The Effects of Enhanced Disclosure Requirements on Management Guidance Quality

Transcription

1 The Effects of Enhanced Disclosure Requirements on Management Guidance Quality Gregory S. Miller University of Michigan Nayana Reiter University of Michigan October 2015 Preliminary Please do not quote or circulate ABSTRACT We investigate the effects of requiring managers to provide contemporaneous explanations with forecasts and to ex-post provide an analyses of their guidance error by examining the impact of a 2010 change in Brazilian regulation that required both. We posit that these new requirements increase accuracy and credibility of management forecasts by affecting managerial accountability. That is, we expect managers to be less prone to disclose biased estimates when they have to contemporaneously state their forecasting assumptions, and also commit to later explain why their numbers were inaccurate. Consistent with our predictions, we find that forecasts become more accurate following the regulation change. In addition, investors appear to recognize this change in managerial incentives and react more to management forecasts. We also find that managers decrease the proportion of long term forecasts and shift from point to range forecasts. This suggests that the regulation had the unintended consequence of decreasing precision and horizon. We thank Ted Christensen, David Maber, Reuven Lehavy, Jed Neilson, Jairo Procianoy, Christina Synn, Lloyd Tanlu, Rodrigo Verdi, Hal White and participants at the AAA/Deloitte/J. Michael Cook Doctoral Consortium for their helpful comments and suggestions. We also thank Daniel Gildin and Andre Scholze for research assistance. Nayana Reiter thanks the University of Michigan and the Paton Fellowship for financial support. All remaining errors are our own.

2 1. Introduction This paper investigates the effect of requiring managers to provide both contemporaneous explanations in support of forecasts and ex-post analyses of their subsequent forecast errors. We use data from Brazil, where a 2010 regulation change required managers to provide both types of information. We posit that these new requirements reduce error and increase credibility of management forecasts by making managers more accountable for their forecasting mistakes. That is, we expect managers to be less prone to disclose biased estimates knowing that they have to state their forecasting assumptions and later explain why their numbers were inaccurate. We also expect investors to respond more fully to these forecasts. Consistent with our predictions, we find an increase in forecast accuracy following the regulation. Further, we show that investors response to forecast increases, suggesting they view the forecasts as more credible. While these findings suggest that the regulation was generally effective, we also find a decrease in long term forecasts and a move towards range rather than point forecasts, indicating some unintended consequences of regulation. Managerial guidance has long been an important source of information for market participants, allowing managers to disclose information about firm s future prospects and align investors expectations (Ajinkya and Gift, 1984). Prior research has found that management forecasts have information content and that market s reaction to forecasts varies across firms and disclosure s characteristics (e.g., Jennings 1987; Pownall et al. 1993). Even though forecast characteristics seem to be a key factor in the assessment of disclosure credibility, they are the least understood component of management forecasts in terms of theory and empirical research (Hirst et al. 2008). This is likely due to the fact that forecast characteristics are primarily choice variables, 2

3 making it hard to isolate their effect on forecast quality. In this paper we use a regulation change to help isolate the impact of the disclosure attributes. While management earnings forecasts are voluntary disclosures in most countries, they are impacted by regulatory oversight which can change the decisions managers make. For example, in 1979 and 1996 the SEC passed safe harbor provisions for US firms issuing forecasts to shield them from litigation related to forward-looking disclosures made in good faith. Despite concerns that this could reduce forecast quality, Johnson et al. (2001) document that the accuracy of earnings forecasts increases after Similarly, research shows that following the enactment of Reg FD, managers increase the frequency of public earnings forecasts (Bailey et al. 2003; Heflin et al. 2003). Evidence also suggests that managers changed forecasts precision in order to avoid missing their now public disclosed guidance (Ciconte et al. 2014). We study a Brazilian regulation which is focused on changing the information accompanying and explaining forecasts both via concurrent disclosure and ex-post analyses. These additional disclosure requirements may increase disclosure quality but may also impose higher disclosure costs on firms. As with other regulatory changes focused on forecasting, the managers in this situation face several choices. They can decide to stop public disclosure of guidance after the change in requirements. Alternatively, they can decide to keep disclosing, but change forecasts accuracy, horizon and/or precision. Our setting allows us to explore a discontinuity around the new regulation and look at changes in guidance frequency, characteristics and quality before and after the regulation. In 2010, the Comissão de Valores Mobiliários (Brazilian Securities Exchange Commission) - hereinafter CVM - changed the regulation regarding managerial guidance, requiring the disclosure of contemporaneous and ex-post explanations with forecasts. We use data from 2006 to 2012 to 3

4 explore the impacts of this change. We posit that this mandatory change in the information accompanying forecasts affects management predecision accountability, resulting in increased accuracy and credibility of such voluntary disclosures. Our main assumptions are that managers control (at least partially) the error in their forecasts, and that the act of justifying guidance can change managers behavior, either by inducing more careful forecasting processes or by increasing the reputational costs of being inaccurate, offsetting the economic benefits of opportunistically disclosing biased estimates. We first test whether the regulation had any effect on the quantity and characteristics of management forecasts being disclosed. As with the passage of Reg FD in the US, reduction of public disclosure is an important concern in this setting because higher disclosing costs may discourage some firms from providing guidance altogether. Our definition of management guidance follows the one used by the CVM and comprises quantitative prospective information provided by the company on its future performance (e.g., earnings, EBITDA, revenues, and margins estimates). We find that the proportion of firms disclosing at least one forecast a year has remained virtually unchanged after the regulation, with a similar number of firms stopping and starting forecasts after January However, firms that decided to continue forecasting have, on average, changed their forecasts characteristics by reducing the frequency of long-term forecasts and shifting from point to range initial forecasts. Thus, the evidence suggests that decreases in the choice of precision of forecasts and horizon are associated with the passage of the regulation. We also find that the frequency of confirming forecasts have increased after We next test whether the regulation affected forecast quality. We measure guidance error and market reactions to forecasts before and after the regulation. We find that management forecasts became more accurate after the regulation. This is true for both positive (i.e. overly 4

5 optimistic) and negative (i.e. overly pessimistic) errors, with the decrease in optimistic errors being larger. We also find that the association between forecast news and abnormal returns around guidance disclosure dates have increased. These results are consistent with the market viewing management guidance as more credible after the regulation. Further analyses indicate that both initial forecasts and revisions are more accurate and credible after the regulation. This suggests that managers are not only revising forecasts in an attempt to avoid missing their own guidance, they are also making initial forecasts that are more accurate. All else equal, initial forecasts that are more accurate are beneficial to investors, as market participants have access to higher quality information earlier in the period. We also find that both confirming and non-confirming revisions became more accurate after the regulation. This evidence is consistent with the idea that managers are providing more accurate initial forecasts and revisions, and appropriately confirming forecasts they believe are still valid. Despite finding some reduction in forecast horizon and the choice to move to less precise forecasting metrics, our evidence suggests the regulation did not impact the amount of forecasting being provided and had a positive impact on its accuracy in credibility. On balance, this suggests that the regulation seems to have positively affected the information being disclosed to market participants. In an attempt to shed some light on whether firms in other environments could experience similar effects on forecast quality by voluntarily following such requirements, we test for changes in accuracy and market reactions to forecasts after 2010 for voluntary versus mandatory compliers. Some firms were voluntarily complying with both of the requirements before the regulation change in 2010, following a recommendation made by the Advising Committee for Information Disclosure (CODIM) in The CODIM discusses and suggests best disclosure practices to 5

6 Brazilian firms, but these best practices are not mandatory and many firms do not follow it. Results of subsample tests suggest that mandatory compliers were affected by the regulation of 2010, with increases in accuracy and market reactions, but not voluntary compliers. Voluntary compliers seem to disclose more accurate forecasts both before and after the regulation. These results are consistent with firms that voluntary commit to best disclosure practices having more accurate and credible forecasts regardless of regulation. The Brazilian regulation seem to have affected mostly firms with lower guidance quality that would otherwise not commit to the disclosure of additional information. Our evidence also indicate that firms that were voluntarily following one of the guidelines but not the other (i.e. providing contemporaneous explanations and assumptions, but not explaining errors ex-post, or explaining errors ex-post but not providing contemporaneous explanations) experienced increases in accuracy after the regulation. This is consistent with the combination of both requirements affecting guidance quality. Our study contributes to the literature on voluntary disclosure and management incentives. Prior research has focused on how disclosure characteristics are related to investors reactions to voluntary prospective information. Empirical evidence suggests that investors consider management forecasts containing additional information to be more credible (Hutton et al. 2003, Baginski et al. 2004, Hirst et al. 2007). Yet, we do not have evidence on whether the commitment to provide additional information and ex-post explanations has a disciplinary effect, changing management forecasting behavior ex-ante and resulting in more accurate forecasts. Our study also contributes to the discussion on regulation and corporate disclosure, helping regulators to understand how managers respond to changes in mandatory guidelines for voluntary disclosure. It also helps managers and investors to recognize what types of disclosure commitments may increase the accuracy and credibility of management forecasts. 6

7 This paper proceeds as follows: section 2 discusses the motivation and setting; section 3 describes the sample and variables; section 4 presents the research design; section 5 shows the results; and section 6 concludes. 2. Motivation and Empirical Predictions Managers voluntarily disclose forecasts to reduce the information asymmetry between insiders and capital market agents (Verrecchia 2001). Lower information asymmetry is desirable because it is associated with higher liquidity (Diamond and Verrecchia 1991) and lower cost of capital (Leuz and Verrecchia 2000). Several empirical studies have found that voluntary management forecasts have information content (e.g., Patell 1976; Penman 1980, 1983; Waymire 1984; Ajinkya and Gift 1984; and Pownall and Waymire 1989) and that market s reaction to forecasts differs across firms and disclosure s characteristics (e.g., Jennings 1987; Pownall et al. 1993). Differences in market reactions result from the central premise of voluntary disclosure theory that says that any entity contemplating making a disclosure will only disclose information that is favorable to the entity. Therefore, in order to correctly interpret the statements of the entity making or not making a disclosure, one should anticipate the entity s incentives to behave in the preceding way (Dye 2001). Disclosure credibility concerns follow from the idea that a manager s reporting objective is not known precisely by the market, and therefore, capital market participants can only conjecture about manager s incentives to inflate or deflate expectations (Fischer and Verrecchia 2000). In other words, investors cannot completely distinguish random forecast errors from deliberate management bias. Prior research discusses six forces that affect managers disclosure decisions: capital market transactions, corporate control contests, stock compensation, litigation, proprietary costs, 7

8 and management talent signaling (Healy and Palepu 2001). Thus, key factors that investors seem to consider when assessing disclosure credibility include situational incentives at the time of the disclosure, levels of external and internal assurance, management s credibility, and characteristics of the disclosure itself (Mercer 2004). Since disclosure characteristics appear to be an important factor in the assessment of management incentives and reporting credibility, we begin our empirical investigation by verifying whether a mandatory change in forecast characteristics affects managers incentives. We utilize changes in guidance accuracy (absolute error) and credibility (abnormal returns) as proxies for changes in management forecasting decisions. Managers have incentives to both optimistically and pessimistically bias their forecasts. Motivations to inflate guidance include psychological factors involving the tendency individuals have to overestimate their ability relative to average (Hribar and Yang 2015). Hribar and Yang (2015) examine the effect of overconfidence on management forecasting behavior and find that overconfident managers are more likely to issue optimistically biased forecasts. Cognitive biases, such as overconfidence, can be attenuated by accountability towards one s decisions. Accountability refers to the implicit or explicit expectation that one may be requested to justify one s beliefs, feelings and actions to others (Lerner and Tetlock 1999). The predecision knowledge of future accountability may stimulate more self-critical and effortful thinking (Lerner and Tetlock 1999). In other words, when managers expect to justify their guidance, they prepare themselves by making more careful forecasts. The issuance of overly optimistic guidance can also be motivated by economic factors. Contractual incentives for productive effort help to align incentives of owners and management but may reduce the credibility of voluntary disclosures (Gigler and Hemmer 1998). Managers may inflate forecasted numbers when their wealth depends on the firm s stock price, as it is the case 8

9 when executives have stock-based compensation or face a takeover threat (Hutton et al. 2003). In addition, higher current market valuations may be important because of the possibility that the firm intends to issue additional equity to finance future operations, or as currency in stock-swaps (Verrecchia 2001). Another interesting rationale to consider is that increasing stock price may simply be a heuristic behavior on the manager s part. For instance, managers may maximize current market capitalization because they have been conditioned to believe that they are truly being evaluated based on this benchmark, regardless of what it is in their contracts (Verrecchia 2001). On the other hand, managers may deflate projections to satisfy a desire to beat market s expectations (Matsumoto 2002). In this scenario, managers voluntarily disclose pessimistic news to minimize current firm price, partially to ensure positive price reactions to future mandatory disclosures, and partially to minimize the liability associated with withholding information that negatively affects stock price (Skinner 1994). Also, a lower stock price in the near term can be beneficial to cover employee stock options. Similarly, managers can benefit by driving down price if they intend to engage in a management buyout or if they are about to receive a new option grant (Fischer and Verrecchia 2000; Aboody and Kasznik 2000). Legal and regulatory environments can influence forecast frequency, characteristics and bias. For example, empirical studies that investigate the effects of Reg FD in the US setting show an impact on management forecasting behavior. Before the passage of Reg FD, managers could privately provide guidance to analysts. Following Reg FD, managers face a choice between public disclosing forecasts and not disclosing them at all. Empirical evidence indicates an increase in public earnings forecasts following Reg FD (Bailey et al. 2003; Heflin et al. 2003). Although managers did not decrease the disclosure of forecasts, evidence suggests that they changed 9

10 forecasts characteristics in order to avoid missing their now public disclosed guidance. Ciconte et al. (2014) find an increase in the frequency of range earnings forecasts from an average of 46.9% in the period to 80.5% in the period. They also find that in the period actual earnings hit the upper bound of range forecasts more frequently. This is consistent with managers loss functions regarding earnings surprises becoming increasingly asymmetric and with managers true earnings expectations being close to the upper bound of range forecasts in the post Reg FD period. Evidence from Japan, where management forecasts are effectively mandated and relatively unconstrained by legal or regulatory forces, suggest that the fact that investors can ex-post observe management forecast errors is insufficient to ensure unbiased guidance ex-ante. Kato et al. (2009) find that managers in Japan tend to provide forecasts that are biased upwards and that this optimism is persistent. The regulation change we study in this paper keeps guidance voluntary but requests firms to provide more information to investors and analysts, by disclosing key assumptions and explaining errors. This additional information may be valued by investors since it can improve overall understanding. A 2015 perception study by Corbin Perception with 165 US institutional investors and analysts and 145 IR professionals indicates that 92% of investors and analysts find it helpful when managers provide forecasts key assumptions during earnings calls, but only 25% of IR professionals report their companies adhere to this practice. The enhanced understanding provided by additional information may increase reporting credibility in a setting where forecasts are voluntary. Reputation and reporting credibility costs can offset managers incentives to bias projections. Lee et al. (2012) examine the relation between management forecast accuracy and CEO turnover and find that the probability of CEO turnover is positively related to absolute 10

11 forecast errors when firm performance is poor and that this positive relation holds for both positive and negative forecast errors. Their results suggest that managers bear a cost for issuing inaccurate forecasts. We posit that by increasing management accountability, the regulation increased management costs for issuing inaccurate forecasts. Accountability requires negative consequences for people who cannot provide satisfactory explanations for why they do what they do (Lerner and Tetlock 1999). In this framework, managers may suffer reputation and reporting credibility losses if they fail to satisfactorily explain their estimates. According to Trueman (1986), management forecasts can be viewed as a signal of manager s abilities to anticipate future changes in firm s environment and adjust production accordingly. One consequence of investors analysis of the disclosed information is their reassessment of managerial talent (Nagar 1999). We argue that the ex-ante knowledge that managers will have to publicly justify their guidance increases their concerns about making inaccurate projections. Managers do not want the market to perceive them to be either incapable of making accurate projections or dishonest. The enhanced disclosure requirements make the quality of the forecast more easily assessed, increasing reputation and credibility costs of making inaccurate projections. In other words, it becomes harder to mislead investors. Penalties for providing detectable opportunistically biased forecasts include losses in firm s credibility and in manager s future career prospects. The increase in reputation and disclosure credibility costs partially offsets the benefits of biasing estimates, resulting in more accurate forecasts. Requirement to Disclose Contemporaneous Explanations and Assumptions Several studies have documented that one disclosure characteristic that affects guidance credibility is the provision of additional information with the forecast. According to Baginski et 11

12 al. (2004), managers provide attributions with forecasts by linking forecasted performance to their internal actions and the actions of parties external to the firm. Their results suggest that forecasts containing attributions are more informative than standalone forecasts. The idea is that this additional information helps investors to interpret management forecasts by increasing their understanding of the determinants of firm performance (Baginski et al. 2004). Hutton, Miller and Skinner (2003) find that managers tend to supplement good news forecasts with information that is ex-post verifiable. They also find that investors react to good news forecasts only when the guidance is accompanied by supporting information, such as forecasts of other items of the income statement, suggesting that this information enhances the credibility of the forecast. Similarly, the results of experimental evidence corroborates archival results and indicates that earnings forecasts that included other key line items are judged to be more credible than earnings forecasts disclosed alone (Hirst et al. 2007). In summary, prior research has found that investors consider management forecast containing additional information to be more credible. Prior research also indicates that managers are more likely to bias their forecasts when it is more difficult for investors to detect that they have misstated their projections (Rogers and Stocken 2005). However, prior research does not provide a full understanding of how the provision of supporting information shapes management behavior. We posit that the provision of forecast attributions increases the quality of the disclosure, promoting the understanding of how managers elaborate the estimates. When investors understand how managers make forecasts, it becomes easier to detect biased estimates and harder to later explain discrepancies resulting from serf-serving managerial biases. Consequently, the requirement to disclose supporting information makes managers more accountable for their forecasts. This increased accountability motivates more accurate and credible managerial 12

13 guidance, either by attenuating cognitive biases or by increasing the reputational costs of intentionally disclosing biased estimates. Requirement to Explain Errors Another requirement that can increase the accuracy and credibility of voluntary disclosure is the commitment to ex-post justify deviations. Accountability research finds that predecision commitment improves judgment under uncertainty by increasing self-critical forms of thought (Lerner and Tetlock 1999). As a result of this increase in cognitive effort, managers may be able to more accurately estimate future results. In addition, the commitment to provide explanations should restrain management s incentives to inflate or deflate the forecast, as managers know that they will have to subsequently explain their bias (Miller 2009). The potential failure to provide satisfactory justification for managers actions can increase reputational and disclosure credibility costs associated with biased forecasts. Therefore, we should expect the commitment to explain discrepancies to counteract managers incentives to misrepresent forecasts, increasing the accuracy of the disclosure. Here it is important to note the distinction between a commitment and a voluntary disclosure. A commitment is a disclosure decision made before knowing the content of the information (ex-ante), whereas a voluntary disclosure is a decision made after the firm observes the information to be disclosed (ex-post) (Leuz and Verrecchia 2000). Hence, firms deciding to disclose forecasts after the change in regulation in 2010 are committing to disclose comparisons and explanations before knowing exactly how their forecasted numbers will compare to actual results. In summary, we posit that management prospective disclosure requirements may increase managers accountability for firms forecasts. A setting where the consequences of a mandatory 13

14 change in provision of attributions and commitment to explanations can be assessed is the Brazilian stock market Instruction CVM 480 In December 2009, CVM released Instruction 480, which requires that if the issuer decides to disclose forecasts, they should: I be included, with all the additional information required, in a document called Reference Form, which is disclosed annually, revised every time relevant information in the form changes, and easily accessible from the Sao Paulo Exchange (Bovespa) website; II be disclosed with assumptions, indicating which can be influenced by management and which are beyond its control, parameters, methodology adopted, period and the duration of the forecast; III - be reviewed periodically, in a time interval that cannot exceed one year; IV - the issuer shall, quarterly, in its interim report, confront the forecasts and the actual results for the quarter explaining the reasons for any differences; VI - in the case the issuer has disclosed any forecast in the last 3 fiscal years, the issuer must compare in the reference form the forecast with the actual performance indicators, clearly indicating the reasons for differences. Regarding forecasts for periods still ongoing, state whether the projections are still valid and, if applicable, explain why they were abandoned or replaced; V - modification of forecasts is considered material fact and the firm has to file a form with CVM (similar to an 8-K) and disclose this information broadly to all investors - usually through the company s website and mailing list. 14

15 The previous Brazilian regulation only required firms issuing new equity to disclose forecasts assumptions in their prospectus, and ex-post explain any differences between projected and actual numbers. After 2010, any guidance disclosed at any time must follow these requirements. Conversations with CVM officials suggest that the main goal of the regulation was to make new equity offerings easier for firms, through a shelf registration process. It is important to note that Instruction 480 applies to both earnings forecasts and operating forecasts. The definition of management guidance used in this study follows the one used by the CVM and comprises any quantitative prospective information provided by the company on its future performance (e.g., earnings, EBITDA, revenues, and margins estimates). Examples of forecast disclosures made before and after the regulation are provided in the appendix. We contacted the CVM to better understand how they monitor firms compliance. We were told that verification of compliance can occur in three ways: (i) receipt of complaints; (ii) preventive analysis of the reference form that are executed in accordance to CVM s risk-based supervision biennial plans; and (iii) Bovespa s daily monitoring of news in information vehicles and in documents disclosed by the companies. The penalties for violation of the rules regulated by the CVM range from a warning to a ban on operating in the securities market, a fine is also a possible penalty. The penalties are usually applied to the director of investor relations and the chief executive officer. Despite the fact that CVM s monitoring process seems to be relatively careful, during the hand collection process we found variation in the way firms comply with the regulation. In the case that the regulation is not being adequately enforced, the mandatory new requirements could be viewed as still voluntary in some way. However, even if this is the case, the regulatory and reputational costs of refuting to comply with the new regulation could still be significant enough to change the behavior of firms for which the previous benefit of not providing guidance 15

16 explanations only slightly exceeded the costs. Finding average results in this setting may suggest that even stronger results could be found in environments with stronger enforcement Instruction 480 and Management Forecast Bias: Predictions Given the voluntary and non-audited nature of management guidance, accuracy and credibility are issues for these disclosures. The Brazilian setting allows us to compare the same group of firms before and after a mandatory change in management forecast characteristics, providing a clearer test for an increase in disclosure accuracy and credibility. An important assumption of this paper is that managers control (at least partially) the error in their forecasts. Once managers decide to provide a forecast, they can strive to achieve accurate forecasts or they can strategically forecast to obtain a desired outcome (Hirst et al. 2008). Thus, their forecast errors can be intentional, resulting from economic incentives, or unintentional, resulting from cognitive biases. In both scenarios, it is reasonable to expect the requirements to affect forecast accuracy. Managers concerns about their reputation and the credibility of their disclosures are likely strong enough to partially offset the motivations to disclosed inaccurate guidance. It is plausible to assume that, even before the regulation, managers were questioned about their forecasts assumptions and deviations by analysts and investors. However, this inquiry was likely verbal and not necessarily public. Managers may have chosen not to respond or even argued that regulation limited their ability to provide further insights. After the regulation, managers have to disclose comparisons between the forecast and the actual, clearly indicating the reasons for deviations in the estimates. The disclosure of such comparison has to occur in three different public documents: the earnings announcement, the annual report, and the reference form. These written 16

17 communication channels may be subject to more scrutiny, contributing to an increase in management accountability. Also, the reference form must contain comparisons of forecasts and actual numbers for the last three years. Therefore, it becomes easier for investors to assess the prior accuracy of management forecasts. The obligation to compare the last three years of estimates can also impose more consistency on firms guidance policy. In addition, if a firm changes significantly because of, for example, an acquisition, managers must now provide a comparable number to the previous forecasted item. For instance, in the case of an EBITDA forecast, managers have to disclose what the EBITDA would have been if the acquiring firm had not acquired the other firm, so that investors can assess how the numbers actually diverged. The implicit rationale behind Instruction 480 is that managers should become more accountable for the guidance they provide. The increased accountability should reduce cognitive biases and increase the reputational and disclosure credibility costs of providing inaccurate estimates. In order to test the joint effects of Instruction 480 s requirements in the accuracy of management guidance, we predict the following: P1: management forecasts are more accurate in the period posterior to the regulation than in the period prior to the regulation. P2: management forecasts optimistic errors in the period posterior to the regulation are smaller than management forecasts optimistic errors in the period prior the regulation. Both predictions assume that biasing costs increase after the regulation. P1 posits that both positive and negative errors are reduced, while P2 accounts for a reduction in the average signed error (i.e. errors become less optimistic). 17

18 We also test for the effects of Instruction 480 on forecast credibility. The effects on the perceived credibility of management forecasts may differ from the actual decrease in bias. Assuming investors recognize the change in management incentives and react more to the disclosure of guidance, we predict the following: P3: management forecasts are more credible in the period posterior to the regulation than in the period prior the regulation. There are at least two reasons why we might not find results consistent with our predictions. First, increased accountability may prompt managers cognitive effort toward self-justification rather than self-criticism (Lerner and Tetlock 1999). In this case, managers may focus their mental energy on defensively rationalizing past actions instead of trying to make more accurate predictions. Second, we assume managers do not lack the necessary abilities to make accurate predictions. They only lack the motivation to do so. Therefore, an increase in the cost of disclosing biased estimates will likely result in an increase in accuracy. However, even when motivated to do so, managers may not able to make more accurate forecasts. The aforementioned reasons provide tension for our study and motivate us to empirically test our predictions. 3. Sample and Variable Definition 3.1. Sample Our sample comprises all Brazilian firms with shares trading at Bovespa that filed at least one annual report with the CVM during our sample period. Our period of analysis goes from 2006 to 2012, four years before and three years after the regulation change in We hand collected information about management forecasts from firms disclosure documents available on the CVM website. The documents include material facts, earnings announcements, press releases, corporate 18

19 presentations, annual reports and reference forms. We consider firms disclosing prospective information in the terms established by Instruction 480 to be forecasting firms, and we include not only forecasts of earnings, but also revenues, EBITDA and margins in our sample. We start our hand collection process with a list of firms that had filed at least one annual report with the CVM from 2006 to This list contained 1,027 firms. Out of this group, 578 firms had shares outstanding trading at Bovespa. 1 Our hand collection process revealed that out of 578 trading firms, 83 firms disclosed 904 forecasts during the period. We consider each forecasted item (number) to be a different observation. For example, some firms disclose forecasts for more than one period in one document. Each item constitutes a separate observation in our sample. Table 1 Panel A presents our sample selection process. Since the regulation may work as a screening device, we perform our accuracy and credibility tests using only firms that continued to disclose forecasts after the change, excluding starters and stoppers. Evidence that forecast accuracy and credibility for these firms changed mitigates the problem of comparing samples of firms with different characteristics. After excluding 24 starting firms (144 forecasts) and 22 stopping firms (95 forecasts), our final sample comprises 661 forecasts of 37 firms. It is important to keep in mind that the Brazilian capital market is much smaller than the U.S. capital market, with an average of 366 firms. Having a small setting allowed us to hand collect the sample of forecasts, checking all firms documents filed with the Brazilian SEC during the period, and to complement Datastream data with hand collected financial and stock price data. Therefore our sample is not subject to database coverage biases and closely represents the set of Brazilian firms directly 1 It is common in Brazil for firms to be listed, but not have any actual free float. State owned pension funds or other large shareholders own sizable stakes in these firms, but do not trade shares on the open market. These firms are excluded from our sample. 19

20 affected by the regulation. Table 1 Panel B indicates that proportion of firms disclosing forecasts has increased from 5% in 2006 to 11% in 2008 and remained around 10-11% after that. 2 Table 2 provides descriptive statistics on continuing, starting and stopping forecasters. We can observe that continuing, stopping and starting firms tend to be different in some dimensions: continuing firms have higher analyst following, ROA, proportion of years with stock issuance, and proportion of ADR firms than both stopping and starting firms. Thus, continuing firms seem to have better information environments and more incentives to disclose (more stock issuance and better performance). We believe that using only the sample of continuing firms in our tests for changes in accuracy and market reactions is indeed more appropriate than including all forecasting firms. Table 3 indicates that most forecasts in our sample are range forecasts. Regarding horizon, annual forecasts tend to be more common. Quarterly forecasts are not very common among Brazilian firms. However, firms make revisions of their annual forecasts during the year. We consider subsequent revisions to be either semi-annual or quarterly forecasts, depending on how many quarterly results are already disclosed for that year. Most forecasts are disclosed within the earnings announcement and are disaggregated, such as EBITDA forecasts disclosed with sales forecasts. Management EPS forecasts are rare. Table 3 indicates that the most forecasted income statement items are sales and EBITDA numbers and margins. We do not restrict our sample to earnings forecasts for two related reasons: (i) earnings forecasts are not frequent in this setting and may not represent accurately the effect of the regulation in management incentives, and (ii) firms may be responding to investor and/or analyst demand by mainly disclosing sales and EBITDA 2 While we understand this is a small sample, it does represent the universe of firms impacted by the regulation change. The small sample may impact our power and our ability to provide some cross-sectional tests. However, to the extent we are able to provide statistical results we see no reason that the results should be viewed as biased. 20

21 numbers. This idea would be consistent with the findings in Ng et al. (2014). Using a sample of management forecasts disclosed by firms in 30 countries during , Ng et al. (2014) find that absolute abnormal returns around guidance disclosures are higher for forecast of sales and other line-items above bottom-line net income than for forecasts of net income Variable Definition Our dependent variables are the accuracy (ACC) and credibility (CAR) of management forecasts. We follow Kato et al. (2009) and compute forecast error as the variation of the forecast in comparison to the actual, i.e. forecast minus actual scaled by lagged total assets. Therefore, a positive error means the forecast is greater than the actual and management was overly optimistic, and a negative error means that the forecast is less than the actual and management was overly pessimistic. For margins, we use the percent point difference between the forecasted margin and the actual margin. We compute accuracy (ACC) as the absolute value of error multiplied by minus one. Thus, an increase in accuracy translates into a reduction absolute error. It is important to measure the change in accuracy because the regulation may have motivated a reduction in the average signed error, but not necessarily a reduction in absolute errors. For example, if errors are on average positive (i.e. optimistic), an increase in negative errors (i.e. pessimistic) will reduce average signed errors without necessarily improving overall accuracy. For range forecasts, we use the interval mean value to compute ACC. We compute the credibility of a forecast (CAR) as the market-adjusted stock return cumulated over the three trading days around the forecast release date (days -1 to +1), using the Brazilian Ibovespa index to measure market returns. For the accuracy tests, the independent variable of interest is POST, an indicator variable that equals 1 if the forecast is disclosed after January 1, For the abnormal returns 21

22 test, the variable of interest is the interaction of POST and NEWS. NEWS is forecast minus the prior actual in case of an initial forecast, or the forecast minus the prior management forecast in case of a revision, scaled by lagged total assets. Limited analyst coverage of Brazilian firms according to I/B/E/S prevents us from using analyst forecasts in our NEWS calculations. Control Variables To ensure that our results are not attributable to other firm characteristics, we incorporate several control variables, including characteristics of the forecast (precision, horizon, and revision), firm performance, and additional firm characteristics that prior studies find to be associated with forecast accuracy and credibility. Since the majority of forecasts are annual, our control variables are fiscal year-end values. LMV is the natural log of the market value of equity. ROA is the ratio of net income over total assets. LOSS is an indicator variable for a loss in the year of the disclosure. NINST is the number of institutional investors following the firm according to the firm s annual report 3. INTANGIBLES is the intangible assets over total assets ratio. DISAGGREGATED is an indicator variable that equals 1 if the forecast was disclosed with other income statement line items forecasts. CONFIRMING is an indicator variable that equals 1 if NEWS equal zero. BAD is an indicator variable that equals 1 if NEWS is negative. QUARTER is an indicator variable that equals 1 if the horizon of the forecast is a quarter. SEMI-ANNUAL is an indicator variable that equals 1 if the horizon of the forecast is either two or three quarters. LONG TERM is an indicator variable that equals 1 if the horizon of the forecast is greater than one year. FINAL is an indicator variable that equals 1 if the forecast is the last management forecast for that period, i.e., no revisions were disclosed afterwards. DEARN is an indicator variable that equals 1 3 Brazilian firms are required to disclose the number of institutional investors in their annual reports. We hand collected this information for our sample of forecasting firms. 22

23 if the forecast was disclosed with the earnings announcement. RANGE is an indicator variable that equals 1 if the forecast is disclosed as an interval. PRIOR ACC is the average accuracy of all prior forecasts disclosed by the firm. The continuous independent and dependent variables are winsorized at the 1 st and 99 th percentiles. 4. Research Design Ours research design consists of pretest and posttest sample differences. First, we investigate whether the regulation change is associated with changes in forecasts frequency and characteristics (number of forecasts, precision, horizon, number of revisions, bundling, and sign of the news). These tests are important because the regulation may discourage some firms from disclosing forecasts. It may also motivate managers to change their forecasts characteristics. For example, managers may decide to reduce the horizon and precision of their forecasts to avoid large forecast errors. Then, we examine the effect of Instruction 480 on management forecast accuracy and credibility estimating the following OLS regressions using only continuing forecast firms: ACC i,t = β 0 + β 1 Post t + β k Control Variables i,t (1) + β j Post t Control Variables + Firm FE i + Year FE t + ε i,t i,t CAR i,t = β 0 + β 1 Post t + β 2 Post t News i,t + β k Control Variables i,t +Industry FE i + Year FE t + ε i,t (2) We define continuing firms as firms that have disclosed at least one forecast both in the pre and post periods. ACC, CAR, Post and the control variables are as defined previously. For 23

24 regression 1 a positive coefficient on Post indicates an increase in forecast accuracy. For regression 2, a positive coefficient on Post X News indicates an increase in market reactions to forecast news after the regulation. We also perform three cross-sectional tests. To investigate whether the regulation had an effect on initial forecasts, its revisions, or both, we perform seemingly unrelated regressions (SUR) for first forecasts and revisions. To determine whether confirming forecasts also presented increases in accuracy and credibility, we perform SUR for confirming and non-confirming forecasts. To test whether voluntary compliers also presented increases in accuracy and credibility, we perform SUR for firms that were voluntarily following the guidelines prior to 2010 and firms that were not. 5. Results Forecast Characteristics Table 3 presents the proportion of forecasts with each characteristic disclosed before and after the regulation for the full sample of forecasts, which includes firms that stopped and started forecasting after 2010, and for the sample of continuing forecasters. The total number of forecasts for the full sample is 513 in the pre period and 391 in the post period. It is worth noting that we have four years in our pre period and three years in our post period. The average number of forecasts per year is similar in the pre and post periods (128.2 versus 130.3). These results are consistent with firms that stopped disclosing forecasts being substituted for firms that started forecasting after For the sample of continuing forecasters, there is small increase in the average number of forecasts per year, from 92.2 to The average number of revisions per firm- 24

25 year remained constant in the pre and post periods for both samples, around 2.3 revisions for continuing firms and around 2 revisions for all firms. Regarding forecast horizon changes, firms from both samples have reduced the proportion of long-term forecasts (i.e. longer than an year) and increased the proportion of semi-annual forecasts. The increase in semi-annual forecasts is likely due to an increase in the number of revisions. Firms in our sample usually disclose annual forecasts at the beginning of the year and revise them throughout the year. We code revisions that have horizons of two or three quarters as semi-annual forecasts. Regarding forecast precision, continuing forecasters reduced the proportion of point forecasts and increased the proportion of range forecasts for first forecasts. In the full sample of forecasts, we observe an increase in open-ended forecasts (minimum or maximum). This evidence is consistent with a decrease in precision after the regulation. Regarding bundling decisions, most forecasts are disclosed with earnings and continuing forecasters have increased the proportion of guidance disclosed within earnings announcements from 51.2% in the pre period to 61.6% in the post period. Firms from both samples have drastically reduced the proportion of forecasts disclosed with more than one horizon. This is likely related to the fact that firms reduced long-term forecasts. As most Brazilian firms disclose annual forecasts, forecasts for more than one horizon usually imply disclosing an annual forecast bundled with a long-term forecast. Most forecasts in our sample are disaggregated, meaning that more than one income statement item is forecasted and disclosed in one document. Regarding forecast news, most forecasts in our sample are good news forecasts. However, the proportion of good news forecasts have decreased after the regulation, from 55.8% to 45.9%, with an increase in confirming forecasts from 20.3% to 35.3% for continuing forecasters. We 25

26 consider confirming forecasts to be disclosures where the firm, usually after an earnings announcement, reiterates its commitment to a forecast disclosed earlier in the year. Evidence from Clement et al. (2003) suggests that management forecasts that confirm analysts expectations reduce uncertainty about future earnings and that investors price this reduction of uncertainty. We believe that the same idea can apply to management forecasts that confirm prior management guidance, and therefore we include these forecasts in our sample. Regarding the items being forecasted, continuing firms have increased the proportion of sales forecasts and decreased the proportion of margins forecasts. In summary, the evidence in Table 3 is consistent with continuing firms changing the characteristics of their forecasts after 2010, shifting from long-term forecasts to semi-annual revisions and from point to range initial forecasts. We also observe an increase in the proportion of confirming forecasts. Forecast Accuracy Figure 1 demonstrates how the distribution of management guidance accuracy changes from the pre to the post regulation period. We can observe that the errors in the post period are smaller and more concentrated around zero. Table 4 presents the results for regressions on management forecast accuracy before and after Instruction 480, controlling for firm and forecast characteristics. The results are consistent with our predictions. The positive and statistically coefficient on Post indicates that management forecast accuracy has increased after the regulation. The average forecast error in our sample is positive, meaning that managers were on average overly optimistic in their guidance with forecasts that are larger than actuals. Results from the regressions using the subsamples of positive and negative errors indicate that the increase in accuracy is coming from a reduction in both positive (i.e. overly optimistic) and negative (i.e. overly pessimistic) errors. However, the increase in accuracy for positive errors is larger. Increased 26

27 incentives to meet or beat guidance due to reputation and career concerns could be responsible for this result. This evidence is consistent with the idea that the requirement to explain guidance has motivated more accurate and less biased management forecasts, even after controlling for several factors potentially influencing forecast errors, such as horizon and precision. The results also indicate that quarterly and semi-annual forecasts are more accurate and long-term forecasts are less accurate. This is consistent with the idea that the inaccuracy of an estimate increases with its horizon. Range forecasts are also positively associated with forecast accuracy. Untabulated tests using the top and bottom of the interval indicate that our results are not affected by our choice to use the mid-point of range forecasts to calculate forecast errors, and accuracy is significantly and positively associated with our Post variable in the three specifications. In order to determine whether the regulation is associated with an increase in forecast accuracy for initial forecasts, its subsequent revisions or both, we estimate seemingly unrelated regressions using the subsamples of first forecasts and revisions. Table 5 presents the results. We find a significant and positive association for both subsamples and failed to reject the hypothesis that the association for first forecasts is equal to the association for revisions. These results suggest that managers are not only revising forecasts in an attempt to avoid missing their own guidance, they are also making initial forecasts that are more accurate. From the investors perspective this is a positive effect, as managers make more accurate initial forecasts in the beginning of the year and also more accurate revisions in the subsequent quarters. Forecast Credibility 27

28 For our tests on abnormal returns around guidance disclosure dates we restrict our sample to stand-alone forecasts, i.e. forecasts that are not bundled with earnings announcements. We perform our tests with a three-day event window centered on the guidance disclosure date (from day -1 to day +1). Table 6 presents the results. We find that, after controlling for the magnitude of forecast news, market reactions to guidance news are larger after Instruction 480, as indicated by the coefficient on the interaction of Post and News. Our evidence also suggests that prior accuracy and disaggregation are positively associated with abnormal returns 4. The abnormal returns around semi-annual forecasts, which tend to be initial forecasts or first revisions, are also higher. Results for the subsamples of first forecasts and revisions, although statistically weaker, indicate that the higher market reactions in the post period are observed in the two groups. This result is consistent with our finding on increased accuracy for both initial forecasts and its revisions. Confirming Forecasts We observe an increase in the proportion of confirming forecasts after the regulation. In an attempt to investigate whether this is a positive or neutral effect of the regulation, we perform our accuracy and abnormal returns test on the subsample of confirming and non-confirming forecasts. Table 7 presents the results of our accuracy tests for confirming and non-confirming revisions. We observe an increase in accuracy for both subsamples and fail to reject the hypothesis that the coefficient on Post for confirming revisions is equal to the coefficient on Post for non-confirming revisions. This evidence is consistent with the idea that managers are providing more accurate initial forecasts and revisions, and appropriately confirming the forecasts they believe are still valid. Table 7 presents the results for our market reactions test. There is an increase in market 4 Untabulated tests with Brazilian companies cross-listed in the U.S. indicate that the results hold for firms exposed to the U.S. capital markets environment, partially alleviating external validity concerns. 28

29 reactions to non-confirming forecasts, but not to confirming forecasts. Contrary to our prediction, confirming forecasts seem to have experienced a decrease in market reaction after the regulation. However, it is important to keep in mind that the majority of confirming forecasts are disclosed with earnings announcement, and were therefore excluded from our abnormal returns test. Voluntary versus Mandatory Compliers Some firms were voluntarily complying with one or both of the main requirements before the regulation change in In 2008, the Advising Committee for Information Disclosure (CODIM) released a document proposing some guidelines for management guidance. The CODIM is a committee formed by members of the Brazilian Institute of Investor Relations, the Brazilian Institute for Corporate Governance, the Brazilian Association of Equity Analysts, the Sao Paulo Exchange, the Brazilian Association of Pension Funds, and the Brazilian Institute of Independent Auditors. The committee discusses and suggests best disclosure practices to Brazilian firms, but these best practices are not mandatory and many Brazilian firms do not follow it. The guidelines suggested by the CODIM in 2008 were very similar to those later required by the regulation. We test for the changes in accuracy and market reactions for voluntary and mandatory compliers. Table 9 presents the results. Our evidence suggest that mandatory compliers were affected by the regulation change of 2010, with increases in accuracy and market reactions. We do not observe significant changes for firms voluntarily following both guidelines. Voluntary compliers seem to disclose more accurate forecasts both before and after the regulation. These results are consistent with firms that voluntary commit to best disclosure practices having more accurate and credible forecasts regardless of regulation. The Brazilian regulation seem to have affected mostly firms with lower guidance quality that would otherwise not follow the guidelines. Our evidence also indicate that firms that were voluntarily following one of the guidelines but not the other (i.e. 29

30 providing assumptions ex-ante, but not explaining errors ex-post, or explaining errors ex-post but not providing assumptions ex-ante) experienced increases in accuracy after the regulation. This is consistent with the combination of both requirements affecting guidance quality. Sensitivity Test The financial crisis of 2008 could have affected forecast accuracy and credibility in the pre regulation period because of increased uncertainty. To alleviate this validity concern, we reestimate our regressions excluding the year of Untabulated tests indicate that our results hold in this subsample, and that guidance accuracy and credibility are higher after Instruction Conclusions We investigate the effects of requiring managers to provide contemporaneous explanations with forecasts and ex-post analyses of their guidance errors. Consistent with our predictions, we find that forecasts become more accurate following the regulation change. Investors appear to recognize this change in managerial incentives and react more to management forecasts. Our results suggest that requiring managers to provide contemporaneous and ex-post explanations decreases managers incentives to bias their forecasts, resulting in more accurate and credible guidance. Cross-sectional tests indicate that the regulation affected firms that were not voluntarily following the guidelines and had lower guidance accuracy. Since we do not find evidence consistent with a reduction in the quantity of forecasts being disclosed and our results suggest an increase on the overall quality of guidance, the regulation seem to have positively affected the information being disclosed to market participants. However, we also find that managers decrease 30

31 the proportion of long term forecasts and shift from point to range forecasts. This suggests that regulation also had the unintended consequence of decreasing precision and horizon. These results demonstrate to investors how the requirement of additional information can affect forecast accuracy. To regulators, the results indicate how managers respond to changes in mandatory guidelines for voluntary disclosure. To managers, the results indicate how additional information and commitment to disclose such information can increase the credibility of their forecasts. 31

32 References Aboody, D., and R. Kasznik CEO stock option awards and the timing of voluntary corporate disclosures. Journal of Accounting and Economics 29: Ajinkya, B. B., and M. J. Gift Corporate Managers Earnings Forecasts and Symmetrical Adjustments of Market Expectations. Journal of Accounting Research 22 (2): Baginski, S. P., E. J. Conrad, and J. M. Hassell The effects of management forecast precision on equity pricing and on the assessment of earnings uncertainty. The Accounting Review 68 (4), Baginski, S. P., J. M. Hassell, J. M., and M. D. Kimbrough Why do managers explain their earnings forecasts? Journal of Accounting Research 42 (1): Bailey, W., H. Li, C. Mao, and R. Zhong Regulation Fair Disclosure and earnings information: Market, analyst, and corporate responses. Journal of Finance 58: Ciconte, W., M. Kirk, and J. W. Tucker Does the midpoint of range earnings forecasts = represent managers expectations? Review of Accounting Studies 19: Clement, M. B., R. Frankel, and J. Miller Confirming Management Earnings Forecasts, Earnings Uncertainty, and Stock Returns. Journal of Accounting Research 41 (4): Copeland, R., and R. Marioni Executives' Forecasts of Earnings per Share versus Forecasts of Naïve Models. The Journal of Business 45(4): Diamond, D., and R. E. Verrecchia Disclosure, liquidity, and the cost of capital. The Journal of Finance 46 (4): Dye, R An evaluation of essays on disclosure and the disclosure literature in accounting. Journal of Accounting and Economics, 32 (1 3): Fischer, P. E., and R. E. Verrecchia, Reporting Bias. The Accounting Review 75 (2): Foster, G Stock Market Reactions to Estimates of Earnings per Share by Company Officials. Journal of Accounting Research: Gigler, F., and T. Hemmer. 1998, On the Frequency, Quality, and Informational Role of Mandatory Financial Reports, Journal of Accounting Research 36 Supplement: Healy, P., and K. Palepu Information asymmetry, corporate disclosure, and the capital markets: A review of the empirical disclosure literature. Journal of Accounting and Economics 31:

33 Heflin, F., K. R. Subramanyam, and Y. Zhang Regulation FD and the financial information environment: Early evidence. The Accounting Review 78: Hirst, D. E., L. Koonce, and S. Venkataraman How disaggregation enhances the credibility of management earnings forecasts. Journal of Accounting Research 45 (4): Hirst, D. E., L. Koonce, and S. Venkataraman Management Earnings Forecasts: A Review and Framework. Accounting Horizons 22 (3): Hribar, P., and H. Yang CEO Overconfidence and Management Forecasting. Contemporary Accounting Research, Forthcoming Hutton, A. P., G. S. Miller, D. J. and Skinner The Role of Supplementary Statements with Management Earnings Forecasts. Journal of Accounting Research 41 (5): Jennings, R Unsystematic Security Price Movements, Management Earnings Forecasts, and Revisions in Consensus Analyst Earnings Forecasts. Journal of Accounting Research 25: Johnson, M. F., R. Kasznik, K K. Nelson The Impact of Securities Litigation Reform on the Disclosure of Forward-Looking Information by High Technology Firms. Journal of Accounting Research 39 (2): Kato, K., D. J. Skinner, and M. Kunimura Management Forecasts in Japan: An Empirical Study of Forecasts that are Effectively Mandated. The Accounting Review 84 (5): Lee, S., S. R. Matsunaga, C. W. Park Management Forecast Accuracy and CEO Turnover. The Accounting Review 87 (6): Lerner, J. and P. E. Tetlock Accounting for the effects of accountability. Psychological Bulletin 125: Leuz, C., and R. Verrecchia The economic consequences of increased disclosure. Journal of Accounting Research 38: Matsumoto, D Management s incentives to guide analysts forecasts. The Accounting Review 77(3): McDonald, C. L Examination of the Reliability of Published Predictions of Future Earnings. The Accounting Review July: Merkley, K. J., L. S. Bamber, and T. E. Christensen Detailed Management Earnings Forecasts: Do Analysts Listen? Review of Accounting Studies 18 (2):

34 Miller, G. S Should Managers Provide Forecasts of Earnings? A Review of the Empirical Literature and Normative Policy Recommendations. Retrieved from Miller_Study_on_Earnings_Guidance.pdf. Mercer, M How do Investors Assess the Credibility of Management Disclosures. Accounting Horizons 18 (3): Nagar, V The Role of the Manager's Human Capital in Discretionary Disclosure. Journal of Accounting Research 37 (Supplement): Ng, J., I. Tuna, and R. Verdi Management Forecast Credibility and Underreaction to News. Review of Accounting Studies, 18: Ng, J., A. Tsang, and O. Urcan Disaggregated Management Forecasts and the Legal Environment: International Evidence. Working paper. Patell, J. M Corporate forecasts of earnings per share and stock price behavior: Empirical tests. Journal of Accounting Research 14 (2): Penman, S. H An empirical investigation of the voluntary disclosure of corporate earnings forecasts. Journal of Accounting Research 18 (1): Penman, S. H The predictive content of earnings forecasts and dividends. Journal of Finance: Pownall, G., C. Wasley, and G. Waymire The Stock Price Effects of Alternative Types of Management Earnings Forecasts. The Accounting Review 88 (4): Pownall, G., and G. Waymire Voluntary Disclosure Credibility and Securities Prices: Evidence from Management Earnings Forecasts, Journal of Accounting Research 27 (2): Rogers, J. L., and. P. C. Stocken Credibility of management forecasts. The Accounting Review 80 (4): Ruland, W The accuracy of forecasts by management and by financial analysts. The Accounting Review April: Skinner, D. J Why Firms Voluntarily Disclose Bad News. Journal of Accounting Research 32 (1): Trueman, B Why Do Managers Voluntarily Release Earnings Forecasts? Journal of Accounting and Economics 8 (1): Verrecchia, R Essays on disclosure. Journal of Accounting and Economics 32:

35 Waymire, G Additional evidence on the information content of management earnings forecasts. Journal of Accounting Research 22 (2):

36 APPENDIX - Examples of guidance disclosed before and after Instruction 480. Example of management guidance before the change in regulation TIM Brasil s Guidance for