results from the regression with estimation window of 3 days prior and 3 after, column 2 representing the results of the window [-3,30] and lastly column 3 with the data on the long event
window[-3,60].
The effect of variable SUE is found to be significantly positive in all of the regressions on the CAR, suggesting that the presence of the surprise increases the CAR in the direction of the surprise. Therefore the earnings surprise can be seen as at least partially responsible for the PEAD confirming the relationship between the two in the dataset. Moreover the coefficient of SUE is increasing with the increase in the duration of the event window, implying that the earnings surprise effect persists at least until the 60 days of the announcement. The positive coefficients of SUE from table 2 support the Hypothesis 2: Standardized unexpected earnings are positively related to cumulative abnormal returns in both markets.
Discovering the presence of the PEAD in both markets and the positive effect of the SUE on the CAR for all event windows create ground for testing of the difference in PEAD between the markets. The regressions used in the table 2 are extended firstly by inclusion of the variable Emerging to analyse the difference and then in next of regression by inclusion of
interaction variable Emerging*SUE. The interaction variable is employed to explore the possibility of the effect of earnings surprise being influenced by the development of the market where it occurred. Table 3 summarizes the above mentioned regressions for all the event windows with columns 1-3 providing results from regressions without the interaction variable and 4-6 for the ones where its added.
The coefficient SUE is still found to be significantly positive and increasing in magnitude with the increased duration of the event window. Supporting the Hypothesis 3: The development factor of a market will have negative impact on the cumulative abnormal returns, the Emerging market coefficient is significantly positive along all of the regressions and the interaction variable from the regressions 4-6 shares the same characteristic.
Interaction term is also significant along all regressions but positive only for the two smaller event windows in column 4 and 5, and negative for the last one. Observing the small positive relationship in the smaller event windows provides weak support for the idea that the effect of the earnings surprise is affected by the development of the market as the coefficient values of SUE are affected by inclusion of the interaction variable. The descending value of the interaction coefficient with the increase in duration of the event window could be caused by the earnings surprise effect being priced in at the slower rate by the emerging market than the developed one. The negative value in the last regression could suggest that with enough time the emerging market can accommodate the earnings surprise better than the developed market, but that would not be consistent with existing literature as developed markets are found to be more efficient than the emerging ones. This result does provide some questions and considerations for a future research.
The value of the Emerging market coefficients in both sets of regression is largest for the event window [3,30] with value 0.808 in regression 2 and 816 in regression 5 while the shortest window offers values of 0.0230 and 0.0250 for regression 1 and 4 respectively. Smallest values of 0.209 and 0.206 come from the regressions on the largest window in the column 3 and 6 respectively. The large difference between these event windows could be caused by the delayed response of the investors, who start properly reacting to the news only after some period of time. This delay in response would be consistent with the findings of Bernard (1989). The possible reason why the investors in emerging markets may be delayed in their response more than those from the developed could be the higher transaction costs between the markets. Higher transaction costs may act as deterrent to trade and therefore slow down the rate by which the new information is incorporated into the stock prices. The next section therefore analyzes the effect of transaction costs on the CAR.
5.3 The effect of liquidity
From the results of the Table 3 raised a potential explanatory factor for the presence of the PEAD. As suggested in the section literature review the liquidity was found to certain extent influence the CAR in the event window (Sadka, 2006). The results of the analysis of the effect of liquidity on the CAR for all even windows in form of bid-ask spread is presented in the Table 4. The variable BAS is the measure of liquidity equal to the difference between the daily bid and ask while the constant observes the CAR in case of no bid-ask spread and no standardized unexpected returns.
Coefficient of the BAS is found to be significant only in the shortest of the event windows with negative value of 0.103 implying that the effect of the liquidity on the PEAD is only short term. This finding is not in line with the previous literature and findings of studies like Sadka (2006) or Li (2017) and would require separate examination. The Hypothesis 4: Bid-ask spread has positive significant effect on the cumulative abnormal returns, finds no support in the data presented in Table 4, but can be adjusted to state that the bid-ask spread has negative effect on the cumulative abnormal returns in the short-term around the earnings announcement.
6.Discussion and concluding remarks
This section briefly summarizes the findings from the empirical analysis, tries to draw conclusions with regards to the hypotheses of the paper and finally relates the results to the research question. Last part of this section includes the limitations of this study and
suggestions for further research based on the results.
In order to find the answer to the research question this paper first analyzed the presence of the earnings surprise through the measure of SUE related to the development of the market. The market representing the emerging markets was India and the one representing the developed was US. Prior studies by Bernard (1989), Abarnabell (1992) and others found the presence of the earnings surprise effect in the US market and Sen (2009) as well as
Sehgal(2015) found the earnings surprise effect in Indian market. The Table 2 provided some level of confirmation of the prior studies in the significant positive value of SUE in both markets. This confirmed the second hypotheses of the paper as well as provided partial answer to the research question as the presence of the earnings surprise effect was confirmed. The PEAD effect appeared to persist even until the 60 days after the earnings announcement However to find the difference in the magnitude of this effect between the markets the comparison of coefficients of SUE for each market was made. Assumption based on the liquidity studies like Sadka(2006) provided motivation for the Hypotheses 1 and 3 where the expectation was that the emerging market will experience larger earnings surprises and as a result of that and other factors like liquidity and information asymmetry larger CAR. Results in Table 1 provided some evidence in favor of the Hypothesis 1 and suggested that the forecast analyses, which were basis of SUE measure are less precise in the emerging market. The Hypothesis 3, which most closely relates to the second part of the research question in finding the difference in magnitude of the earnings surprise effect between the markets found support in the results of empirical analysis presented in Table 3. The emerging markets were found to have larger magnitude of PEAD as well as the earnings surprise effect was found to be negatively influenced by the development factor in the short term. While this provided some support for the Hypothesis 3, it did not provide much clarity in respect to difference in magnitude. The sign of the difference was found from the result of this study and therefore
partial answer to the research question can be provided but the actual magnitude of this difference was not closed in on, due to the negative effect of the interaction variable in Table 3 in the long event window. Next the paper attempted to find the reason for this difference in earnings surprise effect between the markets by employing the methodology of papers focused on liquidity and its connection to the PEAD like Sadka(2006) or Sen (2009) and find the effect of liquidity on the CAR. However the result of that analysis presented in Table 4 were not in line with the previous literature as the effect was not only found to be
insignificant, which was found by some studies, but also negative in the 3 day window around the announcement. The Hypothesis 4 was therefore not supported and this particular result could be basis of the future studies.
The answer to the research question based on the support found for the Hypothesis 1,2 and 3 would be that the earnings surprise effect as well as PEAD are present in both the developed and emerging markets. The magnitude of the surprise effect and PEAD was found to be depended on the development factor with negative relationship, however the its magnitude of this effect is still inconclusive. The paper was able to shed some light on the effect of
development factor on the earnings surprise and PEAD and the author believes it contributes to the body of knowledge with confirmation of the existence of the earnings surprise effect in both markets and the significance of the development factor.
6.1 Limitations:
The inconclusiveness of this study and to some extent the accuracy of the results could have been influenced by the choice of the dataset and some of the methodology. The dataset used by this study had only 153 firms from the Indian market for the 3 year period and therefore including more firms and larger time periods could provide better accuracy to the results pertaining to the emerging market. Furthermore the CAPM model was used to calculate expected earnings due to its widespread usage in previous studies, however multifactor models could have brought more accuracy and provide better estimates of the returns. Additionally the SUE was calculated through the forecast analysts predictions, which were sparse for the Indian market threatening the accuracy of the SUE measure. The puzzling results from the Table 4 may have also been caused by the research design as there are more liquidity measures that could be used apart from the bid-ask spread. Lastly the paper was
unable to explain the some of its result pertaining to liquidity like the ones in Table 4, which could be basis for further research. The connection of liquidity to earnings surprise in the comparison between the emerging markets does not have much literature background. Moreover magnitude of the effect of development factor on the earnings surprise and PEAD should also prove to be interesting topic for the further research in the area.
References
1. Abarbanell, J. S., & Bernard, V. L. (1992). Tests of analysts’
overreaction/underreaction to earnings information as an explanation for
anomalous stock price behavior. Journal of Finance, 47(3), pp. 1181-1207.
2. Amihud, Y & Mendelson H. (1986). Asset pricing and the bid-ask spread Journal
of Financial Economics, 6, pp. 223-249
3. Ball, R., & Brown, P. (1968) An empirical evaluation of accounting income
numbers, Journal of Accounting Research. 6, pp. 159-78.
4. Bernard, V. L. & Thomas, J. K. (1990). Evidence that stock prices do not fully
reflect the implications of current earnings for future earnings. Journal of
Accounting and Economics, 13(4), pp. 305-340.
5. Bernard, V. L., & Thomas, J. K. (1989). Post-earnings- announcement drift:
delayed price response or risk premium? Journal of Accounting research, 27, pp.
1-36.
6. Brown, S., Hillegeist, S. A. & Lo, K. (2009). The effect of earnings surprises on
information asymmetry. Journal of Accounting and Economics, 47(3), pp.
208-225.
7. Ekholm, A. G. (2006). How do different types of investors react to new earnings
8. Fama, E.F. (1970) Efficient capital markets: A review of theory and empirical
work. The Journal of Finance. Vol. 25(2), pp. 383-417.
9. Frazzini, A. (2006). The disposition effect and underreaction to news. Journal of
Finance, 61(4), pp. 2017-2046.
10. Kaestner, M. (2006). Anomalous price behavior following earnings surprises:
does representativeness cause overreaction? Finance, 27, pp. 5-31.
11. Li, A. M. & Dempsey, M. (2017). Is post earnings announcement drift a prices
risk factor in emergency markets? Chinese evidence. Archives of Business
Research, 5(6), pp. 29-47.
12. Livnat, J. & Mendenhall, R. R. (2006). Comparing the post-earning
announcement drift for surprises calculated from analyst and time series
forecasts. Journal of Accounting Research, 44(1), pp. 177-205.
13. Malkiel, B. G. (2003). The efficient market hypothesis and its critics. Journal of
Economic Perspectives, 17(1), pp. 59-82.
14. Sadka, R. (2006). Momentum and post-earnings- announcement drift anomalies:
The role of liquidity risk. Journal of Financial Economics, 80(2), pp. 309-349.
15. Sehgal, S. & Bijoy, K. (2015). Stock price reactions to earning announcements:
evidence from India. Vision- The Journal of Business Perspective, 19(1), pp.
25-36.
16. Sen, K. (2009). Earnings surprise and sophisticated investor preferences in India.
Journal of Contemporary Accounting & Economics, 5(1), pp. 1-19.
17. Skinner, D. J. & Sloan, R. G. (2002). Earnings surprises, growth expectations,
and stock returns or don’t let an earnings torpedo sink your portfolio. Review of
