Annual Earnings Guidance and the Smoothing of Analysts’ Multi-Period Forecasts

Abstract

This article examines the use of annual earnings guidance as a mechanism used by managers to reduce the volatility of analyst earnings forecasts and allow them to report smooth earnings without missing quarterly analyst forecasts. Facing the pressure to meet or beat analyst forecasts and driven by the perceived capital market benefits of reporting a smooth earnings path, managers attempting to influence investors’ earnings expectations over a longer horizon can issue annual guidance to smooth the time-series path of analyst forecasts, a strategy we term as “expectation smoothing.” Our empirical results suggest that annual guidance reduces the volatility of analysts’ multi-period forecasts, which in turn contributes to a smoother actual earnings and higher likelihood of meeting analysts’ quarterly forecasts. We also find that issuing quarterly guidance does not affect the smoothness of analysts’ earnings expectations and that managers with longer horizons are more likely to issue annual guidance, consistent with the unique long-term effects of annual earnings guidance.

Keywords

analyst forecasts annual earnings guidance earnings smoothing expectation smoothing

Introduction

The benefits to firms and managers from smoother earnings paths are well documented (Ronen & Sadan, 1981). Investors use analysts’ projected earnings in forming their own expectations, and missing analysts’ expectations has been shown to be costly to firms and managers (Fuller & Jensen, 2002; Skinner & Sloan, 2002). To avert the cost of missing analyst forecasts, managers have been shown to manage reported earnings (Degeorge et al., 1999) and walk down analyst forecasts (Ajinkya & Gift, 1984). However, if such actions are conducted ad hoc—on a quarterly basis as needed depending on analyst forecasts—then whenever the path of analysts’ annual forecasts happens to be volatile, the necessity of meeting quarterly forecasts would force a pattern of reported earnings that is volatile as well. Therefore, managers wishing to report a smooth earnings path without increasing the risk of missing analyst forecasts likely have an incentive to ensure that analyst forecasts follow the long-term earnings path they envision.¹ One way to achieve that goal is to choose a guidance strategy that induces analysts to issue forecasts that would guide investors to project earnings along the path that managers desire to convey. This strategy involves managers using long-horizon guidance to align analyst forecasts for multiple future periods with their desired smooth pattern of expected future reported earnings.

Management earnings guidance is a major channel through which financial information is impounded in stock prices.² To the extent that managers wish to influence stock prices, guidance strategies are an important part of the managerial toolbox. Recent studies highlight the guidance decision as a multi-period decision (e.g., Einhorn & Ziv, 2008; Graham et al., 2005; Myers et al., 2007; Tang, in press), which suggests that a forward-looking manager would gaze beyond the current period and seek to also adjust market expectations for future periods.

In this study, we investigate whether firm managers strategically use annual guidance to align analyst forecasts with the smooth earnings path they envision and whether smoother analyst forecast paths facilitate the reporting by firms of smoother income without increasing the likelihood of missing analyst forecasts. We focus on management guidance that is bundled with earnings announcements, which emerged as the predominant form of guidance after the passage of regulation fair disclosure (Reg FD) in late 2000 (Rogers & Van Buskirk, 2013). We start by verifying our maintained assumption that annual guidance has the ability to affect analysts’ long-term forecasts. Our analysis suggests that analysts strongly react to annual earnings guidance news, revising not only their one-year-ahead earnings forecasts but also their two-year-ahead forecasts.

We then examine whether annual earnings guidance helps smooth analyst forecasts of future earnings by examining the change in the smoothness of analyst forecasts and the revised forecasts following issuance of annual guidance. Smoothness of analyst forecasts is measured as the negative of the standard deviation of analyst forecast deviations from a trend line that reflects analysts’ expectation of how the pattern of past earnings is going to continue in the future.³ Consistent with our expectation, we find the issuance of annual earnings guidance is positively associated with the change in the smoothness of analyst forecasts following the guidance.

Next, we test whether managers’ annual guidance achieves the expected outcome—whether the resulting smoother analyst forecasts in turn lead to the smoothness of future reported earnings without increasing the risk of missing the analyst earnings forecasts. To that end, we adopt a recursive regression to test the existence of the following event sequence: (a) firms having smoothed earnings in the past (for which earnings smoothness is likely to be more important) are more likely to issue annual guidance; (b) issuing annual guidance increases the smoothness of analyst forecasts; (c) such analyst-revised forecasts become the targets for firms’ actual earnings to meet or beat; and (d) after firms report earnings over time, we observe the smoothness of the reported earnings. Our analyses follow this sequence by estimating a set of recursive regressions to help substantiate the conjectured event sequence.

To solidify our inference that it is the long-horizon guidance (annual) and not just any guidance that achieves the predicted and observed results, we test whether the issuance of quarterly guidance exerts similar effects. Our analysis suggests that issuing quarterly guidance does not help smooth analysts’ multi-year earnings forecasts.⁴ These results suggest that annual guidance is an unique tool managers use to achieve smoother analyst forecasts and ultimately a smoother path of actual reported earnings.⁵

Finally, we provide additional evidence that links managers issuing annual guidance with managerial benefits of smooth earnings, in particular, their long-horizon perspective. The longer the CEO horizon at the firm, the more likely she is to value smooth earnings. Using CEO age and regular dividend payment to proxy for managers’ horizon in the firm, we show that CEOs closer to retirement and of non-dividend-paying firms are indeed less likely to issue annual earnings guidance and more likely to issue quarterly guidance. We also show that analysts’ long-term growth (LTG) forecasts are more consistent with the growth rate implied in their multi-period earnings forecasts when managers provide annual guidance.

Our study contributes to the existing literature in three major ways. First, our study bridges two large yet rather separated streams of the literature on earnings smoothing and on meeting and beating analyst forecasts. We show that both objectives play a role in management guidance decisions and provide evidence that managers view both objectives as integrated and connected rather than independent objectives.

Second, we extend the classic “expectation alignment” hypothesis (Ajinkya & Gift, 1984) and the “expectation management” hypothesis (Matsumoto, 2002) from a single-period framework to a multi-period framework. Prior studies that examine the benefits of management guidance (e.g., Kross et al., 2011) mainly focus only on the period explicitly guided for by managers, thus their definition of “expectation” involves just one period. We extend this literature by providing evidence consistent with managers using guidance to affect analyst expectations beyond the guided horizon. More importantly, by linking their “expectations” into a chain over multiple periods, our analysis of analyst forecasts allows us to examine a new dimension of expectation management—the smoothing of the temporal volatility in analyst forecasts for multiple future periods. Expanding the research lens from a single-period to a multi-period framework has generated important insight (e.g., Myers et al., 2007), but unlike our investigation, Myers et al. examined the achievement of a string of earnings increases through earnings management but not the smoothing of forecasts.

Third, while prior studies have separately examined the effect of annual guidance (e.g., Tang & Zhang, 2018) or quarterly guidance (e.g., Tang & Venkataraman, 2018), we provide initial evidence on the differential effects of annual guidance versus quarterly guidance. Our results highlight the importance of explicitly considering the managers horizon when examining the effects of guidance.

Conceptual Framework and Hypothesis Development

In this section, we outline the process through which managers endeavor to achieve multiple earnings goals that may conflict at times. In particular, we focus on managers’ desire to achieve a smooth path of reported earnings as a multi-period, long-term goal without jeopardizing the goal of meeting periodic analysts’ expectations. We utilize the conjectured dynamic process and rely on the literature related to these topics in developing our hypotheses. Figure 1 offers a graphical presentation of this dynamic process.

Figure 1.

A conceptual framework for smoothing and meet-and-beat process.

Earnings Volatility and Earnings Smoothing

Consistent with evidence from Francis et al. (2004) on the association between volatile earnings and risk, Graham et al.’s (2005) survey reports that corporate executives prefer to report a smooth path of earnings. Smooth earnings more reliably align investors’ earnings expectation with those of managers and therefore is more informative (e.g., Barnea et al., 1976; Ronen & Sadan, 1981). Earnings smoothing can also improve the contemporaneous association between stock returns and earnings (Subramanyam, 1996) and the association between current returns and future earnings (Tucker & Zarowin, 2006). Other benefits of a smooth earnings path include reduced idiosyncratic volatility of stock returns (Chen et al., 2012), lower litigation/political costs and perceived risks, and mitigated underinvestment in relationship-specific assets (Dou et al., 2013). On the contrary, some studies suggest that, when conducted perniciously such as via earnings management, earnings smoothing may not necessarily enhance firm value (Rountree et al., 2008) and may exacerbate stock price crash risks when extreme bad news cannot be smoothed out as a firm has exhausted its reserves to smooth earnings in the previous periods (Chen et al., 2017). Overall, the evidence from the prior literature largely supports the view that earnings smoothing is consistent with maximizing firm value (Rountree et al., 2008).

In addition, managers also have personal incentives to smooth their firms’ earnings to meet bonus targets (Healy, 1985) and to avoid negative earnings surprises to secure their employment (DeFond & Park, 1997). In such a case, even if earnings smoothing is not value-maximizing for the firm, managers may still prefer a smoother earnings path. In this study, we build on the premise supported by prior research that managers on average value the benefits of a smooth earnings path (McInnis, 2010).

Meeting and Beating Analyst Expectations

The importance of analysts in forming investors’ expectations on a firm’s performance and the implications for firm valuation is well documented (see Ramnath et al., 2008 for a recent survey on this literature). Firms meeting and beating analyst expectations enjoy higher equity premiums (Bartov et al., 2002) and have lower cost of debt (Jiang, 2008). In contrast, missing analyst forecasts even by a penny can result in a “torpedo effect”—a significant stock price drop (Skinner & Sloan, 2002), which in turn can reduce the managers’ annual bonus (Matsunaga & Park, 2001). Given such substantial consequences, managers strive to avoid negative surprises at earnings announcements (Graham et al., 2005; Matsumoto, 2002). As a result, the frequency of firms reporting earnings that meet or just beat analysts’ consensus forecasts increases (Brown & Caylor, 2005).

Expectation Smoothing

Whereas prior research documents the benefits of the above discussed reporting goals (smooth earnings path and meeting analyst forecast), it does not elaborate nor provide any evidence on whether managers treat the two as independent or connected goals. The two goals potentially may conflict if analyst forecasts follow a path that is inconsistent with and more volatile than the earnings path envisioned by firm managers. To be able to achieve a smooth earnings path without jeopardizing missing the analyst forecasts, managers are likely motivated to release information that would align analysts’ earnings projections with the smooth earnings path that the management desires to convey. This will help analysts incorporate the long-term smooth projections into their quarterly forecasts and would in turn enable managers to achieve the smooth earnings path without increasing the risk of missing analyst forecasts. We refer to this as “expectation smoothing.”

“Expectation smoothing” is related to “earnings smoothing” in the sense that expectation smoothing is a mechanism that is conducive to achieving earnings smoothing in a way that does not jeopardize meeting analyst forecasts. When managers smooth analyst expectations over multiple horizons, meeting and beating such a smoother path of expectations would more naturally result in a smoother path of actual reported earnings than if they try to meet and beat an otherwise more volatile path of earnings expectations.

While the prior literature on expectation management (Cotter et al., 2006) largely focuses on the effect of quarterly guidance on analysts’ earnings projections for the period explicitly guided for, we expect long-term guidance to be more useful than quarterly guidance in conveying managers’ long-term earnings expectations. Thus, our first hypothesis is

Hypothesis 1 (H1): Ceteris paribus, the change in the smoothness of analyst forecast series around the annual earnings announcement is positively associated with the issuance of annual management guidance.

We do not consider expectation smoothing to be a “pre-requisite” to earnings smoothing, but we expect that firms with smoother expectations will find it easier to deliver smoother earnings, given that they seek to meet and beat analyst expectations as documented in prior research. Thus, managers reap less benefit when earnings are smooth but the expectations were volatile, due to the inability to meet or beat some of the expectations not aligned on the smooth earnings path that they report.

Also, similar to the argument that smoother earnings are viewed as more reliable signals of future performance (e.g., Ronen & Sadan, 1981), smoother forecasts by analysts are likely also more helpful to investors in forming their own future earnings expectations. In this sense, the cost to managers when earnings are smooth but the expectations are volatile is the potential injection of uncertainty caused by volatile analyst forecasts into investors’ judgment of the firms’ future performance. Appendix A provides an illustrative example of how firms can smooth analysts’ expectations.

The objective of smoothing analysts’ expectations requires guidance to also affect expectations beyond the period expressly guided for. Hence, we also empirically test this necessary condition by examining analysts’ revisions of their forecasts for horizons beyond that explicitly guided by the managers. Specifically, we test whether annual guidance changes analyst forecasts for the year following the year guided for. In doing so, our novel notion of expectation smoothing incorporates two well-documented reporting phenomena with very different horizons. Namely, whereas managers’ desire for analyst forecasts to be achievable is relatively short-term focused, their preference for delivering smooth paths of earnings is relatively long-term focused. Thus, we expect “expectation smoothing” to be driven by such dual reporting preferences by managers. Note that we expect the dual goals to be equally important to managers and thus we do not assume that managers prioritize one over the other.

The Iterative Nature of Achieving Reporting Objectives

Hirst et al. (2008) point out that the financial reporting process is iterative. Therefore, achieving reporting objectives, such as meeting or beating analyst expectations or smoothing earnings paths, is not over when one period ends. Managers’ decisions to continue with a certain practice at period t depend on its effectiveness in achieving their goals in periods leading up to period t. For example, managers are more likely to stop issuing quarterly guidance if their firms missed analyst forecasts in prior periods (Feng & Koch, 2010). Managers are also more likely to issue quarterly earnings guidance to maintain a string of meeting or beating analyst forecasts established in the previous periods (Kross et al., 2011).

In the context of our study, one benefit of issuing annual guidance is to help managers smooth analysts’ earnings forecasts, which in turn should help achieve a smoother earnings path, without increasing the risk of missing analyst forecasts. Although firms do not have complete flexibility in switching guidance on and off every quarter, especially if they have already established a consistent pattern of issuing guidance (Tang, in press), managers are likely to issue annual guidance only if, ceteris paribus, they perceive a benefit in reporting a smooth earnings path. Thus, the first step in the iterative reporting process is managers’ decision on whether to issue annual guidance, which we expect depends on whether a firm has a perceived benefit from a smooth earnings path, as detailed in H1. We proxy for the firm’s perceived benefits of smooth earnings path with the active earnings smoothing by the firm. We assume that their revealed preference for smooth earnings is likely to persist in the periods following the smoothing period. We expect that issuing guidance will help the managers smooth analysts’ multi-year earnings expectation, which in turn will allow them to accomplish a multi-year smooth earnings path and regularly meet analyst forecasts. This leads to our second hypothesis:

Hypothesis 2 (H2): Ceteris paribus, firms that benefit from a smooth earnings path are more likely to use annual guidance to smooth analyst earnings expectations, and ultimately to report smoother earnings patterns and regularly meet analyst forecasts in the future.

Sample Selection and Research Design

Sample Selection

Our sample consists of post-Reg FD (2000–2015) firms’ annual earnings announcements for which multi-period (at least two years) analyst annual forecasts and their post-announcement revisions are available on Institutional Brokers’ Estimate System (IBES) within 30 days from the announcement date. We use the post-Reg FD period to ensure a stable regulatory environment for the guidance practice. We end our sample in 2015 because we need two more years to measure the smoothness of earnings path after 2015. Kadan et al. (2009) provide evidence that analysts’ behavior has changed following the Global Analysts Research Settlement. Thus, we verify that all our results hold if we start our sample from the year 2003.⁶ In the interest of a homogeneous guidance sample, we focus on guidance bundled with earnings announcement, which has emerged as the predominant form of guidance after the passage of Reg FD in late 2000 (Rogers & Van Buskirk, 2013).

We use management guidance data from IBES. To ensure that we capture the unique effect of annual guidance, we use a sample of annual guiders defined as firm-years issuing only annual guidance. We define a firm as a guiding firm (GUIDER) if the firm issues one-year-ahead annual earnings guidance bundled with the earnings announcement. Among firm-years with no annual guidance, we define a firm-year as a quarterly guider (QGUIDER) if it issues one-quarter-ahead quarterly earnings guidance bundled with the earnings announcement. The rest of the sample consists of firm-years with no guidance issued. To avoid confounding effects, we exclude 6,323 observations in which the firm issues both annual and quarterly guidance. Our full sample consists of 34,348 unique analyst-firm reports that contain both one- and two-year ahead annual earnings forecast revisions. These revisions apply to 8,494 firm-years with annual guidance, 3,681 with quarterly guidance, and 22,173 with no guidance at all.

Table 1 reports the yearly distribution of our sample. The number of observations each year increases during our sample period, as more analysts are issuing multiple-horizon forecasts. Consistent with Rogers and Van Buskirk (2013), the numbers of both annual guiders and quarterly guiders in our sample increase over time (from 44 and 35 in 2000, to 794 and 238 in 2015) as more firms are issuing bundled guidance. Moreover, annual guidance has overtaken quarterly guidance to become the more dominant horizon of forecasts, consistent with Tang et al. (2020). This trend is also consistent with prior studies that document the phenomenon among many firms to terminate quarterly guidance in the face of criticisms of quarterly guidance due to its alleged contribution to managerial opportunism (e.g., Chen et al., 2011).

Table 1.

Sample Distribution.

Year	Annual guider (A)			Quarterly guider (Q)			Non-guider (N)
	Obs	Firms	Analysts	Obs	Firms	Analysts	Obs	Firms	Analysts
2000	40	16	30	35	16	19	303	143	181
2001	103	45	62	76	25	44	596	190	329
2002	77	34	49	101	38	55	614	196	318
2003	250	71	149	155	44	88	856	206	401
2004	255	78	177	161	49	90	949	225	428
2005	385	101	188	188	51	96	906	174	403
2006	422	114	237	178	50	86	1,015	212	464
2007	594	136	300	212	51	104	969	203	444
2008	652	142	276	316	72	102	1,561	284	565
2009	709	146	318	305	65	135	1,695	270	597
2010	833	153	391	371	77	149	1,886	269	626
2011	787	160	340	365	64	131	1,971	279	672
2012	962	156	380	354	64	149	2,120	267	634
2013	774	136	331	302	58	110	2,147	263	742
2014	857	136	328	324	63	131	2,320	249	688
2015	794	135	291	238	52	90	2,265	242	700
All	8,494	1,759	3,847	3,681	839	1,579	22,173	3,672	8,192

Note. This table reports the yearly sample distribution for firms providing only annual guidance (A) and for firms providing only quarterly guidance (Q) or firms providing no guidance (N). The unit of observations is at the firm-year-analyst level. The number of unique firms and unique analysts are also reported for each year and for the full sample. Obs = observations.

Research Design and Variable Definition

We conduct all our analyses at the individual analyst level, because we are interested in analysts’ reactions to guidance as manifested in multi-period forecasts, which are unique to each individual analyst. To test H1, we estimate the following model:

D E P_V A R_{i, j, t} = β_{1} G U I D E R_{i, t} + β_{2} G U I D_S U R_P O S_{i, j, t} + β_{3} G U I D_S U R_N E G_{i, j, t} + β_{4} E A R_S U R_P O S_{i, j, t} + β_{5} E A R_S U R_N E G_{i, j, t} + β_{6} C A R_{i, t} + ε_{i, t},

(1)

where DEP_VAR_i,_j is defined as one of the following, depending on the analysis. First, REVISION1_{i, j}, is the analyst’s revision of year t+1 forecast immediately following year t earnings announcement, measured as the difference between analyst j’s forecast for firm i for year t+1 following the earnings announcement for year t and analyst j’s forecast for firm i for year t+1 before earnings announcement for year t. Second, REVISION2_i,_j, is the analyst’s revision of year t+2 forecast immediately following year t earnings announcement, measured as the difference between analyst j’s forecast for year t+2 after the earnings announcement for year t and analyst j’s forecast for year t+2 before the earnings announcement for year t. Third, CHA_SMOOTHNESS_i,_j is the difference between the post- and pre-earnings announcement earning per share (EPS) forecast smoothness of analyst j for firm i. Smoothness is derived from a series of five years from year t–2 to year t+2 of actual EPS and EPS forecasts and is equal to the negative of the standard deviation of errors from an earnings trend line.⁷GUIDER_i,_t is an indicator variable that takes the value 1 if the firm issues annual earnings guidance at the earnings announcement for year t. We control for both earnings surprise and guidance surprise. GUID_SUR_POS_i,_j, is any positive difference (guidance is higher than the analyst’s forecast) between the earnings guidance for year t+1 and analyst j’s earnings forecast for year t+1 before the earnings announcement for year t. If the difference is negative, then GUID_SUR_POS_i,_j takes the value 0. GUID_SUR_NEG_i,_j is any negative difference (guidance is lower than the analyst’s forecast) between earnings guidance for year t+1 and analyst j’s earnings forecast for year t+1 before the earnings announcement for year t. If the difference is positive, then GUID_SUR_NEG_i,_j takes the value 0. EAR_SUR_POS_i,_j is any positive difference (actual Earnings Per Share(EPS) is higher than the analyst’s forecast) between the actual EPS for year t and analyst j’s earnings forecast for year t before the earnings announcement for year t. If the difference is negative, then EAR_SUR_POS_i,_j takes the value 0. EAR_SUR_NEG_i,_j is any negative difference (actual EPS is lower than the analyst’s forecast) between the actual EPS for year t and analyst j’s earnings forecast for year t before the earnings announcement for year t. If the difference is positive then EAR_SUR_NEG_i,_j take the value 0. We use separate variables for negative and positive surprises, instead of one continuous variable, to allow for differential analysts’ reactions to negative and positive surprises. CAR_i,_t is the market adjusted (value-weighted) cumulative abnormal stock return of the firm measured from one day before the earnings announcement to one day after the earnings announcement. We include this variable because analysts’ post-earnings announcement revisions are likely affected by the market’s reaction to the news.

Across all regressions we include industry-year two-way interactive fixed effects, which control for within-industry peer effects and allow for different cross-industry timing of guidance choice adaptation. We cluster the standard errors by firm-analyst to supplement the industry-year fixed effects, which allows us to account for potential clustering at four levels (i.e., industry, firm, analyst, and year).

To test H2, we consider the iterative nature of the process depicted in Figure 1. We conjecture that firms concerned about reporting a smooth earnings sequence are more likely to issue annual earnings guidance. The annual guidance helps smooth the path of annual earnings expectation. The smooth path of analysts’ expectations, in turn helps achieve the desired goal—to set expectations that enable managers to achieve a smooth earnings path without increasing the risk of missing analyst forecasts. To align our hypothesized process with our empirical model and to mitigate potential endogeneity, we estimate a triangular system of equations. Specifically, we estimate a four-equation recursive model. In each step, we augment the system by including the dependent variable of the previous equation as an independent variable in the immediately following equation which features a new dependent variable that captures the next step in our conceptual sequence. Because all the independent variables are carried over from one regression to the next, in addition to the dependent variable of the prior regression, the residuals are unlikely to be cross-correlated across equations. Thus, the recursive system allows us to estimate each regression separately, independent of the other three regression equations (Greene, 2007, p. 319).

We begin with estimating firms’ choice of whether to issue annual guidance in the first equation. In the second equation, we test the effect of issuing annual guidance on analysts’ EPS forecast smoothness by adding the firms’ choice to provide annual guidance as an independent variable and estimating forecast smoothness as the dependent variable. In the third equation, we follow the sequence by adding forecast smoothness, the dependent variable in the second equation, as an independent variable and test its effect on the likelihood of meeting future quarterly analyst forecasts as the dependent variable. Finally, in the fourth equation, we estimate the impact of meeting future analyst forecasts on actual earnings smoothness.⁸ We choose this particular order to most closely follow the logical sequence of the events. That is, the issuance of guidance should affect analyst forecasts first, and then the analysts’ revised forecasts become the target for firms’ earnings to meet or beat. Meeting and beating analyst forecasts precede the smoothness of actual reported earnings, because the dependent variable in the third equation is observed before the dependent variable in the fourth equation and because meeting and beating a smooth path of analyst earnings forecasts helps achieve the goal of smooth actual earnings. Finally, after firms report actual earnings over multiple years, we observe the smoothness of actual reported earnings.

The first decision we model is the manager’s choice of whether to issue annual earnings guidance. We acknowledge that smoothing analyst expectations may not be the only purpose of management annual guidance. Thus, we include control variables to account for other managerial incentives to issue annual guidance. Specifically, we estimate the following equation:

\begin{matrix} GUIDE R_{i, t} = α_{1} + α_{2} EAR_SMOOTHIN G_{i, t} + α_{3} MTE_HIS T_{i, t} \\ + α_{4} EAR_SUR_PO S_{i, j, t} + α_{5} EAR_SUR_NE {G_{i}}_{, j, t} \\ + α_{6} EXPECTED_GROWT {H_{i}}_{, t} + \sum α CONTRO {L_{i}}_{, t} + {ε_{i}}_{, j, t}, \end{matrix}

(2)

where GUIDER_i,_t is an indicator variable that takes the value 1 if firm i issues annual earnings guidance at year t. EAR_SMOOTHING_i,_t is firm i’s past earnings smoothing activity level. It is aimed at capturing the firm’s perceived importance of a smooth earnings path.⁹ We conjecture that firms that benefit from a smooth earnings path will actively seek to smooth earnings. Thus, we use the active earnings smoothing by the firm as a proxy for the firm’s perceived benefits of a smooth earnings path. We assume that their revealed preference for smooth earnings is likely to persist in the following periods. We predict the coefficient on EAR_SMOOTHING_i,_t to be positive and significant. Following Tucker and Zarowin (2006) we measure this variable as the percentile rank of the correlation between the changes in discretionary accruals and the changes in pre-discretionary income (i.e., earnings minus discretionary accruals) over the five years ending in year t. A negative correlation indicates that managers are engaged in active earnings smoothing. To obtain a measure that is positively correlated with managers’ active smoothing activities, we define EAR_SMOOTHING_i,_t as Tucker and Zarowin’s (2006) correlation rank multiplied by (–1) then plus 1. MTE_HIST_i,_t is the number of quarters that firm i meet or beat analysts’ consensus expectations over the three years prior to the earnings announcement divided by the number of quarters for which IBES data are available (12 for most firms in the sample). Kross et al. (2011) suggest that, in a quarterly guidance setting, firms with a history of meeting or beating forecasts are more likely to issue guidance to ensure a continued meeting or beating string. We control for earnings surprise with EAR_SUR_POS_i,_j and EAR_SUR_NEG_i,_j. To the extent that analysts’ expected earnings growth may affect managers’ decisions to issue guidance to “walk down” analysts’ expectations (Cotter et al., 2006), we control for the expected growth, derived from analysts’ multi-period forecasts (EXPECTED_GROWTH_i,_t).

Finally, we include a set of variables to control for economic factors identified in Ajinkya et al. (2005) as potentially affecting a firm’s likelihood of issuing annual guidance (CONTROL_i,_t). These variables are analysts’ coverage (NUM_ANALYST_i,_t), return volatility (VOLATILITY_i,t), business smoothness (BUSINESS_SMOOTH_i,_t), board independence (BOARDINDEPEND_i,_t), and institutional ownership (INSTHOLDING_i,_t), whether audited by a Big Four auditor (BIGAUDITOR_i,_t), market value of equity (LMVAL_i,_t), whether the firm operates in an industry with high litigation risks (LITIGATION_i,_t), market-to-book ratio of equity (MTB_i,_t), and managerial ability (MA_SCORE_2016_i,_t) (Demerjian et al., 2013). Appendix B provides detailed variable definitions.

In the following equation, we examine the effect of annual earnings guidance on the smoothness of analysts’ earnings forecast paths:

\begin{matrix} POST_EA_SMOOTHNES S_{i, j, t} = β_{1} + β_{2} GUIDE R_{i, t} + β_{3} EAR_SMOOTHIN G_{i, t} \\ + β_{4} MTE_HIS T_{i, t} + β_{5} EAR_SUR_PO S_{i, j, t} \\ + β_{6} EAR_SUR_NE G_{i, j, t} + β_{7} EXPECTED_GROWT {H_{i}}_{, t} \\ + \sum β CONTRO {L_{i}}_{, t} + {y_{i}}_{, j, t}, \end{matrix}

(3)

where POST_EA_SMOOTHNESS_i,_j is the post-announcement forecast smoothness measured over five years from year t–2 to year t+2 around the earnings announcement. It is measured as the negative of the standard deviation of the deviation from an earnings trend line that includes actual EPS for three years starting at year t–2 and two analysts’ EPS forecasts for years t+1 and t+2. We predict a positive coefficient on GUIDER_i,_t.

In the following equation, we examine the effect of the smoothness of analysts’ earnings forecast path on the likelihood of meeting quarterly EPS forecasts in year t+1 and year t+2 as follows:

\begin{matrix} POST_MT E_{i, t} & = δ_{1} + δ_{2} POST_EA_SMOOTHNES S_{i, j, t} + δ_{3} GUIDE R_{i, t} \\ + δ_{4} EAR_SMOOTHIN G_{i, t} + δ_{5} MTE_HIS T_{i, t} + δ_{6} EAR_SUR_PO S_{i, j, t} \\ + δ_{7} EAR_SUR_NE G_{i, j, t} + δ_{8} EXPECTED_GROWT {H_{i}}_{, t} \\ + \sum δ CONTRO {L_{i}}_{, t} + {σ_{i}}_{, j, t}, \end{matrix}

(4)

where POST_MTE_i,_t is the number of quarters that the firm meet or beat analysts’ consensus expectations in the two years following the earnings announcement divided by the number of quarters for which there is IBES data available (eight for most firms in the sample).

Finally, in the following equation, we model the smoothness of actual earnings as follows:

\begin{matrix} ACT_SMOOTHNE S_{i, t} = γ_{1} + γ_{2} POST_MT E_{i, t} + γ_{3} POST_EA_SMOOTHNES S_{i, j, t} \\ + γ_{4} GUIDE R_{i, t} + γ_{5} EAR_SMOOTHIN G_{i, t} + γ_{6} MTE_HIS T_{i, t} \\ + γ_{7} EAR_SUR_PO S_{i, j, t} + γ_{8} EAR_SUR_NE G_{i, j, t} \\ + γ_{9} EXPECTED_GROWT {H_{i}}_{, t} + \sum γ CONTRO {L_{i}}_{, t} + {ω_{i}}_{, j, t}, \end{matrix}

(5)

where ACT_SMOOTHNESS_i,_t is the smoothness of the series of actual EPS measured over five years from year t–2 to year t+2 and is equal to the negative of the standard deviation of deviations from an earnings trend line. Unlike EAR_SMOOTHING_i,_t, which measures managers’ active use of accrual management to smooth the earnings path over time, ACT_SMOOTHNESS_i,_t measures the outcome of the smoothness of the actual earnings path. As the two measures are different both conceptually and empirically, they are modestly correlated with each other and have Pearson (Spearman) correlation of .09 (.08).

Descriptive Statistics

Table 2 reports descriptive statistics for annual guiders, quarterly guiders, and non-guiders separately. We winsorize all continuous variables at the top and bottom 1 percentiles to mitigate the effect of outliers. For all three groups, analysts (on average) revise down their forecast for year t+1 and year t+2 after the earnings announcement of year t, consistent with a relatively larger optimistic bias in analysts’ earlier forecasts (Cotter et al., 2006). REVISION1 and REVISION2 are on average less negative for guiders than for non-guiders with the difference significant at the 1% level.¹⁰ The mean magnitude of revision for year t+1 is larger than for year t+2. Firms that guide are typically larger in size than non-guiders. Consistent with guiders using guidance to “walk down” analysts’ expectations, 58% of annual guidance surprises are negative (4,938 out of 8,494). Annual guiders’ earnings and forecasts are significantly smoother than those of quarterly guiders and of non-guiders. Annual guiders are also more engaged in earnings smoothing albeit active smoothing seems to be pervasive for both quarterly guiders and non-guider. Finally, quarterly guiders exhibit the highest likelihood of meeting or beating analyst forecast while non-guiders exhibit the lowest likelihood of meeting or beating analyst forecasts.

Table 2.

Descriptive Statistics.

	Pooled (P) (N = 34,348)			Annual guider (A) (N = 8,494)			Quarterly guider (Q) (N = 3,681)			Non-guider (N) (N = 22,173)			(A) versus (Q) and (N)
Variable	Mean	Median	StdDev	Mean	Median	StdDev	Mean	Median	StdDev	Mean	Median	StdDev	Difference	t-stat
REVISION1	−0.116	−0.025	0.672	−0.043	−0.010	0.335	−0.129	−0.020	0.809	−0.142	−0.040	0.738	0.097	11.578
REVISION2	−0.092	−0.020	0.621	−0.032	0.005	0.381	−0.066	−0.005	0.652	−0.118	−0.025	0.684	0.079	10.181
PRE_EA_SMOOTHNESS	−0.573	−0.246	1.270	−0.270	−0.135	0.714	−0.525	−0.216	1.371	−0.698	−0.335	1.393	0.403	25.641
POST_EA_SMOOTHNESS	−0.577	−0.260	1.166	−0.279	−0.141	0.723	−0.552	−0.234	1.415	−0.694	−0.345	1.236	0.395	27.349
CHA_SMOOTHNESS	−0.015	−0.004	0.143	−0.011	−0.003	0.079	−0.018	−0.006	0.138	−0.015	−0.004	0.161	0.005	2.813
ACT_SMOOTHNESS	−0.710	−0.352	1.444	−0.356	−0.199	0.612	−0.683	−0.306	1.633	−0.850	−0.446	1.605	0.471	26.326
EAR_SMOOTHING	0.610	0.649	0.253	0.656	0.702	0.256	0.555	0.584	0.273	0.602	0.642	0.246	0.061	19.344
MTE_HIST	0.740	0.750	0.192	0.805	0.833	0.163	0.837	0.833	0.146	0.698	0.750	0.196	0.087	37.176
POST_MTE	0.733	0.750	0.216	0.789	0.875	0.193	0.840	0.875	0.170	0.694	0.750	0.220	0.074	27.874
EAR	2.870	2.030	4.211	3.203	2.680	2.302	2.966	1.680	6.078	2.726	1.800	4.387	0.443	8.421
EAR_SUR	0.034	0.020	0.355	0.034	0.020	0.182	0.061	0.030	0.329	0.029	0.020	0.405	−0.000	−0.063
EAR_SUR_POS	0.089	0.020	0.250	0.060	0.020	0.133	0.093	0.030	0.256	0.100	0.020	0.281	−0.040	−12.766
EAR_SUR_NEG	−0.056	0.000	0.231	−0.026	0.000	0.111	−0.032	0.000	0.191	−0.071	0.000	0.267	0.040	13.743
CAR	0.004	0.002	0.074	0.004	0.003	0.065	0.004	0.007	0.083	0.003	0.001	0.076	0.001	1.334
NUM_ANALYST	17.785	17.000	8.854	16.526	17.000	7.039	16.657	14.000	9.721	18.455	17.000	9.247	−1.674	−15.164
VOLATILITY	0.024	0.021	0.011	0.019	0.017	0.009	0.026	0.024	0.011	0.025	0.022	0.012	−0.006	−42.338
EXPECTED_GROWTH	0.243	0.191	0.794	0.312	0.258	0.481	0.325	0.190	1.045	0.203	0.158	0.839	0.092	9.255
BUSINESS_SMOOTH	−0.255	−0.153	0.364	−0.276	−0.108	0.440	−0.273	−0.164	0.352	−0.244	−0.162	0.332	−0.028	−6.111
BOARDINDEPEND	0.786	0.818	0.122	0.812	0.833	0.112	0.765	0.800	0.127	0.779	0.800	0.123	0.035	22.848
INSTHOLDING	0.791	0.839	0.213	0.814	0.845	0.179	0.837	0.872	0.170	0.774	0.831	0.229	0.031	11.646
BIGAUDITOR	0.971	1.000	0.167	0.964	1.000	0.186	0.956	1.000	0.206	0.976	1.000	0.152	−0.009	−4.479
LMVAL	8.905	8.893	1.673	9.139	9.124	1.557	8.405	8.335	1.498	8.898	8.908	1.724	0.311	14.912
LITIGATION	0.255	0.000	0.436	0.326	0.000	0.469	0.483	0.000	0.500	0.191	0.000	0.393	0.094	17.261
MTB	3.437	2.469	6.435	4.282	3.065	8.518	3.182	2.534	2.504	3.156	2.249	5.912	1.123	13.991
MA_SCORE_2016	0.056	0.002	0.194	0.047	−0.015	0.184	0.040	0.002	0.158	0.063	0.009	0.203	−0.013	−5.165
DIVIDEND	0.535	1.000	0.499	0.577	1.000	0.494	0.354	0.000	0.478	0.549	1.000	0.498	0.056	8.976
CEOAGE	0.252	0.000	0.434	0.257	0.000	0.437	0.233	0.000	0.423	0.253	0.000	0.435	0.007	1.272
ABS_DIFF_LTG ^a	0.218	0.080	0.495	0.095	0.049	0.226	0.353	0.146	0.817	0.268	0.109	0.528	−0.184	−5.999
GUID				3.468	2.870	2.597
GUID_SUR				−0.072	−0.025	0.438
GUID_SUR_POS ^a				0.153	0.075	0.214
GUID_SUR_NEG ^a				−0.225	−0.100	0.493

Note. This table reports the descriptive statistics for the variables across firms providing guidance for different horizons: pooled, annual guiders (A), quarterly guiders (Q), and non-guiders (N). The right-most two columns report and test the difference in the the means between (A) annual guiders and (Q) and (N) quarterly guiders and non-guiders combined. See Appendix B for variable definitions.

For ABS_DIFF_LTG, N = 1,147, 381, 102, and 664 for column group (P), (A), (Q), and (N), respectively. For GUID_SUR_POS, N = 3,257. For GUID_SUR_NEG, N = 4,938.

Empirical Results

Guidance and Analyst Forecasts

Table 3 reports the results for testing H1. H1 implies that firms’ guidance affects analyst forecast for the horizon beyond the horizon for which a firm explicitly guides. Column 1 reports the results for the analysis in which REVISION1 is the dependent variable. This analysis calibrates our subsequent analyses. Analysts revise their estimates for year t+1 following year t earnings announcements. We control for earnings surprise (EAR_SUR_POS and EAR_SUR_NEG) as most guidance is bundled with earnings announcements (e.g., Rogers & Van Buskirk, 2013). We find a positive coefficient on EAR_SUR_NEG and a negative coefficient on EAR_SUR_POS, which suggests that analysts tend to revise their annual earnings forecasts downward, consistent with the downward revision pattern documented in the prior literature (e.g., Ke & Yu, 2006).

Table 3.

Extended Effects of Annual Guidance on Analyst Forecasts.

	(1)	(2)	(3)
Dependent variable (predicted sign)	REVISION1	REVISION2	CHA_SMOOTHNESS
GUIDER (+)	0.065***	0.038***	0.004**
	(10.025)	(4.946)	(2.049)
GUID_SUR_POS	0.630***	0.603***	−0.003
	(11.884)	(10.657)	(0.267)
GUID_SUR_NEG	0.490***	0.342***	0.021*
	(19.229)	(8.005)	(1.908)
EAR_SUR_POS	−0.162***	−0.093**	0.012
	(3.046)	(2.443)	(1.081)
EAR_SUR_NEG	0.521***	0.370***	0.141***
	(7.037)	(6.286)	(7.831)
CAR	1.234***	1.161***	−0.009
	(18.131)	(20.544)	(0.613)
Fixed effects	Industry × Year
N	34,348	34,348	34,348
Model DF	670	670	670
Adjusted R²	.279	.204	.119
Coefficient equality tests
H₀: GUID_SUR_POS = GUID_SUR_NEGF-stat (df₁ = 1, df₂ = 33,677)	8.06	29.75	4.20
p value	.0045	<.0001	.0404
H₀: EAR_SUR_POS = EAR_SUR_NEGF-stat (df₁ = 1, df₂ = 33,677)	1,042.89	510.38	669.83
p value	<.0001	<.0001	<.0001

Note. This table reports the OLS results on the effect of annual guidance on analysts’ annual earnings forecasts for different horizons: Column 1 reports analyst revisions for the year that the firm guided for (REVISION1), and Column 2 reports analyst revisions for the year after that the firm guided for (REVISION2). Column 3 reports the change in the smoothness of the earnings path implied by REVISION1 and REVISION2 along with the three preceding years’ earnings. See Appendix B for variable definitions. t-statistics (in absolute values) are reported in parentheses. F-tests are conducted for the null hypotheses that the coefficients on GUID_SUR_POS and GUID_SUR_NEG are equal and that the coefficients on EAR_SUR_POS and EAR_SUR_NEG are equal.

***

, **, * indicate statistical significance at .01, .05, and .10 levels, respectively.

After controlling for the bundled earnings news, we find that the analyst forecast revision is positively associated with guidance surprise, as the coefficients on guidance surprises (GUID_SUR_POS and GUID_SUR_NEG) are significantly positive. Also, the revision is positively associated with the abnormal returns around a firm’s earnings announcement date, suggesting that the market’s reaction to the announced earnings affects analysts’ expectations beyond the fundamental news (current earnings and guidance) provided by the firm management. This result suggests a feedback effect flowing from the market to analysts’ expectations. Column 2 reports results for REVISION2 and shows that the effects of guidance extend beyond the period for which the company specifically guides.

Because a positive reaction to guidance news could draw analyst forecasts either closer to or further away from future earnings, we also examine whether the distance between analyst forecasts and the actual future earnings is smaller for guiders than for non-guiders. Untabulated results confirm that the difference between analyst forecasts and the actual earnings is smaller for annual guiders than for non-guiders and quarterly guiders, and that guidance indeed draws analyst forecasts closer to future reported earnings.

Column 3 reports the results for the change in smoothness of the series of past earnings and future forecasts. Consistent with H1, we find a significantly positive coefficient on GUIDER (t-stat = 2.049), suggesting that the issuance of annual guidance is associated with smoother analyst forecast paths. The coefficient on GUID_SUR_POS is insignificant whereas the coefficient on GUID_SUR_NEG is positive, suggesting that as the magnitude of negative guidance surprise increases, the analyst forecast path becomes less smooth as large negative surprises constitute a shock that perturbs the smoothness of analyst expectations. An overall positive effect of annual guidance on the smoothness of analyst forecasts supports the conjecture that managers may be motivated to use annual guidance to smooth the path of past earnings and future forecasts. Additional F-tests reveal that the effects of positive and negative guidance news and earnings news on analyst forecast revisions are not symmetric in the sense that their coefficients are statistically significantly different. This finding supports our design of separately evaluating positive and negative guidance news and earnings news. Taken together, the results in Table 3 suggest that the effect of guidance on analyst forecasts extends beyond the horizon the firm guides for, and that guidance reduces uncertainty and helps smooth the past-earnings-future-forecast EPS path.

The Effect of Annual Guidance on Reporting Objectives

To test H2, we estimate a recursive triangular model. In each step, the variable of interest (explanatory variable) is the dependent variable in the previous analysis. For example, the dependent variable of the estimation reported in Column 1 is the variable of interest (explanatory variable) in Column 2. In each analysis, we include industry-year two-way interactive fixed effects. Table 4 reports coefficients and t-stats produced from estimating the recursive model.

Table 4.

Recursive System on Annual Guidance, Expectation Smoothing, Meeting and Beating Analyst Forecasts, and Smoothing of Actual Reported Earnings.

	(1)	(2)	(3)	(4)
Dependent variable (predicted sign)	GUIDER	POST_EA_SMOOTHNESS	POST_MTE	ACT_SMOOTHNESS
POST_MTE (+)				0.111***
				(5.422)
POST_EA_SMOOTHNESS (+)			0.010***	1.112***
			(7.984)	(35.375)
GUIDER (+)		0.045***	0.016***	0.002
		(3.679)	(5.044)	(0.197)
EAR_SMOOTHING (+)	0.176***	0.416***	−0.061***	−0.040**
	(12.390)	(11.068)	(9.820)	(2.259)
MTE_HIST	0.173***	0.319***	0.325***	−0.109***
	(11.379)	(8.008)	(41.257)	(4.176)
EAR_SUR_POS	−0.037***	−1.426***	0.010*	0.030
	(4.947)	(18.786)	(1.852)	(0.681)
EAR_SUR_NEG	0.003	1.543***	0.026***	0.170***
	(0.395)	(23.282)	(4.171)	(2.753)
EXPECTED_GROWTH	−0.021***	−0.180***	−0.016***	−0.262***
	(7.555)	(6.422)	(7.841)	(12.221)
NUM_ANALYST	−0.001***	0.007***	0.001***	−0.000
	(3.339)	(7.758)	(5.246)	(0.505)
VOLATILITY	−7.327***	−0.483	−0.248	1.458**
	(19.151)	(0.385)	(1.311)	(2.163)
BUSINESS_SMOOTH	0.072***	0.142***	0.001	−0.017
	(5.543)	(6.779)	(0.206)	(1.627)
BOARDINDEPEND	0.287***	−0.523***	0.058***	−0.122***
	(12.651)	(8.234)	(4.848)	(3.182)
INSTHOLDING	0.044***	0.522***	0.048***	−0.019
	(4.053)	(11.318)	(7.714)	(0.677)
BIGAUDITOR	−0.067***	−0.137***	−0.035***	−0.026**
	(3.449)	(4.886)	(4.330)	(2.136)
LMVAL	−0.005**	0.012	0.021***	0.035***
	(2.205)	(1.640)	(16.011)	(7.650)
LITIGATION	−0.018	−0.006	−0.002	−0.086***
	(0.624)	(0.351)	(0.284)	(6.951)
MTB	0.001*	0.003***	0.001***	0.002***
	(1.793)	(5.724)	(5.413)	(5.553)
MA_SCORE_2016	−0.250***	−0.293***	−0.041***	−0.103***
	(17.240)	(6.614)	(5.415)	(4.368)
Fixed effects	Industry × Year
N	34,348	34,348	34,348	34,348
Model DF	679	680	681	682
Adjusted R²	.378	.489	.323	.835

Note. This table reports the results of a four-equation recursive system. Column 1 models firms’ decision to issue annual guidance. Column 2 reports the effect of the issuance of annual guidance on the smoothness of analysts’ annual earnings forecasts revised after the earnings announcement. Column 3 reports the effect of the above-mentioned smoothness of analysts’ forecasts on the likelihood of firms’ actual earnings meeting or beating analysts’ forecasts over the two years following the earnings announcement. Column 4 reports the effect of the above-mentioned outcome of firms’ actual earnings meeting or beating analysts’ forecasts on the smoothness of the actual reported annual earnings. The “predicted signs” apply to the corresponding variables in the first column that they appear. See Appendix B for variable definitions. The t-statistics (in absolute values) are reported in parentheses.

***

, **, * indicate statistical significance at .01, .05, and .10 levels, respectively.

Column 1 reports the results for the effect of past smoothing activity on firms’ decision on whether to issue annual guidance for the next year. Consistent with our expectations, the likelihood that a firm would issue annual guidance increases with past efforts to smooth firms’ earnings (coefficient = 0.176, t-stat = 12.390). The likelihood increases by about 4.3 percentage points when EAR_SMOOTHNESS increases by one standard deviation (0.246). This result suggests that firms that value a smooth path of earnings are more likely to issue annual guidance. Consistent with low visibility into future earnings reducing the likelihood of guidance issuance, higher return volatility is associated with a lower likelihood of annual guidance issuance (coefficient = −7.327, t-stat = −19.151). This result holds after controlling for BUSINESS_SMOOTH, which is defined as the negative of the standard deviation of the pre-discretionary income (i.e., earnings minus discretionary income, as in Tucker & Zarowin, 2006 for the past five years, including year t), intended to ascertain that the impact of annual guidance is not due only to inherent smoothness of the underlying operations.

Column 2 reports estimation results for testing the effect of issuing annual guidance on the post-earnings-announcement analyst forecasts smoothness. Consistent with our expectation, we find annual guidance issuance to be positively associated with the smoothness of the series of past earnings and future forecasts (coefficient = 0.045, t-stat = 3.679), which is approximately 13% smoother relative to the median smoothness of non-guiders (–0.345).

Column 3 reports the results on the effect of the smoothness of the pattern of analysts’ annual EPS forecasts on the likelihood of meeting/beating analysts’ quarterly forecasts in the periods following the earnings announcement. Consistent with our expectation, the smoothness of an earnings forecast path does not increase the likelihood of missing analyst forecasts. To the contrary, a smoother earnings forecast pattern is associated with an increased likelihood of meeting or beating quarterly forecasts in the eight quarters that compose the two-year forecast period. The coefficient on forecast smoothness is significantly positive (coefficient = 0.010, t-stat = 7.984). Moreover, the issuance of annual guidance is associated with an increased likelihood of meeting or beating analyst forecasts directly (i.e., independent of the channel of smoother analyst forecasts) (coefficient = 0.016, t-stat = 5.044).

Finally, Column 4 completes the sequence and reports the effect of meeting/beating quarterly analyst forecasts on the smoothness of the firm’s sequence of reported earnings. Consistent with our prediction, the smoothness of actual earnings path (five years centering at the earnings announcement for year t) increases with the frequency of post-earnings-announcement meeting/beating analyst forecasts (coefficient = 0.111, t-stat = 5.422). This result suggests that frequent meeting/beating analyst forecasts, when achieved within a framework of a smooth analyst forecast path, allows firms to stay within the smooth channel of forecasted annual earnings and thus facilitates reporting a smoother actual earnings pattern. The smoothness of analyst forecasts (coefficient = 1.112, t-stat = 35.375) also contributes to the smoothness of actual earnings, with one standard deviation of forecast smoothness (0.723) associated with a 183% increase in earnings smoothness, relative to the median smoothness for non-guiders (0.446).

Overall, the results reported in Table 4 support our H2 on the effect of management annual earnings guidance on helping achieve a smooth path of analyst forecasts and of reported earnings while not sacrificing the meeting of analyst forecast goals.

Annual Guidance and Horizon—Sensitivity Analyses and Cross-Sectional Tests

Throughout the previous sections of this study, we have focused on annual guidance as a means for managers to convey to analysts and investors their expectation of long-term earnings paths. The core of our conjecture is that annual guidance in more likely to be associated with a long-term managerial horizon. In this section, we empirically evaluate whether it is indeed annual guidance (not just any guidance) that drives the observed relations in Table 4 and conduct cross-sectional analyses that connect annual guidance with long-term managerial horizons.

In Table 5, we report the results for the same recursive system as reported in Table 4 with one change: the guider dummy variable in this analysis is defined as equal to 1 if the firm issues only quarterly guidance and 0 otherwise (QGUIDER). We expect quarterly guidance to be aimed at meeting quarterly analysts’ earnings forecast, which is generally considered to be short-term focused in the literature because its outcome is observed immediately at the end of each quarter (e.g., Fuller & Jensen, 2002). We also expect the quarterly guidance to have no effect on forecasts smoothness and actual earnings smoothness.

Table 5.

Recursive System on Quarterly Guidance, Expectation Smoothing, Meeting and Beating Analyst Forecasts, and Smoothing of Actual Reported Earnings.

	(1)	(2)	(3)	(4)
Dependent variable	QGUIDER	POST_EA_SMOOTHNESS	POST_MTE	ACT_SMOOTHNESS
POST_MTE				0.110***
				(5.373)
POST_EA_SMOOTHNESS			0.012***	1.112***
			(8.891)	(35.316)
QGUIDER		−0.133***	0.069***	0.011
		(6.894)	(18.982)	(1.050)
EAR_SMOOTHING	−0.104***	0.410***	−0.051***	−0.039**
	(8.833)	(11.062)	(8.289)	(2.200)
MTE_HIST	0.243***	0.359***	0.311***	−0.111***
	(21.548)	(9.495)	(39.312)	(4.222)
EAR_SUR_POS	−0.013**	−1.429***	0.012**	0.031
	(2.027)	(18.863)	(2.179)	(0.685)
EAR_SUR_NEG	0.005	1.544***	0.024***	0.170***
	(0.847)	(23.285)	(3.892)	(2.746)
EXPECTED_GROWTH	0.015***	−0.179***	−0.017***	−0.262***
	(5.663)	(6.375)	(8.528)	(12.246)
NUM_ANALYST	−0.001	0.007***	0.001***	−0.000
	(1.491)	(7.623)	(5.267)	(0.499)
VOLATILITY	0.523*	−0.744	−0.403**	1.441**
	(1.830)	(0.600)	(2.125)	(2.124)
BUSINESS_SMOOTH	−0.004	0.144***	0.002	−0.017
	(0.398)	(6.989)	(0.466)	(1.612)
BOARDINDEPEND	−0.074***	−0.520***	0.068***	−0.121***
	(3.690)	(8.273)	(5.738)	(3.175)
INSTHOLDING	0.066***	0.533***	0.044***	−0.020
	(7.768)	(11.523)	(7.046)	(0.694)
BIGAUDITOR	−0.013	−0.141***	−0.035***	−0.026**
	(0.701)	(5.097)	(4.392)	(2.146)
LMVAL	−0.016***	0.010	0.022***	0.035***
	(8.184)	(1.319)	(16.721)	(7.551)
LITIGATION	0.000	−0.007	−0.002	−0.086***
	(0.006)	(0.400)	(0.334)	(6.952)
MTB	−0.001***	0.003***	0.001***	0.002***
	(3.436)	(5.649)	(5.686)	(5.579)
MA_SCORE_2016	0.055***	−0.297***	−0.048***	−0.104***
	(4.841)	(6.852)	(6.489)	(4.484)
Fixed effects	Industry × Year
N	34,348	34,348	34,348	34,348
Model DF	679	680	681	682
Adjusted R²	.242	.490	.330	.835

Note. This table reports the results of a four-equation recursive system. Column 1 models firms’ decision to issue quarterly guidance. Column 2 reports the effect of the issuance of quarterly guidance on the smoothness of analysts’ annual earnings forecasts revised after the earnings announcement. Column 3 reports the effect of the above-mentioned smoothness of analysts’ forecasts on the likelihood of firms’ actual earnings meeting or beating analysts’ forecasts over the two years following the earnings announcement. Column 4 reports the effect of the above-mentioned outcome of firms’ actual earnings meeting or beating analysts’ forecasts on the smoothness of the actual reported annual earnings. The “predicted signs” apply to the corresponding variables in the first column that they appear. See Appendix B for variable definitions. The t-statistics (in absolute values) are reported in parentheses.

***

, **, * indicate statistical significance at .01, .05, and .10 levels, respectively.

Consistent with our expectation, in Column 1 of Table 5, the coefficient on past efforts to smooth earnings (EAR_SMOOTHING) is significant but negatively associated with the likelihood of issuing quarterly guidance (QGUIDER) (coefficient = −0.104, t-stat = −8.833), which suggests that unlike annual guidance, the issuance of quarterly guidance is not driven by the desire for a smooth firm earnings path. Column 2 reports a regression aimed at analyzing the effect of issuing quarterly guidance on the smoothness of analysts’ earnings forecasts. Results suggest that issuing quarterly guidance decreases the smoothness of analyst forecasts (coefficient = −0.133, t-stat = −6.894), as the expectation smoothing goal is not congruent with the meeting and beating goal. The issuance of quarterly guidance is associated with about a 37.7% reduction in the smoothness of analyst forecasts relative to the median level for the non-guiders (–0.345). Consistent with Kross et al. (2011), the results in Column 3 suggest that quarterly guidance is associated with a significantly higher likelihood of future meeting/beating analyst forecasts (coefficient = 0.069, t-stat = 18.982). Finally, the coefficient on quarterly guidance in the regression reported in Column 4 is insignificant (coefficient = 0.011, t-stat = 1.050), suggesting that quarterly guidance does not affect the smoothness of actual earnings paths in the future. This contrast between annual guidance and quarterly guidance lends credence to our conjecture that annual guidance is the tool managers use to convey information on their long-term earnings expectations, which are used to smooth the path of analyst forecasts.

To corroborate the evidence on the relation between issuing annual guidance and longer managerial horizon, we consider two factors that likely capture CEO horizon and add them to our prediction models of issuing annual guidance and issuing quarterly guidance. First, we include an indicator variable for regular dividend payers, following DeAngelo et al. (2008).¹¹ Second, we consider the “horizon problem” faced by older CEOs who are more likely to focus on the short term when contemplating retirement (Cassell et al., 2013). In both tests, we control for firms’ LTG and the market-to-book ratio, because they may influence the firms’ payout policy and choice of CEOs. Table 6 reports the results. Consistent with regular dividend payers being more concerned with the long term, we find that they are more likely to issue annual guidance (coefficient = 0.029, t-stat = 4.358) but less likely to issue quarterly guidance (coefficient = −0.024, t-stat = −4.330). Also consistent with our expectation, we find that CEO age that is 60 years or older is negatively associated with the likelihood of issuing annual guidance (coefficient = −0.023, t-stat = −3.842) but positively associated with the likelihood of issuing quarterly guidance (coefficient = 0.014, t-stat = 2.793). Overall, our results in this section support our main argument that managers use annual earnings guidance to smooth the path of analyst forecasts as a means to achieve smooth earnings.

Table 6.

Annual Guidance versus Quarterly Guidance and Managerial Horizon.

	(1)	(2)	(3)	(4)
Dependent variable	GUIDER	QGUIDER	GUIDER	QGUIDER
CEOAGE			−0.023***	0.014***
			(3.842)	(2.793)
DIVIDEND	0.029***	−0.024***
	(4.358)	(4.330)
EAR_SMOOTHING	0.172***	−0.100***	0.178***	−0.106***
	(12.078)	(8.524)	(12.532)	(8.919)
MTE_HIST	0.172***	0.244***	0.173***	0.243***
	(11.233)	(21.668)	(11.322)	(21.567)
EAR_SUR_POS	−0.037***	−0.013**	−0.037***	−0.013**
	(4.993)	(1.991)	(4.881)	(2.063)
EAR_SUR_NEG	0.003	0.005	0.002	0.006
	(0.474)	(0.775)	(0.290)	(0.918)
EXPECTED_GROWTH	−0.019***	0.013***	−0.021***	0.015***
	(6.986)	(5.127)	(7.560)	(5.702)
NUM_ANALYST	−0.001***	−0.001*	−0.001***	−0.001
	(3.007)	(1.805)	(3.486)	(1.391)
VOLATILITY	−6.841***	0.116	−7.412***	0.574**
	(17.761)	(0.401)	(19.328)	(2.007)
BUSINESS_SMOOTH	0.072***	−0.004	0.073***	−0.005
	(5.560)	(0.396)	(5.641)	(0.471)
BOARDINDEPEND	0.278***	−0.068***	0.281***	−0.071***
	(12.318)	(3.335)	(12.472)	(3.543)
INSTHOLDING	0.048***	0.063***	0.046***	0.065***
	(4.409)	(7.337)	(4.150)	(7.677)
BIGAUDITOR	−0.069***	−0.011	−0.072***	−0.010
	(3.514)	(0.635)	(3.699)	(0.530)
LMVAL	−0.007***	−0.014***	−0.005*	−0.016***
	(2.996)	(6.994)	(1.934)	(8.284)
LITIGATION	−0.020	0.002	−0.018	0.000
	(0.693)	(0.090)	(0.625)	(0.007)
MTB	0.001*	−0.001***	0.001	−0.000***
	(1.900)	(3.643)	(1.635)	(3.163)
MA_SCORE_2016	−0.249***	0.054***	−0.250***	0.055***
	(17.161)	(4.784)	(17.361)	(4.872)
Fixed effects	Industry × Year
N	34,348	34,348	34,348	34,348
Model DF	680	680	680	680
Adjusted R²	.379	.243	.378	.242

Note. This table reports the results of linear probability regressions estimating the effect of the CEO horizon (measured as either the firm being a regular dividend payer or the CEO being close to the retirement age) on the likelihood of the firm issuing annual and quarterly guidance. Columns 1 and 2 report the results based on regular dividend for annual and quarterly guidance, respectively. Columns 3 and 4 report the results for annual (quarterly) guidance and CEO age. Predicted signs apply to the first columns in which the corresponding variable appears. See Appendix B for variable definitions. The t-statistics (in absolute values) are reported in parentheses.

***

, **, * indicate statistical significance at .01, .05, and .10 levels, respectively.

How Does Guidance Affect Analyst LTG Forecasts?

The primary interest of this study is to explore the effect of annual guidance on analysts’ multi-period forecasts and the multi-year pattern of earnings. So far, our analyses have focused on analyst forecasts for the upcoming two years. If guidance has an impact on analyst forecasts over a longer horizon, we expect that guidance also affects analysts’ LTG forecasts as well. This notion motivates our investigation of analysts’ provision of LTG forecasts.

We obtain analysts’ LTG forecasts from IBES. For each earnings announcement in our sample, we search for analysts’ LTG forecasts issued within 30 days after the earnings announcement. The incidence of analysts’ LTG forecasts is similar over a 30-day window prior to the earnings announcement. We then compare the LTG rate, expressed as an annual percentage, with the annual growth rate implied from analysts’ two-year-ahead earnings forecasts.¹² If management guidance helps smooth analyst expectation over longer horizons, then we should expect analysts’ LTG forecasts to be more in line with their growth rate implied from the two years ahead EPS forecasts. We measure the absolute value of the difference between analysts’ LTG forecasts and the growth rate implied by analyst forecasts for year t+1 and year t+2 for each observation with sufficient data, and label the percentile rank of this variable as ABS_DIFF_LTG. We regress this variable on an indicator variable, GUIDER, along with other control variables we included in our previous analysis. We expect that the indicator for GUIDER should load negatively in the regression if management guidance helps smooth analysts’ multi-horizon expectations.

Table 7 reports the results. Because analysts issue LTG forecasts relatively infrequently, our sample contains only 1,147 observations with sufficient control variables. Absent any fixed effect, we find that the GUIDER indicator variable is significantly negatively associated with ABS_DIFF_LTG (coefficient = −0.106, t-stat = −3.465) after controlling for the simultaneous release of positive or negative earnings news and guidance news. This finding is consistent with our expectation that management guidance helps align analysts’ LTG forecast with their implied rate of growth from their annual forecasts for the upcoming two years. This association remains statistically significant when we include year fixed effects (t-stat = −3.072), industry fixed effects (t-stat = −2.423), or industry and year fixed effects simultaneously (t-stat = −2.533). When we replace the industry and year fixed effects with industry-year two-way interactive fixed effects, the result is no longer significant (t-stat = −1.344).

Table 7.

Analyst Long-Term Growth (LTG) Forecast Analysis.

	(1)	(2)	(3)	(4)	(5)
Dependent variable (predicted sign)	ABS_DIFF_LTG	ABS_DIFF_LTG	ABS_DIFF_LTG	ABS_DIFF_LTG	ABS_DIFF_LTG
GUIDER (–)	−0.106***	−0.100***	−0.103**	−0.104**	−0.072
	(3.465)	(3.072)	(2.423)	(2.533)	(1.344)
GUID_SUR_POS	−0.133**	−0.066	−0.138	−0.050	−0.094
	(1.981)	(0.899)	(1.246)	(0.514)	(0.912)
GUID_SUR_NEG	0.014	−0.077	−0.003	−0.071	0.138
	(0.176)	(0.827)	(0.027)	(0.642)	(1.033)
EAR_SUR_POS	0.029	0.053	0.007	0.039	−0.036
	(0.309)	(0.552)	(0.089)	(0.444)	(0.435)
EAR_SUR_NEG	−0.051	−0.076*	−0.036	−0.064	0.025
	(1.155)	(1.792)	(0.676)	(1.280)	(0.398)
MTE_HIST	−0.199*	−0.137	−0.131	−0.088	−0.205
	(1.696)	(1.234)	(0.996)	(0.716)	(1.232)
NUM_ANALYST	0.004**	0.004**	0.005**	0.005**	0.004
	(2.121)	(2.108)	(1.995)	(2.054)	(1.327)
VOLATILITY	9.128***	14.087***	7.728***	12.545***	12.909***
	(4.755)	(5.962)	(3.446)	(4.437)	(4.506)
Industry × Year fixed effects	No	No	No	No	Yes
Industry fixed effects	No	No	Yes	Yes	No
Year fixed effects	No	Yes	No	Yes	No
N	1,147	1,147	1,147	1,147	1,147
Model DF	8	23	49	64	317
Adjusted R²	.059	.181	.093	.206	.373

Note. This table reports the results of an analysis testing the effect of annual guidance on the alignment between analyst’s forecasts of long-term earnings growth and the growth implied from the analyst’s one-year ahead and two-year ahead EPS forecasts. Each column includes a different set of fixed effects. See Appendix B for variable definitions. t-statistics (in absolute values) are reported in parentheses.

***

, **, * indicate statistical significance at .01, .05, and .10 levels, respectively.

In summary, our analysis of analysts’ LTG forecasts provides confirming evidence that managements’ annual guidance induces analysts to issue a smooth path of earnings forecasts for future periods. This finding highlights the importance of extending the research scope of “expectation management” beyond the horizon that is explicitly guided by the management.

Conclusion

In this article, we examine whether firms use annual guidance as a tool to achieve a smooth analyst forecast path and ultimately report a smooth earnings path without increasing the risk of missing analyst forecasts. The evidence provided is consistent with the notion that the effect of firms’ annual guidance is not confined to the specific period explicitly guided for, but extends to future periods. Our analyses also suggest that the decision to provide annual guidance is partly driven by the importance of a smooth path of EPS to the firm, and that annual guidance is positively associated with a smoother path of analyst forecasts, which in turn drives a smoother pattern of reported earnings, as managers endeavor to meet/beat analyst forecasts. Unlike annual guidance, we show that short horizon quarterly guidance does not achieve these goals.

Our study extends the existing literature on expectation management to a multi-period setting through the investigation of the multi-period effects of guidance on analyst forecasts. Moreover, we connect this literature with the earnings smoothing literature by documenting a new phenomenon of “expectation smoothing,” that is, the management of analyst expectations so that a smoother path of earnings expectations can be achieved. Our evidence sheds new light on the complex process of management guidance decisions and financial reporting strategies. Although we have focused on the effect of managers’ quantitative guidance, future research can explore how managers use qualitative information, such as linguistic tones and verbal cues, to affect market expectations over multiple periods. Future research can also examine the variations in the benefits to smoother earnings when expectation smoothing is either present or absent. Also, future research could explore the potential interactive effect between “expectation management” and “earnings smoothing,” which we cannot test, as our recursive regression design models them as sequential decisions.

Footnotes

Appendix A

Appendix B

Variable Definitions.

Variable	Definition
REVISION1	An analyst’s first forecast regarding year t+1 after the earnings announcement for year t, minus his or her last forecast regarding year t+1 before the earnings announcement for year t.
REVISION2	An analyst’s first forecast regarding year t+2 after the earnings announcement for year t, minus his or her last forecast regarding year t+2 before the earnings announcement for year t.
PRE_EA_SMOOTHNESS	Minus one (–1) times the standard deviation of the residuals from a five-year trend line of actual earnings from year t–2 and year t–1, and the analyst’s last earnings forecasts for year t, year t+1, and year t+2 issued before the earnings announcement for year t. The trend-line is obtained by estimating an OLS regression.
POST_EA_SMOOTHNESS	Minus one (–1) times the standard deviation of the residuals from a five-year trend-line of actual earnings from year t–2, year t–1, and year t, and the analyst’s first earnings forecasts for year t+1 and year t+2 issued after the earnings announcement for year t. The trend-line is obtained by estimating an OLS regression.
CHA_SMOOTHNESS	POST_EA_SMOOTHNESS minus PRE_EA_SMOOTHNESS.
ACT_SMOOTHNESS	Minus one (–1) times the standard deviation of the residuals from a five-year trend-line of actual earnings from year t–2 to year t+2. The trend-line is obtained by estimating an OLS regression.
EAR	Actual reported earnings per share for year t collected from IBES.
EAR_SUR	EAR minus the analyst’s last earnings forecast for year t before the earnings announcement for year t.
EAR_SUR_POS	The larger of EAR_SUR and 0.
EAR_SUR_NEG	The lesser of EAR_SUR and 0.
GUIDER	An indicator variable set to one if the firm provides an annual earnings guidance for year t+1 at the earnings announcement of year t, and zero otherwise.
GUID	Guidance of earnings in year t+1 provided at the earnings announcement of year t.
GUID_SUR	GUID minus the analyst’s last earnings forecast for year t+1 before the earnings announcement for year t.
GUID_SUR_POS	The larger of GUID_SUR and 0.
GUID_SUR_NEG	The lesser of GUID_SUR and 0.
NUM_ANALYST	Number of analysts covering the firms.
VOLATILITY	Standard deviation of daily raw returns
EAR_SMOOTHING	Earnings smoothing measure based on Tucker and Zarowin (2006), multiplied by (–1) then +1, so that higher values indicate more aggressive income smoothing. The Tucker and Zarowin measure is based on the correlation between percentile ranks of changes in discretionary accruals and percentile ranks of changes in pre-discretionary income (i.e., earnings minus discretionary accruals).
EXPECTED_GROWTH	The expected growth rate proxied by the slope of the pre-earnings announcement analyst forecasts over the next two years.
CAR	Cumulative market adjusted abnormal returns over the three days centering on the earnings announcement for year t.
BUSINESS_SMOOTH	The negative of business volatility of the firm, proxied by the standard deviation of the pre-discretionary income (i.e., earnings minus discretionary income, as in Tucker & Zarowin, 2006) for the past five years, including year t.
BOARDINDEPEND	Percentage of the board of directors that are not also officers of the firm.
INSTHOLDING	Percentage of the company’s aggregate common stock held by institutions.
BIGAUDITOR	Indicator variable set to 1 if the company is audited by one of the Big Four auditors, and 0 otherwise.
LMVAL	Log of the market value of a firm’s common equity at the beginning of the fiscal period.
LITIGATION	Indicator variable set to 1 if the company belongs to a high litigation industry including biotechnology (2,833–2,836 and 8,731–8,734), computers (3,570–3,577 and 7,370–7,374), electronics (3,600–3,674), and retail (5,200–5,961) industries, and 0 otherwise.
MTB	Ratio of market value to book value of common equity at the beginning of the fiscal period.
MA_SCORE_2016	Management ability score obtained from faculty.washington.edu/pdemerj/data.html
DIVIDEND	Indicator variable set to 1 if the company is a regular dividend payer.
CEOAGE	Indicator variable set to 1 if the CEO’s age is 60 years or over.
MTE_HIST	The number of quarters that the firm meet or beat mean analysts’ expectations in the three years prior to the earnings announcement divided by number of quarters for which there is IBES data available (12 for most firms in the sample).
POST_MTE	The number of quarters that the firm meet or beat mean analysts’ expectations in the two years following the earnings announcement divided by number of quarters for which there is IBES data available (eight for most firms in the sample).
ABS_DIFF_LTG	Percentile ranks of the absolute value of the difference between analyst’s long-term growth forecasts minus the growth rate implied by his or her earnings forecasts for year t+1 and year t+2.

Acknowledgements

The author(s) thank Stan Markov (editor) and two anonymous reviewers for their constructive comments. The author(s) are also grateful for helpful comments from Jeffrey Callen, Brian Rountree (discussant), Vedran Capkun, Jerold Zimmerman, and workshop participants at New York University, University of Toronto, HEC Paris, Yeshiva University, Hong Kong Polytechnic University, Rutgers University, and the AAA Annual Meeting. All errors and omissions are authors’ own.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Michael (Minye) Tang

Data Availability

Data are commercially available.

Notes

References

Ajinkya

B. B.

Bhojraj

Sengupta

(2005). The association between outside directors, institutional investors and the properties of management earnings forecasts. Journal of Accounting Research, 43(3), 343-376.

Ajinkya

B. B.

Gift

M. J.

(1984). Corporate managers’ earnings forecasts and symmetrical adjustments of market expectations. Journal of Accounting Research, 22(2), 425-444.

Barnea

Ronen

Sadan

(1976). Classificatory smoothing of income with extraordinary items. The Accounting Review, 51(1), 110-122.

Bartov

Givoly

Hayn

(2002). The rewards to meeting or beating earnings expectations. Journal of Accounting and Economics, 33(2), 173-204.

Brown

L. D.

Caylor

M. L.

(2005). A temporal analysis of quarterly earnings thresholds: Propensities and valuation consequences. The Accounting Review, 80(2), 423-440.

Cassell

C. A.

Huang

S. X.

Sanchez

J. M.

(2013). Forecasting without consequence? Evidence on the properties of retiring ceos’ forecasts of future earnings. The Accounting Review, 88(6), 1909-1937.

Chen

Huang

A. G.

Jha

(2012). Idiosyncratic return volatility and the information quality underlying managerial discretion. Journal of Financial and Quantitative Analysis, 47(4), 873-899.

Chen

Kim

J.-B.

Yao

(2017). Earnings smoothing: Does it exacerbate or constrain stock price crash risk? Journal of Corporate Finance, 42, 36-54.

Chen

Matsumoto

D. A.

Rajgopal

(2011). Is silence golden? An empirical analysis of firms that stop giving quarterly earnings guidance. Journal of Accounting and Economics, 51(1–2), 134-150.

10.

Cotter

Tuna

Wysocki

P. D.

(2006). Expectations management and beatable targets: How do analysts react to explicit earnings guidance? Contemporary Accounting Research, 23(3), 593-624.

11.

DeAngelo

Skinner

D. J.

(2008). Corporate payout policy. Foundations and Trends in Finance, 3(2–3), 95-287.

12.

DeFond

M. L.

Park

C. W.

(1997). Smoothing income in anticipation of future earnings. Journal of Accounting and Economics, 23(2), 115-139.

13.

Degeorge

Patel

Zeckhauser

(1999). Earnings management to exceed thresholds. The Journal of Business, 72(1), 1-33.

14.

Demerjian

P. R.

Lev

Lewis

M. F.

McVay

S. E.

(2013). Managerial ability and earnings quality. The Accounting Review, 88(2), 463-498.

15.

Dou

Hope

O.-K.

Thomas

W. B.

(2013). Relationship-specificity, contract enforceability, and income smoothing. The Accounting Review, 88(5), 1629-1656.

16.

Einhorn

Ziv

(2008). Intertemporal dynamics of corporate voluntary disclosures. Journal of Accounting Research, 46(3), 567-589.

17.

Feng

Koch

A. S.

(2010). Once bitten, twice shy: The relation between outcomes of earnings guidance and management guidance strategy. The Accounting Review, 85(6), 1951-1984.

18.

Francis

LaFond

Olsson

P. M.

Schipper

(2004). Costs of equity and earnings attributes. The Accounting Review, 79(4), 967-1010.

19.

Fuller

Jensen

M. C.

(2002). Just say no to Wall Street: Putting a stop to the earnings game. Journal of Applied Corporate Finance, 14(4), 41-46.

20.

Graham

J. R.

Harvey

C. R.

Rajgopal

(2005). The economic implications of corporate financial reporting. Journal of Accounting and Economics, 40(1–3), 3-73.

21.

Greene

W. H.

(2007). Econometric analysis (6th ed.). Prentice Hall.

22.

Gujarati

D. N.

(1995). Basic econometrics (3rd ed.). McGraw-Hill.

23.

Healy

P. M.

(1985). The effect of bonus schemes on accounting decisions. Journal of Accounting and Economics, 7(1–3), 85-107.

24.

Hirst

D. E.

Koonce

Venkataraman

(2008). Management earnings forecasts: A review and framework. Accounting Horizons, 22(3), 315-338.

25.

Jiang

J. X.

(2008). Beating earnings benchmarks and the cost of debt. The Accounting Review, 83(2), 377-416.

26.

Kadan

Madureira

Wang

Zach

(2009). Conflicts of interest and stock recommendations: The effects of the global settlement and related regulations. The Review of Financial Studies, 22(10), 4189-4217.

27.

(2006). The effect of issuing biased earnings forecasts on analysts’ access to management and survival. Journal of Accounting Research, 44(5), 965-999.

28.

Kross

W. J.

B. T.

Suk

(2011). Consistency in meeting or beating earnings expectations and management earnings forecasts. Journal of Accounting and Economics, 51(1–2), 37-57.

29.

Matsumoto

D. A.

(2002). Management’s incentives to avoid negative earnings surprises. The Accounting Review, 77(3), 483-514.

30.

Matsunaga

S. R.

Park

C. W.

(2001). The effect of missing a quarterly earnings benchmark on the CEO’s annual bonus. The Accounting Review, 76(3), 313-332.

31.

McInnis

J. M.

(2010). Earnings smoothness, average returns, and implied cost of equity capital. The Accounting Review, 85(1), 315-341.

32.

Myers

J. N.

Myers

L. A.

Skinner

D. J.

(2007). Earnings momentum and earnings management. Journal of Accounting, Auditing & Finance, 22(2), 249-284.

33.

Pindyck

R. S.

Rubinfeld

D. L.

(1997). Econometric models and economic forecasts (4th ed.). McGraw-Hill/Irwin.

34.

Ramnath

Rock

Shane

P. B.

(2008). The financial analyst forecasting literature: A taxonomy with suggestions for further research. International Journal of Forecasting, 24(1), 34-75.

35.

Rogers

J. L.

Van Buskirk

(2013). Bundled forecasts in empirical accounting research. Journal of Accounting and Economics, 55(1), 43-65.

36.

Ronen

Sadan

(1981). Smoothing income numbers: Objectives, means, and implications. Addison-Wesley.

37.

Rountree

B. R.

Weston

J. P.

Allayannis

(2008). Do investors value smooth performance? Journal of Financial Economics, 90(3), 237-251.

38.

Skinner

D. J.

Sloan

R. G.

(2002). Earnings surprises, growth expectations, and stock returns or don’t let an earnings torpedo sink your portfolio. Review of Accounting Studies, 7(2–3), 289-312.

39.

Subramanyam

K. R.

(1996). The pricing of discretionary accruals. Journal of Accounting and Economics, 22(1–3), 249-281.

40.

Tang

(in press). Consistency in management earnings guidance patterns. The International Journal of Accounting.

41.

Tang

Venkataraman

(2018). How patterns of past guidance provision affect investor judgments: The joint effect of guidance frequency and guidance pattern consistency. The Accounting Review, 93(3), 327-348.

42.

Tang

Yao

Zarowin

P. A.

(2020). Once is not enough: The determinants and consequences of management updates of annual forecasts [Working paper]. https://www.uts.edu.au/sites/default/files/ADG_Conf2015_Michael%20Tang.pdf

43.

Tang

Zhang

(2018). Range has it: Decoding the information content of forecast ranges. Review of Accounting Studies, 23(2), 589-621.

44.

Tucker

J. W.

Zarowin

P. A.

(2006). Does income smoothing improve earnings informativeness? The Accounting Review, 81(1), 251-270.