The impact of IFRS on the reaction to lower than expected earnings : a European analysis of the difference in market reactions to earlier- versus fourth quarter bad news earnings announcements

Bachelor Thesis

BSc. Economics and Business – Finance and Organization

Field: Finance

The impact of IFRS on the reaction to lower than expected earnings

A European analysis of the difference in market reactions to earlier- versus fourth quarter

bad news earnings announcements.

Jesse Heynens

11047356

University of Amsterdam (UvA)

26

_{of June, 2018}

1

Statement of Originality

This document is written by Student Jesse Heynens 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 thesis analyzes whether there is a difference in stock price reaction between earlier- and fourth quarter bad news earnings announcements for European companies. First, the existence and signs of abnormal returns are tested, after which they are used to analyze if the height of those abnormal returns differs when bad news earnings are disclosed in the earlier quarters compared to the fourth quarter. No evidence is found to support the hypothesis that lower than expected earnings announcements in theearlier quarters prompt a heavier stock price reaction than fourth quarter bad news earnings. As such a difference was found under the US GAAP accounting standards, it can be concluded that the IFRS accounting standards suppress European firms from using earnings management that differentiates fourth quarter- from earlier quarter earnings.

Keywords:

2

Table of content

1. Introduction...3

2. Literature review ...5

2.1 Stock price reactions to quarterly earnings disclosures ...5

2.2 Definition of unexpected earnings ...8

2.3 The manipulation of earnings by managers ...9

2.4 The impact of IFRS on earnings management ...10

2.5 The difference between earlier- and fourth quarter earnings ...11

2.6 Measuring the difference in stock price reaction ...14

2.7 Hypotheses ...15

3. Sample construction and methodology ...15

3.1 Sample construction ...16

3.2 Methodology ...16

3.2.1 Hypotheses 1a and 1b ...17

3.2.2 Hypothesis 2...18

4. Data analysis and discussion ...20

4.1 Descriptive statistics ...20

4.2 Existence and sign of the cumulative abnormal returns...21

4.3 Difference in stock price reaction to bad news earnings ...22

4.4 Robustness check ...23

4.5 Discussion ...25

5. Conclusion ...26

5.1 Summary and conclusion ...26

5.2 Suggested research ...27

Bibliography ...29

3

1. Introduction

The links between unexpected earnings, quarterly earnings disclosures, earnings manipulation, the timing of information releasesand stock price reactions have been topics of numerous articles (Mendenhall and Nichols, 1988; Salamon and Stober, 1994; Bartov et. al, 2002; Das and Shroff, 2002; Lopez and Rees, 2002; Brown et al., 2009;). One of the conclusions formed is that there is a difference in stock price reaction between quarterly earnings announcements in the first three (earlier) quarters and such an announcement in the fourth-quarter (Mendenhall and Nichols, 1988; Salamon and Stober, 1994, Das and Shroff, 2002). The trend is that the markets react less heavily to fourth-quarter earnings announcements, which is due to the following reasons: the market expects that managers delay bad news (to the fourth quarter) and can do so by manipulating their earnings. Therefore, when lower than expected earnings are disclosed in the earlier quarters it is less expected and leads to a heavier reaction (Mendenhall and Nichols, 1988). Secondly, managers are known to try to balance their earnings over a full fiscal-year and their last resort to do so is by manipulating earnings in the fourth quarter. The market is aware of that notion and therefore does not react as heavily to earnings announcements in the fourth quarter compared to earlier quarter announcements (Salamon and Stober, 1994, Das and Shroff, 2002).

Studies into the difference in stock price reactions between quarterly earnings disclosures in earlier- and fourth quarters, have only used United States (US) based companies in their samples (Mendenhall and Nichols, 1988; Salamon and Stober, 1994, Das and Shroff, 2002). As a result, the conclusions from these studies are based on US-specific characteristics, most notably earnings manipulation which can be applied under the US GAAP accounting standards. Therefore, it is the question if these results also hold for European companies which are subjected to the IFRS accounting standards. As good- and bad news disclosures have different characteristics and implications as proven by Kothari et al. (2009), this research will focus on the difference in market reactions to bad news earnings announcements, between earlier- and fourth quarter disclosures.

To form a conclusion on the topic, the following research question should be answered: Is there a difference in stock price reactions to lower than expected earnings disclosures between earlier- and fourth quarter announcements for European companies?

On topics linked to the research question, the following conclusions have been made: earnings management is the manipulation of earnings, used to meet or beat analysts’ forecasts (Roychowdhury, 2006; Zang, 2011); Managers personally gain from delivering higher than expected earnings (good news) and the markets react negatively if earnings are lower than

4

expected (bad news) (Bartov et. al, 2002; Lopez and Rees, 2002) and managers tend to withhold bad news (Kothari et al., 2009).

In the European Union (EU) the mandatory use of the IFRS accounting standards was adopted to improve the information environment (Armstrong et al., 2010). After its implementation, Jeanjean and Stolowy (2008) conclude that the implementation of IFRS did not decrease earnings management and led to an increase in France. Continuing, IFRS may have the unintended effect of increasing real earnings management (Ipino and Parbonetti, 2016). Where real earnings management is defined as the manipulation of earnings through real business activities (e.g. increasing inventories and driving up sales through discounts) (Roychowdhury, 2006). Which is different from the accrual earnings management methods used under US GAAP (Mendenhall and Nichols, 1988; Das and Shroff, 2002; Zang; 2011).

To be able to form hypothesis and come to a conclusion on the maim research question, the following (sub-) questions should be answered: How do stock prices react to earnings announcements and how is this reaction measured? How are unexpected earnings defined? How do managers influence their earnings? How did IFRS influence the use of earnings management in the EU? What is the difference between earlier- and fourth quarter earnings disclosures? And lastly, how can the difference in stock price reactions be measured?

The sample for this research will contain 55 companies, randomly drawn from the 160 biggest companies by market capitalization, incorporated in the EU. The used data will be drawn from 2006 until 2016 and for each quarter for each firm, the following variables will be used: quarterly earnings announcement dates, forecasted earnings per share by analysts’, actual earnings per share, stock price reaction surrounding the earnings disclosure and stock prices.

The cumulative abnormal returns will be calculated using the market-adjusted model, after which they will be analyzed using a t-test (De Jong, 2007). Then the unexpected earnings will be calculated and finally, a regression will be used to analyze the difference in stock price reactions between lower than expected earnings disclosures in the earlier- and fourth quarters. The structure of this thesis is as follows: in section two, previous literature on topics linked to this research will be discussed, where after the hypotheses will be formed. Section three contains the sample set-up and methodology used. The results and analysis will be presented and discussed in section four and the article will conclude in section five, with a summary and suggested research.

5

2. Literature review

In this section, relevant literature needed to answer the research questions will be discussed. To start, the effects of quarterly earnings disclosures on stock prices and how those effects can be measured will be introduced. Thereafter, the surprise factor of such earnings disclosures will be defined. After which, the implications on how managers try to manipulate earnings and how that affects market expectancies will be discussed. Then, the effects of IFRS on earnings manipulation will be discussed. Continuing, the different characteristics of earlier- and fourth quarter earnings will be summarized, which leads to the main hypothesis. Lastly, the methods to measure the difference in market reaction between quarters will be presented. This section will conclude with a summary of all formed hypotheses.

2.1 Stock price reactions to quarterly earnings disclosures

The existence of stock price reactions to earnings announcements is nothing new (Beaver, 1968; Bernard and Thomas, 1990; Bartov et al., 2002; Brown et al., 2009; Landsman et al., 2012). This section will summarize the findings surrounding this topic and analyze how such a price reaction can be measured. Two hypotheses will be formed, where the type of earnings announcements (good- or bad news) will be linked to the sign of the stock price reactions.

Table 1 Event studies literature summarized

Author(s) Country Time

period

Method Result

Beaver (1968) U.S.A. 1961 - 1965 CAPM Abnormal returns exist surrounding earnings disclosures Bernard and Thomas (1990) U.S.A. 1974 - 1986 Market-adjusted model (+ enhancements)

Three day period is needed to incorporate

all information into the stock price Bartov et al.

(2002)

U.S.A. 1983 - 1997 CAPM Greater stock returns are present after analysts’ forecasts

6 Brown et al.

(2009)

U.S.A. 1995 - 2004 Market-adjusted model

The positive relation between earnings and abnormal returns

is greater for lower than expected earnings Kothari et al. (2009) U.S.A. 1995 - 2002 Market-adjusted model

Stock prices react more heavily to bad

news Landsman et al, (2012) 16 countries with IFRS (e.g. Japan, EU, Australia)

2002 - 2007 CAPM Information content of earnings improved

after adoption of IFRS

When stocks react significantly to an event, the effect can be measured through abnormal returns (Fama et al., 1969). Here abnormal returns are the difference between the expected returns and the actual returns of a security (De Jong, 2007). This follows from the efficient market hypothesis, where stock prices are believed to always reflect all available information about the intrinsic value of a stock (Fama, 1965). This theory has later been revised, as Fama (1970) describes how event studies can be used to show how quickly prices adjust to new information, which shows the reaction of the market to an event (Fama, 1991).

The impact of (quarterly) earnings disclosures on stock prices has been the focus of numerous (event) studies (Beaver, 1968; Bernard and Thomas, 1990; Bartov et al., 2002; Brown et al., 2009; Landsman et al., 2012). Beaver (1968) is the first to study the impact of earnings announcements on stock prices and concludes that earnings announcements do contain information that prompts the market to react in both stock prices and amounts of stock traded. Using the research methods from Beaver (1968), Landsman et al. (2012) found that the information content of earnings disclosures increased after the adoption of IFRS, which was shown through higher abnormal returns and increased trading volumes surrounding earnings disclosures.

When considering the difference between the actual earnings and the expected earnings, summarized as unexpected earnings, Bartov et al. (2002) find that stock price reactions are positively related to the unexpected earnings. Afterwards, Brown et al. (2009) conclude that

7

the positive relation between unexpected earnings and abnormal returns is greater for lower than expected earnings compared to positive unexpected earnings for a US based sample. The positive link between unexpected earnings has thus been proven, therefore our hypothesis will be the same for the sample used in this research. It will be interesting to see whether the results from Brown et al. (2009) will also be evident under the IFRS accounting standards. However, that will not be a tested hypothesis of this research.

When analyzing stock price reactions, previous literature has mainly used two models to define the abnormal returns. Either the market-adjusted model, where the abnormal returns are the difference between the stock price return and the return of a market index (Brown et al., 2009; Kothari et al., 2009), or the market model residuals (CAPM) where the abnormal returns are calculated using an ordinary least squared (OLS) regression (Beaver, 1968; Bartov et al., 2002; Landsman et al., 2012). When describing event study methodology, De Jong (2007) explains how both models are valid ways to define abnormal returns.

Lastly, earlier studies analyze the abnormal returns of multiple days surrounding an event. The number of days that are analyzed varies, between 3- (Brown et. al., 2009), 5- (Kothari et al., 2009) and 7 days (Bernard and Thomas, 1990). After investigating the number of days that should be analyzed, Bernard and Thomas (1990) concluded that even though the biggest price shocks take place on the first trading day after an event, it takes up to three days for all information to be fully reflected in the stock price. Also, Kothari et al. (2009) describe how good news earnings announcements tend to leak to the public before the actual disclosure, which leads to a lower shock directly after the announcement. Therefore, when choosing the number of days to be analyzed, one consideration could be to take a low number of days (one day before until one day after the announcement) and focus on the days with the heaviest shocks, or more days surrounding the event could be used, which should include more of the information content of an announcement.

Concluding, the general conclusions from the papers evaluating quarterly earnings announcements and the reaction of stock prices is that, stock prices do react after such disclosures, resulting in abnormal returns. Therefore, it is expected that the data used in this research will show the same results, especially as the sample contains companies under the IFRS rules, where earnings disclosures are concluded to contain a higher degree of information content (Landsman et al., 2012). Following the conclusion from Brown et al. (2009), it is expected that bad news earnings are matched with negative abnormal returns, while good news earnings lead to positive abnormal returns. As stock price reactions need to be present to answer the main research question, the following hypotheses will be tested:

8

Hypothesis 1B: ℎ < 0 < 0

2.2 Definition of unexpected earnings

When analyzing the reaction to earnings announcements, former papers do not analyze the reaction to the actual returns, but analyze the reaction to the difference of expected- versus realized earnings per share (Mendenhall and Nichols, 1988; Affleck-Graves et al., 2002; Lopez and Rees, 2002; Francis et al., 2002; Brown et al., 2009). These papers define the difference between realized and expected earnings per share, as the unexpected earnings. This section elaborates on how those unexpected earnings are measured and why certain parameters are used.

Former papers have used multiple ways to define expected earnings. When analyzing the difference in stock price reactions to earlier- and fourth quarter earnings disclosures, both Mendenhall and Nichols (1988) and Salamon and Stober (1994) used econometric models (e.g. the seasonal random walk- and the three Box-Jenkins model) to calculate the expected earnings. However, at that time, O’Brien (1988) already described how analysts’ forecasts outperform other earnings forecast models. Therefore, both Mendenhall and Nichols (1988) and Salamon and Stober (1994) also tested their models with analysts’ forecasts as the proxy for expected earnings and conclude that the chance in type of expected earnings used, did not influence their outcomes. More recent papers have shifted to only using the consensus analysts’ forecasts as the expected earnings as it is believed to better represent the market beliefs about the expected earnings (Affleck-Graves et al., 2002; Lopez and Rees, 2002; Francis et al., 2002; Brown et al., 2009).

When calculating the unexpected earnings, a deflator is used to standardize the outcomes (Mendenhall and Nichols, 1988; Salamon and Stober, 1994; Lopez and Rees, 2002; Francis et al., 2002; Brown et al., 2009). However, between these papers, there is no consensus on which deflator should be used (e.g. share price before the event, share price at the event date, actual earnings). When studies were tested with multiple deflators, no significantly different outcomes were measured (Mendenhall and Nichols, 1988; Gu and Wu, 2003). It can thus be concluded that the choice of deflators does not have a significant influence on the outcome of an analysis.

To summarize, in articles with similar topics, the unexpected earnings are defined as the difference between actual- and expected earnings. In recent research analysts’ forecasts are used as the expected earnings and the unexpected earnings are standardized using a deflator. Where the type of deflator used, does not significantly influence the outcomes of the analysis.

9

2.3 The manipulation of earnings by managers

The manipulation of earnings is described as one of the main drivers of the differences in expectancies between earlier- and fourth quarter earnings, which ultimately leads to the difference in stock price reaction (Mendenhall and Nichols, 1988; Salamon and Stober, 1994; Das and Shroff, 2002). For example, managers are expected to manipulate their fourth-quarter results, such that the earnings even out over the whole year, the market anticipates this and as result does not react as heavily when unexpected earnings are presented in the fourth quarter (Das and Shroff, 2002). This section will elaborate on how managers influence their results and how that influences the expectancies of the market.

Lo et al. (2017) define earnings management as the manipulation and misreporting of information by management for their own benefit. The incentives to apply earnings management are firm-level benefits (e.g. lower cost of capital and higher stock returns), which translate into personal benefits for managers (e.g. bonuses and stock options) (Bartov et al., 2002; Brown et al., 2009). When researching the implications of earnings management, such as the influence of CEO’s, types of earnings management and how earnings management shows in financial reports, it is all concluded that earnings management still exists (Zang, 2011; Jiang et al., 2010; Lo et al., 2017).

It is hard to quantify the impact of earnings management and measure how often it actually occurs, as papers proving the existence of earnings management do not explicitly quantify their conclusions (Zang, 2011; Jiang et al., 2010; Lo et al., 2017). Some papers do state certain numbers, for example when Das and Shroff (2002), researched the tendency of companies to manipulate and reverse their earnings in the fourth quarter (e.g. a company reporting positive earnings in the first three quarters and a negative fourth quarter) they found that in 27% of the cases such an earnings reversal was present. When testing that number, they found that it is too high to be a coincidence, only it does not become clear which percentage of the 27% is due to earnings management. Therefore, when analyzing previous research, it is difficult to compare quantitative results with one and another.

When looking further into how managers manipulate their earnings, two main methods are used, real- and accrual-based earnings management (Zang, 2011). Where real earnings management is defined as the manipulation of earnings through real activities (e.g. increasing sales through extreme discount and, increasing inventory to lower cost of goods sold) to meet analysts’ earnings forecasts (Roychowdhury, 2006). While accrual-based manipulation is done through accounting practices, for example, the ability to favourably estimate costs under the US GAAP accounting standards (Mendenhall and Nichols, 1988).

10

When analyzing the use of real- versus accrual-based earnings management, Zang (2011) concludes that managers make a trade-off between the two based on their relative costs and use them as substitutes. As earlier research into the differences in earlier- versus fourth-quarter earnings announcements is heavily based on the use of accrual-based earnings management under US GAAP (Mendenhall and Nichols, 1988; Das and Shroff, 2002), it is thus important to analyze what trade-off is being made within the EU and if that trade-off is affected by the implementation of IFRS. Zang (2011) also finds that managers adjust their level of accrual-based earnings manipulation after they have analyzed the impact of their real earnings management activities. This could be interpreted as managers always realizing their maximum level of earnings manipulations if they prefer to do so. It then becomes the question, if they can under the IFRS accounting standards in the EU. This question will be answered in section 2.4.

In short, there are multiple sources confirming that earnings management is still present and that managers use both real- and accrual-based earnings management to manipulate their earnings. However, as previous research is heavily based on accrual-based earnings management under US GAAP, it is necessary to analyze if this trend is also present within the EU, which will be done in the next section.

2.4 The impact of IFRS on earnings management

After the implementation of IFRS, its effect on earnings management became a topic of research (Tendeloo and Vanstraelen, 2005; Jeanjean and Stolowy, 2008; Ipino and Parbonetti, 2017). This section will analyze those studies and see how IFRS influenced the amount- and type of earnings manipulation used. Thereafter, a comparison will be made between the characteristics of earnings management in the EU and the characteristics of earnings management of samples used in earlier studies linked to the research question.

In 2005 the EU adopted, the mandatory use of the IFRS accounting standards for all companies incorporated within the union. These standards were adopted by lawmakers to implemented uniform reporting standards across the EU, improve the quality of reporting compared to earlier standards (e.g. IAS, German GAAP) and to enhance the overall information environment (Armstrong et al., 2010). Concluding from the market reactions surrounding specific events linked to the implementation of IFRS, investors seemed to have a positive attitude towards the new rules and its implications (Armstrong et al., 2010). Also, Landsman et al. (2012) concluded that the information content (as measured by abnormal returns) of quarterly earnings announcement improved after the implementation of IFRS. It is thus safe to say, that investors believe that after the adoption of IFRS, earnings disclosures provide a better overview of the objective state of a company.

11

The impact of IFRS on the amount of earnings management has been questioned. When Tendeloo and Vanstraelen (2005) studied the use of earnings management under companies using IFRS versus companies using German GAAP (the German accounting standard before IFRS), they concluded that there is no proof, that firms who use IFRS apply less earnings management. This was later also confirmed by Jeanjean and Stolowy (2008) who wrote an article about the effect of mandatory IFRS on earnings management in the U.K., Australia and France. Their conclusion was that the presence of earnings management did not decline and even led to an increase in France.

When it was concluded that IFRS did not decrease the use of earnings management, it then became the question what the effect of IFRS on the type of earnings management applied is. Ipino and Parbonetti (2017) describe the trends in earnings management after IFRS has been implemented. They find that the probability that a firm uses accrual-based earnings management to report a small profit decreases, while the probability that a small profit is reported while using real earnings management increases, by which the overall probability stays the same. The conclusion is that the shift from accrual-based- to real earnings management is especially present in countries with strict institutional enforcement, which characterizes the EU (Ipino and Parbonetti, 2017).

The movement of EU companies towards real earnings management is contrary to the beliefs of earlier research, investigating the relationships between earnings management and the difference in market reaction to earlier- and fourth quarter (bad news) earnings announcements (Mendenhall and Nichols, 1988; Das and Shroff, 2002). As these papers focused on accrual-based manipulation applied over the different quarters. Therefore, it is clear that the sample used in this thesis is not exactly alike samples used in earlier research. As a result, it could be the case that the conclusion from this research also shows the effect of real earnings management on the difference between market reactions to bad news earnings announcements in earlier- versus fourth quarters.

To summarize, two conclusions can be made. First, earnings management did not decline and is still present in the EU after the adoption of IFRS. Therefore, it is relevant to investigate its effect on the price reactions to lower than expected earnings and if those differ within the different quarters of the year. And secondly, that European managers are substituting real earnings management for accrual-based earnings management, which is contrary to the practices analyzed in former papers surrounding the main topic of this thesis.

2.5 The difference between earlier- and fourth quarter earnings

Fourth quarter earnings announcements are usually more volatile and lead to a lower market reaction compared to earlier quarter earnings announcements (Jacob and Jorgensen, 2007).

12

Previous research describes how earnings management, in combination with the tendency of managers to delay bad news and the evidence that managers try to even-out their earnings in the fourth quarter, lead to the market reacting less heavily to fourth quarter (bad news) earnings disclosures (Mendenhall and Nichols, 1988; Salamon and Stober, 1994; Das and Shroff, 2002).

Table 2 studies into the difference in quarterly earnings summarized

Author(s) Country Time period Method Result Mendenhall

and Nichols (1988)

U.S.A. 1968 - 1986 Regression Lower than expected earnings prompt a heavier

stock price reaction in earlier quarters compared

to the fourth quarter Salamon and

Stober (1994)

U.S.A. 1981 - 1991 Regression Earnings announcements in peak sales quarters prompt a larger stock price reaction, and the response to fourth quarter earnings is

lower than for earlier quarter earnings Das and

Shroff (2002)

U.S.A. 1985 - 1998 Regression Firms use fourth quarter earnings to even out their

full year earnings. Therefore, do fourth quarter

earnings lead to lower stock price reactions than

earlier quarter earnings

When looking specifically at the difference in market reaction to bad news earnings announcements in earlier- compared to fourth quarters, Mendenhall and Nichols (1988) conclude that markets react heavier to bad news earnings announcements in the earlier quarters. Their explanation is that managers tend to delay bad news earnings and use earnings management to do so. Therefore, the investors expect that lower than expected earnings disclosures are more often announced in the fourth quarter and consequently, if lower than

13

expected earnings are disclosed in the earlier quarters the market is more surprised, which prompts a heavier stock price reaction (Mendenhall and Nichols, 1988).

The tendency of managers to delay bad news is first studied by Chambers and Penman (1984), who conclude that when earnings are disclosed later than expected, they usually contain bad news and when earnings are disclosed earlier than expected, they are characterized as good news. Mendenhall and Nichols (1988) interpret that as managers always trying to disclose bad news later, so rather in the fourth-quarter than in earlier quarters. Kothari et al. (2009) later confirmed that managers on average still delay bad news, when looking into the announcements of dividend pay-out rate changes. They also find that the magnitude of negative price reactions to bad news, greater is than the magnitude of the positive price reactions to good news, on average 1,3% after good news and -3,5% after bad news in the five trading days surrounding an announcement. This might come forward when analyzing the cumulative abnormal returns after earnings disclosures in the used sample.

The last difference between earlier- and fourth quarters is that managers use earnings management in the fourth quarter to even out their earnings (Salamon and Stober, 1994; Das and Shroff, 2002). Das and Shroff (2002) explain this as follows: if a company performed higher than expected, the manager would like to temper its earnings in the fourth quarter, so he can again present a higher than expected performance the year after. The same is the case when a firm underperforms, and a manager would like to higher the fourth quarter earnings, so the results are even over the full fiscal-year (Das and Shroff, 2002). The market anticipates the manipulation of earnings by managers to even their earnings in the fourth quarter, and as a result react less heavily when the fourth quarter earnings are presented (Salamon and Stober, 1994; Das and Shroff, 2002).

With the tendency of managers to delay bad news and the belief that managers use the fourth quarter to even out their earnings, it only leaves the question if managers still have the ability to do so under the IFRS accounting standards. Given the conclusions from Ipino and Parbonneti (2017) and Zang (2011), it is a given that managers in the EU have an increased focus on real earnings management. However, Zang (2011) notes that managers still apply accrual-based earnings management after they have analyzed the effect of the real manipulations. It is thus logical to think that if managers try to delay bad news earnings to later quarters, but fail to do so using real activities, that they would still implement accrual-based manipulation to delay such news. Therefore, the expectancy is that managers are still most likely to disclose lower than expected earnings in the fourth quarter. Considering that the market has that information, it is expected to lead to a greater stock price reaction when lower than expected earnings are presented in the earlier quarters, compared to such a disclosure

14

in the fourth quarter. In that regard, our hypothesis follows the notions from Mendenhall and Nichols (1988), Salamon and Stober (1994) and Das and Shroff (2002) with the expectancy that the market reaction to bad news earnings announcements will be lower in the fourth quarter compared to the earlier quarters.

Hypothesis 2:

> ℎ

2.6 Measuring the difference in stock price reaction

The methodologies used to measure the difference in market reactions between different quarters, differs in the studies presented in this literature review. Mendenhall and Nichols (1988) use a regression that solely focusses on the difference in price reaction to bad news earnings announcements. Salamon and Stober (1994) use the same type of regression but add variables to study the implications for peak-sales- and non-peak-sales quarters. Lastly, Das and Shroff (2002), use an even more evolved regression were earnings reversals in the fourth quarter and the differences between negative and positive earnings are also incorporated. The different methods will be discussed and analyzed in this section.

Mendenhall and Nichols (1988) regression uses the cumulative abnormal returns as the dependent variable and uses multiple binary variables combined with the unexpected earnings as the explanatory variables. The binary variables are used to distinguish what type of earnings announcement is explained, a fourth-quarter lower than expected earnings announcement, an earlier quarter lower than expected earnings announcement and a higher than expected earnings announcement, where a lower than expected fourth-quarter announcement is the base case (Mendenhall and Nichols, 1988). As the market reactions to higher than expected earnings announcements are not part of the hypotheses, they are only added in the regression to control for potential biases (Mendenhall and Nichols, 1988). This regression thus solely focuses on the relation between the unexpected earnings and the height of the abnormal returns and how those differ between the earlier- and fourth quarters.

Both Salamon and Stober (1994) and Das and Shroff (2002) use the same setup as Mendenhall and Nichols (1988), with Abnormal returns as the dependent variable and unexpected earnings and binary variables as the independent variables, however, both add extra explanatory variables to also study other relations. Salamon and Stober (1994) add variables to control for peak-sales quarters and as a result, explain how those characteristics have an influence on the abnormal returns. Das and Shroff (2002) add multiple binary variables to distinguish between positive and negative earnings and to see how the market reacts to earnings reversals in the fourth quarter.

15

It thus seems that Mendenhall and Nichols (1988) laid out a model which focuses solely on the relation between lower than expected earnings and the abnormal returns and the differences between earlier- and fourth quarters, which can be enhanced to study other explanations of variances in abnormal returns over different quarters (e.g. positive unexpected returns and the effect of peak-sales quarters) and has been used to do so more recently (Salamon and Stober, 1994; Das and Shroff, 2002).

2.7 Hypotheses

Combining all reviewed literature, it is possible to suggest that managers indeed are more likely to present lower than expected earnings in the fourth quarter compared to earlier quarters. That, because of the tendencies of managers to delay bad news, use the fourth quarter earnings to even out their full fiscal-year earnings and use both real- and accrual-based earnings management to reach their desired level of earnings manipulation. Also, following previous literature, the expectancy is that stock prices will react to earnings disclosures, which should result in abnormal returns. When considering the sign of the unexpected earnings, the expectancy is that the sign of the abnormal returns and the sign of the unexpected earnings will match. Thus, the following hypotheses are formed regarding the existence of abnormal returns surrounding quarterly earnings disclosures:

Hypothesis 1A: ℎ > 0 > 0

Hypothesis 1B: ℎ < 0 < 0

Continuing, considering the habits of managers, regardless of the implementation of IFRS, the expectancy is that bad news earnings are still most likely to occur in the fourth quarter and are thus less expected in the earlier quarters. Which, is expected to result in a heavier (more negative) market reaction when bad news earnings are disclosed in the earlier quarters, compared to the same disclosure in the fourth quarters. This results in the following hypothesis:

Hypothesis 2: > ℎ

As this research focuses solely on bad news earnings announcements, a symmetric hypothesis for good news earnings disclosures is not formed. This would also be illogical, as the hypothesis is partially based on the expectancy that managers tend to delay bad news, such behavior is not expected for good news and proven to not exist (Kothari et al., 2009).

3. Sample construction and methodology

This section contains a description on how the data is constructed and on how the hypotheses will be tested. First, the sample set-up will be discussed, including all variables and sources.

16

Afterwards, the choice for certain measures and the tests for all hypotheses will be explained. All tests noted in the methodology section will be completed using the STATA software.

3.1 Sample construction

The sample consists of 55 companies incorporated in the EU, with quarterly earnings disclosures from 2006 until 2016. The year when IFRS was implemented, 2005, has been excluded from the sample. This to control for the possible biases of the transition year (Ipino and Parbonetti, 2017). The needed data is collected from the IBES and Compustat – Capital

IQ databases and the event studies are done through the WRDS event study tool, which

calculates the cumulative abnormal returns surrounding the earnings disclosures. An overview of the variables and sources is shown in table 3. For a quarterly earnings disclosure to be included in the sample, all variables needed to be available in the databases, this resulted in

n = 1905 observations.

Table 3 Variables and sources

Variables Source

Quarterly earnings announcement dates I/B/E/S database

Actual EPS I/B/E/S database

Forecasted EPS (consensus by analysts) I/B/E/S database

Share price Compustat – Global database

Cumulative abnormal returns Event study by WRDS tool – with data from the Compustat - Global database

3.2 Methodology

This section will elaborate on the exact methodologies chosen to test the hypotheses formed in the literature review. All hypotheses, including appropriate methodology, are summarized in table 4.

17

Table 4 Hypotheses

Hypothesis: Method: Tested by:

Hypothesis 1a:

!,# > 0 > 0

Calculated using the Market-adjusted model

One-sided t-test

Hypothesis 1a:

!,# < 0 < 0

Calculated using the Market-adjusted model One-sided t-test Hypothesis 2: > ℎ Calculated using an ordinary least squared regression Testing the _$

coefficient

with a

one-sided t-test

3.2.1 Hypotheses 1a and 1b

For the first research question, the hypotheses are formed using cumulative abnormal returns, which are the sums of the abnormal returns surrounding a quarterly earnings disclosure. The abnormal returns will be calculated from three days before- until three days after the disclosure (t = -3 until t = 3), using the market adjusted model (De Jong, 2007). Where the market returns are the country-specific indexes matched to the company (e.g. AEX, DAX). The choice to use the seven days surrounding the earnings disclosure is made, to include the full information content of the announcement (Bernard and Thomas, 1990). This includes the information that managers leak before announcements (Kothari et al., 2009) and the full information processed after the announcement (Bernard and Thomas, 1990). The variables are defined as follows:

!,#

=

−

ℎ

=

_*

*

− 1

= .



=



*

=

/



=

0

/

18



= ℎ

0

/

ℎ

0

 _!,#

= ℎ

ℎ

To form a conclusion, the existence and sign of the cumulative abnormal returns will be tested. The sample will be split up into two parts: unexpected earnings higher than zero and unexpected earnings below zero, both will be tested using a one-sided t-test. The following assumptions are made to be able to use the t-tests: the abnormal returns are cross-sectionally independent and the abnormal returns are approximately normally distributed (Central limit theorem) (De Jong, 2007). The following hypotheses will be tested using a one-sided t-test:

H

:

> 0 =

< 0 = 0

H

:

> 0 > 0

H

:

< 0 < 0

For all above-stated cases the t-test is defined as (De Jong, 2007):

1223 = √5 ≈ 5 0, Where:



=

∑



= ℎ

ℎ

3.2.2 Hypothesis 2

To answer the main research question and find whether there is a difference in market price reaction to bad news in earlier quarters compared to fourth quarters, a regression will be used. First, however, the unexpected earnings (UQE) component needs to be defined:

9:

=

:

_?

− Ê



9:

= ℎ

0

ℎ



:

= ℎ

@

0

ℎ



Ê

= ℎ

ℎ

0

0



?

=

: ℎ

ℎ

ℎ

19

In line with more recent research, only the analysts’ forecasts will be used as expected earnings (Affleck-Graves et al., 2002; Lopez and Rees, 2002; Francis et al., 2002; Brown et al., 2009). For the deflator, the methodology from the articles, who also study the difference in stock price reactions over different quarters, will be followed and the share price at announcement date will be used to standardize the unexpected earnings (Mendenhall and Nichols, 1988; Salamon and Stober, 1994; Das and Shroff, 2002). However, the choice for this deflator should not influence the outcomes significantly, as concluded by Mendenhall and Nichols (1988) and Gu and Wu (2003).

The regression that will be used is based on the original methodology from Mendenhall and Nichols (1988). This choice is based on the fact that later articles use that same regression, only with small enhancements made to the model (Salamon and Stober, 1994; Das and Shroff, 2002). Those enhancements were only made to study other influences on the differences in abnormal returns (e.g. peak sales season, the reversals of earnings), however the conclusions about the difference in earlier- and fourth quarter disclosures came from the same base regression as that of Mendenhall and Nichols (1988) (Salamon and Stober, 1994; Das and Shroff, 2002). As this thesis only focuses on the difference in market reactions between lower than expected earnings in the earlier- and fourth quarters it seems most appropriate to remove the enhancements used for other links and again use the base regression.

The ordinary least squared (OLS) regression used, based on the methodology from Mendenhall and Nichols (1988), is defined as follows, including binary variables:

;#

= B +

D 9:

E +

D

9:

E +

D

9:

E +

Where the binary variables are:



= 1:

=



ℎ

9:

< 0, ℎ

0



= 1: 9:

> 0, ℎ

0

This regression works as follows: _, shows the relation between the height of a fourth quarter bad news earnings announcement and the height of the cumulative abnormal returns, this is the “base-case”, _$ shows the difference in the height of the cumulative abnormal returns between fourth quarter- and earlier quarter bad news earnings announcements, and _F shows the difference between fourth-quarter bad news- and all good news earnings announcements. The latter is only added to control for potential biases, which could be the case if only negative unexpected earnings were included in the regression (Mendenhall and Nichols, 1988).

The coefficient _$

will show the difference in stock price reaction between fourth-quarter

and earlier quarter bad news earnings announcements. Thus, it is this variable that will

20

answer the main research question. As the hypothesis is that the stock price reaction

to earlier quarter bad news earnings disclosures will be greater than for fourth quarter

bad news earnings disclosures, the expectancy is that

will be positive and

significantly different from zero. Therefore hypothesis 2 will be rewritten as follows:

H

:

= 0

H

:

> 0

The coefficient of _$ will be tested with a one-sided t-test to see if it is significantly different from zero.

Following the guidelines from MacKinlay (1997), there is no reason to expect the residuals from the regression to be homoscedastic, so heteroskedasticity is assumed. Therefore, when testing to parameters of the regression, robust standard errors will be used to correct for serial-correlation and heteroskedasticity.

4. Data analysis and discussion

In this section, the results from the tests described in the methodology section will be presented and discussed. First, the descriptive statistics will be presented. Thereafter, all the hypotheses will be tested, and the results will be presented and analyzed. Following, a robustness check will be done with a modified sample. Lastly, the results will be discussed and interpreted.

4.1 Descriptive statistics

To give an insight into the sample, the descriptive statistics are presented in table 5. For this research, it is interesting to see that in the fourth quarter a 58% of the disclosed earnings do not meet the forecasts, while this number is 47% in the first three quarters. This could suggest that managers indeed withhold bad news. However, other factors could also play a role in these findings.

21

Table 5 Descriptive statistics

Observations Unexpected earnings Mean cumulative

Abnormal returns

Total events n = 1905 Positive UQE n(UQE > 0) =

951 Full sample (SD) 0.0012 (0.0482) Non-fourth quarter events =G,,$,F = 1432 47% UQE < 0

Negative UQE n(UQE < 0) = 954 UQE > 0 (SD) 0.0058 (0.0448) Fourth-quarter =GH = 473 58% UQE < 0 Average UQE (standardized) -0.0732 UQE < 0 (SD) -0.0034 (0.0510) Standard Deviation (SD) 0.303768

4.2 Existence and sign of the cumulative abnormal returns

For both hypotheses 1a and 1b (see table 4) a one-sided t-test is used to test whether cumulative abnormal returns are present after quarterly earnings announcements and if the sign of the cumulative abnormal returns matches that of the unexpected returns. The results are presented in table 6.1 and 6.2 respectively.

Table 6.1 Hypothesis 1a: One sided t-test CAR(UQE > 0) (n = 951)

t-value p-value Cumulative Abnormal Returns M SD 0.0058 (0.0448) 4.0041 0.0000**

Note: Result with *** is rejected at 1% significance level

Table 6.2 Hypothesis 1b: One sided t-test CAR(UQE < 0) (n = 954)

t-value p-value Cumulative Abnormal Returns M SD -0.0034 (0.0510) -2.0428 0.0207*

22

From both table 6.1 and 6.2, it is clear that abnormal returns are present surrounding quarterly earnings disclosures and that the cumulative abnormal returns do match the sign of the unexpected earnings. As both tests are rejected at a 5% significance level and the sample with positive unexpected earnings is even rejected at a 1% significance level. Therefore, both hypotheses 1a and 1b are proven to be correct. Also, it is interesting to see that mean abnormal returns for good news earnings (0.0058) are show to be higher than the absolute value of the mean abnormal returns of bad news earnings (0.0034). This is contrary to the believe of Brown et al. (2009) and Kothari et al. (2009), who concluded, that the stock price reaction to earnings announcements is greater when negative unexpected earnings are presented by US companies. This was not a formal hypothesis, however it could be a sign of investors trust in the IFRS accounting standards compared to US GAAP.

4.3 Difference in stock price reaction to bad news earnings

With the existence of stock price reactions proven, the cumulative abnormal returns can be used to continue the analysis. To answer the final research question an OLS regression is used to see whether there is a difference between the height of cumulative abnormal returns after a lower than expected earnings announcement in the first three quarters compared to such a disclosure in the fourth quarter. The results are presented in table 7.1.

Table 7.1 Hypothesis 2

Regression: _;#= B + _,D 9:_;<#E + _$D _, 9:_;<#E + _FD _$ 9:_;<#E + _;#

IJKLM Coefficient Robust

Standard Error t P , 0.0163 0.0070 2.33 0.020** $ 0.0021 0.0087 0.24 0.808 . F -0.0167 0.0094 -1.77 0.076* Constant (B) 0.0019 0.0011 1.67 0.094*

Note: Results with * are rejected at 10%- and with ** at a 5% significance level

As stated in the methodology, the binary variables in the regression are defined as follows: _, is 1 if 9:_;<#< 0 and the event is a non-fourth quarter disclosure, otherwise 0 and _$ is 1 if

9:;<# > 0 and otherwise 0. The results should be interpreted as follows: the , coefficient shows a positive relation between fourth quarter unexpected earnings and the cumulative abnormal returns, which is significantly different from zero at a 5% significance level. This is to be expected, considering the results from the abnormal return tests, where is proven that the abnormal returns match the sign of the unexpected earnings. Secondly, _$ is the coefficient that shows the difference in abnormal returns, per unit of unexpected earnings, between fourth-

23

and earlier quarter announcements. The coefficient is calculated at 0.0021, which is not significantly different from zero. It is the _$ variable that is the focus of this research, as it is not significantly different from zero, it means that the null hypothesis cannot be rejected based on this test. _F reflects the difference in abnormal returns between fourth quarter bad news- and all good news earnings announcements. The value is negative and rejected at a 10% significance level. A negative value might seem illogical, as the abnormal returns after good news earnings announcements are expected to be higher than after bad news earnings announcements. However, when considering that _, reflects the positive relation between fourth quarter bad news earnings announcements and abnormal returns (the lower UQE goes, the lower CAR becomes), _F can be interpreted as the correction when unexpected earnings are above zero. Lastly, the constant shows a positive coefficient which is significant at a 10% level. Table 7.2 Hypothesis 2 Regression: _;#= B + _,D 9:_;<#E + _$D _, 9:_;<#E + _FD _$ 9:_;<#E + _;# IJKLM F(3, 1901) R-squared n = 1905 6.01 p-value = 0.0004*** 0.0071

Note: Results with *** are rejected at 1% significance level

As evident from table 7.2 the F-statistic is rejected at a 1% significance level, so the joint slope coefficients are not equal to zero, which means the model can be used. The table also states that the R-squared value is very low. However, as it is not the objective to explain the dependent variable with this regression, it should not matter for our analysis that the R-squared value is that low.

Based on the results from the regression (table 7.1), there is no evidence that supports the hypothesis that there is a difference in stock price price reaction between bad news earnings in earlier- and fourth quarters, were a heavier reaction is expected in the non-fourth quarters. To make sure these results are not due to the used sample, a robustness check is being presented in section 4.4 below.

4.4 Robustness check

For a robustness check, the used sample will be modified to check whether the results are not due to specific company characteristics. As is suggested in previous research, financial- and utility (e.g. gas and electricity) companies are less likely to apply earnings management, as they face heavier regulation (Van Tendeloo and Vanstraelen, 2005; Roychowdhury, 2006; Kothari et al., 2009; Ipino and Parbonetti, 2017; Lo et al., 2017). This could influence the

24

results, as earnings management is the main driver behind the hypothesis that lower than expected earnings announcements are more likely in the fourth quarter, what results in a heavier reaction when bad news earnings are presented in the earlier quarters.

To check if the presence of financial- and utility companies influence the results founded in section 4.3, the sample is modified and all financial- and utility companies are omitted. This resulted in a sample of 30 companies and n = 1153 observed quarterly earnings announcements. This sample will be used to test hypothesis 2 again, the results are presented in table 8.1 and 8.2 below.

Table 8.1 Hypothesis 2 with financial- and utility companies excluded

Regression: _;#= B + _,D 9:_;<#E + _$D _, 9:_;<#E + _FD _$ 9:_;<#E + _;#

IJKLM Coefficient Robust

Standard Error t P , 0.0106 0.0168 0.63 0.530 $ 0.0187 0.0268 0.70 0.485 F 0.0456 0.0214 2.13 0.033** Constant (

B)

Note: Result with ** is rejected at a 5% significance level

Table 8.2 Hypothesis 2 with financial- and utility companies excluded

Regression: _;#= B + _,D 9:_;<#E + _$D _, 9:_;<#E + _FD _$ 9:_;<#E + _;#

IJKLM F(3, 1149) R-squared

n = 1153

6.64

p-value = 0.0002*** 0.0123

Note: Results with *** are rejected at 1% significance level

The results from the robustness test do show some differences from table 7.1. As table 8.1 shows that the constant and _, are not significantly different from zero anymore. Also, _F changed from a negative value in table 7.1 to a positive value in table 8.1, this implies that the difference in stock price reaction between fourth quarter bad news earnings announcements and all good news earnings announcements is larger for non-financial- and non-utility companies. However, as this is not the focus of this research, no hypothesis was formed on that issue.

The main variable for hypothesis 2, _$, did increase a little compared to the original sample. As the value was 0.0021 in the original test and 0.0187 for the robustness test. However, as the coefficient from table 8.1 is still not significantly higher than zero, it cannot be concluded

25

that the lack of difference in market reaction between earlier- and fourth quarter bad news earnings announcement is due to the inclusion of utility- and financial companies. Therefore, this robustness test is taken as a confirmation of the original results, which concludes that there is no evidence that supports the existence of a difference in stock price reaction between earlier- and fourth quarter bad news earnings announcements.

4.5 Discussion

The results for all hypotheses are summarized in table 9 below.

Table 9 Hypotheses and results

Hypothesis: Method: Tested

by: Results Hypothesis 1a: !,# > 0 > 0 Calculated using the Market-adjusted model One-sided t-test Hypothesis significantly proven at 1% level Hypothesis 1a: !,# < 0 < 0 Calculated using the Market-adjusted model One-sided t-test Hypothesis significantly proven at 5% level Hypothesis 2: $> 0 ( > ℎ ) Calculated using OLS regression One-sided t-test Hypothesis not proven in regular- and robustness tests

The results of hypotheses 1a and 1b are not surprising, as they are both intuitive and in line with previous research. However, the conclusion of hypothesis 2 is surprising and does provide new insights, as the results do not lead to the expected results and are not in line with previous research. As there is no proof of a difference in market reaction between earlier- and fourth quarter lower than expected earnings, it would mean that the market does not believe that bad

26

news earnings announcements are more likely in the fourth quarter. There could be many explanations for this outcome, a few will be presented below.

The first explanation could be that managers do not want to withhold bad news until the fourth quarter. That explanation would contradict the conclusions from Chambers and Penman (1984) and Kothari et al. (2009), where is concluded that managers always have the tendency to delay bad news. So, if managers could delay earnings until the fourth quarter, it is expected that they would. Therefore, this explanation does not seem likely.

A second explanation would be that the conclusion is a result of managers increased focus on real earnings management, where they might not have the ability to delay earnings until the fourth quarter. The use of real earnings management could have an impact. However, it is also noted by Zang (2011), that managers adjust their level of accrual-based earnings manipulation after they have analyzed the effectiveness of their real earnings manipulation. This would suggest that if they wanted to delay bad news until the fourth quarter and they could by adjusting their level of accrual-based earnings management, they would happily do so. This brings up the last explanation discussed in this thesis. That is, that managers do not have the ability to delay lower than expected earnings until the fourth quarter within the EU under the IFRS accounting standards. The market is aware of that notion and therefore does not expect a higher frequency of lower than expected earnings announcements in fourth quarters. As the existence of earnings management in the EU is proven, it could imply that IFRS smoothened the use of earnings management during the year, which could be an extra explanation for the positive market reaction to its implementation.

So, from the results, it is evident that there is no proof of a difference in market reaction to lower than expected earnings announcements between earlier- and fourth quarters. When these empirical results are combined with the analyzed literature, the most likely explanation for the inexistence of a difference is due to managers not being able to delay earnings to the fourth quarter under the IFRS standards. However, there could be other factors that influence the findings, therefore suggestions for further research are made in section 5.

5. Conclusion

5.1 Summary and conclusion

In this research, the difference in stock price reactions between earlier- and fourth quarter bad news earnings announcements have been analyzed for European companies. Earlier research concluded that such a difference exists for US-based companies. More specifically, that the stock price reaction following an earlier quarter bad news earnings announcement was

27

heavier, than following such an announcement in the fourth quarter (Mendenhall and Nichols, 1988; Salamon and Stober, 1994; Das and Shroff, 2002). Therefore, it was the main focus of this thesis to study if those implications also hold for companies in the EU after the adoption of IFRS.

First, stock price reactions after quarterly earnings disclosures were proven to exist, through the presence of cumulative abnormal returns surrounding the disclosures. It was also proven that the signs of the cumulative abnormal returns match the signs of the unexpected earnings presented in the quarterly disclosures, as both one-sided t-tests were rejected at a 5% significance level.

As the cumulative abnormal returns were proven to exist, they were used to analyze if there is a difference in stock price reactions between earlier- and fourth quarter bad news earnings disclosures. Both the regular regression and the robustness test, with financial- and utility companies omitted from the sample, did not show a significant difference between the earlier- and fourth quarter reactions. Therefore, it is the conclusion that the hypothesis, that earlier quarter bad news earnings announcements would prompt a heavier stock price reaction than fourth quarter bad news earnings announcements for EU companies, is rejected as no proof of such a difference is found.

As no difference in the reaction between earlier- and fourth quarter bad news earnings is found, it can be concluded that IFRS plays an instrumental role in reducing the difference in earnings management between earlier- and fourth quarters and therefore improves the information environment within the EU.

5.2 Suggested research

Further research should study if other conclusions made, about differences in quarterly earnings between earlier- and fourth quarters under the US GAAP rules, also do not hold for companies that fall under the IFRS accounting standards. As proven in this research, it is evident that the implications of US GAAP and IFRS differ when it comes to the ability of managers to manipulate earnings differently over the different quarters. As a result, it is also the question if the conclusion that managers reverse their earnings in the fourth quarter also holds for European companies.

Secondly, it would be interesting to test whether the conclusions from this research differs before- and after the adoption of IFRS and whether the hypothesis from this thesis does hold for countries where IFRS is adopted but who do not have as strict institutional enforcement as the EU. As a result, it would become clearer if it is the IFRS accounting standards, the institutional enforcement or other characteristics of the countries in the EU, that prohibits

28

managers from manipulating their earnings differently between earlier and fourth quarters. Lastly, it would be interesting to zoom in on how real earnings management is distributed over a company’s fiscal-year and how that affects the market expectancies.

29

Bibliography

Affleck‐Graves, J., Callahan, C. M., & Chipalkatti, N. (2002). Earnings predictability,

information asymmetry, and market liquidity. Journal of Accounting Research, 40(3), 561-583.

Armstrong, C. S., Barth, M. E., Jagolinzer, A. D., & Riedl, E. J. (2010). Market reaction to the

adoption of IFRS in Europe. The accounting review, 85(1), 31-61.

Bartov, E., Givoly, D., & Hayn, C. (2002). The rewards to meeting or beating earnings

expectations. Journal of accounting and economics, 33(2), 173-204.

Beaver, W. H. (1968). The information content of annual earnings announcements. Journal

of accounting research, 67-92.

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), 305-340.

Brown, S., Hillegeist, S. A., & Lo, K. (2009). The effect of earnings surprises on information

asymmetry. Journal of Accounting and Economics, 47(3), 208-225.

Chambers, A. E., & Penman, S. H. (1984). Timeliness of reporting and the stock price

reaction to earnings announcements. Journal of accounting research, 21-47.

Das, S., Shroff, P.K., 2002. Fourth quarter reversals in earnings changes and earnings management. Working paper, University of Minnesota.

De Jong, F. (2007). Event studies methodology. Lecture Notes.

Fama. E. F. (1965). The behavior of Stock-Market Prices. The Journal of Business, 38(1), 43-105.

Fama, E. F., Fisher, L., Jensen, M. C., & Roll, R. (1969). The adjustment of stock prices to

new information. International economic review, 10(1), 1-21.

Fama. E. F. (1970). Efficient Capital Markets: A Review of Theory and Empirical Work. The

Journal of Finance, 25(2), 383-417.

Fama. E. F. (1991). Efficient Capital Markets: II. The Journal of Finance, 46(5), 1575-1617. Francis, J., Schipper, K., & Vincent, L. (2002). Expanded disclosures and the increased usefulness of earnings announcements. The Accounting Review, 77(3), 515-546. Gu, Z., & Wu, J. S. (2003). Earnings skewness and analyst forecast bias. Journal of

Accounting and Economics, 35(1), 5-29.

Ipino, E., & Parbonetti, A. (2017). Mandatory IFRS adoption: the trade-off between accrual

based and real earnings management. Accounting and Business Research, 47(1), 91

121.

Jacob, J., & Jorgensen, B. N. (2007). Earnings management and accounting income

aggregation. Journal of Accounting and Economics, 43(2-3), 369-390.

Jeanjean, T., & Stolowy, H. (2008). Do accounting standards matter? An exploratory analysis

of earnings management before and after IFRS adoption. Journal of accounting and

30

Jiang, J. X., Petroni, K. R., & Wang, I. Y. (2010). CFOs and CEOs: Who have the most

influence on earnings management?. Journal of Financial Economics, 96(3),

513-526.

Kothari, S. P., Shu, S., & Wysocki, P. D. (2009). Do managers withhold bad news?. Journal

of Accounting Research, 47(1), 241-276.

Landsman, W. R., Maydew, E. L., & Thornock, J. R. (2012). The information content of

annual earnings announcements and mandatory adoption of IFRS. Journal of

Accounting and Economics, 53(1-2), 34-54.

Lo, K., Ramos, F., & Rogo, R. (2017). Earnings management and annual report

readability. Journal of Accounting and Economics, 63(1), 1-25.

Lopez, T. J., & Rees, L. (2002). The effect of beating and missing analysts' forecasts on the information content of unexpected earnings. Journal of Accounting, Auditing &

Finance, 17(2), 155-184.

MacKinlay, A. C. (1997). Event studies in economics and finance. Journal of economic

literature, 35(1), 13-39.

Mendenhall, R., & Nichols, W. (1988). Bad News and Differential Market Reactions to

Announcements of Earlier-Quarters Versus Fourth-Quarter Earnings. Journal of

Accounting Research, 26, 63-86

O'brien, P. C. (1988). Analysts' forecasts as earnings expectations. Journal of accounting

and Economics, 10(1), 53-83.

Roychowdhury, S. (2006). Earnings management through real activities

manipulation. Journal of accounting and economics, 42(3), 335-370.

Salamon, G. L., & Stober, T. L. (1994). Cross‐Quarter Differences in Stock Price Responses

to Earnings Announcements: Fourth‐Quarter and Seasonality

Influences. Contemporary Accounting Research, 11(1), 297-330.

Van Tendeloo, B., & Vanstraelen, A. (2005). Earnings management under German GAAP

versus IFRS. European Accounting Review, 14(1), 155-180.

Zang, A. Y. (2011). Evidence on the trade-off between real activities manipulation and

31

Appendix

Companies included in the sample: ENEL ASSI GENERALI ALLIANZ INTESA SANPAOLO BARCLAYS PLC BAYER AG BANCO SANTANDER BMW BNP PARIBAS UNICREDITO ITAL. DAIMLER DEUTSCHE BANK DEUTSCHE TELEKOM ENI DEUTSCHE POST MUNICH RE GROUP OMV AG

ROYAL BANK OF SCOTLAND REPSOL

RWE

ROYAL DUTCH SHELL SIEMENS SOCIETE GENERALE SANOFI-AVENTIS TEF THYSSEN KRUPP TOTAL

LLOYDS BANKING GROUP UNILEVER BANCO BILBAO E. ON VOLKSWAGEN AIR FRANCE KLM ANGLO AMERICAN POLSKI KONSERN AEGON BRITISH AMERICA BOUYGUES BT GROUP GAS NATURAL MAPFRE COMMERZBANK CONTINENTAL DEXIA ENBW ENERGIE FRESENIUS VIVENDI GLAXOSMITHKLINE IBERDROLA LUFTHANSA NORDEA BANK TELECOM ITALIA PHILIPS RANDSTAD HOLDING UNIPOL GRUPPO