Management_Earnings_Forecasts_A_Review_and_Framework1

Management Earnings Forecasts:A Reviewand FrameworkD.Eric Hirst,Lisa Koonce,and Shankar V enkataramanSYNOPSIS:In this paper,we provide a framework in which to view management earn-ings forecasts.Specifically,we categorize earnings forecasts as having threecomponents—antecedents,characteristics,and consequences—that roughly corre-spond to the timeline associated with an earnings forecast.By evaluating managementearnings forecast research within the context of this framework,we render three con-clusions.First,forecast characteristics appear to be the least understood component ofearnings forecasts—both in terms of theory and empirical research—even though it isthe component over which managers have the most control.Second,much of the priorresearch focuses on how one forecast antecedent or characteristic influences forecastconsequences and does not study potential interactions among the three components.Third,much of the prior research ignores the iterative nature of management earningsforecasts—that is,forecast consequences of the current period influence antecedentsand chosen characteristics in subsequent periods.Implications for researchers,educa-tors,managers,investors,and regulators are provided.INTRODUCTIONManagement earnings forecasts are voluntary disclosures that provide information about expected earnings for a particular firm.Such forecasts represent one of the key volun-tary disclosure mechanisms by which managers establish or alter market earnings ex-pectations,preempt litigation concerns,and influence their reputation for transparent and accurate reporting.Management earnings forecasts are influential;they have been shown to affect stock prices ͑Pownall et al.1993͒,analysts’forecasts ͑Baginski and Hassell 1990͒,and bid-ask spreads ͑Coller and Yohn 1997͒.The purpose of this paper is twofold.First,we provide an organizing framework in which to evaluate research on management earnings forecasts.This framework characterizes management forecasts as having three components—antecedents,characteristics,and D.Eric Hirst is a Professor and Lisa Koonce is a Professor,both at The University of Texas at Austin,and Shankar Venkataraman is an Assistant Professor at the Georgia Institute of Technology.We thank the Department of Accounting at The University of Texas for funding for this project,and Steve Baginski,John Hassell,Greg Miller,Dave Ziebart ͑editor ͒,and two anonymous referees for helpful comments.Accounting HorizonsV ol.22,No.3American Accounting Association 2008DOI:10.2308/acch.2008.22.3.315pp.315–338Submitted:August 2006Accepted:January 2008Published Online:September 2008Corresponding author:Lisa KoonceEmail:Lisa.Koonce@ 315316Hirst,Koonce,and V enkataraman consequences.We organize a large number of studies on management earnings forecasts within this framework.1Second,we draw on this framework to identify areas and topics where additional management earnings forecast research would be most productive.This paper makes several contributions to the literature.First,although prior reviews of the management earnings forecast literature exist͑Cameron1986;King et al.1990͒,they predate important changes in the environment in which management earnings forecasts are produced and consumed.For example,the passage of Regulation Fair Disclosure͑Reg FD͒and the increased use of earnings benchmarks significantly influence both the frequency and content of management earnings forecasts͑Bailey et al.2003;Brown and Caylor2005͒.Because significant research has been conducted since those prior reviews,an up-to-date evaluation of the literature is warranted. Second,by viewing the forecasting process as involving three related components,we can more readily organize the existing literature and draw insights from that research.The decomposition of the literature also enables us to identify where theory development and additional empirical re-search are warranted.Given the importance of voluntary management earnings forecasts to the functioning of capital markets͑Healy and Palepu2001͒,we believe that a re-evaluation of the literature is in order.The framework we use herein is adapted from one developed by Wiedman͑2000͒who characterizes both voluntary and mandatory disclosure as involving three components—antecedents,characteristics,and consequences.2We tailor her framework to the management earn-ings forecast domain and identify the elements comprising the three components.Specifically, antecedents are the environmental andfirm-specific characteristics,such as the legal setting and the incentives of managers that influence the likelihood of a forecast being issued.Forecast characteristics embody the choices that a manager makes relating to the content of the forecast itself,such as its form,horizon,and level of detail.Consequences are the reactions to the forecast, such as stock price changes and analyst behavior.Not surprisingly,these consequences are a function of antecedents and forecast characteristics.Looking at the published research over the last several decades through the lens of this framework yields three important implications for researchers in the management earnings fore-cast area.First,gaining a better understanding of the choices that managers make once they decide to issue an earnings forecast is an important direction for both theory development and empirical research.Existing theories largely focus on why managers choose to issue a forecast and the likely consequences of those decisions͑e.g.,Ajinkya and Gift1984;Skinner1994;Stocken2000; Verrecchia2001͒.That is,theories primarily address antecedents and consequences.Thus,oppor-tunities exist for theory development on the choices that managers make related to forecast char-acteristics.Because of fewer theories,it is perhaps no surprise that relative to the vast literature on forecast antecedents and consequences,comparatively fewer studies examine how managers choose the characteristics of their earnings forecasts.The lack of theory may explain why the research that is conducted on forecast characteristics is fairly recent͑at least compared with research on forecast antecedents and consequences͒.To the extent that forecast characteristics are examined in the literature,they are primarily treated as exogenous variables͑Baginski et al.2004͒. Given that managers have substantially greater control over forecast characteristics than they have over forecast antecedents and consequences,it is striking that the decisions managers make about such characteristics are comparatively less understood͑Choi et al.2006͒.For example,we know relatively little about why managers decide to issue forecasts with external versus internal 1By intention,we do not survey every paper on the topic of management earnings forecasts.Our goal is to describe a sufficient number of studies within the context of our framework to be able to legitimately render our conclusions.2We use different component labels than Wiedman͑2000͒who uses the terms environment,attributes,and impact.Accounting Horizons September2008 American Accounting AssociationManagement Earnings Forecasts:A Review and Framework317 attributions,why they issue them in conjunction with other disclosures,and what the nature is of those other disclosures͑Baginski et al.2004,17͒.Second,our review of the literature highlights that the typical study focuses on the main effect of one or more forecast antecedents or characteristics on forecast consequences.Because main effect results are unlikely to hold under all conditions,we argue that researchers should identify and test possible interactions among antecedents or characteristics.Given the large number of studies looking at main effects,interaction tests will push forward our knowledge and understand-ing of such forecasts͑Libby et al.2002͒.Although such interaction tests have been more prevalent in recent research͑e.g.,Rogers and Stocken2005;Hutton et al.2003;Wang2007͒,ample oppor-tunities exist for further study.For example,managers with strong incentives tied to thefirm’s stock price͑a forecast antecedent͒are viewed as issuing self-serving forecasts͑Rogers and Stocken2005͒.What is less clear-cut,however,is how other factors might moderate the influence of these incentives.For example,does providing forecasts more frequently moderate the influence of incentives on how forecasts are viewed by market participants?Furthermore,interaction tests are useful in reconciling conflictingfindings in the literature. Including a theoretically motivated conditioning,or moderator,variable often allows the re-searcher to identify where the effect holds͑or where it holds in a different way͒.For example, Baginski et al.͑1993͒find that stock price reactions to earnings forecasts are contingent on forecast form;point forecasts lead to greater stock price reactions relative to range forecasts.In contrast,Pownall et al.͑1993͒and Atiase et al.͑2005a͒find no variation in stock price reactions conditional on forecast form.One possible reason for these mixed results may originate from Hirst et al.͑1999͒who examine how prior forecast accuracy may moderate the effects of forecast form. They experimentally show that only when prior forecast accuracy is high do investors consider forecast form.When prior accuracy is low,forecast form does not influence investor judgments. That is,prior accuracy is shown to be a significant moderator variable.Careful consideration of other such interactions within and among antecedents,characteristics,and consequences could help establish important boundary conditions to main-effect results.Third,our review reveals a multi-period aspect of management earnings forecasts,yet the existing literature does not completely leverage this feature͑see Miller͓2002͔for an exception͒. Forecast consequences from one period can later become forecast antecedents and may influence subsequent choices about forecast characteristics.For example,accurate forecasts in prior periods enhance afirm’s reputation for forecasting accuracy͑a consequence͒,which,in turn,becomes an antecedent for a current period forecast.This antecedent,in turn,can significantly influence whether managers issue subsequent forecasts and,if they do,the chosen characteristics of those forecasts as well as the market’s reaction to them.A challenge in leveraging this multi-period feature is identifying the manner in which consequences from one period affect antecedents and forecasts in subsequent periods.For example,afirm’s reputation for accuracy is normally built over an extended time and is generally not quickly changeable,while remedying information asymmetries may occur more readily.Thus,recognizing the multi-period and iterative nature of the forecasting process and identifying how forecast consequences become antecedents are im-portant to extend our understanding of the impact of forecasts beyond immediate,one-time reac-tions.In sum,our paper has numerous implications for those who conduct research on manage-ment earnings forecasts.Our paper also has implications forfinancial reporting educators,managers,investors,and regulators.For those who teachfinancial reporting,our paper provides a succinct appraisal of the scholarly literature on management forecasts.Mostfinancial reporting textbooks focus on manda-toryfinancial reporting and provide little,if any,coverage on important voluntary disclosures like management earnings forecasts.This focus is puzzling given the important role of earnings forecasts in the operation of capital markets.We believe our paper will inform these Accounting Horizons September2008American Accounting Association318Hirst,Koonce,and V enkataraman scholars/educators,thereby facilitating classroom coverage.Doing so will enable students to un-derstand,for example,why the stock market may react very little when the mandatory earnings reports are released͑i.e.,the market already anticipates the news via management earnings fore-casts͒or why managers manipulate earnings in response to previous forecasts.This knowledge is not only helpful forfinancial reporting classes,but also for auditing courses.By recognizing the expectations that managers create with their earnings forecasts,auditing students͑and practitio-ners͒can improve their ability to identify when certain accounting practices may be questionable.Managers can benefit from our paper by becoming more knowledgeable about scholarly researchfindings regarding these forecasts.This information should improve their ability to make informed decisions about whether and when to issue a forecast and what characteristics it should possess.For example,from our review,managers can better distinguish between the changes they might make to their earnings forecasts to achieve immediate responses from market participants and the changes that would not be quickly appreciated.Managers also can gain insight into the forecast characteristics that increase the credibility of good-news forecasts͑e.g.,verifiable infor-mation͒and the characteristics that do not͑e.g.,soft talk;Hutton et al.2003͒.Investors shouldfind our paper useful because it draws attention to the full range of forecast characteristics that managers can choose and the consequences that may follow.The distinct characteristics of forecasts are important because not all choices that managers make are equally relevant to investors.For example,a pre-commitment to disclosure may be viewed as a more-powerful cue to the believability of a forecast than its form.Our review also allows investors to fully appreciate the various types of reactions that management earnings forecasts can produce. While many investors might solely focus on the stock price reaction to a management earnings forecast,our review highlights the important role of other consequences,such as management’s reputation for clear and transparent disclosure.A better understanding,through the framework,of the relations among antecedents,characteristics,and consequences also can inform investors as they lobbyfirms to make changes in their disclosures.Finally,our paper informs regulators who think about the quality,frequency,and effectiveness of management earnings forecasts͑House Committee on Financial Services2006͒.Since1973, when the Securities and Exchange Commission͑SEC͒lifted its prohibition against forward-looking information,the SEC and other regulators have been concerned with the credibility of such information.Thus,they should be interested in our systematic review of the management earnings forecast research literature,particularly where empiricalfindings have͑or have not͒changed with new regulations.For example,we discuss research showing that although Reg FD does not decrease the amount of forward-looking informationfirms provide,it apparently de-creases the quality of such information͑Bailey et al.2003;Irani and Karamanou2003͒.The rest of the paper is organized as follows.In the next section,we define management earnings forecasts and provide an overview of our framework.In the following three sections,we summarize the major themes in the earnings forecast literature within the context of the three components of our framework—namely,antecedents,characteristics,and consequences.Our goal is not to provide an exhaustive survey of individual studies,but to identify broad themes in the literature.In the penultimate section,we reflect on our review of the earnings forecast research in light of our framework and identify areas and topics where additional earnings forecast research would be most productive.Thefinal section provides conclusions.FRAMEWORKKing et al.͑1990͒define management earnings forecasts as voluntary managerial disclosures predicting earnings prior to the expected reporting date.The term earnings guidance often is used synonymously with earnings forecasts,both in the popular press͑Zuckerman2005͒and in the Accounting Horizons September2008 American Accounting AssociationManagement Earnings Forecasts:A Review and Framework319 academic literature͑Atiase et al.2005a;Hutton2005͒.3Although earnings forecasts are commonly issued well in advance of quarterly and annual earnings releases,they are sometimes provided after the accounting period has ended but before the earnings are announced.These latter forecasts are typically referred to as earnings preannouncements.4When management forecasts indicate substantial shortfall from expected earnings,they are commonly termed earnings warnings ͑Kasznik and Lev1995͒.Figure1provides a pictorial representation of the antecedent,characteristics,and conse-quences framework that we use to discuss management earnings forecasts.After providing an overview of our framework,we then discuss each component of the framework in greater detail.Because earnings forecasts are voluntary disclosures,thefirst question confronting managers is whether to issue a forecast.The answer is shaped by a combination of the environment faced by thefirm andfirm-specific characteristics.In our framework,we label these factors as antecedents because they are precursors to the actual forecast decision.Having chosen to issue an earnings forecast,the manager then faces a broad array of choices regarding the attributes of that forecast. These choices involve,for example,the form of the forecast͑e.g.,point,range,qualitative,etc.͒, the horizon͑e.g.,quarterly versus annual͒,and the information to accompany the forecast͑e.g.,the presence or absence of attributions͒.We label these choices as forecast characteristics because they are attributes associated with the forecast.The third component of our framework,forecast consequences,captures events and reactions that occur after the forecast is issued,such as the stock market reaction to the forecasted earnings information.Not surprisingly,these consequences are a function of the antecedents and forecast characteristics.For example,management forecasts lead to larger earnings forecast revisions by analysts when they originate fromfirms with high prior forecast accuracy͑Williams1996͒.Although our framework is organized chronologically—that is,consequences follow the choice to issue a forecast with particular characteristics,which,in turn,follow forecast antecedents—it is important to recognize that the framework is iterative.That is,forecast conse-quences in one period become forecast antecedents in subsequent periods.For example,Healy et al.͑1999͒document that expanded disclosures result in increased institutional ownership and analyst coverage.In other words,they view institutional ownership as a consequence of disclo-sure,whereas Ajinkya et al.͑2005͒argue that higher institutional ownership leads to more fre-quent and more precise forecasts,implying that institutional ownership is an antecedent.Reflect-ing the natural iterative aspect of the forecasting process,some elements are listed under more than one component of our framework͑see Figure1͒.5Forecast AntecedentsWe define antecedents as factors that exist at the time the manager decides to issue a forecast. Indeed,antecedents influence whether the manager will issue a forecast and,as explained later, 3Earnings guidance represents any manager-provided information that guides outsiders in their assessment of afirm’s future earnings,both directly and indirectly͑Miller2002͒.Thus,earnings guidance might include,but need not be limited to,earnings forecasts.For instance,afirm’s comments on its prospects in a new product market might be construed as indirect earnings guidance.All references to earnings guidance in this paper refer to earnings forecasts only.4Even though preannouncements are technically earnings forecasts,most of the literature treats them as early earnings announcements rather than late earnings forecasts.Accordingly,we do not discuss preannouncements in this paper.5In theory,all elements under characteristics and consequences could be listed under antecedents.That is,historical stock market reaction͑a consequence͒to afirm’s forecasts could be an antecedent.The historical choices that afirm makes about forecast form͑a characteristic͒could be viewed as an antecedent.Although we do not list all of these items under the antecedents component,the reader is encouraged to think about whether and how other characteristics and conse-quences might be profitably viewed as antecedents.Accounting Horizons September2008American Accounting AssociationF IG U R E 1A F r a m e w o r k f o r A n a l y z i n g M a n a g e m e n t E a r n i n g s F o r e c a s ts 320Hirst,Koonce,and V enkataraman Accounting HorizonsSeptember 2008American Accounting AssociationManagement Earnings Forecasts:A Review and Framework321 can influence the manager’s choice of forecast characteristics and the forecast’s consequences.As shown in Figure1,we classify antecedents into two broad categories:͑1͒forecast environment—that is,features of the legal and regulatory environment and features of the analyst and investor environment,and͑2͒forecaster characteristics—that is,information asymmetry,pre-commitment to disclosure,firm-specific litigation,managerial incentives,prior forecasting behavior,and pro-prietary costs.Firms often have little or no control over certain antecedents in the short term, although they can influence some of those antecedents in the longer term.For example,afirm’s prior forecasting accuracy is an antecedent.A reputation for accuracy is normally built over an extended time and is generally not quickly changeable.We organize our discussion of antecedents as follows.Wefirst discuss the two components of the forecasting environment.Then,we discuss the forecaster characteristics.Forecast EnvironmentLegal and regulatory environment.Although management earnings forecasts are voluntary disclosures,legal and regulatory environments can influence the type of forecasts thatfirms make and the channels through which they are disseminated.6Four significant changes in the U.S. regulatory environment occurred over the last four decades.In1973,the SEC allowedfirms to include forward-looking information in their regulatoryfilings.In1979,the SEC provided safe harbor tofirms issuing forecasts to shield them from frivolous litigation related to forward-looking disclosures made in good faith.Then,in1996the Private Securities Litigation Reform Act ͑PSLRA͒extended the safe harbor so thatfirms could not be sued easily for forecasts that do not materialize.Finally,Reg FD,passed in2000,mandated that material information could not be disclosed selectively.7Thefirst three regulatory changes are largely aimed at encouraging companies to more freely disclose forward-looking information͑including earnings forecasts͒.These regulatory changes are controversial.For example,the primary concern with the PSLRA is that companies might issue self-serving,opportunistic forecasts with impunity given the expanded safe harbor.Johnson et al.͑2001͒document that the PSLRA does not adversely affect the quality of management earnings forecasts,at least for high-techfirms.In fact,theyfind that the accuracy of earnings forecasts increases after the passage of the PSLRA.Reg FD raises concerns of a different nature.Before the passage of Reg FD,managers could privately provide earnings guidance to selected analysts.Prior research documents that many managers used this private channel extensively͑Ajinkya and Gift1984;Hutton2005͒.Following Reg FD,managers face a choice between public disclosure and no disclosure at all.Many,includ-ing regulators,fear thatfirms would choose the no-disclosure option because earnings forecasts are voluntary.However,research reveals an increase in public earnings forecasts following the passage of Reg FD,suggesting that concerns about the information-chilling effect of this regula-tion may have been misplaced͑Bailey et al.2003;Heflin et al.2003͒.However,Wang͑2007͒reports that the decision regarding the continuation of guidance after the passage of Reg FD depends onfirm-specific characteristics.Specifically,thosefirms with lower information asymme-try and higher proprietary information costs do,in fact,reduce or eliminate their public disclosures subsequent to Reg FD.6We focus on the U.S.environment.However,the framework could be expanded to cover cross-border issues in other legal and regulatory regimes͑see Baginski et al.2002͒.7Although the SEC does not provide a bright line definition of what constitutes material information,recent enforcement actions indicate that even a discussion confirming a prior forecast could be construed as material information͑SEC 2005͒.Accounting Horizons September2008American Accounting Association322Hirst,Koonce,and V enkataraman The research discussed above compares forecasting behavior before and after a new regula-tion is enacted.In contrast,Baginski et al.͑2002͒study the effect of the legal environment on earnings forecast behavior by examining two legal paring the United States and Canada,they argue that the business environment in the two countries is similar,but the United States is more litigious͑Baginski et al.2002,28–29͒.Specifically,they compare forecasts issued by Canadian versus U.S.managers andfind that Canadian managers behave in a manner that suggests less concern over litigation.Specifically,Canadian managers issue more forecasts in general and,in particular,during periods of rising profits,relative to their U.S.peers.Furthermore, they issue longer-term and more-precise forecasts than their U.S.counterparts.Finally,Canadian managers are less likely to issue interim forecasts during periods of earnings decreases.Analyst and investor environment.Two other environmental factors also affect whether managers issue forecasts,namely the behavior of investors and analysts.Research shows that investors and analysts seek disclosure of forward-looking information,such as earnings forecasts, and they tend to invest in and provide coverage of companies that have more forthcoming disclo-sure policies͑Ajinkya et al.2005;Healy et al.1999͒.Both institutional investment and analyst coverage have changed over time.Specifically, institutional investors’share of the U.S.stock market capitalization has expanded from under30 percent in1980to over50percent in2002͑Gompers and Metrick2001;Gillan and Starks2003͒. Furthermore,analyst coverage of companies has increased over time.Barber et al.͑2001,2003͒report that the number offirms followed by analysts rose from1,841in1986to5,786in2001. Arguably,because of these changes in institutional investment and in analyst coverage,the number offirms providing public earnings guidance also increased from approximately10to15percent in the mid-1990s to approximately50percent in2004͑Anilowski et al.2007͒.Forecaster CharacteristicsBecause the legal/regulatory and investor/analyst antecedents largely vary longitudinally,they arguably affect U.S.firms similarly.For example,all public U.S.firms are subject to the forces of Reg FD in the same fashion.However,the other antecedent,forecaster characteristics,exhibits significant cross-sectional variation because it refers to unique aspects of the forecastingfirm.We discuss a number of these important forecast antecedents next͑see Figure1͒.Information asymmetry.Many of the motivations managers have for issuing earnings fore-casts are congruent with those of shareholders.That is,the supply of and the demand for forecasts is assumed to be largely driven by stock price considerations,with managers issuing forecasts͑and analysts and investors demanding them͒to reduce the asymmetry in information between manag-ers and analysts and current or potential investors͑Ajinkya and Gift1984;Verrecchia2001͒. Lower information asymmetry is viewed as desirable because it is associated with higher liquidity ͑Diamond and Verrecchia1991͒and lower cost-of-capital͑Leuz and Verrecchia2000͒.Prior research supports the idea that there is a relationship between information asymmetry andfirm behavior.Specifically,firms issuing management earnings forecasts are noted to have higher information asymmetry͑measured by bid-ask spreads prior to the forecast͒compared with firms that do not issue such forecasts͑Coller and Yohn1997͒.More importantly,that studyfinds that the information asymmetry is reduced after the issuance of the forecasts,suggesting that forecasts are highly influential.Interestingly,though,managers may not always desire to reduce information asymmetry.For example,if lower information asymmetry leads to greater monitoring ͑Shleifer and Vishny1989͒,managers may not necessarily seek to voluntarily reduce it.Pre-commitment to disclosure.Economic theory suggests that pre-commitment to continued disclosure also reduces the information asymmetry component of afirm’s cost-of-capital͑Leuz and Verrecchia2000͒.Although managers can proffer their commitment to disclosure,they can only credibly signal such commitment by providing disclosures more frequently͑Botosan andAccounting Horizons September2008 American Accounting Association。