Intra-industry Information Transfers - Profit Warnings

Page |1

Table of Contents List of Tables.............................................................................................................................2 List of Figures...........................................................................................................................2 Chapter 1: Introduction...........................................................................................................3 Chapter 2: Literature Review.................................................................................................6 2.1. Information transfers.......................................................................................................7 2.2. Profit Warnings................................................................................................................8 Chapter 3: Hypotheses Development...................................................................................13 3.1. Market Reaction to Profit Warnings..............................................................................13 3.2. Factors That May Influence the Effects of Profit Warnings on the Industry.................14 Chapter 4: Sample Data and Methodology..........................................................................16 4.1. Sample Data...................................................................................................................16 4.2. Event Study ..................................................................................................................17 4.3. Cross-Sectional Regression Analysis............................................................................19 Chapter 5: Descriptive Statistics...........................................................................................21 5.1. Characteristics of Profit Warnings.................................................................................21 Chapter 6: Results..................................................................................................................29 6.1. Event Study Results – Full Sample...............................................................................29 6.2. Event Study Results – Industry-Wide Sub Sample.......................................................34 6.3. Event Study Results – Firm-specific Sub Sample.........................................................38 6.4. Multivariate Cross-Sectional Analysis – Full Sample...................................................42 6.5. Multivariate Cross-Sectional Analysis – Industry Wide Sub-Sample...........................44 6.6. Multivariate Cross-Sectional Analysis – Firm Specific Sub-Sample............................46 Chapter 7: Conclusion...........................................................................................................47 7.1. Conclusions and Implications........................................................................................47 7.2. Limitations of the Study................................................................................................51 7.3. Suggestions for Further Research..................................................................................52 Bibliography...........................................................................................................................54 Appendix A: Profit Warning Classification.......................................................................A.1 A.1. Characteristics of Industry-Wide and Firm Specific Profit Warnings........................A.1 A.2. Examples of Profit Warnings and their Classification................................................A.1 Appendix B: Distribution of Average Abnormal Returns................................................B.1

List of Tables Table 1: Profit Warnings segmented by source of warning......................................................21 Table 2: Frequency Distribution of industries for announcing and non-announcing firms.....23

Page |1

Table 3: Summary Statistics and Pearson correlations - Full Sample......................................26 Table 4: Summary Statistics and Pearson Correlations - Industry Wide Sub-sample..............26 Table 5: Summary Statistics and Pearson Correlations – Firm Specific Sub-sample..............26 Table 6: Test of Significance between ACAR of Announcing and Non-Announcing Firms...27 Table 7: Accounting Ratios for Companies making Profit Warnings......................................28 Table 8: Effects on Returns in Response to Announcements - Full Sample............................30 Table 9: Effects on Returns in Response to Announcements – Industry-Wide Sub Sample. . .34 Table 10: Effects on Returns in Response to Announcements – Firm-Specific Sub Sample. .38 Table 11: Cross-Sectional Multivariate Model Results - Full Sample.....................................42 Table 12: Multivariate Cross-Sectional Analysis - Industry Wide Sub-Sample Results..........44 Table 13: Multivariate Cross-Sectional Analysis – Firm Specific Sub-Sample - Results.......46

List of Figures Figure 1: AAR and ACAR for Full Sample of Announcing Firms..........................................32 Figure 2: AAR and ACAR for Industry-Wide Sub Sample of Announcing Firms..................36 Figure 3: AAR and ACAR for Industry-Wide Sub Sample of Announcing Firms..................37 Figure 4: AAR and ACAR for Firm-Specific Sub Sample of Announcing Firms...................40 Figure 5: AAR for Firm-Specific Sub Sample of Non-Announcing Firms.............................41

Page |1

Page |3

Chapter 1: Introduction Perhaps the simplest way of rephrasing the heading of this study is as follows: Is there a difference between the share price behaviour of non-announcing firms in response to firm specific and industry-wide announcements as pertaining to profit warnings. This document is a study of intra-industry information transfer between the returns of firms releasing information, through profit warnings, and those of non-announcing firms in the same industry. Many intra-industry information transfer studies have been conducted in the past but in different contexts such as; mergers, stock repurchases, dividend omissions and bankruptcy announcements. There have also been prior studies which studied the market reaction to profit warnings in the announcing firm, however little research has been done in the area of industry wide information transfers due to profit warnings and thus this study intends to increase our knowledge in this area. This study aims to evaluate the impact of a profit warning on the announcing company’s own share price behaviour and the share price behaviour of other companies in the same industry. Jin () notes that a significant shortcoming of previous studies is that they failed to distinguish between firm specific factors and industry-wide factors; as a result this study treats industry factors (profit warnings) and firm specific factors (profit warnings) separately in an attempt to accurately describe the intra-industry information transfer and hence derive proper meaningful conclusions. The results of this study suggest that announcements that convey industry wide information cause the announcing company and the non-announcing companies in the industry significant negative cumulative abnormal returns. This observation implies that the information conveyed in the warning was new information to the market, for both the announcing company and the non-announcing companies, and thus caused the abnormal movement of the share price. Our research also showed that profit warnings for companies conveying company specific information caused negative abnormal returns for the announcing company; however it did not have the same negative abnormal effect for non-announcing companies in the same industry. This indicates that the market only gains new information about the announcing company from the company specific profit warning as shown by the insignificant abnormal returns for the non-announcing companies, as expected. The cross-sectional regression of the full sample demonstrates the intra-industry effect of profit warnings since it shows that the size of the effect of the profit warning on the announcing company affects the size of the abnormal returns for non-announcing companies

Page |3

in the industry. The larger the market reaction for the announcing companies the lager the market reaction for non-announcing companies. We also found that in the full sample the intra-industry effect is more prevalent depending on the Earnings to Price ratio for the announcing company; the higher the EPR the more negative the cumulative abnormal returns in non-announcing companies. The regression of the industry wide information sub-sample shows that the intra-industry effect is influenced by not only the size of the market response on the share price and the market capitalisation of the announcing firm but also the homogeneity of the services/ product offered of the industry. With respect to the market reaction, the greater the market response on the announcing company’s share price, the greater the affect on the nonannouncing companies’ share price. With respect to the size of market capitalisation, the greater the size of the announcing company the greater the abnormal returns for the nonannouncing companies. The homogeneity of the industry looks at how similar the services/ product offered is and it was found that, for the industry wide sub-sample, the more homogenous the industry the less the abnormal returns were found to be for the nonannouncing companies. This finding for the homogeneous industry group is against our expectations and suggests that homogeneous industries provide an environment conducive to competitive reactions. For the company specific profit warning sub-sample it was found that there was no real significant relationship between the variables we tested, except for the size of the announcing company, thus suggesting the lack of an intra-industry effect. This seems plausible since there is no real reason for abnormal returns of the non-announcing companies for problems directly related to another company only. The results did however show that the greater the size of the announcing company the less the cumulative abnormal returns on the non-announcing companies, this could be due to more information generally being available for large companies so less of a market response.

The following chapters of this study are structured as follows: Chapter 2 comprises the literature review, encompassing a summary of literature as pertaining to profit warnings and intra-industry information transfers. Chapter 3 deals with the hypothesis development, in this section our hypotheses are derived and defined. Chapter 4 relates to the methodology used in the data collection process as well as the hypothesis testing procedures. Chapter 5 contains the descriptive statistics followed by the detailed results of the study (in Chapter 6). Chapter 7 contains the detailed conclusions as well as recommendations for further studies.

Page |3

___________________________________________________________________________

Page |3

Page |6

Chapter 2: Literature Review Many studies have reported an association between the returns of firms releasing information and those of non-announcing firms in the same industry. This relation is known as intraindustry information transfer and has been documented in different contexts such as mergers, stock repurchases(), dividend omissions and initiations(), and bankruptcy announcements(), but little or no research has been done in relation to profit warnings. Several studies in the past have investigated intra-industry information transfer of earnings related information such as those by Jin () and Baginski (), but neither of these considered the effects of the announcements on announcing and non-announcing firms in the industry separately. Jackson and Madura () and Elayan, Meyer and Sun ()found that there is a strong negative response to profit warnings. They also noted that the share price of the announcing firm began to be affected about five days before the warning is issued and continued to decrease for up to five days after, with little or no overreaction to the announcement. Previous studies by Lang and Stulz() and Caton, Goh and Kohers ()also show that, on average, the information transfer preserves the implication of the news, that is; if a firm releases good (bad) news, the market perceives the non-announcing firms to have good (bad) news. Several studies show that there are considerable incentives for management to announce profit warnings in a timely manner despite the fact that they are considered as voluntary disclosures left to the management’s discretion. Skinner (), Lang and Lundholm() and Richardson, Teoh and Wysocki() all give different reasons why it is in the managements interests to disclose information to the public.

Page |6

Literature Review

P a g e | 12

2.1. Information transfers Intra-industry information transfer is described as the process whereby information conveyed to the market about one firm (the announcing firm) conveys value-relevant information about other non-announcing firms operating in that industry.1 Firth’s () study marked the beginning of the information transfer literature which grew out of Ball and Brown’s () results. Firth () finds that other information sources, besides conventional sources (such as interim reports) used to anticipate the firm’s annual earnings announcement included the financial results of closely competing firms. Firth () finds that investors used information contained in the announcement of financial results to re-evaluate the share prices not only of the company whose results are being announced, but also of closely competing companies in the same industry. Prior research has also focused on the information transfer associated with management earnings forecasts. Baginski () used earnings forecast data to test the hypothesis that the sign of the earnings of announcing firms conveyed value relevant information to shareholders of non-announcing firms. The result is consistent with that of Firth () in terms of earnings forecast data. Other studies have studied information transfers resulting from events other than earnings announcements and management forecasts. Caton, Goh and Kohers () find that a dividend omission announcement transmits unfavourable information across the announcing company's industry that affects cash flow expectations and ultimately stock prices. Other studies have found evidence of spill over effects associated with mergers, stock repurchases, and nuclear and chemical plant accidents. As stated earlier Jin () determines that a common shortcoming of earlier studies is that they failed to distinguish industry common factors from firm specific factors. Industry-wide factors are those that are common to the entire industry, and the effects of which will be felt among all firms in that industry, while firm specific factors are factors that affect only the one firm. Since it is reasonable to assume that industry common factors cause intra-industry information transfers, co-mingling industry factors with firm specific factors may lead to an inaccurate description of the intra-industry information transfer issue and hence improper conclusions about it. A signal about industry common factors should affect security prices of the announcing firm as well as other firms in the same industry. But a signal about firm specific factors (with no resultant competitive shift) should affect security prices of the announcing firm only. 1

Usually resulting in a material movement in the reference price of the non-announcing firmswithin the industry

Literature Review

P a g e | 12

2.2. Profit Warnings A profit warning is an announcement by a publically traded company in which the company advises that its earnings won't meet analyst expectations. Profit warnings form part of management’s voluntary financial disclosures, and as such they are not mandated by any regulatory body, and it is purely up to the discretion of management as whether to make profit warning announcements or not. 2.2.1. Management’s Incentives to Disclose information Disclosure in this context implies the act of releasing all relevant information pertaining to a company that may influence an investment decision. To make investing as fair as possible for everyone, companies must disclose both good and bad information. In the past, selective disclosure was a serious problem for investors because insiders would frequently take advantage of information for their own gain - at the expense of the general investing public. Disclosure of financial information with regards to a publically listed company will lead to a reduction in the asymmetric information between company insiders and its investors (or the general public). Asymmetric information is a situation in which one party in a transaction has more or superior information compared to another. This often happens in transactions where the seller knows more than the buyer; this could be a harmful situation because one party can take advantage of the other party’s lack of knowledge(). With increased advancements in technology, asymmetric information has been on the decline as a result of more and more people being able to easily access all types of information. While it may be clear as to why management might want to disclose ‘good news’ there is also benefit that can be obtained by disclosing ‘bad news’ to the public as well. Managers may damage their reputations if they consistently fail to disclose bad news in a timely and appropriate manner (). Failing to disclose bad news consistently will also have further negative consequences, analysts will become less likely to follow the firm, thus reducing liquidity and hence the firm’s stock price will suffer. Lang and Lundholm() find that increased voluntary disclosure lowers the cost of information acquisition for analysts and therefore increases their supply; This results in increased investor following, reduced information asymmetry and greater demand for a firm’s shares leading to a lower cost of capital, thus a net benefit for the firm despite the ‘bad news’. Another incentive or benefit that managers may derive from disclosure of bad news relates to an attempt by management to “walk down” financial analysts’ forecasts of

Literature Review

P a g e | 12

earnings to beatable targets. The argument is that managers release disclosures during the year that guide financial analysts to believe that end-of-period earnings will be lower than the actual announced earnings. This would generate good news on the official announcement date since actual earnings will be above the recent expectations (based on the profit warning statement) the result would be a positive share price reaction at the earnings announcement date. The intentional “walk down” of analysts’ forecasts may be justified by management’s expectations that the capital market punishment for reporting a negative surprise at announcement date is greater than the reward from reporting a positive surprise. Richardson, Teoh and Wysocki(), a U.S. study provides evidence that firms walk down analysts (it should perhaps be noted that no relevant studies were found for the South African market – but this does not imply that firms do not walk down analysts in South Africa). There is also a legal incentive for management to disclose information to the public in that profit warnings act to help mitigate the legal liability of managers. Shareholders may sue when there are consistent large stock price declines on earnings announcement days, since shareholders can allege that managers failed to disclose adverse earnings news promptly and appropriately(); and as a result they can claim that due to the manager’s failure to promptly disclose material bad news they bought overvalued stocks that devalued after management revealed this information.

Literature Review

P a g e | 12

The Johannesburg Stock Exchange (JSE) also obliges companies to keep investors informed in a timely manner about any material, price-relevant information2. The incentive behind such disclosure regulations is to encourage informed decision making by all parties to securities transactions and hence to decrease asymmetric information between insiders and outsiders. Compliance to the listing rules of the JSE is mandatory for all listed companies and failure to comply may result in a civil or criminal liability and may result in a number of sanctions including suspension of trading or de-listing of the company. It should be noted here that Profit warnings are considered as voluntary price sensitive announcements and as a result companies have no obligation to issue them, unless not doing so would breach the listing requirements (as given at the bottom of the previous page). It should also be noted that according to Skinner’s () findings and those of Collett () both the number of negative trading disclosures and their absolute impact is much higher than the number and impact of positive announcements in the United States. This could be explained by the fact that Shareholders are more likely to mount a class action if bad news is withheld than good news. Similarly, fund managers will look more favourably on directors who deliver company performance above expectation. Another incentive for disclosure is the existence of Employee Stock Option Plans (ESOPs). As Aboody and Kaznik () suggest, management makes opportunistic voluntary disclosure decisions that maximize their stock option compensation, this is done by disclosing bad news before the award of ESOPs in order to depress the share price and hence the option exercise price, which is set at the market price at the date of award.

2

JSE Listing Rule 3.4 – under General Obligation of Disclosure states that the following provisions apply in respect of material price sensitive information:

With the exception of trading statements, an issuer must, without delay, unless the information is kept confidential for a limited period of time in terms of paragraph 3.6, release an announcement providing details of any development(s) in such issuer’s sphere of activity that is/are not public knowledge and which may, by virtue of its/their effect(s), lead to material movements of the reference price of such issuer’s listed securities. With regards to Cautionary announcements Cautionary announcements listing rule 3.9 states that: Immediately after an issuer acquires knowledge of any material price sensitive information and the necessary degree of confidentiality of such information cannot be maintained or if the issuer suspects that confidentiality has or may have been breached, an issuer must publish a cautionary announcement (complying with paragraph 11.40). An issuer that has published a cautionary announcement must provide updates thereon in the required manner and within the time limits prescribed in paragraph 11.41.

Literature Review

P a g e | 12

2.2.2. Market Reaction to Profit Warnings Since profit warnings issued by firms generally indicate “bad news” in relation to the previously anticipated firm’s annual earnings it is logical that the market reaction to a profit warning announcement will be negative. Many studies such as the one by Jackson and Madura ()confirm this expectation.3 Jackson and Madura’s study also confirms the presence of information leakage upto five days before (t - 5) the profit warning announcement, then during the two-day announcement period (t – 1) and ( t ) there is a significant further decline in share price, followed by a further decline for upto seven days (t + 7) after the announcement. This implies that the effects of the profit announcement were not instantaneous and hence any study relating to the profit announcement must be conducted over the entire period of at least thirteen days for which the share price has been found to be affected by the announcement.4 Jackson and Madura() found no evidence of significant reversal of the share price after the four day post-announcemnt period5. 2.2.3. Profit warnings and Information Transfers Besides the fact that many studies confirm the relationship between profit warnings and negative revaluation of the announcing firm, it is also expected that investors use information contained in firm specific and industry specific information to revaluate competing companies in the same industry as the announcing firm. If a firm issues a profit warning whose underlying cause is firm specific, the effects of the warning may only be limited to the announcing firm. An alternative argument as posed by Lang and Schultz () is that the bad news announced by one firm may create opportunities for rival firms in the same industry6. A positive reaction in the share prices of non-announcing firms to a negative announcement by a firm in that industry is called 3

Jackson and Madura confirmed that profit warnings are associated with a negative revaluation of the firm. They found that in the US firms that issued profit warnings had an average decline in share price of 17.1% over a four-day pre-warning period and over the two-day announcement period).

Jackson and Madura found that the average abnormal returns were negative at the 0.1% level for days: (t – 3) and (t - 2) – In the pre warning period (t + 1) and (t + 4) – In the post warning period

4

Jackson and Madura’s study found that there was a a general underreaction on the announcement date, hence the further decline in share prices upto seven days after the announcement. They found that the Cummulative abnormal return from day (t + 11) to day (t + 60) was 1.49% and that it is statistically significant. Thus the unusually strong negative share price response to profit warnings does not appear to represent an overreaction.

5

Such competitive effects have been tested and detected for specific types of bankruptcy events by Lang and Stulz() and for specific types of dividend reductions by Laux, Starks, and Yoon().

6

Literature Review

P a g e | 12

the competitive effect. According to Lang and Stulz (), the competitive effect is the change in the value of competitors that can be attributable to wealth redistribution. Tawatnuntachai and D’Mello () found that the competitive effect is more pronounced in industries with imperfect competition, where the announcement of an event reveals comparative information about other firms in the industry.7 If on the other hand a firm issues a profit warning whose underlying cause is not only firm specific, but rather includes industry-wide factors the bad news effects may spill over into other firms in the industry.8 Tawatnuntachai and D’Mello () found that contagion and competitive effects are not mutually exclusive, and thus the observed stock price reaction is the sum of the two effects. They found that a significant negative change in non-announcing firms’ stock prices indicates that profit warning announcements convey negative industry-wide information. Studies have confirmed that profit warnings do result in negative industry effects – albeit to different degrees, but as Jin () notes no previous studies distinguished between industry-wide factors from firm specific factors. In Jin’s view the combining of these two types of factors may lead to an erroneous picture regarding intra-industry information transfer effects and hence any conclusions drawn from such studies would be flawed.

7

A competitive effect could, for example occur where there is a drop in production efficiency for the announcing firm resulting in higher marginal costs and hence higher prices or lower profit for that firm. In this case, demand for competitors’ products could increase because their products would be substitutes for the now more expensive products of the announcing firm.

Known as contagion

8

Literature Review

P a g e | 12

P a g e | 13

Chapter 3: Hypotheses Development The content and effects of profit warnings on the industry are dependent on the degree of asymmetric information within that industry, as well as to how firm specific events will affect the industry as a whole. To the extent that profit warnings provide valuable information regarding the industry, there are several issues that need to be examined. 3.1. Market Reaction to Profit Warnings When management issues a profit warning it is presumably doing so because it believes that the information is important in order to reduce asymmetric distribution of information between investors and insiders. Management does so because they believe that the information may lead to material movements of the reference price of the listed securities. A profit warning indicating that profits will fall short of previously expected levels is an announcement with negative implications from the investor’s perspective and thus it is expected that a profit warning will result in a negative market response. This leads to the following hypothesis: H1:

Firms that make profit warning announcements will experience negative revaluations of their share prices at the time of the announcement.

The nature of the profit warning may be due to a firm-specific event, or it may be due to industry wide factors that would affect most firms in the industry. Firms whose management does not disclose the industry wide effects on profits through the use of profit warnings may experience negative valuation effects through what is known as the contagion effect.9 This contagion effect appears because firms that operate as direct competitors in the same industry will have strong correlation in the value of their investments; as such the warning announcement conveys bad news about non-announcing firms as well as the announcing firm. Based on the expected contagion effect, the following hypothesis was posed: H2:

Non announcing firms experience negative revaluations of their share prices as a reaction to profit warnings that convey industry wide information.

The third hypothesis that was developed, due to previously composed contradicting reports concerning the intra-industry effects on non-announcing firms’ stock prices. Hertzel () finds that firm initiated announcements of stock repurchases have negligible effects on industry rivals, while Tawatnuntachai and D’Mello (), Lang and Sultz () and 9

The contagion effect implies that If the underlying conditions reflect industry-wide factors, the effects of the warning may spill over to industry rivals.

P a g e | 13

Baginski () all find evidence of contagion and/or competition effects on non-announcing firms followed by firm specific announcements by announcing firms. In order to confirm the presence and effects of intra-industry effects of profit warnings in South Africa the following hypothesis was tested: H3:

Non-announcing firms do not respond to announcements of profit warnings that convey firm specific information only.

3.2. Factors That May Influence the Effects of Profit Warnings on the Industry Past research indicates that there may be a number of factors that could affect the impact that profit warnings have on the industry. 3.2.1. General Market Reaction to the Profit Warning Announcement If a new profit warning reveals information that is previously unknown to the market and the information is industry wide, there should be an effect on all competing firms in that industry. Firth () argues that when an announcement contains a strong element of surprise, the relatively large abnormal return for the announcing firm is likely to be mirrored, to a lesser extent, in the magnitude of the abnormal returns for industry-related firms. Thus the following hypothesis is postulated: H4:

There exists a positive relationship between the average cumulative abnormal returns of non-announcing firms and the average cumulative abnormal returns of announcing firms.

P a g e | 13

3.2.2. The Size of Firm That Issues the Announcement “The size of the announcing firm could be an indication of the level of influence, power, and leadership of this company in the industry” (). As noted by Kohers in the previous statement the size of the firm could be a significant factor that will affect how announcements issued by that firm will be “taken” by the market. Thus the argument is that a profit warning issued by a relatively large firm may be more likely to send industry-wide signals which negatively affect other firms that regard the announcing firm as an important industry leader.10 This leads to the following hypothesis: H5:There exists a negative relationship between the average cumulative abnormal returns of the non-announcing firms and the size of the announcing firm. 3.2.3. Degree of Industry Homogeneity Most studies relating to intra-industry information transfers indicate that industries which are characterized by a high degree of homogeneity are more likely to exhibit intraindustry information transfers to a larger extent.11 This implies that in an industry comprised of firms that are relatively similar, a profit warning announcement by one firm may result in a greater overall industry reaction.12 The higher the degree of industry homogeneity, the more positively correlated the returns of the announcing firm would be with those of the other firms in the industry, thus strengthening the information transfer. Hypothesis 6 therefore tests whether industry homogeneity has an effect on the cumulative abnormal returns of non-announcing firms in the industry. H6:There is a negative relation between the average cumulative abnormal returns of the non-announcing firms and the degree of industry homogeneity. ___________________________________________________________________________

10

It is expected that the larger the announcing firm, the more negative reaction of the non-announcing firms will be. 11

Studies such as: Kohers (1999); Tawatnutntachai and D’Mello (2002) and Baginski (1987) all noted this phenomenon

12

This is because the factors which give rise to the announcement of the profit warning will affect other firms in the homogeneous industry and thus there will be a more significant negative reaction in the non\announcing firms than there would have been in a less homogenous industry.

P a g e | 13

P a g e | 16

Chapter 4: Sample Data and Methodology This section of the report deals with how the sample data was obtained, as well as with the methodology that was applied in testing the hypotheses that were developed in chapter 3. 4.1. Sample Data The sample data consists of well publicised profit warning announcements. The sample of profit warnings, as well as the daily stock price information was obtained from McGregor BFA and DataStream. The dataset includes profit warnings made by JSE listed firms between May 1999 and February 2004. All disclosures with implications for annual earnings will be included, not only quantitative statements, but also qualitative statements, as they convey the direction of earnings changes to investors. 4.1.1. Profit Warning Classification The profit warnings will be classified as either containing firm specific information or industry-wide common information based on the reasons provided for in the profit warning document.13 4.1.2. Data Selection Criteria - Filters All well publicised profit warnings were considered, but those with the following properties were excluded: Simultaneous announcements (within the industry) and announcements made within two days of each other are excluded to ensure that the profit warning of the announcing firm is the only factor driving the observed market reactions around the warning announcement.

13

See appendix A - The Appendix provides the basis on which profit warnings were classified and examples of the actual announcements.

P a g e | 16

Sample Data and Methodology

P a g e | 20

Additionally, to narrow the focus on the share price response to profit warnings, we exclude announcements mentioning several events, such as: •

Announcements that provide half year earnings along with a warning about future earnings are removed from the sample.

•

Announcements that include other information, such as dividend announcements are removed, as the market reaction could not be fully attributable to the warning announcement itself.

•

Firms are excluded from the dataset for the presence of confounding events, such as share repurchase announcements or acquisition announcements within a two day window of the profit warning.

•

The Diversified Financials industry has also been excluded since we do not expect a strong information transfer effect to occur in this industry.

•

Announcing firms that do not trade on the day of a profit warning have also been excluded from the sample since we do not expect there to be an information transfer if the announcing firm does not react to its own profit warning.

4.2. Event Study 14 In order to identify the stock price response of the announcing firm to the announcement of a profit warning and to capture informative announcements, the event study methodology will be used to estimate daily abnormal returns for the 11 day window (t - 5, t + 5) to test hypothesis 1. The event day (day t) is the date of the profit warning announcement. The event window was assumed to consist of 250 days. Profit warnings by other firms in the same industry around the warning announcement (days t - 2 to t + 2) will be excluded to ensure that the profit warning of the announcing firm rather than profit warnings of the other firms is driving the observed market reactions around the warning announcement. The index model is used to estimate abnormal returns.

14

Formulas used to conduct the event study were obtained from “Investments” By Bodie, Kane and Marcus, Sixth Edition.

Sample Data and Methodology

P a g e | 20

The index model parameters are estimated using ordinary least squares method in the following model:

Rit = αi + βi Rmt + εit (1) Where: •

Rit is the return for stock i in time t,

•

Rmt is the market return during the period t measured by the All Share Index,

•

αi is the average rate of return the stock would realise in the period with a zero market return,

•

βi is the coefficient of volatility of stock i’s return in relation to the market return,

•

εit is the regression residuals.

Normal returns are then calculated as follows:

E(Rit) = αi + βiRmt

(2)

Where: •

E(Rit) is the expected return on stock i in time t,

•

αi and βi are the parameters of the market model,

•

Rmt is the market return in time t measured by the All Share Index.

To estimate the effects of the profit warning on the warning firm’s returns, abnormal returns are estimated as follows:

A(Rit) = Rit –E(Rit) Where: •

A(Rit) is the abnormal stock return for stock i at time t,

(3)

Sample Data and Methodology

P a g e | 20

Average abnormal returns for each day and cumulative abnormal returns are estimated over the event period and a t-test is used to verify whether mean and cumulative abnormal returns are statistically significant or not.

AARt = 1 n  ARit n i 1

(4) ACARx,y = y

 AAR tx

t

(5) Where: •

AARt is the average abnormal return at time t,

•

CARxy is the average cumulative abnormal return from time x to time y.

To compute the effect on the reference share price of a profit warning on the announcing firm’s competitors, average abnormal returns will be estimated for a random sample of firms in the industry for the event window. A minimum of two competing firms and a maximum of ten competing firms will be selected for each profit warning released by the announcing firm.15. The above analysis will be conducted for each classification of profit warning, firm specific and industry-wide in order to test hypotheses 2 and 3. 4.3. Cross-Sectional Regression Analysis A cross-sectional regression is employed to more comprehensively examine the variables that influence the industry effects of profit warnings and to provide evidence of hypotheses 4 through 6. The two-day (days t to t + 1) cumulative abnormal returns to non-announcing firms are regressed against measures of: the market reaction of the announcing firm, the size of the warning firm, the degree of competition and the control variables such as growth of the announcing firm and market sentiment. We control for a firm’s growth prospects, since a profit warning may reflect a more severe signal to a high growth industry. 16 We also control for market sentiment at the time of each profit warning since a profit warning may 15

If there are not at least three competing firms in the industry, the effects of profit warning will not be considered. Companies have been divided into industries according to the industry they are listed in on the JSE. 16

The announcing firm’s growth (measured by the earnings-price ratio (EPR)) is used as a proxy for industry growth.

Sample Data and Methodology

P a g e | 20

reflect a more severe signal in declining markets. The non-announcing firms’ negative market reactions in rising markets may be attenuated since investors could believe that the industry outlook as a whole is positive and that future period’s profits will rise. The following model is estimated to examine the variables that influence the industry effects of profit warnings: CAR = α + β1CARA + β2SIZE + β3HOMO + β4GROWTH + β5SENTIMENT + ε (6) Where: •

CAR is the two-day (days 0 to +1) cumulative abnormal return for the nonannouncing firms;

•

CARA is the two-day (days 0 to +1) cumulative abnormal return for the firm announcing the profit warning;

•

SIZE is the size of the firm issuing the profit warning (measured by the natural log of the market value of equity) on day t - 20 (relative to the profit warning);

•

HOMO is a dummy variable which takes the value of 1 one for homogeneous industries (e.g. Banking, Energy and Utilities industries) and 0 otherwise. Where homogeneous industries are industries that are similar in terms of the product or service offering or the general operating environment. 17

•

GROWTH is a measure of growth of the announcing firm and is a proxy for the growth of the industry and is measured using the earnings-price ratio (higher earnings-price ratio implies lower growth);

•

SENTIMENT is an indicator of market sentiment. It is a measure of the underlying feeling in the market during the 20 day period prior to the profit warning and is measured by the holding period return on the JSE All Shares Index;

•

ε is the regression error term.

__________________________________________________________________________

17

Specifically, the Banking and Utility industries are characterised as homogeneous, because of the regulatory constraints which greatly limit the ability to diversify. The Energy industry is characterised as homogeneous due to the relatively standardised products and services with little differentiation potential.

Sample Data and Methodology

P a g e | 20

Chapter 5: Descriptive Statistics The subsequent section is to be used as a summary of the data that has been collected during this study. It includes a breakdown of the types of profit warnings that were issued in various industries as well as the various correlations that were conducted. 5.1. Characteristics of Profit Warnings After applying the filters as discussed in section 4.1.2 to the 124 profit warnings that were recorded over the study period the final sample consists of 51 profit warnings as shown in Table 1: Profit Warnings. This is a substantial reduction in the size of the sample data, but if the filters were not included the results of the study would have not been a true representation of intra-industry information transfers, which this report aims to investigate.18

Table 1: Profit Warnings segmented by source of warning

1999 Sample by source of warning Total Sample 12 Industry-Wide 3 Firm Specific 9

2000

2001

2002

2003

Total

23 7 16

9 1 8

4 0 4

3 2 1

51 13 38

Additional information is shown in Table 2 which contains a frequency distribution of the announcing and non-announcing firms corresponding to each profit warning observation in the sample on an industry basis. A total of 28 industries are represented. The distribution suggests substantial dispersion in the industries of profit warning firms. No single industry accounts for more than 9.09% of the sample observations and only six industries contribute more than 5% of the observations in the sample. The least represented industries are; Banks, Buildings and Construction materials, Chemicals, Diamond, Electrical Equipment, Farming and Fishing, Food and Drug Retailers, Investment Companies, Other Construction and Publishing and Printing as well as the various retailers with just one profit warning each in the final sample.19 As noted above, the minimum number of competitors is two nonannouncing firms and the maximum is ten non-announcing firms. The mean number of One should also note that it was found that relative to other stock markets there is a lower disclosure record of profit warnings in South Africa – For the same period a total of 246 profit warnings were available on ASX listed companies, this however does not indicate poor disclosure practices on the part of South Africa. This may be explained by different economic conditions in the two counties.

18

19

This may be attributed to the narrow classification of industries that was employed, but this has been deemed necessary in order to ensure that only strict competitors’ returns within the same industry are compared.

Sample Data and Methodology

P a g e | 20

competitors per profit warning is 4.53 and the median is 4.50. There are a total of 231 nonannouncing firms and Table 2 shows that the sample is well diversified.

Descriptive Statistics

P a g e | 24

Table 2: Frequency Distribution of industries for announcing and non-announcing firms

Industry

No. of Announcing Firms

Percentage of Total Announcing Firms

Average No. of Rival Firms per Announcement

Total No. of Rival Firms

Percentage of Total Rival Firms

Auto Parts

2

3.64%

2

3

1.30%

Banks

1

1.82%

8

8

3.46%

Building & Construction Materials

1

1.82%

10

10

4.33%

Business Support Services

3

5.45%

3

10

4.33%

Chemicals – Speciality

1

1.82%

5

5

2.16%

Computer Services

4

7.27%

4

15

6.49%

Containers & Packaging

1*

5.45%

4

4

1.73%

Diamond

1

1.82%

10

10

4.33%

Diversified Industrials

2

9.09%

3

6

2.60%

Electrical Equipment

1*

1.82%

8

8

3.46%

Electronic Equipment

2

3.64%

3

5

2.16%

Engineering – General

3

5.45%

2

5

2.16%

Farming & Fishing

1

1.82%

10

10

4.33%

Food & Drug Retailers

1

1.82%

4

4

1.73%

Gold Mining

2

3.64%

9

17

7.36%

Insurance - Non-Life

2

3.64%

2

3

1.30%

Investment Banks

4

7.27%

3

12

5.19%

Investment Companies

1

1.82%

10

10

4.33%

Metals & Minerals

2

3.64%

9

18

7.79%

Other Construction

1

1.82%

10

10

4.33%

Other Financial

4

7.27%

2

7

3.03%

Publishing & Printing

1

1.82%

5

5

2.16%

Real Estate Holdings & Development

2

3.64%

10

20

8.66%

Restaurants and Pubs

2

3.64%

2

3

1.30%

Descriptive Statistics

P a g e | 24

No. of Announcing Firms

Percentage of Total Announcing Firms

Average No. of Rival Firms per Announcement

Total No. of Rival Firms

Percentage of Total Rival Firms

Retailers – Hardlines

1

1.82%

4

4

1.73%

Retailers - Multi Department

2

3.64%

3

6

2.60%

Retailers - Soft Goods

1

1.82%

8

8

3.46%

Shipping & Ports

1

1.82%

5

5

2.16%

Total: 28 defined industries

51

100%

4.53

231

100%

Industry

* indicates situations where there was more than one profit warning – in these cases the second profit warning and the issuing firm were excluded from the study, but there were enough non-announcing competitors to allow the initial profit warning to be considered.

Descriptive Statistics

P a g e | 24

Descriptive Statistics

P a g e | 28

Tables 3, 4, and 5 provide the Pearson correlations between the Average Cumulative Abnormal Return (ACAR) for announcing and non-announcing firms from day t = 0 and day t = +1. For the full sample, average cumulative abnormal returns for announcing firms are negative (at -7.97%) and those for portfolios of non-announcing firms are also negative (albeit to a lesser degree of -0.10%). For the industry-wide sub sample, average cumulative abnormal returns for announcing firms are negative (-8.49%) and those for non-announcing firms are also negative (-0.33%). For the firm specific sub sample, average cumulative abnormal returns for announcing firms are negative (at -7.80%) and those of nonannouncing firms are slightly negative (-0.02%). This suggests that the profit warnings included in the sample tend, on average, to convey bad news not only for the announcing firms, but also for non-announcing firms (at varying degrees depending on the type of announcement). The Pearson correlation coefficient between ACAR (0 to +1) for announcing firms and ACAR (0 to +1) for non-announcing firms is positive (0.123) and significant at the 0.01 level for the full sample. For the industry-wide sub sample, the Pearson correlation between ACAR (0 to +1) for announcing and non-announcing firms is positive (0.344) and significant at the 0.01 level. For the firm specific sub sample, the Pearson correlation between ACAR (0 to +1) for announcing and non-announcing firms is small and positive (at 0.048), but not significant. These correlations provide preliminary evidence of an association between announcing firms’ ACAR and the stock returns of competitors in the same industry for profit warnings conveying industry-wide information (as expected) and for the full sample, but not for firm specific announcements. Furthermore, Table 6 shows that the difference between the two correlation coefficients is significant at the 0.01 level (zscore = 1.608)20.

Z-statistic testing was applied because despite only having 13 values for the industry specific subsample, it has been determined in Appendix B that the daily returns of the full sample follows a normal distribution, hence any sample drawn from this population can be tested by the z-statistic.

20

Descriptive Statistics

P a g e | 28

Table 3: Summary Statistics and Pearson correlations - Full Sample

Full Sample Summary Statistics Mean (%) Standard Deviation N Pearson Correlation Announcing ACAR 0 to +1 Significance (2-tailed)

ACAR (0 to +1) Announcing

ACAR (0 to +1) Non-Announcing

-7.97% 0.161 51

-0.10% 0.134 51

1

0.123 0.015

Table 4: Summary Statistics and Pearson Correlations - Industry Wide Sub-sample

ACAR (0 to +1) Announcing I.W. Sub-Sample Summary Statistics Mean (%) -8.49% Standard Deviation 0.058 N 13 Pearson Correlation Announcing ACAR 0 to +1 1 Significance (2-tailed)

ACAR (0 to +1) Non-Announcing -0.33% 0.097 13 0.344 0.044

Table 5: Summary Statistics and Pearson Correlations – Firm Specific Sub-sample

ACAR (0 to +1) Announcing F.S. Sub-Sample Summary Statistics Mean (%) -7.80% Standard Deviation 0.175 N 38 Pearson Correlation Announcing ACAR 0 to +1 1 Significance (2-tailed)

ACAR (0 to +1) Non-Announcing -0.02% 0.080 38 0.048 0.313

Descriptive Statistics

P a g e | 28

Table 6: Test of Significance between Correlation Coefficients of ACAR for Announcing and Non-Announcing Firms21

Industry-Wide Sub Sample (R1)

Correlation Coefficient Between Announcing and Non-Announcing Firms' ACAR (0 to +1) 0.344

13

Firm Specific Sub Sample (R2)

0.048

38

R1 - R 2 z-score Significance (2-tailed)

0.296 1.608 0.108

N

Table 7 shows accounting ratios for companies making profit warnings segmented by industry. Return on Equity (ROE) is a key indication of the company’s performance as it provides information on how well managers are employing funds invested by the shareholders to generate returns. The positive earnings per share (EPS) and ROE ratios indicate good recent performance for the companies making profit warnings. The EPS for the Banking is the highest at 332.12 cents/share and the ROE for the Food and Drug Retailers is the highest at 43.30%. The average market sentiment of 0.123 % suggests that the overall outlook prior the announcement of the profit warnings was positive

21

This table yields the result of a test of the hypothesis that the two correlation coefficients obtained from the relationship between the average cumulative abnormal return (0 to +1) for the announcing firms and non-announcing firms are equal between the industry-wide and firm-specific sub samples.

Descriptive Statistics

P a g e | 28

Table 7: Accounting Ratios for Companies making Profit Warnings

Industry Auto Parts Banks Building & Construction Materials Business Support Services Chemicals - Speciality Computer Services Containers & Packaging Diamond Diversified Industrials Electrical Equipment Electronic Equipment Engineering - General Farming & Fishing Food & Drug Retailers Gold Mining Insurance - Non-Life Investment Banks Investment Companies Metals & Minerals Other Construction Other Financial Publishing & Printing Real Estate Holdings & Development Restaurants and Pubs Retailers - Hardlines Retailers - Multi Department Retailers - Soft Goods Shipping & Ports Average

Average MCAP (ZAR m)

Average PER

Average ROE

Average EPS (cents)

Average Sentiment *

484.86 21296.4 1

5.78

11.90

239.53

0.82

8.04

21.45

332.12

0.78

882.99 2887.04 1589.90 429.33 1631.48 588.38 60.80 1505.46 38.79 159.31 601.65 3757.54 7922.33 3943.29 3047.23 131.42 41660.5 0 1253.15 554.16 1027.54

6.42 8.91 8.66 7.07 -4.33 4.21 8.15 5.19 -3.94 1.78 4.35 18.31 4.47 18.54 5.02 0.30

-14.75 -7.66 4.36 -13.90 18.23 6.19 7.98 10.44 -25.38 7.63 15.00 43.30 12.36 15.60 23.59 -55.04

142.00 98.45 95.48 30.61 39.60 95.90 228.20 140.94 -1.35 67.53 71.90 47.60 301.20 263.07 196.00 12.65

0.22 0.06 -0.14 -0.10 -0.50 -0.10 -0.30 0.09 0.02 0.54 -0.20 1.40 -0.04 0.02 -0.03 0.09

5.18 4.07 7.69 11.37

21.40 2.20 -76.60 0.50

281.23 72.76 12.28 191.00

-0.09 0.04 -0.01 -0.10

318.71 105.76 1237.76

5.96 0.65 8.57

2.00 13.30 17.92

71.08 19.73 246.43

-0.02 -0.10 0.36

1826.97 716.62 3432.92

7.88 10.64 5.98

21.84 1.91 29.02

23.80 2.31 195.98

1.05 -1.08 0.76

3681.87

6.25

4.10

125.64

0.123

Note: MCAP = (Number of shares issued at end year) x (share price at year end). PER = (Closing share price on last day of company's financial year / (Earnings per share) ROE = (Net profit after tax before abnormal, less outside equity interests / (diluted weighted number of shares outstanding during the year). EPS = (Net Profits After Tax) / (shareholders equity - outside equity interests). *

Holding period return over 20 days prior to the profit warning.

Descriptive Statistics

P a g e | 28

All accounting ratios are for the year end closest to the announcement of the profit warning.

Descriptive Statistics

P a g e | 28

Results

P a g e | 46

Chapter 6: Results This section of the report describes the results that have been obtained (from the event study and the multivariate regression) in a graphical manner through the use of tables and graphs, there is also a detailed explanation of the information that can been drawn from each set of results that has been obtained. 6.1. Event Study Results – Full Sample Table 8 presents the event study results for the full sample of profit warnings for announcing and non-announcing firms for the period of five days before, to five days after the announcement day. Cumulative abnormal returns are also reported in Table 8 for days -1 to 0; 0 to +1; -1 to +1 and -2 to +2.

Results

P a g e | 46

Table 8: Effects on Returns in Response to Profit Warning Announcements - Full Sample22

Average Abnormal Return (%) Day Relative to Announcement -5 -4 -3 -2 -1 0 1 2 3 4 5 Day Relative to Announcement -1 to 0 0 to +1 -1 to +1 -2 to +2 N

22

Announcing Firms

Non-Announcing Firms

0.90% 0.87% 0.34% -0.19% -0.62% -4.89% -3.09% -1.25% -0.52% -0.21% 0.22% Average Cumulative Abnormal Return (%) Announcing Firms -5.51% -7.97% -8.60% -10.04% 51

1.37% 2.11% -0.42% 1.03% 0.07% -0.31% 0.21% -0.55% 0.31% -1.22% 0.15%

Non-Announcing Firms -0.25% -0.10% -0.04% 0.44% 51

This table reports mean abnormal returns and mean cumulative abnormal returns around the announcements of 51 profit warnings between January 1999 and December 2003. Non-announcing average abnormal returns are the equally-weighted average abnormal returns of portfolios of other firms with the same industry classification as announcing firms. Abnormal returns are calculated using the market model. Average Cumulative Abnormal Return is the sum of the average abnormal returns for the days specified (-1 to 0, 0 to +1, -1 to +1, and -2 to +2).

Results

P a g e | 46

6.1.1. Effects on Announcing Firms – Full Sample The largest average abnormal return for announcing firms occurs on day t = 0 (-4.89%) followed by day t = +1 (-3.09%) and the average cumulative abnormal returns from day t = -1 to t = +1 (-8.60%) and from day t = -2 to t = +2 (-10.04%), all of which are highly significant with p-value = 0.00. These negative results are consistent with the expectation that share prices in announcing firms will fall following a profit warning. These results therefore provide support for hypothesis 1 (H1 on pg. ), which stipulated a negative reaction in the announcing firms to the announcement of a profit warning. On the days prior to the announcement there is presence of significant and negative average abnormal returns which is consistent with the findings of Elayan, Meyer and Sun (2002). There is a relatively gentle decline and a gradual increase in negative average abnormal returns on the days before the profit warning compared to the large declines on day t = 0 and day t = +1. This provides support for the notion that investors have identified firms with poorer than expected earnings and have started to downgrade market prices or there was some information leakage prior to the announcement of the profit warnings. There are no significant daily average abnormal returns after the two-day announcement period (days t = 0 to t = +1) suggesting that the majority of the response to the profit warnings occurs over this period. The significant negative reaction on the day following the warning (-3.09%) could be the result of some market participants delaying their trading until they observe the assessment, based on research bulletins or trading, of informed investors. Figure 1 overleaf, provides a visual depiction of average abnormal returns and average cumulative abnormal returns over the event window for the full sample of announcing firms.

Results

P a g e | 46

Figure 1: AAR and ACAR for Full Sample of Announcing Firms23

The large negative spike in average abnormal returns on day t = 0 can clearly be seen. Figure 1 also shows that in the days prior to the profit warning the market starts a downward anticipatory movement. There is then a sharp drop over days t = -1 to t = 0 and a further fall from day t = 0 up to day t = +1 and thereafter a recovery as shown by the average cumulative abnormal return trend line. 6.1.2. Effects on Non-Announcing Firms – Full Sample The average abnormal return for non-announcing firms on day t = 0 (-0.31%) is significant at the 5% level (p-value = 0.038). The average cumulative abnormal returns from days t = -1 to t = +1 (-0.04%) are significant at the 1% level (p-values = 0.01)24. By observation alone it is not clear whether the decline on day t = 0 is significant, but the pvalue for the t-test performed on the differences of returns from zero show that it is significant at the 1% level (P-value = 0.02) These results provide evidence of an intraindustry information transfer from firms announcing profit warnings to non-announcing firms. 6.1.3. Full Sample Results – Summary of Findings The contagion effect dominates the competitive effect on average which results in negative abnormal returns for the announcement-period. Thus, the unfavourable information conveyed by profit warning announcements has a net negative effect on the equity value of the other firms in the industry. Therefore, the finding of previous studies that the market revises the announcing and non-announcing firms’ values in the same direction in the context of different corporate events also extends to profit warnings and confirms Hypothesis 1.

23

This Figure shows the relation between Days Relative to the Profit Warning announcement (x-axis), Average Abnormal Return (AAR) and the Average Cumulative Abnormal Return (ACAR) for the full sample of profit warnings (N = 51) for announcing firms over the event window (t = –5 up to day t = +5).

24

p-values for t-tests of the differences of the returns from zero. p-values display the smallest level of significance for which the hypothesis can be rejected ().

Results

P a g e | 46

6.2. Event Study Results – Industry-Wide Sub Sample Table 9 presents the event study results for the industry-wide information sub sample of profit warnings for announcing and non-announcing firms for the period of five days before, to five days after the announcement day. Cumulative abnormal returns are also reported in Table 9 for days -1 to 0, 0 to +1, -1 to +1 and -2 to +2. Table 9: Effects on Returns in Response to Profit Warning Announcements – Industry-Wide Sub Sample

Average Abnormal Return (%) Day Relative to Announcement -5 -4 -3 -2 -1 0 1 2 3 4 5 Day Relative to Announcement -1 to 0 0 to +1 -1 to +1 -2 to +2 N

Announcing Firms

Non-Announcing Firms

0.32% 0.37% 0.27% -0.74% -1.40% -3.84% -4.64% -0.45% -0.34% -1.20% -0.24% Average Cumulative Abnormal Return (%) Announcing Firms -5.24% -8.49% -9.89% -11.07% 13

0.02% -0.09% -0.98% 0.78% -1.75% -1.66% 1.33% -1.02% -0.64% -0.88% -0.09%

Non-Announcing Firms -3.41% -0.33% -2.08% -2.31% 13

Results

P a g e | 46

6.2.1. Effects on Announcing Firms – Industry Wide Sub Sample The largest average abnormal return for announcing firms occurs on day t = 1 (-4.64%) (p-value = 0.00) followed by day t = 0 (-3.84%)25 (p value = 0.01). The average cumulative abnormal returns from days t = -1 to t = +1 (-9.89%) and from days t = -2 to t = +2 (-11.07%), both with p-value = 0.00. This shows a negative reaction in announcing firms to the announcement of profit warnings that convey industry-wide information. This was the expected reaction as the profit warnings convey negative implications on all firms in the industry from the market’s perspective. There are also significant and negative average abnormal returns on day t = -2 (-0.74%, p-value = 0.03) and day t = -1 (-1.40%, p-value = 0.01). These negative average abnormal returns could be attributed to some of the factors discussed previously in section 6.1.1 or investors may have been able to predict adverse industry conditions from alternative sources, such as from forecasts of a drought for the agriculture industry, before a profit warning was issued. The presence of further decline in AAR up to five days after the announcement period suggesting that the full response to the profit warning conveying industry wide information occurs over an extended period of time. This could be because of investors continually re-adjusting their valuations according to how the market reacts to the news for the industry in the days after the announcement. As above, for the full sample, the significant negative reaction on the day following the warning (-4.64%, p-value = 0.01) could be the result of some market participants delaying their trading until they observe the assessment, based on research bulletins or trading, of informed investors. Figure 2: AAR and ACAR for Industry-Wide Sub Sample of Announcing Firms

Figure 2 provides a visual depiction of average abnormal returns and average cumulative abnormal returns over the event window for the industry-wide information sub sample of 13 announcing firms. The large negative spike in average abnormal returns on days t = 0 and t = 1 is clearly shown. Figure 2 also shows that in the days prior to the release of the profit warning, there is a downward anticipatory movement. There is then a sharp drop over days t = 0 to t = 1 and a further fall from day t = 1 up to day t = +5. This result was not expected, some possible explanations are examined later in this section

25

Results

P a g e | 46

6.2.2. Effects on Non-Announcing Firms – Industry Wide Sub Sample Figure 3 plots the average abnormal returns and the average cumulative abnormal returns in non-announcing firms against the day relative to the profit warning for the industrywide information sub sample. At first glance there seems to be a somewhat mixed reaction in the AAR of non announcing firms; in the pre-announcement period, there is a significant and negative average abnormal return for non-announcing firms on day t = -3 (-0.98%), followed by a smaller but positive AAR on day t = -2 (0.78%), but which was found to be insignificant (p-value = 0.21). By observing the trend line of the ACAR shown in Figure 3 there is a gradual increase in the negative abnormal return up until its maximum point of -1.75% on day t = -1 followed by a further general declining trend similar to that of the announcing firms shown in Figure 2 26. The significant average abnormal returns, which show a negative trend up to and after the announcement could be explained by one of the same factors for announcing firms prior to the announcement of a profit warning, as described in section 6.2.1 (on pg. ). Figure 3: AAR and ACAR for Industry-Wide Sub Sample of Non-Announcing Firms

6.2.3. Industry-Wide Sub Sample Summary of Results Table 9 shows that the average abnormal return for non-announcing firms on day t = 0 (-1.66%) is significant at the 1% level (p-value = 0.01). The average cumulative abnormal returns from days t = -1 to t = +1 (-2.08%) and from days t = -2 to t = +2 (-2.31%) are also significant at the 1% level (p-values = 0.01). These results provide evidence of an intra-industry information transfer from firms announcing profit warnings that convey industry-wide information to non-announcing firms. There is a net contagion effect resulting in a negative effect on the equity value of other firms in the industry. Thus, if the underlying conditions that give rise to profit warnings reflect industry-wide factors, there is a spill over effect to the rest of the industry. These results therefore provide support for hypothesis 2 (H2) which states that there is a negative reaction in nonannouncing firms to the announcement of profit warnings that convey industry-wide information.

26

The increase in positive AAR on day t = 1, has been evidenced but after applying the t-tests was found not to be significant (p-value = 0.26), despite this point there is a clear negative trend over the period after the announcement for up to days (t = 5), the AAR for days t = 3 and t = 4 have been found to be significant using the t-tests.

Results

P a g e | 46

6.3. Event Study Results – Firm-specific Sub Sample Table 10 presents the event study results for the firm-specific information sub sample of profit warnings for announcing and non-announcing firms for the period of five days before, to five days after the announcement day. Cumulative abnormal returns are also reported in Table 10 for days -1 to 0, 0 to +1, -1 to +1 and -2 to +2. Table 10: Effects on Returns in Response to Profit Warning Announcements – Firm-Specific Sub Sample

Average Abnormal Return (%) Day Relative to Announcing Firms Non-Announcing Firms Announcement -5 1.10% 1.83% -4 1.04% 2.87% -3 0.36% -0.22% -2 -0.01% 1.11% -1 -0.36% 0.69% 0 -5.24% 0.15% 1 -2.56% -0.17% 2 -1.53% -0.39% 3 -0.58% 0.64% 4 0.13% -1.34% 5 0.38% -0.29% Average Cumulative Abnormal Return (%) Day Relative to Announcing Firms Non-Announcing Firms Announcement -1 to 0 -5.60% 0.83% 0 to +1 -7.80% -0.02% -1 to +1 -8.16% 0.66% -2 to +2 -9.69% 1.39% N 38 38 6.3.1. Effects on Announcing Firms – Firm-Specific Sub Sample The largest average abnormal return for announcing firms occurs on day t = 0 (-5.24%) followed by day t = +1 (-2.56%), both highly significant with p-values of 0.00. The average cumulative abnormal returns from days t = -1 to t = +1 (-8.16%) and from days t = -2 to t = +2 (-9.69%) are also significant with p-values of 0.00. Thus, there is a negative reaction in announcing firms to the announcement of profit warnings that convey firm specific information. It is interesting to note that the maximum negative average abnormal return for firm specific announcements (-5.24% on day t = 0) is greater than that of industry-wide announcements (-4.64% on day t = 1). However, this

Results

P a g e | 46

difference is not statistically significant (p-value = 0.271). Thus, there is no differential reaction to industry-wide announcements and firm specific announcements for the firms making the announcement. There are significant and negative average abnormal returns on the days prior to the announcement of the profit warnings suggesting either an information leakage prior to the announcement of the profit warnings or investors may have identified firms that have specific problems before the release of the profit warnings and downgraded market prices before day t = 0. There is a positive and significant average abnormal return on day t = + 4 (0.13%, pvalue = 0.03), possibly suggesting a small correction to over-reaction to the profit warning. After this correction there are gradual adjustments in the equity value until the stock price reaches its true equilibrium value. Figure 4 provides a visual depiction of average abnormal returns and average cumulative abnormal returns over the event window. Figure 4: AAR and ACAR for Firm-Specific Sub Sample of Announcing Firms

The large negative spike in average abnormal returns on day t = 0 is clearly shown. Figure 4 also shows that in the days prior to the release of the profit warning (days t = -2 and t = -1), there is a downward anticipatory movement. There is then a sharp drop on day t = 0 followed by further decline up to day t = +3 thereafter a recovery as shown by the average cumulative abnormal return trend line.

Results

P a g e | 46

6.3.2. Effects on Non-Announcing Firms – Firm-Specific Sub Sample Figure 5shows the relation between days relative to the profit warning (x-axis) and average abnormal return (y-axis) for the firm specific sub sample of profit warnings (N = 38) for non-announcing firms over the event window (day –5 to day +5). Figure 5: AAR for Firm-Specific Sub Sample of Non-Announcing Firms

The daily ACAR have not been included in the figure due to the findings which implied that there are no significant average abnormal returns for non-announcing firms over the entire event window. The average abnormal return on day t = 0 and the average cumulative abnormal return over the periods t = -1 to t = +1 and t = -2 to t = +2 are statistically insignificant.

Results

P a g e | 46

6.3.3. Firm-Specific Sub Sample Summary of Results The results for this section suggest that profit warnings containing firm specific information have no significant effect on their industry rivals and hence imply that there is no evidence of either a contagion or competitive intra-industry effect. These results provide support for hypothesis 3 (H3 on pg. ) which stated that non-announcing firms do not respond to the announcement of profit warnings that convey only firm specific information. By inspection of Figure 5 it can be seen that there is no clear finding of abnormal returns around the announcement day for non-announcing firms in the firm specific sub sample, which is confirmed by tests for statistical significance27. 6.4. Multivariate Cross-Sectional Analysis – Full Sample Five independent variables are used to explain the market reaction in non-announcing firms to profit warnings and provide evidence on the three hypotheses previously developed. The two day (days t = 0 to t = +1) ACAR of the non-announcing firms is the dependent variable. Since negative Price-Earnings Ratios (PER) are not very meaningful, observations with negative EPRs have been removed from the sample and the regression has been run with only positive EPRs in the model28. The regression results for the sample are shown in Table 11. Table 11: Cross-Sectional Multivariate Model Results - Full Sample

Variable

Value 0.1223 0.0835 -0.0141 0.0215 0.2633 -0.0021

Intercept ACARA LNMVE HOMO SENTIMENT EPR Regression Summary Statistics R-Squared 0.0582 Adj. R-Squared 0.0420 F-statistic 3.4809 N 50

p-value 0.3280 0.0792 0.4905 0.5976 0.3320 0.0938

27

This observation has been backed up by statistical tests. i.e. Determining the p-values for t-tests of the differences of the average abnormal returns from zero

28

A negative PER implies that investors will be given money to buy a company's earnings, which is not the case.

Results

P a g e | 46

Table 11 provides the parameter estimates of the independent variables The Model is significant at the 1% level (F-statistic = 3.4809). 6.4.1. Coefficient - ACARA – Full Sample The coefficient for the average cumulative abnormal return for announcing firms (ACARA) is positive and significant at the 10% level in the Model (p-value = 0.0792). This conforms to our expectations and provides support for hypothesis 4 (H4 on pg. ) for the full sample of profit warnings. The β1 coefficients are less than one, indicating that stock price reactions of non-announcers will be less than the abnormal return of the profit warning firms. 6.4.2. Coefficient – LNMVE – Full Sample The coefficient for the natural log of the size of the announcing firm (LNMVE) is negative but insignificant. This is contrary to our expectations and indicates that the size of the announcing firm is not a determinant of the stock price reaction of nonannouncing firms for the full sample of profit warnings. Thus hypothesis 5 (H5 on pg. ) is not supported for the full sample of profit warnings. 6.4.3. Coefficient - HOMO – Full Sample The coefficient for the homogeneous industry (HOMO) which is a dummy variable that takes the value of one for homogeneous industries29 (e.g. Banking, Energy and Utilities industries), has been found to be positive but insignificant. Thus, hypothesis 6 (H6 on pg. ) is not supported for the full sample of profit warnings. 6.4.4. Coefficient - SENTIMENT – Full Sample The coefficient for the market sentiment variable (SENTIMENT) is positive but insignificant. This provides some evidence that the negative stock price reaction in nonannouncing firms is attenuated when the recent market sentiment has been positive. 6.4.5. Coefficient - EPR – Full Sample The coefficient for the earnings-price ratio (EPR) is negative and significant at the 10% level (p-value = 0.0938). Thus the EPR is a significant explanatory factor for the stock price reaction of non-announcing firms. Homogeneous industries are industries that are similar in terms of the product or service offering or the general operating environment. Specifically, the Banking and Utility industries are characterised as homogeneous, because of the regulatory constraints which greatly limit the ability to diversify. The Energy industry is characterised as homogeneous due to the relatively standardised products and services with little differentiation potential. 29

Results

P a g e | 46

6.5. Multivariate Cross-Sectional Analysis – Industry Wide Sub-Sample The same regression as the one discussed above was then carried out for the Industry Wide Sub-Sample, and the results are summarized in the table below. Table 12: Multivariate Cross-Sectional Analysis - Industry Wide Sub-Sample Results

Variable Value Intercept 0.0563 ACARA 0.0588 LNMVE -0.0078 HOMO 0.0104 SENTIMENT -0.0058 EPR 0.0000 Regression Summary Statistics R-Squared 0.7283 Adj. R-Squared 0.5585 F-statistic 4.2894 N 13

p-value 0.3979 0.0247 0.0952 0.0787 0.5359 0.7762

The adjusted R-squared for the industry-wide sub sample (0.5585) is higher than that of the full sample (0.0420) indicating that the independent variables in the models explain more of the variation in the ACAR of the non-announcing firms for the profit warnings conveying industry-wide information than for profit warnings conveying industry-wide and firm specific information as in the full sample.

Results

P a g e | 46

6.5.1. Coefficient - ACARA – Industry Wide Sub-Sample The coefficient for the average cumulative abnormal returns for announcing firms (ACARA) is positive and significant at the 5% level in the model (p-value = 0.0247). This indicates that for profit warnings conveying industry-wide information there is a negative reaction in non-announcing firms, however to a lesser extent than in the announcing firms since the coefficient is less than one. This provides support for hypothesis 4. 6.5.2. Coefficient - LNMVE – Industry Wide Sub-Sample The coefficient for the natural log of the size of the announcing firm (LNMVE) is negative and significant at the 10% level (p-value = 0.0952). This indicates that for profit warnings that convey industry-wide information, the larger the size of the announcing firms, the more negative is the ACAR of the non-announcing firms. This supports hypothesis 5 for profit warnings that convey industry-wide information. 6.5.3. Coefficient - HOMO – Industry Wide Sub-Sample The coefficient for the homogeneous industry variable (HOMO) is positive and significant at the 10% level (p-value = 0.0787). This provides evidence that the higher the degree of industry homogeneity, the less negative is the cumulative abnormal returns of non-announcing firms, contrary to our expectations. 6.5.4. Coefficient - SENTIMENT – Industry Wide Sub-Sample The coefficient for the market sentiment variable (SENTIMENT) is positive but insignificant in the model. This provides some evidence that the negative stock price reaction in non-announcing firms, following the announcement of a profit warning that conveys industry-wide information, is attenuated when the recent market sentiment has been positive. 6.5.5. Coefficient - EPR – Industry Wide Sub-Sample The coefficient for the earnings-price ratio (EPR) is negative, but insignificant thus is not an explanatory factor for the stock price reaction of non-announcing firms in response to profit warnings that convey industry-wide information.

Results

P a g e | 46

6.6. Multivariate Cross-Sectional Analysis – Firm Specific Sub-Sample The regression models for the firm specific sub sample, presented in Table 13, are not statistically significant (F-statistic = 0.8567). The adjusted R-squared for the Model is negative (-0.0089) indicating a poor fit of the regression. Table 13: Multivariate Cross-Sectional Analysis – Firm Specific Sub-Sample - Results

Variable Intercept ACARA LNMVE HOMO SENTIMENT EPR Regression Summary Statistics R-Squared Adj. R-Squared F-statistic N

Value -0.0687 0.0734 -0.0126 -0.0037 0.0216 0.0000

p-value 0.5051 0.1148 0.0986 0.9548 0.9370 0.2397

0.0543 -0.0089 0.8567 37

The coefficient for the natural log of the size of the announcing firm (LNMVE) is positive and significant at the 10% level (p-value = 0.0986). This indicates that the greater the size of the announcing firm, the smaller is the negative average cumulative abnormal return in the non-announcing firms for profit warnings that convey firm specific information to the market. The remaining coefficients were found to be statistically insignificant. __________________________________________________________________________

Results

P a g e | 46

P a g e | 47

Chapter 7: Conclusion This section of the report will discuss the implications of the findings as derived from the results, followed by a discussion regarding the limitations of the study and closing with recommendations for future studies. 7.1. Conclusions and Implications Investors utilise publicly available information, including information provided in profit warnings, in their decisions about capital allocation. For this reason, the J.S.E. obliges listed companies to keep investors informed in a timely manner about material price-relevant information. Thus the main contribution of this study lies in the answer to the question of whether there is a difference between the share price behaviour of non-announcing firms in response to firm specific announcements and industry-wide common announcements. 7.1.1. Conclusions Drawn From the Event Study For profit warnings conveying industry-wide information the announcing firms and the rivals of the announcing firms incur significant negative abnormal returns. This reaction implies that participants of financial markets gather new information about industry conditions for rivals as well as for announcing firms from profit warnings that are issued due to industry-wide factors. For profit warnings conveying firm specific information the announcing firms incur significant negative abnormal returns, whereas, the rivals of the announcing firms do not. The significant negative reaction shows that market participants gather new information about the announcing firms from firm specific announcements but not for the rival firms, shown by the insignificant average abnormal returns. This is because firm specific factors are problems unique to a particular firm and therefore should have no effect on rival firms. By separately documenting the intra-industry information transfer process for the industry-wide information and firm specific information sub-samples, this study overcomes a shortcoming of previous studies in the area of information transfers which lead to an inaccurate description of the intra-industry information transfer process. The finding of significant negative abnormal returns in the days leading up to the announcement day provides evidence of some information leakage. This is consistent with the results of Jackson and Madura () in the U.S. In contrast Helbok and Walker() and Collett () did not find the same in the U.K. which could suggest a lower level of leakage in periods running up to trading updates in the U.K. than in the U.S. and South

P a g e | 47

Africa. Alternatively, investors in South Africa and the U.S. may have identified firms with poorer than expected earnings and had started to downgrade market prices prior to the profit warning. This could be due to investors in South Africa and the U.S. utilising other sources of information to predict the negative effects before the profit warning is issued. 7.1.2. Conclusions regarding Full Sample Multivariate Regression For the full sample of profit warnings the average cumulative abnormal return of the announcing firms was a significant contributor to the cumulative abnormal return of the non-announcing firms. Thus, if the profit warning contains new information that affects the whole industry the market will inflict a more negative price adjustment on all the firms in the industry. The positive market sentiment coefficient for the full sample of profit warnings provides evidence that the negative stock price reaction in non-announcing firms is attenuated when the recent market sentiment has been positive. That is, the market punishes competing firms to a lesser extent when market sentiment is relatively favourable. Once the negative EPRs are excluded from the cross-section regression model, the EPR becomes a significant explanatory factor for the stock price reaction of non-announcing firms. The negative coefficient indicates that the higher the EPR (lower industry growth), the more negative is the stock price reaction in non-announcing firms. This could be explained by the notion that firms that are not growing do not have the ability to expand into areas not affected by the reasons given for the profit warning, resulting in greater negative average cumulative abnormal returns in non-announcing firms. The size of the announcing firms is not a significant contributor to the cumulative abnormal return of the announcing firms for the full sample of profit warnings. Perhaps investors did not distinguish between large and small firms that announced a profit warning or perhaps large firms do not necessarily have more power, influence or leadership in their respective industries. The degree of industry homogeneity had no significant impact on the cumulative abnormal return of the announcing firms for the full sample of profit warnings. Possibly the homogeneous industries used in this study were not of a sufficient degree of homogeneity.

P a g e | 47

7.1.3. Conclusions regarding Industry Wide Sub-Sample Multivariate Regression For the industry-wide sub sample of profit warnings, the average cumulative abnormal return of the announcing firms was a significant contributor to the cumulative abnormal return of the non-announcing firms; the implications of this are the same as those discussed in section 7.1.2. The size of the announcing firm was a significant contributor to the cumulative abnormal return of the non-announcing firms. The negative coefficient implies that for profit warnings issued due to industry-wide factors, larger firms are seen as leaders in the industry who have more power and influence over smaller firms. In homogeneous industries, there is a less negative reaction in non-announcing firms to the announcement of a profit warning. This provides evidence that, contrary to what we expected, homogeneous industries provide an environment conducive to competitive reactions. That is, rival firms in homogeneous industries react less negatively to profit warnings that are issued for industry-wide reasons than do firms in less homogeneous industries. The growth and Sentiment of the industry were found to be insignificant contributors to the cumulative abnormal return of the non-announcing firms for profit warnings issued due to industry-wide factors.

P a g e | 47

7.1.4. Conclusions regarding Firm Specific Sub-Sample Multivariate Regression For the firm specific information sub sample of profit warnings, the size of the announcing firms was a significant contributor to the cumulative abnormal return of the non-announcing firms. Graham and King () found that information is typically more readily available for large firms than information about smaller firms. Therefore, the positive coefficient for firm size of the announcing firms could imply that the information was already available to investors in announcing firms and investors in nonannouncing firms resulting in less negative abnormal returns in non-announcing firms. The average cumulative abnormal return of the announcing firms was not a significant contributor to the cumulative abnormal return of the non-announcing firms. This could be due to the nature of firm specific information, that bad news specific to the has no inference for the non-announcing firms. The homogeneous industry variable and the market sentiment variable are not significant contributors to the cumulative abnormal return of the non-announcing firms. This again could be due to the nature of firm specific information even in homogeneous industries and in all conditions of recent market sentiment.

P a g e | 47

7.2. Limitations of the Study The most significant limitation of this study is the scarcity with which profit warnings are issued in South Africa. This factor resulted in a very small Industry-wide Sub-sample30 of 13 profit warnings, after the filters were applied, the results of which may need further confirmation. Most previous studies that conducted similar tests used samples in excess of 100 announcements, however the analysis of the data generally provided results in line with our expectations. Although this study provides evidence of intra-industry information transfers, the negative average and cumulative average abnormal returns detected are relatively small compared to previous studies in the area of information transfers. A possible reason for this is due to the limitations of the industry definitions that were applied. These limitations include the level of aggregation at which some industries are defined and the diversification strategies adopted by many firms. The classification may have been too specific not allowed some competing companies’ returns to be compared, but this was done to ensure that only the returns of direct competitors are considered. This however could have reduced the magnitude of information transfers observed relative to those that would be observed using alternative industry definitions. Another limitation of this study stems from the difficulty in classifying some of the profit warnings into their respective groups, that is, those that convey industry-wide information and those that convey only firm specific information. In some profit warnings, the reasons given for the warning are somewhat ambiguous which made it difficult to clearly classify them which may have resulted in the incorrect classification of some profit warnings. Another possible limitation is that managers may attempt to blame their expected fall in profitability on industry-wide factors and hide firm specific problems from the market. That is, if a company’s profit warning contains only industry-wide reasons for the warning, but there are firm specific problems that are not mentioned, the negative reaction in the announcing firm may be attenuated.

The Average Abnormal Returns were found to follow a Normal type distribution, hence even with the small sample it was possible to apply hypothesis testing to the sample in order to evaluate the developed hypotheses. – See Appendix B 30

P a g e | 47

7.3. Suggestions for Further Research The results of this study have shown that profit warnings provide new information about competitors in the same industry. Refinements and extensions of the study are indicated to provide a better understanding about the information transfer effect. One possible improvement would be to address the limitation of the industry definitions mentioned above. For example a homogeneous line of business classification could also be used in order to refine this study – this is the approach that has been adopted by some previous studies on intra-industry information transfers. Firms could be grouped in terms of end product similarity and only those firms with at least 50 percent of their revenue from the same line of business could be included in the sample. This procedure will reduce the amount of diversification in each industry and thus increase the ability of the test to detect intra-industry information transfers. Different variables could also be included in the cross-sectional regressions to investigate what types of factors are useful in explaining the industry reactions. For example, the level of competition within an industry could explain, in part, the effects of a firm’s profit warning on other firms. A profit warning made by a firm in a highly competitive industry could cause the firm’s rivals to positively re-evaluate their competitive standings within the industry. Rival firms may then re-assess the announcing firm as a less prominent and less threatening industry competitor. Thus, the herfindahl index, a measure of the degree of industry concentration (lower herfindahl index implies a higher level of competition), could be included in the regression models to provide evidence of whether the level of competition within an industry is a significant explanatory factor of the effects of a profit warning on other firms. The state of the economy could impact the effects of the average abnormal returns, in that it may be possible to note greater impact on share prices when the economy is growing stronger. This report took this into account in the SENTIMENT variable in the multivariate regression, but the variable only focused on a period of 20 days, perhaps a longer period is required in order to take the effects of the economy into account, or an entirely new variable could be introduced. The decision of whether to disclose as well as whether to disclose accurately and fully may be influenced by many factors. Further research may also investigate the determinants of disclosure decisions. For example, managers rewarded with share options may have the incentive to release bad news prior to stock option award dates, thus pushing down the price

P a g e | 47

of the option. Aboody and Kaznik () provide evidence for this practice. The threat of legal action is also likely to have an impact on disclosure behaviour. It would be interesting to assess this determinant of the decision to disclose bad news in South Africa and compare it to the more litigious environment of the U.S. It is likely that the threat of litigation will have less of an influence on managers’ behaviour in South Africa than in the U.S. but further study needs to be performed in order to verify this.

P a g e | 61

Bibliography

P a g e | 62

Aboody, D. & Kaznik, R. (2000). CEO Stock Option Awards and The Timing of Corporate Voluntary Disclosures.

P a g e | 63

Journal of Accounting Economics , 29, 73 - 100.

P a g e | 64

Baginski, S. P. (1987). Intra-Industry Information Transfers Associated with Management Forecasts of Earnings.

P a g e | 65

Journal of Accounting Research , vol. 25, no. 2, 196.

P a g e | 66

Ball, R. & Brown, P. (1968, vol. 6 no. 2). An Empirical Evaluation of Accounting Income Numbers. The Journal of

P a g e | 67

Accounting Research , pp. 159-178.

P a g e | 68

Bodie, Z. Kane, A. & Marcus, A. (2005). Investments, Sixth Edition, International Edition. Mc Graw Hill.

P a g e | 69

Caton, G. L., Kohers, N. & Goh, J. (2003). Dividend Omissions and Intra-Industry Information Transfers . The

P a g e | 70

Journal of Financial Research , vol 26, no. 1, 51.

P a g e | 71

CFA Institute. (2008). Hypothesis Testing. In Ethical and Professional Standards and Quantitative Methods

P a g e | 72

(pp.455 - 495). Boston: Pearson Custom Publishing.

P a g e | 73

Collet, N. (2004). Reactions of the London Stock Exchange to Company Trading Statement Announcements.

P a g e | 74

Journal of Business, Finance and Accounting , vol. 31, no 1, 3-35.

P a g e | 75

P a g e | 76

P a g e | 77

P a g e | 78

P a g e | 79

P a g e | 80

P a g e | 81

P a g e | 82

P a g e | 83

P a g e | 84

P a g e | 85

P a g e | 86

P a g e | 87

P a g e | 88

P a g e | 89

P a g e | 90

P a g e | 91

P a g e | 92

P a g e | 93

P a g e | 94

P a g e | 95

P a g e | 96

P a g e | 97

P a g e | 98

P a g e | 99

P a g e | 100

P a g e | 101

P a g e | 102

P a g e | 103

P a g e | 104

P a g e | 105

P a g e | 106

P a g e | 107

Appendix A

P a g e |A.108

Appendix A: Profit Warning Classification A.1. Characteristics of Industry-Wide and Firm Specific Profit Warnings Table A1 broadly identifies the characteristics that have been used to classify profit warnings into the industry-wide and firm specific groups based on the reasons stated in the announcement released to the market. Table A. 1 Classification of Profit Warnings into Industry-Wide and Firm Specific Groups

Firm Specific Production & Quality problems Accounting Errors/Difficulties Litigation costs Delays to contracts and negotiations Loss of major customer Restructuring costs Acquisition/mergers costs Problems with contracts and negotiations Management and systems problems Fraud/Theft Repair/remedial works

Industry-Wide Difficult Market/Trading Conditions Regulation/Legislation Adverse weather conditions Sales short of forecasts

Appendix A

P a g e |A.109

Appendix A

P a g e |A.110

A.2. Examples of Profit Warnings and their Classification E.g. 1: On the 27th September 2000 Stella vista technologies limited released the Following statement: Announcement: trading statement and forecast warning The directors of Stella vista, listed in the development capital market sector of the Johannesburg stock exchange, take this opportunity to update shareholders on the current trading position. Transactions falling outside the financial year as announced on 28 July 2000 the company received an order valued at R24 million for final delivery in December 2000. This contract, which had taken almost two year to conclude, involves the delivery of 40 led displays for installation on service station forecourts around the country. As at Stella vista's year-end, 31 august 2000, some R3,9 million of this contract had been completed and invoiced. Production is currently on scheduled and delivery will be completed by due date. In august the company received an order for R18,5 million, for which the contract had been signed and the products completed and through the factory acceptance test by the year-end. However, the customer has experienced difficulty in arranging finance and, as payment has not been received in terms of the contract, Stella vista has withheld delivery and consider that it would not be prudent to account for the contract in the financial year ended 31 august 2000. Should the purchaser secure the finance necessary to conclude the transaction, shareholders will be advised and the transaction included in the current financial year. Forecast warning Had either of the above contracts been completed within the financial year ended 31 august 2000, the company would have exceeded forecast turnover and profit. However, as advised above, this did not happen. Accordingly, shareholders are advised that, while the company will achieve a profit for the year ended 31 august 2000, it will not achieve the profit as forecast in the prospectus.

Appendix A

P a g e |A.111

Current prospects The directors are satisfied that the company is well positioned to be successful in the current financial year ending 31 august 2001. The company has the balance of the forecourt display order, amounting to R20, 1 million, to complete and has secured orders for in excess of a further R3.7 million. Total orders on hand at the end of September 2000, the first month of the new financial year, therefore already exceed the turnover for the previous financial year. The R18.5 million orders may or may not be fulfilled, and the company is in the process of disposing of the inventory on hand or using it to fulfil other orders. The development of the new generation led display and control system will be completed during this calendar year. This should place Stella vista in the top five companies worldwide in term of this technology. In order for Stella vista to compete and be active in the largest market for Led displays, it has established an office in the United States during the year. It has already exhibited products at trade shows in Los Angeles and Florida. The international market for led displays as supplied by Stella vista is growing faster than was initially expected and the company expects positive results from its offshore operations in the near future. The audited preliminary results for the financial year ended 31 august 2000 will be released in mid-November and the financial statements distributed to shareholders shortly after that date. Discussion regarding example 1 This warning is firm specific as it addresses problems that are only relevant to Stella Vista namely the two uncompleted contracts. These problems effect Stella Vista only are we do not expect that it will affect the other companies in the industry.

Appendix A

P a g e |A.112

E.g. 2: On the 14th March 2000 Concor limited released the following profit warning: Profit warning Shareholders in Concor are advised that the group will report a loss for the six months ended 31 December 1999 despite a satisfactory performance from all core operations. This result is below forecast and market expectations and has arisen because of - continued losses on the Lesotho joint venture project; - Higher than expected closure costs of the electrical division; and - A valuation shortfall on a shopping centre in Wellington. Agreement has been reached with our international Lesotho joint venture partners, limiting the total loss, to the amount already provided by the group. The group's balance sheet remains strong and the provisions created for these losses have no immediate impact on the cash. The interim results for the six months ended 31 December 1999 will be published on 29 March 2000. By order of the board. Discussion regarding example 2 The above statement has been classified as firm specific which will lead to a decrease in profit for Concor. This classification is due to firm specific problems given: continued losses on the Lesotho joint venture project, higher than expected closure costs of the electrical division and a valuation shortfall on a shopping centre in Wellington. As such it is not expected that these factors will convey any information about other firms in the industry.

Appendix A

P a g e |A.113

E.g.3. On the 16th July 1999 Woolworths holdings limited released the following statement: Statement regarding profit The difficult economic conditions in the segment of the retail market in which the group's South African business, Woolworths, operates, together with a very mild winter, has impacted negatively on the profitability of Woolworths. We wish to confirm to shareholders the market expectation that headline earnings per share (HEPS) for the year ended 30 June 1999 will be lower than those for the comparable period last year. Further details will be announced with the release of the results for the year ended 30 June 1999 on 19 august 1999 and later when the annual financial statements are issued. Plans presently being implemented are expected to result in an improvement in trading performance in the forthcoming year. Discussion Regarding Example 3: The above profit warning has been classified as conveying industry-wide information in that the information it conveys is relevant to all companies in the retail industry. As such the difficult economic conditions and the mild winter will have a negative effect for all companies in the retail industry.

P a g e |B.114

Appendix B: Distribution of Average Abnormal Returns The following figures serve as evidence of Normal-like distributions of returns that have been recorded for 5 days prior to and five days after the Profit warning announcement is made. They show the distributions of abnormal returns for each day, for the full population of announcing firms (N = 54). Since the Full Population of announcing Firms exhibits a normal-like distribution of returns on each of the individual days, it is possible for one to apply t-statistic testing for significance on samples of the population even when the sample contains less than 30 values (). Please note – Frequency distributions of days t = -5, t = -1, t = 0, t = 1 and t = 5 are the only ones displayed in this appendix (in order to avoid unnecessary information), but all of the days exhibit the same general shape of frequency distribution.