Docstoc

Materiality and Disclosure

Document Sample
Materiality and Disclosure Powered By Docstoc
					                      A Unified Framework of Management Earnings Forecasts:
                      Voluntary, Opportunistic and Disclose or Abstain Incentives



                                                 Edward Li
                                       edward.li@simon.rochester.edu


                                              Charles Wasley
                                  charles.wasley@simon.rochester.edu


                                             Jerold Zimmerman
                                   jerry.zimmerman@simon.rochester.edu


                                         Simon School of Business
                                          University of Rochester




                                                ABSTRACT

Extant literature views management earnings forecasts (MEFs) as voluntary corporate disclosures
designed to increase transparency or allow managers to trade opportunistically in their firm’s stock based
on inside information. We offer and test a unified framework of management guidance that incorporates
these two reasons along with a manager’s affirmative duty to disclose material information or abstain
from trading in the firm’s stock. Since prior tests of the theories, rationales and economic effects of
management’s decision to issue a MEF assume all such forecasts are issued voluntarily such tests are mis-
specified due to identification problems and correlated omitted variables. An empirical model combining
all three rationales for managers to issue forecasts reveals that the typical variables of interest in MEF
studies exhibit different explanatory power depending on which of the three underlying rationales is most
likely driving the manager’s decision to issue guidance.



                                        Current draft: January 7, 2010


                  Preliminary and incomplete, do not quote or circulate without permission




We gratefully acknowledge the financial support provided by the Simon School at the University of
Rochester.
                       A Unified Framework of Management Earnings Forecasts:
                        Voluntary, Strategic, and Disclose or Abstain Incentives

1. Introduction

        Healy and Palepu (2001), in reviewing the empirical disclosure literature, call for research into
the various factors that affect management’s disclosure choices. We provide evidence on this question by
investigating the effects of voluntary disclosure incentives, strategic/opportunistic motives related to
insider trading, and the disclose or abstain (DOA) provisions underlying the securities laws and exchange
requirements have on managers’ decision to disclose a management earnings forecast (MEF). 1
        A commonly held view in the disclosure literature is that managers voluntarily issue MEFs to
improve transparency (which presumably serves to increase liquidity in the firm’s stock by reducing
information asymmetry and increasing average investor information precision), and to lower the firm’s
cost of capital (see, e.g., Diamond and Verrecchia, 1991; Ajinkya and Gift, 1984; Kasznik and Lev, 1995;
Frankel et al., 1995; Coller and Yohn, 1997, Williams, 1996; Graham et al., 2005). Other researchers
hypothesize and document that management guidance is strategic and allows managers to
opportunistically trade in their firm’s stock (Noe 1999; Rogers and Stocken 2005; Cheng and Lo 2006).
        An additional rationale for managers to issue MEFs has received much less attention in the
literature. In particular, insider trading regulations imposed by the Securities Acts and stock exchange
listing requirements require managers to disclose material information prior to trading in their firms’
stock (Heitzman, et al., 2010). Specifically, SEC Rule 10b-5 prohibits insiders from trading based of
material information. The insider must either disclose the information or abstain from trading (“disclose
or abstain”, hereafter DOA). Thus, an additional rationale for managers to release MEFs is to comply
with insider trading regulations since such disclosures serve as a mechanism for managers to disclose
material information prior to executing their trades. Merely delaying trading until quarterly earnings are
announced does not reveal the manager’s privately held information about next quarter’s earnings. As a
result, managers wishing to comply with insider trading laws will disclose their material private
information about future quarters’ earnings, perhaps by issuing a MEF along with the earnings
announcement.
        While the disclosure literature recognizes that MEFs are issued for a variety of reasons, no paper
views the decision to issue a MEF comprehensively. Rather, a typical study assumes one motive for why
MEFs are released (e.g., to reduce the threat of shareholder litigation or to capture cost of capital benefits)

1
  We use the terms “MEFs,” “management guidance,” and “management forecasts” interchangeably. “Management
warnings” and “earnings preannouncements” are also used interchangeably. Warnings (preannouncements) are
defined as management disclosures about the current quarter earnings issued from 3 weeks before the close of the
fiscal quarter up to the announcement of actual earnings for the fiscal quarter, whereas MEFs are disclosures about
current fiscal period earnings issued prior to the 3 weeks before the end of the fiscal period.

                                                                                                                  1
and then either introduces control variables to capture other disclosure incentives or ignores other
incentives altogether. Moreover, no paper tests the extent to which MEFs are released because of the
DOA requirements. Clearly, MEFs are made for a variety of reasons. Thus, including all MEFs in a
given study, testing say their disclosure effects on the firm’s cost of capital (see, e.g., Coller and Yohn
1997), at best reduces the power of the test by including some MEFs that were not issued voluntarily, and
at worst introduces a correlated omitted-variables problem that may compromise the inferences drawn if
some of the independent variables used to measure the “cost of capital” effects are correlated with other
MEF disclosure incentives such as DOA requirements. More generally, not controlling for how DOA
incentives motivate managers to issue MEFs opens such papers up to the criticism of failing to control for
alternative hypotheses that might also explain the results.
        To address the shortcomings in prior MEF research, we offer a unified framework of MEFs and
perform empirical tests that incorporate competing economic reasons for why managers issue earnings
guidance. The competing reasons are voluntary disclosure benefits, strategic/opportunistic trading in their
firm’s stock, and compliance with the DOA requirements. To conduct such tests we develop an algorithm
that separates MEFs into three mutually exclusive subsets based on the above three reasons managers
issue guidance. Such an algorithm will be useful to future researchers interested in creating more
powerful tests of the economic determinants and consequences of management earnings guidance. 2
        Our unified framework of MEF disclosure starts with the premise that throughout the fiscal
period managers continuously decide whether to issue guidance and whether to trade in their firm’s
securities. Since the actual motives driving these decisions are unobservable, researchers cannot know
the direction of causality (or if any causal relation exists) between these two decisions for any given MEF
observation or insider trading transaction. For example, a manager might be faced with a large liquidity
shock (i.e., a divorce settlement) and must sell some stock. Wanting to comply with the insider trading
requirements because the manager is in possession of material information about current quarter earnings,
he/she issues a MEF and then trades. In this case the desire to trade causes the release of the MEF.
Alternatively, a manager observes that the market’s consensus forecast of current quarter’s earnings is too
high and wishing to maintain the firm’s reputation for transparency decides to reveal his/her private
information via a MEF. However, if the manager believes that the market “over-reacted” to the MEF and
thus believes the firm is now under-valued, he/she only then decides to trade. Here, issuance of the MEF
“caused” the insider trading. Finally, scenarios can exist where there is no causal relation between a
MEF and any insider trading if the manager behaves opportunistically (in the ex-ante sense) by buying


2
   Our analysis focuses on MEFs issued before three weeks prior to the end of the fiscal period. That is, we do not
focus on the managerial decision to issue earnings warnings or preannouncements. Warnings/preannouncements are
likely made to reduce expected litigation costs (Skinner, 1994 and 1997).

                                                                                                                 2
stock before issuing a good news forecast. In this case, the manager jointly decided (ex-ante) to issue a
forecast and to trade.
        To distinguish between the opportunistic and voluntary incentives for issuing MEFs, it is
important to differentiate between managers’ ex-ante opportunism and ex-post opportunism with respect
to insider trading and MEF disclosure decisions. Ex- ante opportunism occurs when managers withhold
material private information or release biased material disclosures. “Biased” disclosures are those that
differ from the manager’s private information. Ex-ante opportunism transfers wealth from existing share-
holders to the manager. Such behavior is fraudulent and subjects the manager to civil and criminal
penalties if detected. If the manager discloses unbiased information, observes the market reaction, and
then trades with the belief that the market mis-prices the news, such a manager is behaving
opportunistically, but the opportunism is ex-post. This is because the manager faithfully (unbiasedly)
represented his/her private information, and by trading is simply betting the market is wrong.
Importantly, here the manager fulfilled his/her fiduciary responsibilities to the market by faithfully
disclosing unbiased information prior to trading. Simply put, the marginal investor and the manager have
different expectations of firm value. No wealth transfers result until the actual results are known. While
it is straight-forward to describe these alternative scenarios conceptually, it is very difficult to empirically
identify ex-ante opportunism from ex-post opportunism because one cannot observe biased disclosures
except in rare cases of proven or admitted fraud.
        Not knowing the exact motivations for insider trading and the decision to issue a MEF
necessitates indirect tests of our unified framework. In particular, we first sort all MEFs into three
mutually exclusive sub-samples based on characteristics of the forecast (e.g., market reaction to the MEF,
forecast error, etc). Each sub-sample is constructed to contain forecasts most likely to have been issued
because of voluntary disclosure incentives, strategic/opportunistic motives, or because of DOA
requirements. We then estimate a model that includes all three sub-samples to predict the disclosure of a
MEF in a given firm quarter. The model contains independent variables that capture the voluntary
disclosure, opportunistic and DOA incentives. We predict that the voluntary disclosure variables should
have greater explanatory power in the voluntary disclosure sub-sample than in the other two sub-samples.
Likewise, the independent variables that proxy for opportunistic and DOA incentives will have greater
explanatory power in those sub-samples, respectively.
        Using 23,144 MEFs from the First Call CIG database spanning the 1998 – 2007 period, we
partition MEFs into three mutually exclusive and exhaustive categories: voluntary forecasts (n=15,751),
opportunistic forecasts (n=3,156) and DOA forecasts (n=4,237). While our algorithm classifies roughly
70% of all forecasts as voluntary, a substantial portion (30%) is not. Univariate tests provide evidence
consistent with our three-way categorization of MEFs, and hence add construct validity to our

                                                                                                               3
classification algorithm. First, MEFs classified as DOA are more likely to be issued concurrently with an
earnings announcement (76%) than opportunistic MEFs (52%) or voluntary MEFs (59%). Second, DOA
MEFs are more likely to have insider trading in the five days following the release of the MEF than either
opportunistic or voluntary MEFs. Third, given there is insider trading in the five days following the
release of the MEF, the intensity of that insider trading following DOA MEFs is greater than that in the
same five day window following the release of either a voluntary or an opportunistic MEF.
         The central feature of our empirical tests is the estimation of a multinomial probit model using
our three sub-samples of MEFs where the dependent variable is one, two, or three (depending on the
subsample of the MEF) if a management forecast(s) was issued in the fiscal quarter, or zero otherwise.
This allows the coefficients on the independent variables included in the model to differ across the MEF
subsample partitions. As predicted by our unified MEF framework, the voluntary disclosure proxies
have greater explanatory power in the sub-sample of voluntary forecasts than in the other two sub-
samples, the opportunistic disclosure proxy variables have greater explanatory power in the sub-sample of
opportunistic forecasts than in the other two sub-samples, and the DOA proxy variables have greater
explanatory power in the sub-sample of DOA forecasts than in the other two sub-samples.
         Our results have implications for prior and future research using management guidance to test
voluntary disclosure incentives. First, while our classification suggests that roughly 70% of management
guidance appears to be voluntary, the balance is not. As a result, studies treating all MEFs as if they are
homogenous and can be grouped into a “one-size-fits all” classification weakens the underlying tests in
such studies. Moreover, as discussed later, some of the independent variables used to proxy for voluntary
disclosure incentives also proxy for DOA incentives which brings into question the inferences regarding
the predicted affects of the voluntary disclosure theory tested in such studies. Second, the well-accepted
association between voluntary disclosures such as MEFs and a firm’s cost of capital need not be causal
(see, e.g., Heitzman, et al., 2010). A widely-held belief and cited result in the disclosure literature is that
disclosure causes the lower cost of capital. An alternative interpretation is that firms with a lower cost of
capital have stock prices that are more sensitive to earnings information and as a result such firms have
lower materiality thresholds. More precisely, because firms with a low cost of capital have lower
materiality thresholds they have to discuss more in general, and in specific, are more likely to issue a
MEF to comply with DOA provisions requiring a manager to disclose all material information before
trading in the firm’s stock. 3   Prior papers infer that independent variables capturing voluntary disclosure


3
  The intuition for why firms with a lower cost of capital have lower materiality thresholds is as follows. The cost
of capital is positively related to a firm’s beta and a firm’s beta is negatively related to its earnings response
coefficient (ERC) which measures the sensitivity of the firm’s stock price to accounting information (Collins and
Kothari, 1989). So firms with higher ERCs have lower betas, and lower beta firms have a lower cost of capital.
Since the stock prices of firms with a lower cost of capital are more sensitive to accounting information (because as

                                                                                                                        4
incentives are consistent with voluntary disclosure theory. However, many of those same variables also
capture the materiality of the disclosure and hence measure DOA incentives, leading to an identification
problem in many prior MEF studies.
         Our study is not without limitations. It is important to emphasize that we are testing a joint
hypothesis, namely, the accuracy of the algorithm we use to classify the underlying motives of the
managers issuing the MEFs in our sample and our unified framework of management forecasts. Any
algorithm attempting to classify unobservable managerial motives will produce classification errors. This
problem exists in studies requiring measures of earnings management, conservatism, abnormal returns,
and unexpected earnings. If the MEF misclassifications are purely random, this would simply reduce the
power of the test to distinguish between the competing motives for managers to issue forecasts that we
outline in our unified framework. On the other hand, if the misclassifications are not random, our
inferences about the relative importance of the competing motives for managers to issue forecasts will be
misleading. In the end, we realistically expect that our algorithm will produce some misclassifications.
As a result, we do not predict that our independent variables used to explain MEF disclosure will only be
significant in their corresponding sub-sample. Rather, classification errors may lead an independent
variable only predicted to explain voluntary MEFs to also appear significant in say the opportunistic sub-
sample model if enough voluntary forecasts are misclassified as opportunistic.
         The remainder of the paper is organized as follows. Section 2 summarizes the relevant literature
on MEFs and insider trading. Section 3 develops our unified framework of MEFs, describes an algorithm
for sorting MEFs into three mutually exclusive and exhaustive subsets and lays out the empirical model
used to test our unified framework. Section 4 describes our data and provides summary statistics. Section
5 reports the results and Section 6 concludes.


2. Literature Review
         This section summarizes prior research and is organized into four sub-sections. The first
summarizes research on MEFs, while the second and third discuss insider trading research and research
focusing jointly on MEFs and insider trading, respectively. The fourth subsection summarizes the
securities regulation underlying managers’ affirmative duty to disclose their material inside information
or abstain from trading.




just discussed, ERCs are decreasing in the cost of capital), firms with a lower cost of capital have lower materiality
thresholds, and hence are more likely to disclose a given piece of information such as a MEF because the
information is material.

                                                                                                                         5
2.1. Research on Management Earnings Forecasts
        There is a very large and diverse literature studying various aspects of MEFs (see Hirst et al.,
2008). Since our paper focuses on managers’ incentives to issue forecasts, we begin our literature review
with the various reasons posited for why managers issue forecasts. Managerial incentives to supply
voluntary disclosures like MEFs include signaling good firm performance (Verrecchia 1983; Dye 1985a
and b; Lev and Penman, 1990), reducing litigation risk (Skinner 1994; Kasznik and Lev 1995),
facilitating access to capital markets and reducing the cost of capital (Frankel, McNichols, and Wilson
1995; Lang and Lundholm 2000; Healy, Hutton, and Palepu 1999), adjusting analysts’ and investors’
expectations (Ajinkya and Gift 1984), affecting stock option compensation (Aboody and Kasznik 2000),
and signaling managerial talent (Trueman 1986).
        One of the main predictions of the theoretical disclosure literature is that when firms’ bear
proprietary costs from disclosure or when investors are uncertain about managers’ information, firms will
voluntarily disclose good news and withhold less favorable news (Verrecchia 1983; Dye 1985 a and b;
Verrecchia 2001). Early empirical evidence on MEFs is consistent with this prediction (Penman 1980;
Lev and Penman 1990). Subsequent empirical evidence established the importance of litigation risk as a
driver of MEFs. Skinner (1994) and Kasznik and Lev (1995) document that MEFs are more likely to
convey bad news, consistent with managers being concerned with litigation risk and issuing preemptive
MEFs to adjust investor expectations downward. More recently, Rogers and Van Buskirk (2009) report
that firms involved in class-action litigation reduced the amount of information (e.g., MEFs and
conference calls) post-litigation. Using a sample of MEFs from 1994-2003, Anikowski et al. (2007)
conclude that the proportion of firms issuing guidance has increased over time from less than 10% in the
mid-1990s to around 25% in 2001–2003. Firms issuing guidance now represent approximately 45% of
Compustat on a value-weighted basis. Anikowski et al. (2007) document that manager’s release neutral
MEFs early in a quarter, downward MEFs toward the quarter end and upward MEFs at quarter end.
        Besides focusing on managers’ incentives to issue guidance, a number of, primarily descriptive
studies, examine various characteristics of the MEFs such as the form, horizon, or level of specificity.
Some studies (Baginski et al., 1993 and Pownall et al., 1993) examine how forecast precision/form (i.e.,
point, range, minimum, maximum MEFs) affects the informativeness of MEFs, and generally find that
more precise forecast forms are more informative. Miller (2002) documents that firms expecting to
sustain an earnings increase in the next year issue more long-term (i.e., annual) MEFs while firms
expecting to have an earnings decline in the next year issue more short-term (i.e., quarterly) MEFs, a
choice that enables poorly performing firms to focus on the current positive news and avoid discussion of
the longer-term decline. Clement et al. (2003) argue that confirming MEFs reduce market uncertainty.
They find that the market reaction to such forecasts is positive and, that while consensus analysts’

                                                                                                           6
earnings forecasts do not change after confirming MEFs, the dispersion in analysts’ forecasts significantly
declines. Papers by Hutton et al. (2003) and Baginski et al. (2004) study the actions managers take to
enhance the credibility of MEFs such as supplementing them with qualitative and/or verifiable forward-
looking statements or explaining their MEFs. Another line of research finds that disaggregated MEFs are
more informative than aggregated MEFs (see e.g., Lansford et al. 2009) where an aggregated MEF is a
forecast of earnings alone while a disaggregated MEF includes the various line items underlying the
aggregate forecast (e.g., revenues and various expense items).
        Managers’ willingness to bias their MEFs has been studied by McNichols (1989), Rogers and
Stocken (2005) and Ajinkya et al. (2005) among others. Rogers and Stocken (2005) find that when
earnings are more difficult to forecast, managers are more willing to bias their MEFs while Ajinkya et al.
(2005) find that firms with more outside directors and institutional ownership are more likely to issue a
forecast, to forecast more frequently, that their MEFs tend to be more specific, accurate and less
optimistically biased.
        Another line of inquiry examines the economic effects of MEFs, typically the stock market
effects. The expectations adjustment hypothesis advanced by Ajinkya and Gift (1984) and King, Pownall,
and Waymire (1990) posits that managers release MEFs to align analysts’ and investors’ earnings
expectations with their own. By reducing information asymmetry, managers’ issuance of an earnings
forecast reduces incentives for costly private information search which improves shareholder welfare ex-
ante (see Diamond, 1985). The release of MEFs, and the corresponding expected reduction in information
asymmetry, is widely accepted in the disclosure literature as a means for firms to lower their cost of
capital. Consistent with this view, prior research concludes there is a relation between disclosure and a
firm’s cost of capital (see, e.g., Coller and Yohn, 1997; Healy et al., 1999; Lang and Lundholm, 1993 and
1996; Botosan, 1997; Botosan and Plumlee, 2002; Frankel et al., 1995; and Verrecchia and Weber, 2006).
For example, Coller and Yohn (1997) find that firms issuing MEFs have lower bid-ask spreads following
the MEF’s release, which they interpret as evidence that issuance of MEFs leads to a lower cost of capital
because theoretical research demonstrates that bid-ask spreads are positively related to the cost of capital.
Healy, et al. (1999) find that 97 firms with sustained increases in disclosure exhibit lower bid-ask spreads,
greater stock liquidity and greater analyst following, all three of which lower firms’ cost of capital. In a
related vein, Frankel, et al., (1995) present evidence that MEFs are part of a longer-term disclosure policy
where firms raising more external capital are more likely to issue MEFs than non-financing firms.
Finally, Lennox and Park (2006) argue that voluntary disclosures such as MEFs reduce information
asymmetry and find that firms whose stock returns are more sensitive earnings (i.e., firms with larger
ERCs) issue MEFs more frequently, evidence that managers voluntarily issue MEFs to reduce
information asymmetry.

                                                                                                               7
         Few studies mention managers’ affirmative duty under the securities regulations to disclose
private material information (e.g., MEFs) or abstain from trading as a motive for releasing a MEF. While
much of the MEF literature summarized above is primarily descriptive, most researchers seem to agree
that MEFs are released because of either voluntary disclosure incentives (to adjust analysts’ and
investors’ expectations, reduce information asymmetries, facilitate access to capital markets, or lower the
firm’s cost of capital) or opportunistically (to signal good firm performance or to increase stock option
compensation). Papers focusing on MEFs to reduce litigation risk examine MEFs issued late in the fiscal
quarter (warnings) or after the fiscal quarter ends but prior to the quarter’s earnings announcement
(earnings preannouncements). We exclude warnings and preannouncements from MEFs studied in this
paper. Based on the above cited studies, we simplify the various motives of managers to issue MEFs
posited by prior researchers to just two: voluntary disclosure incentives whereby managers seek to lower
their firm’s cost of capital and opportunism. We ignore other hypothesized motives such as signaling
managerial talent.


2.2. Research on Insider Trading
         The Securities and Exchange Act of 1934 argued that insider-trading restrictions increase public
confidence in capital markets and lower firms' cost of capital (Seyhun, 1992). The Federal Insider
Trading Sanctions Act of 1984 and the 1988 Insider Trading and Securities Fraud Enforcement Act
increased criminal fines from $10,000 to $1 million and jail sentences from five years to ten years,
allowed recovery of treble damages, created a bounty program for informants, and made top management
legally responsible for insider trading by any of the firm's employees. Seyhun (1992, p. 176) concludes
that following these acts and subsequent court cases, “insiders displayed a greater reluctance to exploit
earnings announcements and takeover information.” 4
         Lakonishok and Lee (2001) study whether one can learn anything about prices from observing an
insider’s trade. While they find little market reaction around actual trades and trade disclosure dates, they
do find that firms where CEOs are buying tend to have stronger future returns compared to CEOs that are
selling, although selling CEOs do not necessarily have negative future returns. With regard to insiders’
purchases, they find that purchases predict future price movements, but only for small firms. Ke et al.
(2003) examine insider trading patterns leading up to a “break” in earnings increases. They find an
increase in the frequency of net insider sales in the ninth through third quarters before an earnings break,


4
  For example, in 2003 the SEC settled a civil insider-trading case with the former chief operating officer of Waste
Management. He was charged with failing to disclose that earnings contained significant nonrecurring gains in a
press release and in a conference call. The COO agreed to pay $3.7 million, which includes disgorgement, interest,
and penalties, and is barred for five years from serving as an officer or director of a public company (see Taub,
2003).

                                                                                                                       8
a selling pattern that is stronger in the firm-quarters drawn from growth firms, those that precede a longer
break or a greater earnings decline at the break. Frankel and Li (2004) use insider trading profitability
and intensity as proxies for information asymmetry between insiders and outsiders to examine the ability
of insider trading to predict subsequent returns. They find that the total number of insider purchase
transactions is negatively associated with financial statement informativeness and analyst following, but
positively related to company news. Piotroski and Roulstone (2005) examine whether insider-trading
driven by contrarian beliefs versus insider-trading driven by superior knowledge of cash flows are
incremental to each other. They find that insider purchases are positively related to future earnings
performance, positively related to book-to-market ratios, and inversely related to past returns. Each of
these relations has incremental explanatory power for insider purchases, suggesting that insiders trade on
the basis of both contrarian beliefs and private information about future cash flows. Huddart et al. (2007)
explore how insiders condition their trades on knowledge of upcoming firm disclosures by focusing on
insider trading in short periods before and after firms make public disclosures. In the 20 days before a
quarterly or annual earnings announcement, there is little association between insider trades and the
announcement return. However, in the period following the earnings announcement and before the 10-Q
or 10-K is filed, there is a strong association between the frequency and value of insider trades and the
filing return, suggesting that insiders trade to profit from information made public at the filing date.
        Yermack (1997, p. 449) finds “that patterns of companies’ quarterly earnings announcements are
consistent with the interpretation that CEOs receive stock option awards shortly before favorable
corporate news,” which he interprets as evidence that CEOs influence of the timing of stock option
awards to benefit from foreknowledge of impending improvements in firm performance. Bettis, et al.
(2000) report that over 90% of their surveyed firms have policies restricting insider trading and over 75%
have explicit blackout periods prohibiting insider trading. Roulstone (2003) finds that firms restricting
insider trading pay more total compensation and use more bonus, stock options and restricted stock.
        The preceding studies document several empirical regularities with one important implication for
our unified framework of MEFs. While the magnitude of the insider trading profits and the types of firms
where insiders profit is open to some debate, there is general consensus in the literature that insider
trading does exist and can be profitable.


2.3. Research Jointly Examining Management Earnings Forecasts and Insider Trading
        Some papers examine whether managers time MEFs opportunistically to affect wealth transfers
from shareholders. Aboody and Kasznik (2000, p. 73) conclude that that their findings “suggest that
CEOs make opportunistic voluntary disclosure decisions (i.e., MEFs, emphasis ours) that maximize their
stock option compensation.” Rogers (2008) explores how insider trading affects disclosure quality

                                                                                                             9
(where disclosure quality is measured indirectly using bid-ask spreads and market depth) and finds that
managers provide higher quality disclosures (i.e., (MEFs) before selling shares than they provide in the
absence of trading. While the results in Rogers (2008) are consistent with managers complying with
DOA requirements, unlike our study, he does not directly focus on the role of DOA requirements as a
determinant of MEFs. Rees et al. (2008) find that while managers provide more pessimistic guidance
prior to stock option awards than afterwards, they also report that managers do not distort pre-grant
guidance and that firms awarding option grants issue more frequent forecasts and have smaller absolute
analyst forecast errors than other firms.
         Two papers addressing the relation between insider trading and MEFs that are most relevant to
our study are Noe (1999) and Cheng and Lo (2006). At a general level, both studies document that
managers buy more shares following bad news MEFs than after good news MEFs, and sell more after
good news MEFs than after bad news MEFs, results that both studies interpret as evidence of managers
strategically timing their insider trading with the disclosure of MEFs. In particular, Noe (1999) studies
insider trading before and after the release of MEFs to investigate whether managers are opportunistic ex-
ante in the sense that that buy or sell shares before they release a MEF to exploit the information revealed
by the subsequent MEF. The evidence is inconsistent with such ex-ante opportunism. Noe (1999) also
investigates how managers trade after the release of MEFs and finds that managers generally behave like
contrarians in that they buy if the stock price reaction to the MEF is negative or sell if the stock price
reaction to the MEF is positive. Noe (1999) focuses exclusively on whether managers behave
strategically when they trade before or after MEFs, which means he does not consider (or recognize) that
managers could be issuing MEFs to comply with DOA requirements. Noe’s (1999) finding that there is
significant buying and selling after MEFs is also consistent with our alternative hypothesis that managers’
issue MEFs to comply with DOA requirements. Moreover, under the alternative hypothesis, insiders
execute their trades after disclosing an MEF because it means they have disclosed their material
information prior to trading as required by rule 10b-5. 5
         Cheng and Lo (2006) extend Noe (1999) by treating MEFs as a strategic decision made by
mangers to trade opportunistically in their firm’s stock. Cheng and Lo (2006) find that mangers increase
the frequency of bad news MEFs when purchasing shares (presumably to reduce the purchase price),
however managers do not alter their MEF forecasting behavior when they are selling. Cheng and Lo
(2006) discuss DOA requirements only in passing when they argue that their findings are inconsistent


5
  Another interpretation of Noe’s (1999) results is that managers are strategic ex-post, but not ex-ante in the sense
that they issue an MEF to reduce the cost of capital or adjust analyst expectations, etc., and only decide to trade after
they observe the market response to the forecast. Here, the ex-ante act of issuing the MEF could be unrelated to the
decision to trade.


                                                                                                                      10
with mangers violating such requirements. Unlike our study, however, Cheng and Lo (2006) do not
consider DOA requirements, voluntary disclosure incentives, and opportunistic managerial behavior as
competing alternative explanations for why managers choose to release MEFs.


2.4. Management’s Affirmative Duty to Disclose Material Information or Abstain from Trading
        A large literature in securities regulation addresses insider trading regulation under Rule 10b-5 of
the U.S. Securities Exchange Act of 1934. In general managers have no affirmative duty to disclose
material events under the securities laws unless “(1) a Commission statute or rule requires disclosure, (2)
an ‘insider’ …is trading, or (3) a previous disclosure is or becomes inaccurate, incomplete, or misleading”
(see Loss and Seligman 2004, p. 3510-3511). Of special interest to our study is that if insiders want to
trade their firm’s securities, Rule 10b-5 prohibits such trading on the basis of material information. The
insider is required to either disclose the information or abstain from trading (DOA).
                                                                                         6
        The SEC routinely brings between 40 to 60 insider trading actions annually. Besides SEC
enforcement of insider trading, the Department of Justice pursues criminal cases, and FINRA (Financial
Industry Regulatory Authority) oversees ten U.S. exchanges using state-of-the-art insider trading
surveillance and investigation programs for all listed securities in the United States. One of the more
prominent recent cases involved the 2007 conviction of Joe Nacchio, former CEO of Qwest, on 19 counts
of illegal insider trading connected to his sale of $52 million in Qwest stock in early 2001. The Justice
Department alleged Nacchio sold Qwest stock while knowing of the company’s deteriorating financial
condition. He was sentenced to six years in prison and ordered to pay $71 million in fines and
               7
forfeitures.
        As noted in the legal literature, there is little clear cut guidance in the securities laws or SEC
regulations as to what constitutes material information. For example, while Staff Accounting Bulletin
(SAB) 99 and Regulation FD list types of information that may help determine materiality, neither goes
further than that in terms of providing specific guidance on the defintion of material information. Such
lack of clarity creates substantial uncertainty as to when managers must make a disclosure. Consistent
with this, Heminway (2003) argues that: “while corporate issuers and their directors and officers know
that they cannot trade when they are in possession of material undisclosed information, the imprecise


6
 http://sec.gov/spotlight/insidertrading.shtml.
7
 “Supreme Court again denies ex-Qwest CEO Nacchio,” Associated Press (11/30/2009). Other cases involve SEC
v. Thomas J. Bucknum, (D. Mass. Jan. 12, 2006) where a director paid over $2 million in disgorgement and penalty
because he sold stock while in possession of inside negative news about the company; SEC v. Kopsky (Civil Action
No. 4:07-CV-0379) where the company president tipped a friend regarding future earnings announcements; and
SEC v. Cao (Case No. 2:06-cv-1269 DSF-RC C.D. Ca.) where executives at Countrywide Financial were ordered to
pay penalties and disgorge profits because they sold shares short prior to releasing quarterly earnings below
projections.

                                                                                                             11
existing legal standard defining what is ‘material’ makes it difficult for those issuers, directors and
officers to understand their legal obligations.” Hemingway (2003) also observes that “section 10(b) and
Rule 10b-5 are antifraud provisions, not mandatory disclosure rules - rules that call for specific
substantive disclosure based on a transactional or periodic reporting requirement. … In the insider
trading context, however, the ‘disclose or abstain’ rule has the effect of a macro mandatory disclosure
rule. The existence of a duty to disclose arising out of issuer or insider securities trading activity compels
disclosure of all material nonpublic information before any trade is made. … Accordingly, it is
appropriate to refer to the regulation of insider trading under Rule 10b-5 as a form of disclosure
regulation (emphasis added).” One way for insiders to deal with this uncertainty is to issue a MEF
before trading in their firm’s stock. Indeed, in Marx v. Computer Sciences Corp. [507 F.2d 485, 489 (9th
Cir. 1974] the court noted that “generally earnings projections (emphasis added) of a company constitute
a prime factor in estimating the worth of its stock, especially when made close to the end of the fiscal
year” and therefore are material. In fact, risk-averse insiders will likely set a very low materiality
threshold, which will result in them issuing some MEFs that are immaterial.
         The preceding review of the securities regulation literature suggests that one way insiders wishing
to trade in their firm’s stock and be in compliance with Rule 10b-5 of 1934 Act is to issue a MEF.


3. A Unified Framework of Management Earnings Forecasts (MEFs)
         To provide a unified framework as to why managers issue earning guidance we synthesize the
various motives researchers have used to explain the existence of MEFs. Our objective is simply to show
that no single model explains the release of all MEFs. Based on the extant literature, we categorize the
existing reasons into three mutually exclusive categories. First, MEFs issued as part of a firm’s overall
voluntary disclosure policy to improve the firm’s financial transparency, thereby (presumably) lowering
its cost of capital. Second, managers strategic/opportunistic issuance of MEFs so they can profitably
trade in their firm’s stock. Third, MEFs issued because managers wish to trade in their firm’s stock and
want to comply with the insider trading laws that impose upon them an affirmative duty to disclose their
private information prior to trading (DOA). The last two reasons to issue MEFs (opportunistic and DOA)
are not voluntary disclosures in the sense that managers are not disclosing MEFs as part of their firm’s
voluntary disclosure policy to improve transparency. Rather, MEFs are released because managers want
to trade, either opportunistically or to comply with insider trading laws. 8




8
  As noted above, we are not focusing on the decision of managers to issue earnings warnings (management
guidance made shortly before or after the close of the fiscal period but prior to the release of the fiscal period’s
earnings). Our analysis only applies to MEFs issued prior to the end of the fiscal period.

                                                                                                                       12
         Like the earnings management literature where the true motives of managers are not observable
and researchers must estimate proxy variables for earnings management such as discretionary accruals or
meet or beat benchmarks, we must construct variables to proxy for managers’ various MEF disclosure
motives. 9 To recognize that MEFs are issued for different reasons, we construct an algorithm that
separates all MEFs into one of three mutually exclusive and exhaustive subsamples: (i) those MEFs
where, a priori, we have reason to believe were issued voluntarily, (ii) those MEFs where, a priori, we
have reason to believe were issued strategically to allow the manager to make opportunistic trades and
(iii) those MEFs where, a priori, managers likely had an affirmative duty to disclose material information
(details of the algorithm are provided below in section 3.1).
         Our empirical analysis uses the following MEF prediction model:
         Pr(MEF) = f(V1, V2, … Vl, O1, O2, …, Om, D1, D2, … Dn),                                             (1)
where MEF= 1 if a management forecast is issued in the quarter and zero otherwise, Vi = ith voluntary
disclosure determinant, Oi = ith opportunistic disclosure determinant and Di = ith disclose or abstain
(materiality) disclosure determinant. Equation (1) is estimated using all three subsamples (voluntary
MEFs, opportunistic MEFs and DOA MEFs). We predict that the coefficients on the Vi variables have
more explanatory power for voluntary MEFs than for non-voluntary MEFs. Likewise, for the
opportunistic MEFs subsample we predict that the coefficients on the Oi variables are larger in absolute
value than the coefficients on the Oi variables estimated in the voluntary and DOA subsamples. Finally,
for the DOA subsample, the coefficients on the Di variables are predicted to be larger in absolute value
than the coefficients on the Di variables estimated in the voluntary and opportunistic subsamples. An
important maintained assumption underlying our tests is that one and only one motive underlies each
MEF. For example, we rule out situations where a particular MEF was released because managers
wanted to trade and still comply with insider trading restrictions (DOA) and the forecast still would have
been issued for voluntary disclosure reasons had the manager not traded. To the extent that dual motives
for MEFs exist, the power of our tests will be reduced.


3.1      An algorithm to form mutually exclusive subsamples of MEFs
         In a given fiscal period we assume managers make two inter-related decisions: the decision
whether to issue management guidance (e.g., an MEF) and the decision whether to buy/sell their firm’s
stock (insider trading). As briefly discussed above, the direction of causality can run in both directions or
in neither direction. For example, in some quarters managers may want to trade and so this causes the


9
  Other similar situations of having to construct empirical proxies for unobservable variables exist throughout the
accounting literature. For example, conservatism, market expectations of earnings and expected stock returns are
just a few examples.

                                                                                                                      13
release of an MEF to satisfy insider trading prohibitions (i.e., DOA requirements). In other quarters,
managers may want to trade opportunistically and so issue a biased MEF. It is also possible that
managers may issue a MEF voluntarily in some quarters with no intention of trading, but after observing
the market reaction to the guidance the manager believes the stock is mispriced so he/she only then
decides to trade. Finally, a manager may decide to issue an MEF voluntarily and to trade, but each
decision is independent of the other. Since we cannot observe the actual motives underlying managers’
disclosure and trading decisions we must make some simplifying assumptions to construct the three
subsamples of MEFs of interest to our study. Accordingly, based on observed characteristics of
managers’ trading behavior and of their MEFs, we classify firm-quarters with MEFs as voluntary,
opportunistic or DOA, and then test whether this classification scheme captures the underlying economics
of the three motives for issuing guidance by estimating equation (1) and testing whether the coefficients
vary in predictable ways across the subsamples.


3.1.1 Sub-sample 1: Voluntary MEFs
        Voluntary disclosure theory predicts that MEFs are issued because the benefits to the firm of
issuing the forecasts exceed the costs. A maintained assumption of voluntary disclosure models is
managers maximize shareholder value (i.e., there are no conflicts of interest between shareholders and
managers). This maintained assumption means that MEFs issued voluntarily are unbiased (biased MEFs
are considered in subsample #2) and that the market price adjustment to the new information contained in
an unbiased MEF is also unbiased (i.e., the market is efficient). Empirically, however, not all post MEF
stock prices correctly capture the information contained in the forecast. While, on average, the post MEF
price is correct (i.e., the market is efficient), some prices may be “too high” and others “too low” in the
sense that they are above or below the managers’ assessment of the firm’s equity value. As a result, after
observing the post MEF stock price, and based on their private knowledge of the firm’s future cash flows,
managers have expectations about whether their firm is mispriced. If the price is above the managers’
valuation they sell some of their firm’s stock, while if it is below their valuation they will buy stock. It is
important to appreciate that in this scenario there is no ex ante managerial opportunism in the sense that
good news forecasts (MEF announcement CAR > 0) are not optimistic (i.e., MEF ≤ actual earnings) and
likewise, bad news forecasts (MEF announcement CARs < 0) are not pessimistic (i.e., MEF ≥ actual
earnings). Simply put, firms with positive abnormal MEF announcement returns (CARs) likely contain
more over priced stocks and firms with negative MEF announcement CARs likely contain more under
priced stocks.
        Following the reasoning above we select observations for the voluntary disclosure subsample of
MEFs using the following algorithm:

                                                                                                              14
         A1a: There is no insider trading in the 61 day window centered on release of the MEF. These
               MEFs are voluntary because the absence of insider trading in the 30 days before and after
               the MEF’s release is a likely indication that the manager had no affirmative duty to
               disclose and/or was not behaving opportunistically, 10 OR
         A1b: Large good news MEFs (large positive CAR defined as a 3-day size-adjusted abnormal
               returns centered on the MEF announcement day of at least +5% followed by insider selling
               within 30 days and (MEF - actual earnings) ≤ 0. 11 Here the market likely “over-priced”
               these MEFs even though they were not optimistic, OR
         A1c: Large bad news MEFs (large negative CARs defined as a 3-day size-adjusted abnormal
               returns centered on the MEF announcement day of at least -5% followed by insider buying
               within 30 days and (MEF - actual earnings) ≥ 0. Here the market likely “under-priced”
               these MEFs even though they were not pessimistic, OR
         A1d: Insider purchasing occurs in the 30 days prior to release of bad news MEFs or insider
               selling occurs in the 30 days prior to release of good news MEFs.
         Steps A1b and A1c might misclassify some affirmative duty to disclose (DOA) MEFs (subsample
3) as voluntary (subsample 1) because steps A1b and A1c classify all large news MEFs that are not
opportunistic as voluntary MEFs. Moreover, some managers may have issued a MEF because they
wanted to trade (i.e., DOA) and such MEFs resulted in large price changes. If we could observe
managers’ actual motives, such MEFs would be included subsample 3 not subsample 1. All of our
inferences are robust after redoing our tests where MEFs meeting the criteria in A1b and A1c were
instead classified as DOA (subsample 3).


3.1.2 Sub-sample 2: Opportunistic MEFs and Insider Trading
         Ex ante managerial opportunism can manifest itself in two ways: (i) managers can sell prior to the
release of material bad news or buy prior to the release of material good news or (ii) they can issue
optimistic MEFs prior to selling or issue pessimistic MEFs prior to buying. Situation (i) clearly places
the manager at risk of violating insider trading prohibitions. Alternatively, if their forecast is unbiased, it
is not strategic and voluntary disclosure (subsample #1) exists.
         To select observations for subsample #2 the following algorithm is used:


10
   This algorithm likely mis-classifies some MEFs. For example, suppose managers issue biased MEFs in hopes of
trading strategically (i.e., they issue optimistic MEFs before selling and pessimistic MEFs before buying), only to
observe (ex post) that the market reaction to the MEF is smaller than expected, making it unprofitable to trade
strategically. If such scenarios actually occur in the data, our sample of voluntary MEFs will contain some forecasts
issued opportunistically ex ante, but which did not result in any insider trading ex post.
11
   Our inferences are robust to using +3% and +10% cutoffs for defining “large” positive and negative CARs.

                                                                                                                   15
        A2a: Insider purchasing occurs in the 30 days prior to release of good news MEFs or insider
              selling occurs in the 30 days prior to release of bad news MEFs, OR
        A2b: At the release of the MEF there is a large positive CAR (as defined above), there is insider
              selling in the next 30 days and the (MEF - actual earnings) ≥ 0, OR
        A2c: At the release of the MEF there is a large negative CAR (as defined above), there is insider
              buying in the next 30 days, and (MEF - actual earnings) ≤ 0.
        Clearly, not all observations classified by steps A2a, b, and c need to be strategic. For example,
in some cases insiders ex ante trade with no intention of issuing a subsequent MEF because they do not
believe they are in possession of inside information. However, following the trade they learn something
material about the firm’s earnings so they disclose it via a MEF and the disclosure results in a large stock
price reaction. So it can be by chance the MEF CAR is positive, the insider sells, and the guidance was
optimistic. Thus, similar to subsample #1 we expect subsample #2 to contain some observations that
belong in the other two subsamples.


3.1.3 Sub-sample 3: Disclose or Abstain (DOA)
        In this scenario insider trading occurs because managers have personal liquidity reasons or
because they want to accumulate shares. Liquidity shocks include paying college tuition bills, income
taxes or divorce settlements, buying homes, or exercising options (to pay the exercise price and taxes).
Managers may want to purchase shares because they believe the stock is overvalued, they want to signal
their belief it is overvalued, or to meet share ownership guidelines imposed by the firm’s board of
directors. Many companies have formal or informal ownership policies requiring top managers to own
either a fixed number of shares or a multiple of their base salary (usually between one times to five times
base salary depending on rank). Managers can meet these ownership guidelines via exercising options
and holding the shares, via restricted stock grants, and by buying the stock in the open market.
        To comply with insider trading requirements, managers have an affirmative duty to disclose
material inside information prior to trading (DOA). Since these liquidity shocks or accumulation
incentives are assumed random, there is no association between the sign of the MEF announcement CAR
and insider buying/selling. If managers wish to avoid the appearance of insider trading, they will issue
MEFs that have both large and small positive and negative announcement CARs. However, the more
private information the manager has (as measured by the difference between the consensus analyst
forecast and the MEF), the more likely the manager perceives he/she has an affirmative duty to disclose,
and hence the more likely the manager issues a MEF. The liquidity shock or stock accumulation
requirements cause insider trading and issuance of the MEF. Thus, such MEFs are not “voluntary” in the
sense that the manager is not pursuing a disclosure policy to increase the firm’s financial transparency.

                                                                                                            16
Since we cannot observe managers’ random liquidity shocks, their demand to hold shares arising from
formal and informal ownership guidelines, or their perceived duty to disclose prior to trading the
following algorithm is used to select observations for subsample #3:
        A3a: All MEFs not classified as voluntary (subsample #1) or opportunistic (subsample #2) that
               have insider trading in the 30 days following the MEF’s release are classified as DOA. 12


3.2     An Empirical MEF Prediction Model
        Having detailed the formation of the three MEF subsamples, we now describe the empirical
model used to predict whether an MEF issued in a given quarter is more consistent with the voluntary
disclosure incentive variables (the Vi’s), the opportunistic incentive variables (the Oi’s) or the DOA
incentive variables (the Di’s). Specifically, we use the following multinomial probit model as our MEF
prediction:
MEFDUM = α0 + α1ERC + α2EARN_SURPRISE + α3SIZE + α4MB + α5RTNVOL +
                 α6EARNVOL + α7HI_TECH + α8REGULATION + α9HABITUAL +
                 α10RESTATE+ α11BACKDATE,                                                              (2)
where MEFDUM takes the value of 1, 2 or 3 if the firm-quarter observation is classified as voluntary,
strategic or DOA, respectively, and zero for all quarters without a MEF. Since MEFs are often issued
concurrently with an earnings announcement, the model is estimated using: (i) all observations in a given
subsample, (ii) only those MEFs in a given subsample issued without an earnings announcement and (iii)
only those MEFs in a given subsample where the MEF is bundled with an earnings announcement. 13


3.2.1. Voluntary determinants of MEFs
        The first nine independent variables in equation (2) (ERC, EARN_SURPRISE, SIZE, MB,
RTNVOL, EARNVOL, HI_TECH, REGULATION and HABITUAL) are variables suggested by
voluntary disclosure theory to capture voluntary disclosure incentives (see, e.g., Lennox and Park, 2006).
These eight variables are designated as the voluntary determinates (i.e., Vi‘s) in equation (1). The
intuition underlying each variable is discussed in turn below (see the Appendix for detailed variable
definitions and measurement techniques).
        ERC is the firm’s earnings response coefficient. Managers of firms with higher sensitivity of
stock prices to earnings-related information have stronger incentives to voluntarily issue MEFs.
EARN_SURPRISE proxies for management’s private information (Kasznik and Lev, 1995) and is the


12
 As in the previous subsamples we expect subsample #3 to also contain some misclassified firms.
13
 Firm-quarters with only an earnings warning or earnings preannouncement are dropped (i.e., not counted as
MEFDUM = 0) to avoid confounding our inferences from the estimation.

                                                                                                             17
absolute value of the difference between the most recently measured consensus analyst earnings forecast
three weeks before the fiscal quarter end and that measured just prior to the earnings announcement date
of the prior quarter. 14 Under voluntary disclosure theory, managers wishing to appear transparent have
incentives to release information in a timely fashion. SIZE is typically used in the literature to capture the
incentives larger firms have to issue more frequent MEFs due to greater economies of scale in
information production, greater demand for information by investors and analysts, or because they face
greater litigation risk (Lang and Lundholm, 1993; Kasznik and Lev, 1995; Frankel et al., 1995). If a firm
operates in a risky legal environment, managers may issue MEFs to reduce expected litigation costs
(Skinner, 1994 and 1997). RTNVOL, EARNVOL, HI_TECH, and REGULATION all control for
litigation risk. HABITUAL is one if the firm has issued a MEF in six of the last eight quarters and is
designed to capture the tendency of managers that have more transparent disclosure policies to issue
MEFs more frequently.


3.2.2 Opportunistic determinants of MEFs
          Managers issuing MEFs strategically face weaker control systems that permit such behavior.
However, attempts to measure the strength of firms’ governance systems encounter both empirical and
theoretical challenges (see, e.g., Larcker et al., 2007 and Brickley and Zimmerman, 2009). Instead of
using characteristics of firms such as board size or antitakeover provisions to differentiate “strong” from
“weak” control systems, we use outcome measures. In particular, RESTATE equals one if the firm has
restated earnings at any time over the sample period and BACKDATE is one if the firm is listed in The
Wall Street Journal's "Options Scorecard" on November 28, 2006 as being under scrutiny for option grant
backdating. RESTATE and BACKDATE comprise our opportunistic determinates of disclosure (i.e., the
Oi’s) in equation (1) and capture firms with weak control systems. 15


3.2.3    Disclose or abstain determinants of MEFs
          Managers wishing to trade in their firm’s securities and wanting to fully comply with insider
trading requirements will disclose all material information prior to trading. Merely waiting to trade until
after releasing the firm’s current quarter earnings will not inform the market of the manager’s material
inside information about the current and future quarters. The current judicial standard of materiality in


14
   Since we exclude all earnings warnings and earnings preannouncements from our study, to prevent
EARN_SURPRISE from being contaminated by any such announcements we use the consensus analyst forecast
outstanding prior to the last three weeks of the quarter instead of the actual earnings for the quarter.
15
   While Ajinkya et al. (2005) report that firms with more outside directors issue less optimistically biased MEFs, they recognize
that endogeneity makes it difficult to draw inferences about causality. Wishing to avoid such problems and to avoid further
sample size restrictions due to missing data on board composition, we use outcome measures rather than firm characteristics to
proxy for governance strength.

                                                                                                                                18
securities law describes an item as material if there is a “substantial likelihood that the disclosure of the
omitted fact would have been viewed by the reasonable investor as having significantly altered the ‘total
mix’ of information made available.” (TSC Industries v. Northway, Inc., 1976). Wishing to comply with
this standard, two factors likely affect managers’ assessment of whether they are in possession of material
inside information: the magnitude of the news and sensitivity of a reasonable investor to a “unit” of news
(Heitzman, et al., 2010). The magnitude of the insider’s information depends on the market’s
expectation. We assume that the materiality of a MEF depends on (i) the magnitude of the item relative
to analysts’ consensus forecast and (ii) the sensitivity of the firm’s stock price to accounting earnings (i.e.,
the firm’s earnings response coefficient, or ERC). EARN_SURPRISE (defined above) proxies for the
magnitude of the manager’s private earnings information relative to the market’s expectation (consensus
analyst forecast prior to the management’s forecast). Since we cannot observe the manager’s forecast
unless the manager discloses it, we assume that the consensus forecast 21 days before the end of the
quarter proxies for the manager’s expectation of the quarter’s earnings.
        While the current judicial standard of materiality calls for a substantial likelihood a reasonable
investor would be affected by the disclosure, Heminway (2003) argues that the lack of clear judicial or
regulatory guidance as to what constitutes material information causes significant uncertainty on
managers having to comply. As a result, risk-averse managers, those concerned about their reputation, or
those in firms with large ERCs might disclose a MEF even if it confirms the market’s expectation.
        Table 1 summarizes the independent variables hypothesized to predict MEF disclosure in the
three subsamples (i.e., the Vi, Oi, and Di’s). While the voluntary disclosure and DOA disclosure
independent variables (Vi’s and Di’s) share two common explanatory variables (ERC and
EARN_SURPRISE), the voluntary disclosure model also includes several other explanatory variables that
should be more important (only important) in the voluntary subsample. The opportunistic independent
variables (i.e., the Oi’s) should only have explanatory power in the strategic subsample and the estimated
coefficients on such variables should be smaller in absolute value (or insignificant) in the other two
subsamples.


4. Data and Summary Statistics
4.1   Sample selection and data sources
        We begin with 244,986 unique firm-quarters of data from First Call’s CIG database between
1998 and 2007 with an actual EPS figure expressed in US dollars. Merging the First Call data with the
“Merged COMPUSTAT/CRSP database” available from WRDS results in a loss of 31,226 observations.
We next exclude 20,256 observations by limiting our sample to firms covered by Thompson Financial’s
Insider Trading database. We lose another 142,125 observations due to missing data to calculate all of

                                                                                                                19
the independent variables underlying our multinomial prediction model (estimated at the firm-quarter
level). ERC and EARNVOL cause most of the loss because they require 16 consecutive quarters of data.
Excluding 3.095 earnings warnings and preannouncements leaves us with 48,284 firm-quarter
observations which we classify into two categories, MEF quarters and non-MEF quarters, based on
whether there is at least one MEF issued during the period from the last quarter’s earnings announcement
date up through one day before the current quarter’s earnings announcement date. Of the 48,284 firm-
quarters, 13,796 are MEF quarters containing 23,144 unique MEFs.
        Using the algorithm described in Section 3.1, we assign each MEF into one of three mutually
exclusive categories (DOA, Opportunistic, and Voluntary MEFs) based on: (i) the existence of insider
trading within the -30 to +30 trading day interval relative to the MEF announcement date (i.e., day 0), (ii)
the cumulative abnormal return at the MEF announcement date and (iii) the MEF error. Our definition of
insider trading includes all open market purchase or sale transactions conducted by the CEO, CFO,
Chairman, President, Executive VP or Senior VP. We define good (bad) news MEF as those with a
positive (negative) two-day (i.e., day 0 and +1) cumulative market-adjusted return. MEF news is deemd
“large” if the absolute value of two-day cumulative market-adjusted return is greater than 5%. We
calculate the MEF error by subtracting the management earnings forecast from actual reported EPS.
        Table 2 reports the number of MEFs categorized into each of the three sub-samples using the
algorithm described in Section 3. The numbers of MEFs in the Voluntary, Opportunistic, and DOA
subsamples are 15,751, 3,156, 4,237, and respectively. The algorithm classifies 68.1% of the overall
sample of MEFs as voluntary, 13.6% as opportunistic and 18.3% DOA. More specifically, step A1a of
the algorithm classifies 12,146 MEFs as voluntary because there is no insider trading in the ± 30 days
surrounding the MEF. Similarly, step A2a (insider purchasing occurs in the 30 days prior to release of
good news MEFs or insider selling occurs in the 30 days prior to release of bad news MEFs) categorizes
2,580 MEFs as opportunistic. Table 2 also provides insight into which of the various steps of the
algorithm are responsible for generating the most observations in each subsample. Clearly, steps A1a and
A2a are the most important in terms of classifying 14,726 of the 23,144 MEFs. As noted above, since we
cannot directly observe managers DOA decisions, MEFs not classified as voluntary or opportunistic are
placed in the DOA subsample.
        In summary, each firm-quarter observation in the overall sample is classified into one of four
mutually exclusive types (DOA, Opportunistic, Voluntary or no MEF). Since some firm-quarters contain
multiple MEFs, to avoid any ambiguity in the test, we drop 1,104 quarters if the firm-quarter contained
MEFs classified in more than one type (e.g., the firm issued both a voluntary and a DOA MEF in the
same quarter or if the firm issued an opportunistic and a voluntary MEF in the same quarter). This final
data filter results in a final sample of 47,180 firm-quarter observations for use in our empirical tests.

                                                                                                            20
4.2   Summary statistics
        Table 3 Panel A presents frequency distributions of the firm-quarter observations for the three
MEF subsamples (DOA, Opportunistic and Voluntary) from 1998 to 2007. Panel A indicates that the
number of firm-quarter observations with MEFs (excluding warnings and preannouncements) increased
nearly six fold from 1998 to 2007. Within subsamples, over time there is a slight increase in the
percentage of MEFs classified as DOA (from 17.1% in 1998 to 18.2% in 2007), a slight decrease in the
percentage of MEFs classified as opportunistic (from 15.6% in 1998 to 13.1% in 2007) and a slight
increase followed by a similar decrease in the percentage of MEFs classified as voluntary. Overall, the
data reveal that the proportion MEFs assigned to the subsamples are relatively stationary over time.
        Panel A also indicates that MEFs classified as DOA are fairly uniformly distributed over the four
fiscal quarters, while MEFs classified as opportunistic or voluntary tend to be more heavily concentrated
in the fourth quarter. The final set of data reported in Panel A breaks out “Bundled” MEFs (an MEF
issued within ±1 day of an actual earnings announcement), “Unbundled” MEFs (an MEF not issued
within ±1 day of an actual earnings announcement) and “Both” (a firm-quarter containing both bundled
and unbundled MEFs). Interestingly, and consistent with the notion that some MEFs are issued to
comply with insider trading restrictions, 76% of DOA MEF firm-quarters are issued concurrently with
actual earnings announcements compared to 52.2% and 59.1% of MEFs classified as opportunistic or
voluntary, respectively.
        Panel B of Table 3 reports descriptive statistics (by subsample) about the number of MEFs (as
opposed to firm-quarter MEFs) including MEF horizon (quarterly or annual), MEF form (point, range,
end, or qualitative) and news content (bad, good, or neutral). 16 While Panel A is based on firm-quarter
observations, Panel B is based on MEF observations. The percentage of firm-quarters classified as
voluntary MEFs is 70.8% (see Panel A) whereas the percentage of all MEFs classified as voluntary is
68.1% (see Panel B). The data in Panel B indicates there is little difference across subsamples in terms
of forecast horizon (quarterly versus annual) except for opportunistic MEFs, where 57.4% of
opportunistic MEFs are associated with quarterly horizons, versus 55.9% (55.6%) for DOA (voluntary)
MEFs. While there is a slight tendency for opportunistic MEFs to be issued for longer forecast horizons,
the differences are not large. The data in Panel B also indicates that there is little difference in the
forecast form (point, range, end, qualitative) across subsamples. Range MEFs are the most prevalent
comprising roughly 74% to 77% of the subsamples, followed by point estimates which comprise roughly
16% to 17% of the MEFs in the subsamples. Turning to the “news” in the sample of MEFs, opportunistic



16
   We use the algorithm in Anilowski et al. (2007) to classify MEFs into good, bad, neutral, and mixed news based
on First Call’s CIG code, the value of the MEF, and the most recent analyst consensus forecast.

                                                                                                                21
MEFs have a higher concentration of bad news (54.4%) compared to either DOA MEFs (47.2%) or
voluntary MEFs (50.2%). 17


5. Results
5.1   Evidence on the timing of MEFs and insider trading
        Before presenting our main findings based on the multinomial probit model, we first present
some results on the timing of insider trading relative to the MEF release dates for each of the three
subsamples. Panel A (B) of Table 4 reports on the timing of insider sales (purchases) around MEFs using
five-day event windows for the overall -30 to +30 trading day period relative to the MEF release date
(day 0). In the DOA subsample, while only 1.0% of the firm-quarters have insider sales on the MEF
release date, 33.2% of the DOA firm-quarters have insider sales in the +1 to +5 trading day window
(Panel A). The frequency of insider sales falls monotonically from 33.2% for this period to 16.1% in the
+26 to +30 day window. By construction, there are no insider sales in the DOA subsample during days -
30 to -1 because any MEF with insider trading in the 30 days preceding an MEF is classified as either
opportunistic or voluntary depending on the direction and magnitude of the news and insider trading (see
Table 2). Consistent with our expectations, the DOA subsample has a greater frequency of insider sales
in the five days immediately following the MEF (33.2%) than both the opportunistic MEF subsample
(22.7%) and the voluntary MEF subsample (7.6%). While the lower incidence of insider trading in the
voluntary subsample is induced by the selection criteria that firm-quarters with no insider trading in the -
30 to +30 day window (A1a) are classified as voluntary MEFs, there is no obvious mechanical relation
inducing the finding that DOA MEFs are more likely to have trading than opportunistic MEFs in +1 to +5
day window. Managers issuing MEFs classified as DOA are more likely to trade immediately following
the MEF than the managers issuing opportunistic MEFs. This finding adds construct validity to our MEF
classification scheme.
        The six right most columns of Panel A report the intensity of insider selling given insiders sold in
a particular five day window. There are 1,407 firm-quarters with insider sales in the +1 to +5 trading day
window following an MEF classified as DOA. Stated differently, 30.6% of all insider shares sold during
that year were sold in this five day window, which is larger than the fraction sold in any other five day
window following the release of a MEF classified as DOA. In addition, the intensity of insider selling
(mean percentage of annual shares sold) is greater in the +1 to +5 day window following MEFs classified


17
   From the insider trading literature we know that insiders sell more frequently than they buy. From Table 2, we
also know that about 82% of MEFs classified as opportunistic are generated by algorithm A2a. Since algorithm A2a
selects MEFs as opportunistic if there are insider sales in the 30 days prior to the MEF and the MEF is bad news, it
is not surprising that the opportunistic MEF subsample has a higher proportion of bad news MEFs than the other
subsamples.

                                                                                                                 22
as DOA (30.6%) when compared to the same window for both MEFs classified as opportunistic (20.2%)
and MEFs classified as voluntary (20.6%). One interpretation of this pattern is that, consistent with
managers wishing to comply with insider trading restrictions, MEFs classified as DOA are more likely to
have insider sales immediately following the MEF, and a larger fraction of the shares sold in the
corresponding year (30.6%) are sold in the five days immediately following such MEFs when compared
to the other two subsamples. These findings on the intensity of insider selling add additional construct
validity to the classification scheme we use to assign MEFs into subsamples.
         Table 4 Panel B repeats the analysis for MEFs with insider purchases. An obvious difference
from Panel A is that there are far fewer insider purchases than sales. For example, in the +1 to +5
window only 4.8% of the firm-quarters have an insider purchase following a MEF classified as DOA
compared to 33.2% of firm-quarters with insider sales in the same window (Panel A). Despite the smaller
incidence of insider purchases vis-à-vis insider sales, we continue to observe the same declining
frequency of insider purchases and intensity of shares purchased following the release of MEFs as we
observed in Panel A. As before, MEFs classified as DOA have a higher incidence of firm-quarters with
insider purchases in the +1 to +5 trading day window (4.8%) than both MEFs classified as opportunistic
(3.3%) and MEFs classified as voluntary (0.8%). More importantly, 75.1% all shares purchased during
the year are bought in the +1 to +5 trading day window following a MEF classified as DOA, and this
percentage is larger than that of any other 5 trading day window over the next 25 days. In addition,
insiders purchase 74.2% of all the shares purchased during the corresponding year in the five trading day
window following the release of MEFs classified as opportunistic, a proportion that is larger than the
corresponding insider purchases following MEFs classified as voluntary (65.4%).
         Table 5 reports the means and medians of all the independent variables used in the multinomial
probit model (i.e., the Dk’s, Oj’s and Vi’s). Since there are 11 variables for which we calculate means and
medians per subsample, there are 33 tests for mean and median differences reported in Table 5. Of that
total, we observe 21 (22) significantly different mean (median) differences at least at the 10% level. 18
The primary takeaway from Table 5, which is consistent with our unified framework of MEFs, is that
roughly two thirds of the 11 independent variables exhibit significantly different means and medians
across MEF subsamples. If our algorithm to partition MEFs into DOA, opportunistic and voluntary
subsamples was simply assigning MEFs to subsamples at random, we would have observed no significant
differences in means or medians across the subsamples.




18
  The chance of observing 21 or 22 successes (i.e., differences) where the likelihood of success (a difference) is
10% out of 33 trials is 0.000.

                                                                                                                     23
5.2   Primary empirical tests of the managerial decision to issue a MEF
        Table 6 reports the results of estimating the multinomial probit using all firm-quarter observations
(i.e., with or without an actual earnings announcement, but excluding quarters with earnings warnings or
preannouncements). The dependent variable is zero if the firm-quarter contains no MEF, 1 if it contains
only DOA MEFs, 2 if it contains only opportunistic MEFs and 3 if it contains only voluntary MEFs.
Separate coefficient estimates are generated for each independent variable and subsample. The symbol ‡
to the right of a given coefficient indicates whether it is consistent with the predictions of our unified
MEF disclosure theory (see Table 1). If a given variable is predicted to have a positive sign (e.g., ERC)
then a ‡ denotes that the estimated coefficient is larger in that subsample than in the subsample where that
variable is not expected to explain MEF disclosure. Likewise, if the coefficient is predicted to be
negative, then a ‡ denotes that the estimated coefficient is smaller in that subsample than in the subsample
where that variable is not expected to explain MEF disclosure.
        Examination of the results reported in panel A of table 6 reveals that the rank of the firm’s ERC is
significant in predicting all three types of MEFs. If we could partition MEFs into subsamples without
classification errors we would only expect the ERC variable to be significant in the DOA and voluntary
MEF subsamples because they are the only subsamples where our unified theory predicts that ERCs
should explain MEF disclosure choice (see Table 1). However, misclassifications will arise because we
cannot observe managers’ actual MEF disclosure motives; and our empirical analysis involves a joint test
of our classification algorithm and our unified framework of MEF disclosure choice. Since we recognize
that subsample classification errors exist, our joint test involves comparing the estimated coefficients on
the ERC variable across the three subsamples. In particular, the ERC variable should be more important
when predicting MEF disclosure choice in the DOA and voluntary MEFs subsamples when compared to
predicting opportunistic MEFs. Consistent with this prediction we observe that the estimated coefficients
are larger in the DOA and voluntary MEF subsamples (0.108 and 0.109) when compared to the
opportunistic MEF subsample (0.088). In a similar vein, we expect the coefficients on the
EARN_SURPRISE variable to be larger in the DOA and voluntary MEF subsamples when compared to
the opportunistic MEF subsample. The results show that the coefficient on EARN_SURPRISE is larger
in the voluntary MEF subsample (0.101) when compared to the opportunistic MEF subsample (0.014),
while the coefficient for the DOA subsample (0.011) is smaller than that in the opportunistic MEF
subsample (0.014).
        The next seven coefficients in Table 6 (LNMVE, MB, RTNVOL, EARNVOL, HI_TECH,
REGULATION, and HABITUAL) are predicted to be more important (i.e., have a greater affect on MEF
disclosure choice) in the voluntary MEF subsample than in the other two subsamples (i.e., if a given
coefficient is predicted to positive it should be larger than that for the other two subsamples, and if the

                                                                                                              24
coefficient is predicted to be negative it should be smaller than in the other two subsamples). Turning to
the results we find that, with the exception of LNMVE and MB, the five estimated coefficients expected
to capture voluntary disclosure incentives are consistently larger (if the coefficient is predicted to be
positive) or smaller (if the coefficient is predicted to be negative) in the voluntary subsample than in at
least one of the other two subsamples. The last two coefficients (RESTATE and BACKDATE) are
designed to capture weak corporate governance and are predicted to be larger (i.e., more important in
predicting MEF disclosure, note that both coefficients are predicted to be positive) in the opportunistic
subsample than in the other two subsamples. Consistent with our predictions (see table 2), the results
reveal that both coefficients are larger than the corresponding coefficients in the other two subsamples.
        With 11 variables in each model’s estimation and two pair-wise comparisons between models to
make (e.g., DOA versus the other two subsamples), there are 22 possible sign comparisons to make in
Table 6, where a sign comparison is based on the predictions shown in Table 1. For example, one sign
comparison involves whether the estimated coefficient on ERC is larger in the DOA subsample than in
the opportunistic subsample. Ignoring the two sign comparisons involving the coefficient on MB
(because it has the wrong sign in the voluntary MEF subsample) leaves 20 sign comparisons. Of these,
fourteen of the sign comparisons are in the predicted direction (as denoted by a ‡ in the table). The
likelihood that observing 14 differences out of 20 comparisons is due to chance is 0.06 (one-tailed test).
Overall, this evidence is consistent with our unified MEF disclosure framework predicting that commonly
used variables in the MEF disclosure literature exhibit differential explanatory power for managers’
decision to issue MEFs for DOA, opportunistic and voluntary reasons.
        In addition to the sign comparisons reported above, we also performed six Wald tests to examine
the equality of a set of coefficients across models. For example, our unified MEF disclosure framework
predicts that the coefficients on ERC and EARN_SURPRISE should be larger (i.e., more important) in
the DOA subsample than in the opportunistic subsample. Panel B of Table 6 reports the results. Based
on a Chi-square statistic 1.02 we are unable to reject the null that the coefficients on ERC and
EARN_SURPRISE are different between the DOA and opportunistic MEF subsamples. However, the
test indicates the coefficients are different in the voluntary and opportunistic MEF subsamples (Chi-
square = 27.96, p-value < 0.001). The Wald tests comparing all seven variables between the voluntary
and DOA subsamples, and the voluntary and opportunistic subsamples reveal significant differences (Chi-
square statistics of 29.8 and 78.48, respectively, p-values < 0.001). The final two Wald tests reported in
Panel B compare the two variables expected to explain opportunistic MEFs (RESTATE and
BACKDATE). Only one of the two Wald tests (opportunistic vs. DOA, Chi-square = 5.19, p-value =
0.075) rejects the null hypothesis of equality. In summary, four of the six Wald tests reject the null of
equality.

                                                                                                              25
           Overall, we interpret the evidence in Table 6 to be generally supportive of our unified MEF
disclosure framework as described in section 3. The essence of that framework is that “one size does not
fit all” in the sense that a given independent variable varies in its ability to predict MEF disclosure across
our three MEF subsamples. Furthermore, the estimated coefficients vary in predictable ways across the
three subsamples, consistent with our unified MEF disclosure framework, and consistent with the notion
that some MEFs are issued because managers behave opportunistically, others are issued because
managers have incentives to make voluntary disclosures (like MEFs) to increase firm transparency, while
some MEFs are released because managers want to trade in their firm’s securities and want to comply
with insider trading restrictions requiring them to disclose their private information or to abstain from
trading.


5.3   Robustness tests
           To insure the robustness of the inferences drawn from Table 6 we perform several sensitivity
tests. First, we replaced the ±5% cutoff for defining “large” vs. “small” CARs for classifying voluntary
and opportunistic MEFs (see Table 2) with ±3% and ±10% cutoffs. Our inferences are unchanged under
these alternative thresholds. Second, the analysis in Table 6 pools all MEFs (whether they were issued
concurrently with an actual earnings announcement or not). We re-estimated the multinomial probit
model using only bundled MEF firm-quarters (results reported in Table 7) and only unbundled MEF firm-
quarters (results reported in table 8). 19
           The results reported in Panel B of Table 7 reveal three out of six significant Wald tests compared
to four of six significant Wald tests in Table 6. In addition, while Table 6 reported 14 out of 20 sign
comparisons consistent with our predictions, Table 7 reports 14 out of 18 consistent sign comparisons
(see Panel A of Table 7). 20 The results reported in Table 8 show that two of the six Wald tests are
significant (see Panel B of Table 8) and that 10 out of 16 sign comparisons are significant (see Panel A of
Table 8). 21 The Wald test results in Table 8 are weaker than those in Table 6 which is likely due to
having fewer MEF firm-quarters containing only unbundled MEFs (3,897 unbundled MEF firm-quarters
in Table 8 vs. 12,692 MEF firm-quarters in Table 6). Overall, the results reported in Tables 7 and 8
indicate that Table 6’s results are not driven by either bundled or unbundled MEFs. Moreover, the sign


19
   Firm-quarters containing both bundled and unbundled MEFs are excluded from the estimations reported in
Tables 7 and 8, but were included in Table 6’s estimations. Consistent with Rogers and Van Buskirk (2009), our
sample contains 7,767 firm-quarters classified as only bundled MEFs and 3,897 unbundled MEF firm-quarters (see
Table 3 Panel A).
20
   Only 18 sign comparisons are possible in Table 7 because LNMVE and MB have the wrong sign in the voluntary
MEF subsamples.
21
   Only 16 sign comparisons are possible in Table 8 because ERC has the wrong sign in both the DOA and voluntary
MEF subsamples and MB and HI_TECH have the wrong signs in the voluntary MEF subsample.

                                                                                                             26
comparisons are similar across the two multinomial probit estimations using either just bundled or
unbundled MEFs and the fewer significant Wald tests in Tables 7 and 8 are consistent with lower power
tests. As a result, we conclude that our inferences from Table 6 are robust to bundled and unbundled
MEFs.
        Finally, steps A1b (i.e., MEF CAR ≥ +5% with insider selling in the [+1,+30] trading day interval
and (MEF – Actual EPS) ≤ 0) and A1c (MEF CAR ≤ -5% with insider buying in the [+1,+30] trading
day interval and (MEF- Actual EPS) ≥ 0) of our MEF classification algorithm classify all large news
MEFs that are not opportunistic as voluntary. We recognize that either of these two steps might
misclassify some DOA MEFs as voluntary. For example, some managers may have issued a MEF
because they wanted to trade (DOA) and some of these MEFs resulted in large price changes. As a
robustness check, we reran our tests with all A1b and A1c MEFs reclassified as DOA and all prior
inferences remain unchanged or improve. For example, the significance of the Wald tests increases in
Tables 6 and 7 and in Table 8 three of the six Wald tests are significant whereas before only two of the six
Wald tests reject the null hypothesis of equal coefficients.


6. Conclusions
        Despite an extensive literature on MEFs, few studies recognize, and none explore the role played
by the securities laws and stock exchange listing agreements as a determinant of managers’ decision to
issue MEFs. Moreover, the literature has yet to fully recognize that the decision to issue a MEF can be
driven by a manager’s desire to comply with insider trading restrictions prior to trading in his/her firm’s
securities. Rather, most studies assume that MEFs are voluntary disclosures made to reduce information
asymmetries or are opportunistic disclosures where managers disclose good news before selling stock or
they disclose bad news before selling stock or receiving stock options. This paper attempts to remedy this
deficiency by providing a framework that incorporates three motives for why MEFs are disclosed: (i) pure
voluntary disclosure incentives, (ii) opportunistic motives and (iii) to comply with insider trading
prohibitions such as the disclose or abstain rule underlying SEC rule 10b-5. We hypothesize that each of
the three reasons to issue MEFs will have a set of independent variables that explains whether that
particular type of MEF will be disclosed in a given firm quarter, and that those independent variables
differ across the three types of MEFs.
        Like other areas of accounting research (e.g., earnings management or conservatism) we cannot
observe managers’ actual motives to issue MEFs. As a result, we must rely on indirect measures to
capture such motives. Consistent with this, we first develop an algorithm that assigns MEFs into one of
three mutually exclusive and exhaustive subsamples (voluntary MEFs, opportunistic MEFs and disclose
or abstain MEFs) based on characteristics of the MEFs such as the stock price reaction to the MEF, the

                                                                                                              27
existence of insider purchases and sales and whether the MEF was optimistic or pessimistic. The
algorithm indicates that about 68% of all MEFs are voluntary, 14% are opportunistically motivated, and
18% are DOA.
        Since our empirical analysis is a joint test of our MEF partitioning algorithm and our unified
MEF framework we provide some evidence lending construct validity to our classification scheme. First,
the MEFs we classify as DOA are more likely to be issued concurrently with earnings announcements
than are MEFs classified as voluntary or opportunistically motivated. If DOA MEFs are issued because
managers want to insure they have disclosed all of their private material information, then we argue that
they will report actual earnings and a concurrent forecast of future earnings prior to trading. Second, we
find that the existence and intensity of insider trading is heaviest in the five-day trading window following
the release of an MEF classified as DOA when compared to any other five-day trading window in the 30
days preceding and following the MEF’s release. Third, the likelihood and intensity of insider trading is
greatest among MEFs classified as DOA when compared to MEFs classified as voluntary or
opportunistic. After categorizing each firm-quarter with a MEF into the voluntary, opportunistic, or DOA
subsamples, we estimate a pooled multinomial probit model with all three subsets using independent
variables predicted to explain the disclosure MEFs. Consistent with our expectations, the independent
variables predicted to explain a particular type of MEF (e.g., voluntary MEFs) have greater explanatory
power for the MEFs in that particular subsample (e.g., the subsample of voluntary MEFs) than in the
other two MEF subsamples (e.g., opportunistic and DOA).
        Based on this evidence we infer that “not all MEFs are created equal” in the sense that no single,
simple model, such as a voluntary disclosure theory based on presumed cost of capital benefits, explains
managers’ rather complex decision process to issue a MEF. Simply put, some MEFs are issued
voluntarily, while others are issued because the manager is opportunistically motivated, and yet others are
issued because the manager is complying with insider trading restrictions and is issuing the MEF prior to
selling or buying the firm’s stock.
        In this paper we identified an overlooked reason for why managers issue guidance – not wanting
to violate insider trading regulations, but still wanting to trade in their firm’s securities, they issue
earnings guidance (i.e., an MEF). Because the MEF literature has overlooked this economic motive to
issue MEFs, namely, a disclose or abstain from trading rationale, our findings have implications for prior
and future MEF research. First, many MEFs treated as voluntary disclosures in prior research were not
truly voluntary in the sense they are not motivated by managers’ desire to improve their firm’s
transparency, reduce information asymmetries and perhaps lower the firm’s cost of capital. As a result,
including MEFs issued for opportunistic or DOA reasons reduces the power of the test in settings
designed to test pure voluntary disclosure theories. Furthermore, prior studies treating all MEFs as if they

                                                                                                            28
are all being made voluntarily misclassifies some MEFs as voluntary that are likely being made for
opportunistic or DOA reasons. Such misclassifications introduce an identification problem into voluntary
disclosure studies potentially leading to incorrect inferences regarding voluntary disclosure theories. This
is because some of the variables used to predict purely voluntary MEFs are the same variables that are
likely to predict MEFs being issued for DOA reasons. The algorithm we develop to identify MEFs issued
voluntarily (and for other reasons as well) should help future researchers structure better specified tests of
pure voluntary disclosure theories.




                                                                                                            29
Appendix

Variable Definition and Measurement

MEFDUM equals one if there is at least one MEF issued during the period from the earnings
announcement date of the last quarter up to one day before the current quarter’s earnings announcement
date and zero otherwise.
ERC is the quartile rank of the firm’s estimated earnings response coefficient. We estimate an ERC for
each firm-quarter by regressing the two-day (i.e., day 0 and +1) earnings announcement period market-
adjusted return on unexpected earnings using the 16 most recent quarters (we require non-missing data for
all 16 past quarters). Unexpected earnings is calculated as actual EPS minus the most recent consensus
analyst earnings forecast issued prior to the earnings announcement date, deflated by the stock price at
day -1.
EARN_SURPRISE is the quartile rank of the absolute value of the difference between the most recent
consensus analyst earnings forecast issued prior to three weeks before the end of the fiscal quarter and
that issued prior to the actual earnings announcement date of the prior quarter, deflated by the stock price
one day before the prior quarter’s earnings announcement date.
SIZE is the natural logarithm of the market value of common equity at the beginning of the quarter.
MB is the market-to-book value of common equity ratio at the beginning of the quarter.
RTNVOL is the standard deviation of daily raw returns over the 250 trading days before the beginning of
the quarter (a minimum of 100 days is required).
EARNVOL is the standard deviation of the seasonal change in actual quarterly EPS scaled by assets per
share as of the beginning of the quarter based on the 16 most recent quarters of data (we require non-
missing data for all 16 past quarters).
HI_TECH equals one if the COMPUSTAT SIC code at the end of the quarter is 2833–2836 (Drugs),
8731–8734 (R&D services), 7371–7379 (Programming), 3570–3577 (Computers), or 3600–3674
(Electronics) and zero otherwise.
REGULATION equals one if the COMPUSTAT SIC code at the end of the quarter is 4812–4813
(Telephone), 4833 (TV), 4841 (Cable), 4811–4899 (Communications), 4922–4924 (Gas), 4931
(Electricity), 4941 (Water), or 6021–6023, 6035–6036, 6141, 6311, 6321, 6331 (Financial firms) and zero
otherwise.
HABITUAL equals one if the firm issued a MEF in at least six of the last eight quarters.
%MOB is the firm’s percentage of Meet-or-Beat (MOB) quarters in the 16 most recent quarters. A
quarter is a MOB quarter if actual EPS equals or exceeds the most recent consensus analyst earnings
forecast by at most two cents.
RESTATE equals one if the firm has at least one financial restatement during the period 1998-2006 per
the U.S. Government Accountability Office (GAO) Financial Restatement database and zero otherwise.
BACKDATE equals one if the firm is listed in The Wall Street Journal's "Options Scorecard" on
November 28, 2006 as being under scrutiny for option grant backdating and zero otherwise.




                                                                                                          30
                                                References

Aboody, D., and R. Kasznik, 2000, “CEO Stock Option Awards and the Timing of Corporate Voluntary
      Disclosures,” Journal of Accounting and Economics 29, pp. 73-100.
Ajinkya, B., S. Bhojraj, and P. Sengupta, 2005, “The Association Between Outside Directors,
       Institutional Investors and the Properties of Management Earnings Forecasts,” Journal of
       Accounting Research 43, pp. 343-376.
Ajinkya, B and and M. Gift, 1984, “Corporate Managers’ Earnings Forecasts and Symmetrical
       Adjustments of Market Expectations,” Journal of Accounting Research 22, pp. 425–444.
Anilowski, C., M. Feng, and D. Skinner, 2007, “Does Earnings Guidance Affect Market Returns? The
       Nature and Information Content of Aggregate Earnings Guidance,” Journal of Accounting and
       Economics 44, pp. 36-63.
Baginski, S., E. Conrad, and J. Hassell, 1993, “The Effects of Management Forecast Precision on Equity
       Pricing and on the Assessment of Earnings Uncertainty,” Accounting Review 68, pp. 913-927.
Baginski, S., J. Hassell, and M. Kimbrough, 2004, “Why Do Managers Explain Their Earnings
       Forecasts?" Journal of Accounting Research 42, pp. 1-29.
Bettis, J., J. Coles, M. Lemmon, 2000, “Corporate Policies Restricting Trading by Insiders,” Journal of
         Financial Economics 57, pp. 191-220.
Botosan, C., 1997, “Disclosure Level and the Cost of Equity Capital,” Accounting Review 72 (3), pp. 323-
       349.
Botosan, C. and M. Plumlee, 2002, “A Re-examination of Disclosure Level and the Expected Cost of
       Equity Capital,” Journal of Accounting Research 40 (1), pp. 21-40.
Brickley, J., and J. Zimmerman, 2009, “Corporate Governance Myths: Some Comments on Armstrong,
        Guay, and Weber,” Working paper University of Rochester.
Cheng, Q., and K. Lo, 2006, “Insider Trading and Voluntary Disclosures,” Journal of Accounting
       Research 44, pp. 814-48.
Clement, M., R. Frankel, and J. Miller, 2003, “Confirming Management Earnings Forecasts, Earnings
       Uncertainty, and Stock Returns,” Journal of Accounting Research 41, pp. 653-679.
Coller, M., and T. Yohn, 1997, “Management Forecasts and Information Asymmetry: An Examination of
        Bid-ask Spreads,” Journal of Accounting Research 35, pp. 181-191.
Collins, D., and S. Kothari, 1989, “An Analysis of the Cross-sectional and Intertemporal Determinants of
         Earnings Response Coefficients,” Journal of Accounting and Economics 11, pp. 143-181.
Diamond, D., 1985, “Optimal Release of Information by Firms,” Journal of Finance, pp. 1071-94.
Diamond, D., and R. Verrecchia, 1991, “Disclosure, Liquidity, and the Cost of Capital,” Journal of
      Finance 46, pp. 1325-1355.
Dye, R., 1985a, “Disclosure of nonproprietary information” Journal of Accounting Research, 23,
        pp. 123–145.
Dye, R., 1985b, “Strategic accounting choice and the effects of alternative financial reporting
        Requirements,” Journal of Accounting Research 23, pp. 544–574.




                                                                                                          31
Frankel, Richard and Xu Li, 2004, “Characteristics of a Firm’s Information Environment and the
        Information Asymmetry Between Insiders and Outsiders,” Journal of Accounting and Economics
        37, pp. 229-259.
Frankel, R., M. McNichols, and G. Wilson, 1995 “Discretionary Disclosure and External Financing,”
        Accounting Review 70, pp. 135-150.
Graham, J., C. Harvey, and S. Rajgopal, 2005, “The Economic Implications of Corporate Financial
      Reporting,” Journal of Accounting and Economics 40, pp. 3-73.
Healy, P., A. Hutton, and K. Palepu, 1999, “Stock Performance and Intermediation Changes Surrounding
        Sustained Increases in Disclosure,” Contemporary Accounting Research 16, pp. 485-520.
Healy, P., and K. Palepu, 2001, “Information Asymmetry, Corporate Disclosure, and the Capital Markets:
        A Review of the Empirical Disclosure Literature,” Journal of Accounting and Economics 31, pp.
        405-440.
Heitzman, S., C. Wasley, and J. Zimmerman, 2010, “The Joint Effects of Materiality Thresholds and
       Voluntary Disclosure Incentives on Firms’ Disclosure Decisions,” forthcoming Journal of
       Accounting and Economics.
Heminway, J., 2003. “Materiality Guidance in the Context of Insider Trading: A Call for Action” 52 AM.
      U. L. REV. 1131 reprinted at 36 SEC. L. REV. 448 (2004).
Hirst, D., L. Koonce, and S. Venkataraman, 2008. Management Earnings Forecast: A Review and
        Framework. Accounting Horizons 22, pp. 315-38.
Huddart, S., B. Ke, and C. Shi, 2007, “Jeopardy, Non-public Information, and Insider Trading Around
       SEC 10-K and 10-Q Filings,” Journal of Accounting and Economics 43, pp. 3-36.
Hutton, A., G. Miller, and D. Skinner, 2003, “The Role of Supplementary Statements with Management
        Earnings Forecasts,” Journal of Accounting Research 41, pp. 867-890.
Kasznik, R., and B. Lev, 1995, “To Warn or Not to Warn: Management Disclosures in the Face of an
       Earnings Surprise,” Accounting Review 70, pp. 113-134.
Ke, B., S. Huddart and K. Petroni, 2003, “What Insiders Know about Future Earnings and How They Use
        It: Evidence from Insider Trades,” Journal of Accounting and Economics 35, pp. 315-346.
King, R., G. Pownall and G. Waymire, 1990, “Expectations Adjustment via Timely Management
       Forecasts: Review, Synthesis and Suggestions for Future Research,” Journal of Accounting
       Literature 9, pp. 113–144.
Lakonishok, J. and I. Lee, 2001. Are insider trades informative? Review of Financial Studies 14, 79-111.
Lang, M. and R. Lundholm, 1993, “Cross-sectional Determinants of Analyst Ratings of Corporate
       Disclosures,” Journal of Accounting Research 31, pp. 246-271.
Lang, M. and R. Lundholm, 2000, “Voluntary Disclosure During Equity Offerings: Reducing Information
       Asymmetry or Hyping the Stock?” Contemporary Accounting Research 17, pp. 623-663.
Lansford, B., B. Lev, and J. Tucker, 2009, “Causes and Consequences of Disaggregating Earnings
       Guidance,” Working paper Northwestern University.
Larcker, D., S. Richardson, I. Tuna, 2007, “Corporate governance, accounting outcomes, and
        organizational performance,” Accounting Review 82, pp. 913-1008.
Lennox, C., and C. Park, 2006, “The Informativeness of Earnings and Management’s Issuance of
       Earnings Forecasts,” Journal of Accounting and Economics 42, pp. 439-458.


                                                                                                       32
Lev, B. and S. Penman, 1990, “Voluntary Forecast Disclosure, Nondisclosure, and Stock Prices,” Journal
        of Accounting Research 28, pp. 49-76.
Loss, L., and J. Seligman, 2004, Securities Regulation, 5th ed. (Aspen Publishers).
McNichols, M., 1989, “Evidence of Informational Asymmetries from Management Earnings Forecasts
      and Stock Returns,” Accounting Review 64, pp. 1-27.
Noe, C., 1999, “Voluntary Disclosures and Insider Transactions,” Journal of Accounting and Economics
        27, pp. 305-326.
Penman, S., 1980, “An Empirical Investigation of the Voluntary Disclosure of Corporate Earnings
      Forecasts,” Journal of Accounting Research 18, pp. 132-160.
Piotroski, J., and D.Roulstone. 2005, “Do Insider Trades Reflect both Contrarian Beliefs and Superior
        Knowledge about Future Cash Flow Realizations?” Journal of Accounting and Economics 39, pp.
        55-81.
Pownall, G., C. Wasley, and G. Waymire, 1993, “The Stock Price Effects of Alternative Types of
       Management Earnings Forecasts,” Accounting Review 68, pp. 896-912.
Rees, L., A. Srivastava, and S. Tse. 2008, “An Examination of the Accuracy and Usefulness of
        Management Earnings Guidance around Stock Option Grants,” Working paper Texas A&M.
Rogers, J., 2008, “Disclosure Quality and Management Trading Incentives,” Journal of Accounting
       Research 46, pp 1265-1296.
Rogers, J. and P. Stocken, 2005, “The Credibility of Management Forecasts,” Accounting Review 80, pp.
        1233-1260.
Rogers, J. and A. Van Buskirk, 2009, “Bundled Forecasts and Selective Disclosure of Good News,”
        Working Paper University of Chicago.
Roulstone, D., 2003, “The Relation Between Insider-trading Restrictions and Executive Compensation,”
       Journal of Accounting Research 41, pp. 525-551.
Seyhun, N., 1992, “The Effectiveness of the Insider-Trading Sanctions,” Journal of Law & Economics
       35, pp. 149- 182.
Skinner, D., 1994, “Why Firms Voluntarily Disclose Bad News,” Journal of Accounting Research 32, pp.
        38-60.
Skinner, D., 1997, “Earnings Disclosures and Stockholder Lawsuits,” Journal of Accounting and
        Economics 23, pp. 249-282.
Taub, S., 2003, “Former Waste Management Executives Settle Insider-Trading Charges,” CFO.com
        (October 24).
Trueman, B., 1986, “Why do managers voluntarily release earnings forecasts?” Journal of Accounting
      and Economics 8, pp. 53–72.
Verrecchia, R., 1982, “The use of mathematical models in financial accounting,” Journal of Accounting
       Research 20, pp. 1–55.
Verrecchia, R., 1983, “Discretionary Disclosure,” Journal of Accounting and Economics 5, pp. 179-94.
Verrecchia, R., 2001, “Essays on Disclosure,” Journal of Accounting and Economics 32, pp. 97-180.
Verrecchia, R., and J. Weber, 2006, “Redacted Disclosure,” Journal of Accounting Research 2006, pp.
       791-814.



                                                                                                        33
Williams, P., 1996, “The Relation between a Prior Earnings Forecast by Management and Analyst
       Response to a Current Management Forecast,” Accounting Review 71, pp. 103-113.
Yermack, D.,1997, “Good Timing: CEO Stock Option Awards and Company News Announcements,”
      Journal of Finance 52, pp. 449-476.




                                                                                                34
Table 1
Independent Variables Predicting Management Earnings Forecast Disclosure in the Voluntary Disclosure,
Opportunistic Disclosure and Disclose or Abstain Disclosure Subsamples

                Voluntary Disclosure          Opportunistic        Disclose or Abstain
                  Subsample (Vi)             Subsample (Oi)         Subsample (Mi)
              ERC (+)                      RESTATE (+)           ERC (+)
              EARN_SURPRISE (+)            BACKDATE (+)          EARN_SURPRISE (+)
              SIZE (+)
              MB (-)
              RTNVOL (-)
              EARNVOL (-)
              HI_TECH (+)
              REGULATION (-)
              HABITUAL (+)


          Predicted signs in parentheses




                                                                                                   35
Table 2
Classification of Voluntary, Opportunistic and Disclose or Abstain Management Earnings Forecast
Subsamples


                                                                                                  Percent
                                                                                    Percent of
  Subsample                                Algorithm                       N                        of
                                                                                    Subsample
                                                                                                  sample

                            No Insider Trading (purchases or sales)
                            in the [-30,+30] interval and no suspect
                 A1a        opportunistic MEF.                           12,146        77.1%
                            CAR on MEF announcement date ≥ +5%
                            followed by insider sales in the [+1,+30]
                 A1b        interval and MEF ≤ Actual EPS.                947          6.0%
   Voluntary                CAR on MEF announcement date ≤ -5%
                            followed by insider purchases in the
                 A1c        [+1,+30] interval and MEF ≥ Actual.           202          1.3%
                            Insider purchases in the [-30,-1] interval
                            prior to a bad news MEF or insider sales
                            in the [-30,-1] interval prior to a good
                 A1d        news MEF.                                    2,456         15.6%

                            Total Voluntary                              15,751       100.0%      68.1%

                            Insider purchases in the [-30,-1] interval
                            prior to a good news MEF or insider
                            sales in the [-30,-1] interval prior to a
                 A2a        bad news MEF.                                2,580         81.7%
                            CAR on MEF announcement date ≥+5%
 Opportunistic
                            followed by insider sales in the [+1,+30]
                 A2b        interval and MEF > Actual EPS.                431          13.7%
                            CAR on MEF announcement date ≤ -5%
                            followed by insider purchases in the
                 A2c        [+1,+30] and MEF < Actual EPS.                145          4.6%

                             Total Opportunistic                         3,156        100.0%      13.6%

  Disclose or
   Abstain
    (DOA)        A3a        All other MEFs                               4,237        100.0%      18.3%

     Total                                                               23,144                   100.0%




                                                                                                  36
Table 3
Descriptive Information on Management Earnings Forecasts

Panel A: Number of firm-quarters

                                          Quarters with MEFs                          Quarters
                                                                                      without
                                       Opportunistic                         Total
                      DOA MEFs                             Voluntary MEFs              MEFs
                                          MEFs                               MEFs
Year                   N       %        N        %           N         %      N          N
1998                    59   17.1%        54 15.6%             233   67.3%      346          2,384
1999                    86   18.6%        73 15.8%             303   65.6%      462          2,787
2000                    94   14.8%        83 13.1%             456   72.0%      633          2,910
2001                   158   14.0%       159 14.1%             810   71.9%    1,127          2,671
2002                   199   14.8%       150 11.1%             999   74.1%    1,348          3,125
2003                   261   17.8%       140 9.5%            1,066   72.7%    1,467          3,657
2004                   354   20.0%       170 9.6%            1,243   70.3%    1,767          3,885
2005                   316   17.6%       213 11.8%           1,271   70.6%    1,800          4,272
2006                   344   18.0%       218 11.4%           1,346   70.5%    1,908          4,291
2007                   334   18.2%       240 13.1%           1,260   68.7%    1,834          4,506

Quarter                N       %        N         %          N         %      N          N
1                      564   25.6%       342    22.8%        2,033   22.6%    2,939          8,046
2                      544   24.7%       368    24.5%        2,125   23.6%    3,037          9,064
3                      529   24.0%       378    25.2%        2,375   26.4%    3,282          8,824
4                      568   25.8%       412    27.5%        2,454   27.3%    3,434          8,554

Timing of MEFs relative to earnings announcement
                     N        %        N      %              N       %        N          N
Bundled Only       1,675 76.0%          783 52.2%            5,309 59.1%      7,767               -
Unbundled Only       468 21.2%          652 43.5%            2,777 30.9%      3,897               -
Both                  62 2.8%            65 4.3%               901 10.0%      1,028               -

Total                2,205 17.4%        1,500 11.8%          8,987 70.8%     12,692       34,488




                                                                                                 37
 Panel B: Number of MEFs
                                                 Categories of MEFs
                               DOA                   Opportunistic               Voluntary               Total
 Horizon:                N            %             N         %                N         %                N
 Quarterly             2,369        55.9%         1,811     57.4%            8,754     55.6%            12,934
 Annual                1,868        44.1%         1,345     42.6%            6,997     44.4%            10,210
 Form:
 Point                  663         15.7%          543         17.2%        2,453         15.6%         3,659
 Range                 3,148        74.4%         2,403        76.2%        12,103        76.9%         17,654
 End                    231         5.5%           91          2.9%          593          3.8%           915
 Qualitative            192         4.5%           116         3.7%          587          3.7%           895
 News:
 Bad                   2,000        47.2%         1,714        54.4%         7,907        50.2%         11,621
 Good                  1,528        36.1%          972         30.8%         5,684        36.1%         8,184
 Neutral                685         16.2%          460         14.6%         2,085        13.2%         3,230
 Mixed                  24          0.6%           10          0.3%           75          0.5%           109

 Total                 4,237        18.3%         3,156        13.6%        15,751        68.1%         23,144


The table is based on a sample of 47,173 firm-quarter observations from the period 1998-2007. Panel A presents the
number of firm-quarter observations in each category by year and quarter. Quarters with (without) MEFs include
firm-quarter observations that have at least one (no) MEF issued during the period from the last quarter’s earnings
announcement date up to one day before the current quarter’s earnings announcement. A quarter is defined as DOA,
Opportunistic or Voluntary quarter if all MEFs issued in the quarter are classified as DOA, Opportunistic or
Voluntary, respectively. A quarter is classified as Bundled Only (Unbundled Only) if all MEFs in the quarter are
made (not made) in conjunction with (i.e., within [-1,+1] trading day relative to) an earnings announcement. If a
quarter contains both bundled and non-bundled MEFs, it is classified as Both. Panel B presents the number of
MEFs in each category by forecasting horizon (i.e., quarterly or annual), form (i.e., point, range, end, or qualitative
estimates) and news. A MEF is classified as “Voluntary” if it meets any of the four criteria: (1) there is no insider
trading within the [-30,+30] trading day window centered on the MEF release date; (2) it is a “large” good news
MEF (two-day CARs > 5%) followed by insider selling within 30 trading days and the amount of the MEF does not
exceed actual earnings; (3) it is a “large” bad news MEF (two-day CARs < -5%) followed by insider buying within
30 trading days and actual earnings does not exceed the amount of the MEF; (4) insider buying occurs in the 30 days
prior to a MEF with negative two-day announcement period CARs or insider selling occurs in the 30 days prior to a
MEF with positive two-day announcement period CARs. A MEF is classified as “Opportunistic” if it meets any of
the three criteria: (1) insider selling occurs in the 30 days prior to a MEF with negative two-day announcement
period CARs or insider buying occurs in the 30 days prior to a MEF with positive two-day announcement period
CARs; (2) it is a “large” good news MEF (two-day CARs > 5%) followed by insider selling within 30 trading days
and the amount of the MEF exceeds actual earnings; (3) it is a “large” bad news MEF (two-day CARs < -5%)
followed by insider buying within 30 trading days and actual earnings exceeds the amount of the MEF. All MEFs
not classified as “Voluntary” or “Opportunistic” are classified as “DOA” (i.e., disclose or abstain). Following
Anilowski et al (2007) we classify MEFs into good, bad, neutral and mixed news based on First Call CIG code, the
amount of the MEF and the most recent analyst consensus forecast.




                                                                                                                    38
Table 4
Incidence and Intensity of Insider Trading around Management Earnings Forecasts
Panel A: Insider Sales
                     % of MEFs with insider sales in each window                  Mean percentage as of total annual insider sales
 Trading Day         DOA            Opportunistic        Voluntary
                                                                                      DOA          Opportunistic       Voluntary
   Window          (N=4,237)         (N=3,156)          (N=15,751)
                       %                  %                 %                     N         %       N        %        N         %
[-30,-26]              -                26.7%              4.8%                  -           -      844    18.2%       752    16.9%
[-25,-21]              -                23.5%              4.5%                  -           -      743    19.2%       702    17.4%
[-20,-16]              -                24.8%              4.6%                  -           -      782    17.4%       723    15.3%
[-15,-11]              -                23.3%              4.2%                  -           -      736    14.0%       669    16.0%
[-10, -6]              -                19.3%              3.5%                  -           -      610    15.2%       556    14.3%
[ -5, -1]              -                16.0%              2.9%                  -           -      505    14.1%       453    13.8%
[ 0 ]                1.0%               4.1%               1.5%                 43      20.7%       128     8.7%       243    16.5%
[ +1, +5]            33.2%              22.7%              7.6%              1,407      30.6%       717    20.2%     1,197    20.6%
[ +6,+10]            26.3%              20.0%              5.4%              1,116      28.4%       630    15.6%       856    14.7%
[+11,+15]            22.4%              17.3%              4.6%                948      26.3%       546    12.0%       722    13.1%
[+16,+20]            20.4%              15.3%              4.3%                863      26.6%       484    12.0%       676    13.1%
[+21,+25]            18.6%              14.8%              4.4%                789      22.5%       467    11.7%       688    13.4%
[+26,+30]            16.1%              16.0%              4.0%                683      26.8%       505    12.8%       632    13.3%




                                                                                                                                      39
 Table 4 Panel B: Insider
 Purchases
                     % of MEFs with insider purchases in each window                    Mean percentage as of total annual insider purchases
 Trading Day            DOA              Opportunistic           Voluntary
                                                                                              DOA            Opportunistic         Voluntary
   Window             (N=4,237)           (N=3,156)             (N=15,751)
                          %                    %                     %                    N         %         N         %         N         %
 [-30,-26]                 -                  2.4%                  0.4%                     -         -     76       48.4%      66       42.7%
 [-25,-21]                 -                  2.8%                  0.5%                     -         -     88       54.1%      78       42.1%
 [-20,-16]                 -                  2.0%                  0.3%                     -         -     64       50.4%      55       61.2%
 [-15,-11]                 -                  1.9%                  0.4%                     -         -     61       34.2%      61       50.9%
 [-10, -6]                 -                  1.9%                  0.3%                     -         -     59       52.4%      41       46.6%
 [ -5, -1]                 -                  1.6%                  0.2%                     -         -     50       61.7%      38       41.1%
 [ 0 ]                   0.2%                 0.7%                  0.2%                    10    46.0%      21       19.5%      27       53.9%
 [ +1, +5]               4.8%                 3.3%                  0.8%                   204    75.1%      104      74.2%      128      65.4%
 [ +6,+10]               3.8%                 2.2%                  0.5%                   162    64.7%      69       47.9%      76       48.1%
 [+11,+15]               3.2%                 1.6%                  0.3%                   134    67.4%      52       54.9%      51       36.7%
 [+16,+20]               2.3%                 1.2%                  0.4%                    99    69.2%      37       37.9%      66       50.9%
 [+21,+25]               1.9%                 1.1%                  0.3%                    79    70.5%      34       43.6%      52       46.8%
 [+26,+30]               1.5%                 1.2%                  0.3%                    65    58.6%      39       39.0%      45       56.9%

The table is based on a sample of 23,144 MEFs from the period 1998-2007. Panel A (B) reports on insider sales (purchases) made within the [-30,+30]
trading day window centered on the MEF release date. See tables 1 and 2 for a description of the algorithm used to classify MEFs into the DOA,
Opportunistic and Voluntary subsamples. For each trading day window, the percent of MEFs with insider sales (purchases) is calculated as number of
MEFs that have insider selling (buying) in the window divided by total number of MEFs. For each trading day window, the mean percentage of total
annual insider sales (purchases) is computed on the sub-sample of MEFs with insider selling (buying) in the window. Percentage of total annual insider
sales (purchases) is calculated as number of shares sold (purchased) by insiders in the window divided by the annual total number of shares sold
(purchased) by insiders of the firm.




                                                                                                                                                         40
Table 5
Descriptive Statistics for the Independent Variables Used in the Multinomial Probit Model Estimation

                                                                  Quarters with MEFs                         Quarters
                                                                                                             without
 Variable:                                           DOA             Opportunistic        Voluntary           MEFs


 ERC                                Mean            2.830 *             2.759              2.773 ‡‡               2.395
                                    Median          3.000 *             3.000              3.000 ‡‡               2.000
 EARN_SURPRISE                      Mean           2.365               2.347     †††       2.601    ‡‡‡          2.485
                                    Median         2.000               2.000     †††       3.000    ‡‡‡          2.000
 LNMVE                              Mean           7.884     *         7.975     †††       7.498    ‡‡‡          7.337
                                    Median         7.702     *         7.816     †††       7.375    ‡‡‡          7.233
 MB                                 Mean           3.431     ***       3.916     †††       3.119    ‡‡‡          3.024
                                    Median         2.574     ***       2.859     †††       2.330    ‡‡‡          2.182
 RTNVOL                             Mean         0.00074             0.00078     †††     0.00085    ‡‡‡        0.00087
                                    Median       0.00047     **      0.00050     †††     0.00055    ‡‡‡        0.00054
 EARNVOL                            Mean         0.00121             0.00110     ††      0.00086    ‡‡‡        0.00112
                                    Median       0.00007     **      0.00009     ††      0.00008               0.00008
 HI_TECH                            Mean           0.348               0.332               0.326    ‡‡           0.227
                                    Median         0.000               0.000               0.000    ‡‡           0.000
 REGULATION                         Mean           0.075     **        0.058               0.069                 0.100
                                    Median         0.000     **        0.000               0.000                 0.000
 HABITUAL                           Mean           0.437     ***       0.355               0.368    ‡‡‡          0.026
                                    Median         0.000     ***       0.000               0.000    ‡‡‡          0.000
 RESTATE                            Mean           0.282     ***       0.322               0.308    ‡‡‡          0.260
                                    Median         0.000     ***       0.000               0.000    ‡‡‡          0.000
 BACKDATE                           Mean           0.278               0.322     †         0.307                 0.260
                                    Median         0.000               0.000     †         0.000                 0.000
 N                                                 2,205               1,500               8,987                34,488

The table is based on a sample of 47,173 firm-quarter observations for the period 1998-2007. Quarters with (without)
MEFs include firm-quarters that have at least one (no) MEF issued during the period from the last quarter’s earnings
announcement date up to one day before the current quarter’s earnings announcement. A quarter is defined as a DOA,
Opportunistic or Voluntary quarter if all MEFs issued in the quarter are classified as DOA, Opportunistic or
Voluntary, respectively. See tables 1 and 2 for a description of the algorithm used to classify MEFs into the DOA,
Opportunistic and Voluntary subsamples. See the appendix for variable definitions.

*, **, *** indicate p-values of 10%, 5%, and 1%, respectively, two-tailed tests. t-tests (Wilcoxon tests) are used to
test for differences in means (medians).

*’s under DOA are used to indicate whether the DOA sample differs from the Opportunistic sample.
†’s under Opportunistic are used to indicate whether the Opportunistic sample differs from the Voluntary sample.
‡’s under Voluntary are used to indicate whether the Voluntary sample differs from the DOA sample.




                                                                                                                 41
Table 6
Multinomial Probit Regression of the Decision to Issue Management Earnings Forecasts (All MEFs including
those with and without an actual earnings announcement)

 Panel A:                                                          MEF Subsample

                                             DOA                    Opportunistic              Voluntary
Intercept                                      -4.170                      -4.266                 -3.219
                                            (-19.08) ***                 (-18.87)    ***        (-19.53)    ***
ERC                                    +         0.108 ‡                     0.088         +        0.109   ‡
                                                (6.12) ***                  (4.74)   ***           (7.34)   ***
EARN_SURPRISE                          +         0.011                       0.014         +        0.101   ‡
                                                (0.68)                      (0.76)                 (7.71)   ***
LNMVE                                            0.121                       0.126         +        0.060
                                                (7.30) ***                  (7.94)   ***           (4.36)   ***
MB                                             -0.002                        0.021         -        0.000
                                              (-0.34)                       (3.40)   ***           (0.06)
RTNVOL                                       -58.404                      -33.116          -     -42.241    ‡
                                              (-2.01) **                   (-1.17)                (-2.02)   **
EARNVOL                                          7.816                     -4.139          -      -8.783    ‡‡
                                                (1.40)                     (-0.84)                (-1.91)   *
HI_TECH                                          0.303                       0.238         +        0.284   ‡
                                                (5.76) ***                  (4.23)   ***           (6.29)   ***
REGULATION                                     -0.175                      -0.306          -      -0.237    ‡
                                              (-2.04) **                   (-3.41)   ***          (-3.23)   ***
HABITUAL                                         2.305                       2.094         +        2.388   ‡‡
                                             (43.21) ***                  (36.04)    ***         (48.11)    ***
RESTATE                                          0.058       +               0.172   ‡‡             0.147
                                                (1.14)                      (3.13)   ***           (3.35)   ***
BACKDATE                                         0.171       +               0.236   ‡‡             0.187
                                                (1.68) *                    (2.51)   **            (2.23)   **
Coefficients and z-statistics on year-quarter indicators have been suppressed.
Chi-square                                                              5,051.68
p-value                                                                   0.000
N                                                                        47,180




                                                                                                   42
 Table 6 (continued)

 Panel B: Wald tests for equality of coefficients:

 (1) ERC, EARN_SURPRISE:
 DOA vs. Opportunistic                    Chi-square2         =    1.02 p-value          =      0.600
 Voluntary vs. Opportunistic              Chi-square2         =    27.72 p-value         =      0.000

 (2) LNMVE, MB, RTNVOL, EARNVOL, HI_TECH, REGULATION, HABITUAL:
 Voluntary vs. DOA           Chi-square7 = 29.80 p-value  = 0.000
 Voluntary vs. Opportunistic Chi-square7 = 78.48 p-value  = 0.000

 (3) RESTATE, BACKDATE:
 Opportunistic vs. DOA                    Chi-square2         =    5.19    p-value       =    0.075
 Opportunistic vs. Voluntary              Chi-square2         =    0.56    p-value       =    0.756

Quarters with (without) MEFs include firm-quarters that have at least one (no) MEFs issued during the period from last
quarter’s earnings announcement date up to one day before the current quarter’s earnings announcement date. A quarter
is defined as a DOA, Opportunistic or Voluntary quarter if all MEFs issued in the quarter are classified as DOA,
Opportunistic or Voluntary, respectively. See tables 1 and 2 for a description of the algorithm used to classify MEFs into
the DOA, Opportunistic and Voluntary subsamples. See the appendix for variable definitions.

*, **, *** indicate p-values of 10%, 5%, and 1%, respectively, two-tailed tests. ‡ denotes that the coefficient
magnitude differs in the direction predicted in Table 1. Z-statistics for the pooled regressions have been adjusted for
heteroskedasticity and firm-specific clustering.




                                                                                                                  43
Table 7
Multinomial Probit Regression of the Decision to Issue Management Earnings Forecasts (MEFs Issued with an
actual earnings announcement)

Panel A:                                                         MEF Subsample

                                          DOA                   Opportunistic              Voluntary
Intercept                                -4.368                 -4.696                     -3.485
                                        (-16.02)    ***         (-12.85)         ***       (-13.99)    ***
ERC                                +     0.177                  0.209                  +   0.198
                                         (8.24)     ***         (8.54)           ***       (10.19)     ***
EARN_SURPRISE                      +     0.008      ‡           -0.001                 +   0.084       ‡
                                         (0.41)                 (-0.03)                    (5.34)      ***
LNMVE                                    0.058                  0.024                  +   -0.068
                                         (2.80)     ***         (1.22)                     (-4.39)     ***
MB                                       0.002                  0.022                  -   0.002
                                         (0.22)                 (2.91)           ***       (0.25)
RTNVOL                                   -63.465                22.728                 -   -12.172     ‡
                                         (-1.85)    *           (0.68)                     (-0.52)
EARNVOL                                  18.349                 6.002                  -   -0.154      ‡‡
                                         (2.87)     ***         (0.86)                     (-0.03)
HI_TECH                                  0.427                  0.385                  +   0.422       ‡
                                         (6.86)     ***         (5.40)           ***       (7.86)      ***
REGULATION                               -0.035                 -0.075                 -   -0.079      ‡‡
                                         (-0.35)                (-0.65)                    (-0.94)
HABITUAL                                 2.487                  2.350                  +   2.594       ‡‡
                                         (42.79)    ***         (34.84)          ***       (48.73)     ***
RESTATE                                  0.058            +     0.152            ‡‡        0.138
                                         (0.95)                 (2.22)           **        (2.66)      ***
BACKDATE                                 0.250            +     0.293            ‡‡        0.283
                                         (2.14)     **          (2.57)           **        (2.81)      ***
Coefficients and z-statistics on year-quarter indicators have been suppressed.
Chi-square                                                      5,215.97
p-value                                                         0.000
N                                                               42,255




                                                                                           44
Table 7 (continued)

Panel B: Wald tests for equality of coefficients:

 (1) ERC, EARN_SURPRISE:
 DOA vs. Opportunistic                   Chi-square2        = 1.69        p-value      =   0.430
 Voluntary vs. Opportunistic             Chi-square2        = 14.05       p-value      =   0.001

 (2) LNMVE, MB, RTNVOL, EARNVOL, HI_TECH, REGULATION, HABITUAL:
 Voluntary vs. DOA         Chi-square7  = 77.39 p-value  = 0.000
 Voluntary vs. Opportunistic             Chi-square7        = 60.42       p-value      =   0.000

 (3) RESTATE, BACKDATE:
 Opportunistic vs. DOA                   Chi-square2        =    2.17     p-value      =   0.338
 Opportunistic vs. Voluntary             Chi-square2        =    0.06     p-value      =   0.973

Quarters with (without) MEFs include firm-quarters that have at least one (no) MEFs issued during the period from
last quarter’s earnings announcement date up to one day before the current quarter’s earnings announcement date.
A quarter is defined as a DOA, Opportunistic or Voluntary quarter if all MEFs issued in the quarter are classified as
DOA, Opportunistic or Voluntary, respectively. See tables 1 and 2 for a description of the algorithm used to
classify MEFs into the DOA, Opportunistic and Voluntary subsamples. A firm-quarter is included if all MEFs in
the quarter are made in conjunction with an actual earnings announcement (i.e., within [-1,+1] trading days relative
to an actual earnings announcement date). If a quarter contains both bundled and non-bundled MEFs, it is
excluded. See the appendix for variable definitions.

*, **, *** indicate p-values of 10%, 5%, and 1%, respectively, two-tailed tests. ‡ denotes that the coefficient
magnitude differs in the direction predicted in Table 1. Z-statistics for the pooled regressions have been adjusted
for heteroskedasticity and firm-specific clustering.




                                                                                                           45
Table 8
Multinomial Probit Regression of the Decision to Issue Management Earnings Forecasts (MEFs Issued without
an earnings announcement)
                                                                 MEF Subsample

 Panel A:                                   DOA                  Opportunistic            Voluntary
Intercept                                   -4.703                      -4.726                 -4.008
                                           (-15.66) ***                (-16.41) ***        (-19.03) ***
ERC                                 +       -0.056                      -0.060        +        -0.043
                                            (-2.20) **                  (-2.40) **             (-2.31) **
EARN_SURPRISE                       +        0.006                       0.024        +         0.076 ‡
                                             (0.23)                      (0.98)                (4.38) ***
LNMVE                                        0.211                       0.211        +         0.205
                                           (10.25) ***                   (9.97) ***         (11.01) ***
MB                                          -0.003                       0.017        -        -0.005
                                            (-0.31)                      (2.19) **             (-0.69)
RTNVOL                                     -22.951                     -63.257        -     -60.648 ‡
                                            (-0.60)                     (-1.66) *              (-2.03) **
EARNVOL                                    -23.718                     -11.621        -     -15.442 ‡
                                            (-2.66) ***                 (-1.84) *              (-2.21) **
HI_TECH                                     -0.122                      -0.014        +        -0.081
                                            (-1.52)                     (-0.18)                (-1.20)
REGULATION                                  -0.447                      -0.504        -        -0.465 ‡
                                            (-4.05) ***                 (-4.39) ***            (-4.97) ***
HABITUAL                                     1.150                       1.153        +         1.185 ‡‡
                                             (9.54) ***                 (11.15) ***         (12.70) ***
RESTATE                                      0.034          +            0.175 ‡‡               0.128
                                             (0.51)                      (2.40) **             (2.21) **
BACKDATE                                    -0.263          +            0.181 ‡‡               0.039
                                            (-1.56)                      (1.41)                (0.32)
Coefficients and z-statistics on year-quarter indicators have been suppressed.
Chi-square                                                           1,065.25
p-value                                                               0.000
N                                                                     38,385




                                                                                          46
Table 8 (continued)

Panel B: Wald tests for equality of coefficients:

 (1) ERC, EARN_SURPRISE:
 DOA vs. Opportunistic                   Chi-Square2         =     0.34     p-value        = 0.843
 Voluntary vs. Opportunistic             Chi-square2         =     5.45     p-value        = 0.066

 (2) LNMVE, MB, RTNVOL, EARNVOL, HI_TECH, REGULATION, HABITUAL:
 Voluntary vs. DOA         Chi-square7  =    1.65 p-value   = 0.977
 Voluntary vs. Opportunistic             Chi-square7         =     11.13    p-value        = 0.133

 (3) RESTATE, BACKDATE:
 Opportunistic vs. DOA                   Chi-square2         =     5.12     p-value        = 0.077
 Opportunistic vs. Voluntary             Chi-square2         =     0.34     p-value        = 0.340

Quarters with (without) MEFs include firm-quarters that have at least one (no) MEFs issued during the period
from last quarter’s earnings announcement date up to one day before the current quarter’s earnings
announcement date. A quarter is defined as a DOA, Opportunistic or Voluntary quarter if all MEFs issued in
the quarter are classified as DOA, Opportunistic or Voluntary, respectively. See tables 1 and 2 for a description
of the algorithm used to classify MEFs into the DOA, Opportunistic and Voluntary subsamples. Firm-quarters
with MEFs made in conjunction with an actual earnings announcement (i.e., within [-1,+1] trading days relative
to an actual earnings announcement date) are excluded. If a quarter contains both bundled and non-bundled
MEFs, it is excluded. See the appendix for variable definitions.

*, **, *** indicate p-values of 10%, 5%, and 1%, respectively, two-tailed tests. ‡ denotes that the coefficient
magnitude differs in the direction predicted in Table 1. Z-statistics for the pooled regressions have been adjusted
for heteroskedasticity and firm-specific clustering.




                                                                                                           47