READY, AIM, ACQUIRE: IMPRESSION OFFSETTING AND ACQUISITIONS

Drawing on expectancy violation theory, we explore the effects of anticipatory impression management in the context of acquisitions. We introduce impression offsetting, an anticipatory impression management technique organizational leaders employ when they expect a focal event will negatively violate the expectations of external stakeholders. Accordingly, in these situations, organizational leaders will announce the focal event contemporaneously with positive, but unrelated information. We predict impression offsetting will generally occur in the context of acquisitions, but also more frequently for specific acquiring firms and acquisitions that are more likely to lead to an expectancy violation. We also posit that offsetting will effectively inhibit observers ’ perceptions of events as negative expectancy violations by positively influencing shareholder reactions to acquisition announcements. Consistent with our hypotheses, in a sample of publicly traded acquisition targets, we find evidence for impression offsetting, in which characteristics of both acquirers and their announced acquisitions predict its frequency of use. We also find evidence that impression offsetting is effica-cious; on average, it reduces the negative market reaction to acquisition announcements by over 40%, which translates into approximately $246 million in market capitalization.

When external stakeholders interact with a firm, they develop expectancies about organizational leaders' decisions. Meeting these expectations is a means by which organizational leaders maintain their position (Puffer & Weintrop, 1991) and build their reputations (Fombrun, 1996). When organizational leaders act in a manner inconsistent with how stakeholders expect the organization to behave, stakeholders may perceive such inconsistent actions as "expectancy violations." Consequently, expectancy violations trigger a range of organizational impression management activities (see Bolino, Kacmar, Turnley, & Gilstrap, 2008and Elsbach, 2006, 2012. In prior impression management literature, scholars overwhelmingly focused on reactive impression management (Bolino et al., 2008;Elsbach, 2006), which consists of activities initiated in response to an expectancy violation intended to positively influence external perceptions of the organization. Events that trigger reactive impression management are typically (a) negative and (b) unanticipated by organizations. Perhaps counter-intuitively, given its potential for effectiveness, this prior impression management research has only rarely considered anticipatory impression management (AIM Q:1 ), in which organizations anticipate a negative reaction to an event that is not yet known by stakeholders. This form of impression management occurs in anticipation of an event that may lead to an expectancy violation (Elsbach, Sutton, & Principe, 1998). In this study, we examine a previously unexplored form of anticipatory impression management in which organizational leaders seek to offset perceptions of potentially negative expectancy violations-and associated negative market reactions-by contemporaneously releasing positive, but unrelated, announcements.
Firms use AIM techniques to proactively influence firm evaluations. Prior research, however, has examined only two forms of AIM. One form, "strategic noise," occurs when organizational leaders are unclear about the market reaction an event will engender (Graffin, Carpenter, & Boivie, 2011). Strategic noise involves obfuscating the connection between the event and the market reaction by releasing positive and negative information, so observers cannot effectively evaluate the event in isolation, inhibiting perceptions of an expectancy violation. A second form, "big bath" accounting, occurs when organizational leaders announce significant bad news, usually in the form of disappointing earnings, contemporaneously with other negative news, such as write-downs, to exaggerate a negative expectancy violation (e.g., Elliott & Shaw, 1988).
In this study, we introduce "impression offsetting" as an AIM tactic. We define impression offsetting as: organizational actions initiated to positively influence external perceptions of the organization by releasing positive, but unrelated, information, in anticipation of an event becoming known that may negatively violate external stakeholders' expectations. Offsetting thus takes the form of contemporaneously releasing one or more positive, but unrelated, announcements. Since shareholders often react negatively to acquisition announcements (Haleblian, Devers, McNamara, Carpenter, & Davison, 2009), we theorize and test whether organizations release positive information to "offset" the likelihood their announcement engenders a negative expectancy violation. To further test the underlying theoretical mechanisms, we posit that more severe expectancy violations should increase the frequency with which impression offsetting is employed. Specifically, in this context, we suggest that the riskier the acquisition, in terms of the anticipated stock market's stock market Q:2 reaction to the announcement, the more likely organizational leaders will employ this tactic.
Our research has the potential to make several theoretical contributions to the impression management literature. First, we introduce the concept of impression offsetting, which is a previously unexplored AIM tactic. As part of this concept introduction, we organize previous AIM work, which has been conducted in isolation in various literatures, and we explain how impression offsetting differs from other AIM techniques. Second, using expectancy violation theory, we predict a general condition under which negative expectancy violations are likely. Accordingly, we test whether there are prescriptive expectancies, based on general norms, that influence the likelihood that observers will perceive an event-acquisitions-as a negative expectancy violation, and, thus lead to impression offsetting. We then focus our analysis to specific conditions under which observers are more likely to perceive an acquisition as a negative expectancy violation. Hence, we also test if predictive expectancies, which are based on specific actor or situation characteristics, influence the increased likelihood observers will perceive a particular acquisition as a negative expectancy violation. Thus, our pattern of findings, as guided by expectancy violation theory, helps us better understand when and how offsetting is used. Third, we find that impression offsetting influences the stock market reactions to acquisition announcements. Despite the fact that researchers have documented that organizational leaders actively engage in impression management activities, we know very little about the effectiveness of such activities on organizational outcomes (see Bolino et al., 2008 for a recent review; see also Zavyalova, Pfarrer, Reger, & Shapiro, 2012). We find that impression offsetting attenuates the negative stock market reaction to an acquisition announcement, which, we argue, suggests inhibited perceptions of a negative expectancy violation.

Reactive and Anticipatory Impression Management Surrounding Expectancy Violations
Expectancy violation theory suggests observers hold expectations regarding how an actor should behave in a given situation (e.g., Burgoon, 1993;Burgoon, Stern, & Dillman, 1995). According to this theory, the term expectancy refers to an, "enduring pattern of behavior" (Burgoon, 1993: 31) to which an actor is expected to conform. These expectancies are broadly classified as prescriptive or predictive (Burgoon & Hale, 1988;Floyd, Ramirez, & Burgoon, 1999). Prescriptive expectancies are based on general norms for typical behavior (Burgoon & Hale, 1988), whereas predictive expectancies refer to anticipated behaviors for specific actors or for specific situations (Floyd et al., 1999;Kim, 2014). When behavior is consistent with either type of expectancy, it provides observers little new information. In contrast, failure to conform to either type of expectancy results in an expectancy violation that is "distracting and redirects attention toward the actor and the violation" (Burgoon, 1993: 35). Expectancy violations are positive or negative, depending on if an actor's behavior exceeds or violates the expectation (e.g., firm exceeds or misses earnings expectations). Research also suggests that when observers perceive intentionality on the part of the actor, the magnitude of an expectancy violation, whether positive or negative, is amplified (Bachman & Guerrero, 2006). 1 In an effort to manage perceptions of-and reactions to-negative expectancy violations, firms engage in impression management (Elsbach, 2006). Elsbach (2012) notes that two important factors in organizing and understanding impression management activities relate to its timing (i.e., anticipatory or reactive) and its valence (i.e., positive, neutral, or negative). Using Elsbach's (2012) factors, we review prior research on impression management. Reactive impression management occurs when organizational leaders engage in impression management activities after a negative expectancy violation is known to stakeholders with the intent of reducing stakeholders' negative reactions to the event itself. Hence, the goal of reactive impression management is to, "provide clear, rational explanations for an organization's actions. . . and focus on visible changes the organization has instituted to prevent a similar crisis from occurring in the future" (Elsbach, 2012: 467). This tactic involves providing additional, typically self-serving, information about an expectancy violation in an effort to lessen the negativity of ongoing stakeholder reactions (e.g., Porac, Wade, & Pollock, 1999). For instance, researchers found that organizational leaders tend to blame external forces for poor performance (Bettman & Weitz, 1983;Staw, McKechnie, & Puffer, 1983) and provide self-serving accounts to justify compensatory decisions in response to the release of executive compensation information (Porac et al., 1999;Zajac & Westphal, 1995). Reactive impression management thus involves the following sequence: (1) an event becomes known; (2) stakeholders perceive a negative expectancy violation (i.e., evidenced by a strong negative reaction to the violation); and (3) organizational leaders engage in impression management tactics in an attempt to influence ongoing stakeholder reactions by providing additional information about the violation. The goal is for leaders to reduce the perceived severity of the expectancy violating event by providing a justification or explanation for their actions (Elsbach, 2006(Elsbach, , 2012. In contrast to reactive impression management, we focus on AIM techniques. Consistent with prior research (e.g., Graffin et al., 2011) we define AIM as activities that are undertaken in anticipation of, or contemporaneously with, an event that organizational leaders believe may be perceived as a negative expectancy violation. Unlike reactive impression management, which typically provides additional information about a known expectancy violation (e.g., Zavyalova et al., 2012), AIM provides information that is unrelated to the focal event. Indeed, if organizational leaders provide event-related information before observers know about the violation, the observers may retrospectively view that information as self-serving, which could negatively affect stakeholder impressions (Hovland, Janis, & Kelley, 1953). The contemporaneous aspect of AIM ensures that stakeholders must consider multiple pieces of information at the same time, which prevents an evaluation of the focal event in isolation (Graffin et al., 2011).
AIM involves the following sequence: (1) organizational leaders become aware of a potential expectancy violation before it becomes known to external stakeholders; (2) organizational leaders engage in impression management tactics either prior to or contemporaneously with the event; (3) stakeholders react to both the event and the other information the organization released. Table 1 provides a summary of reactive impression management and AIM. We first discuss two forms of AIM that have been studied in prior work, and then introduce a third form: impression offsetting.
Strategic noise involves releasing positive and negative information prior to or contemporaneously with a focal event to which the anticipated market reaction is unclear (e.g., Graffin et al., 2011). The only empirical study to examine strategic noise did so within the context of CEO successions. Graffin and colleagues suggested the goal of this tactic is to "minimize direct scrutiny of the event" (2011: 749). Executives reported their motivation for employing strategic noise was to allow more time to evaluate the newly appointed CEO on his or her own merits rather than being "unduly constrained by what the market 'said' about their potential" (Graffin et al., 2011: 751). Thus, the goal for strategic noise is to obfuscate the link between a focal event and the market reaction in order to give the event time to unfold and be evaluated on its own. Organizational leaders could thus employ this tactic in other anticipated occurrences to which the expected market reaction is uncertain such as entering a new geographic or product market.
Big bath accounting occurs when organizational leaders anticipate a potential negative expectancy violation-typically disappointing performanceby engaging in a one-time effort to stimulate perceptions of a negative expectancy violation and overstate the negativity of the event by releasing other negative announcements prior to or contemporaneously with it (e.g., Elliott & Shaw, 1988). Specifically, firms will "under-report earnings by the maximum amount possible, preferring to take a big bath in the current period in order to report higher earnings in the future" (Kirschenheiter & Melumad, 2002: 762). All bad news is released at once because "when circumstances are bad, making things just a little worse by cleaning out the rubbish does little harm to either reputation or prospects" (Walsh, Craig, & Clarke, 1991: 174). This approach allows managers to reserve funds for future periods to demonstrate a comparative increase in profit (Walsh et al., 1991), signal to shareholders that they have aggressively dealt with past problems (Strong & Meyer, 1987: 644), and to lower the future performance benchmark (Elliott & Shaw, 1988 Q:3 ). Other examples of negative expectancy violations that may trigger the use of big bath accounting include bankruptcy or earnings misstatements.
Finally, we introduce a previously unexplored form of AIM-impression offsetting. We suggest that organizational leaders will engage in impression offsetting when they anticipate an upcoming event may lead to a negative expectancy violation. In an effort to reduce the perceived size of this potential negative expectancy violation, organizational leaders may release positive, but unrelated, information prior to or contemporaneously with the event becoming known to external stakeholders (see Appendix A for the list of offsetting announcements in our sample). Impression offsetting is thus an attempt to manage impressions positively by offsetting an anticipated negative reaction to an event, which reduces the likelihood that observers will perceive it as a negative expectancy violation. Expectancy violation theory suggests violations lead to increased scrutiny for the violator and cause observers to seek additional information regarding the violation (Burgoon, 1993). Thus, the primary motivation for offsetting is to inhibit perceptions of a negative expectancy violation by reducing the perceived size of a negative reaction. Although releasing this unrelated positive information may not directly change how investors react to the acquisition announcement, it does force market participants to react to multiple pieces of news. Thus, by compelling investors to react both to a potential negative expectancy violation as well as to the offsetting announcement(s), the net effect is to reduce the overall negativity of the market reaction to the acquiring firm by requiring market participants to react to multiple pieces of news simultaneously. By doing so, organizational leaders may also diminish the additional scrutiny and undesirable attention that typically follows negative expectancy violations.
Releasing positive information that is unrelated helps avoid the potential pitfalls associated with providing further information to the focal event as prior impression management research suggests audiences may view attempts to justify a focal event with skepticism (Tedeschi & Reiss, 1981). A key aspect of a justification is an argument that the event's perceived negative consequences are actually positive. A danger in such a justification is that audiences perceive organizational leaders as "protesting too much" and thus view the organization more negatively than they would had the event not been justified (see Ashforth & Gibbs, 1990). More generally, if audiences question the truthfulness of the justification, they may interpret it as deceptive and perceive the organization as dishonest (e.g., Schneider, 1981). As such, releasing unrelated positive information may allow organizational leaders to offset a potentially negative shareholder reaction without engendering the skepticism related to justifications. Table 2 provides a summary comparison of these three AIM tactics.
As we noted earlier, observer expectancies can be broadly classified as prescriptive (i.e., general expectancies), or as predictive (i.e., specific to an actor or situation) (Burgoon & Hale, 1988). In the context of firms, investors hold the prescriptive expectancy that organizational leaders should act in a manner that increases profits (Friedman, 1970) and enhances shareholder value (e.g., Fligstein, 1990). When organizational leaders behave in a manner that violates this prescriptive expectancy, a negative expectancy violation occurs. For instance, Zavyalova and colleagues (2012) note that a product recall is a form of a negative organizational expectancy violation because it reduces the positive media coverage firms receive, triggering firms to actively engage in reactive impression management. Further, if the negative expectancy violation is viewed as intentional by observers, the magnitude of the perceived negative violation will be amplified (Bachman & Guerrero, 2006). For example, if a firm fails to meets its earnings expectations, but the reason(s) for its disappointing performance is unclear or ambiguous, a minor negative expectancy violation may occur. If a firm intentionally engages in a behavior that observers associate with impairing firm performance, however, the observers will perceive a more significant negative expectancy violation.
We expect organizational leaders to engage in impression offsetting in an effort to reduce the likelihood observers perceive an event as a negative expectancy violation. We test our theory in the context of acquisition announcements because research suggests the market typically reacts negatively to acquisition announcements (Haleblian et al., 2009) and this reaction is more consistently negative for larger, publicly-held, targets (Moeller, Schlingemann, & Stulz, 2004). Thus, when firms undertake large public acquisitions, shareholders will likely react negatively to such announcements because they trigger a prescriptive negative expectancy violation. Due to widely-reported figures suggesting the failure rate of acquisitions is between 70 and 90% (e.g., Christensen, Atlton, Rising, & Waldeck, 2011 Q:4 ), as well as academic research suggesting acquisitions generally destroy shareholder value (e.g., Haleblian et al., 2009), managers likely anticipate a negative shareholder reaction. Further, due to the analyses of valuation and strategic fit, as well as the required due diligence that precedes large public acquisitions, shareholders will almost certainly interpret acquisitions as intentional, which likely amplifies the prescriptive negative expectancy violation (Bachman & Guerrero, 2006). In sum, this research motivates organizational leaders to inhibit the likelihood shareholders will perceive an acquisition announcement as a negative expectancy violation and their advanced knowledge regarding the announcement of an acquisition gives them the opportunity to engage in impression offsetting.
At the same time, however, reactive impression management techniques will not likely be effective in this context. Specifically, organizational leaders will likely not be able to positively sway the market's reaction to an acquisition announcement by providing a post-hoc justification that attempts to cast the acquisition in a more favorable light. Indeed, as we noted earlier, when observers impute selfinterest, after the fact justifications do not produce positive results (Schneider, 1981). In our context, organizational leaders may not be able to justify an acquisition by reactively providing additional relevant information, because most of the information is already publicly available.
AIM techniques may be more effective, however. Decision makers could positively bias stakeholders' overall impression of the organization by announcing positive unrelated news such as an increase in dividends or by announcing a new product contemporaneously with the acquisition announcement. This positive bias may, in turn, inhibit perceptions of a prescriptive negative expectancy violation and reduce the amount of additional scrutiny and attention that typically follows expectancy violations by reducing the magnitude of a perceived violation. We thus suggest: Hypothesis 1. Organizational leaders will release unrelated positive information around acquisition announcements more often than is predicted by organizations' baseline count of positive announcements.

Suppressing the Negative Expectancy Violation
Inherent in the logic associated with the arguments above is that the use of impression offsetting will positively influence the market reaction to acquisition announcements. We contend that unrelated positive information released through impression offsetting generates additional information in the media. This additional information forces investors to assess and likely "average" multiple pieces of information simultaneously. As such, the unrelated positive information reduces the magnitude of the negative reaction to the expectancy violation, in the form of an acquisition announcement. More broadly, despite numerous studies examining the use of organizational impression management, there is very little evidence regarding its effectiveness (Bolino et al., 2008). Our context, however, allows us to examine a quantifiable measure of the effectiveness of impression offsetting: the abnormal market return for the acquiring organization following the announcement of an acquisition. Indeed, as we noted earlier, when actors attempt to manage stakeholder impressions by releasing information that is directly relevant to a focal event, such announcements are typically met with skepticism (e.g., Tedeschi & Reiss, 1981). This skepticism results from observers knowing the organization has a vested interest in positively "spinning" news to make observers perceive a focal event more favorably. By releasing unrelated positive information, however, such skepticism may be reduced. Also, the newly released impression offsetting information means the market will have new, positive information to process. By doing so, the net combined reaction to the impression offsetting and the acquisition announcement reduces the likelihood of a strong market repudiation. The reduced magnitude of the reaction, in turn, reduces the likelihood that investors will perceive the acquisition as a negative expectancy violation. We thus expect impression offsetting will positively influence the market reaction to an acquisition announcement by forcing the market to react to the positive offsetting announcement(s) as well as the acquisition announcement and hypothesize: Hypothesis 2. Releasing unrelated positive information will be positively associated with abnormal returns around acquisition announcements.

Impression Offsetting and Predictive Expectancies
Next, we test conditions under which organizational leaders are more or less likely to engage in impression offsetting. We do so to more directly test the theoretical mechanisms that drive the use of this previously unexplored impression management tactic. Since we propose that the anticipation of an expectancy violation drives the use of this tactic, it logically follows that the greater the expected violation, the more likely organizational leaders will engage in impression offsetting.
Such contentions are consistent with expectancy violation theory which proposes that, in addition to prescriptive expectancies, which are based on general normative behavior across firms, observers also hold predictive expectancies that an actor's prior behavior (Burgoon & Hale, 1988) or specific context (Floyd et al., 1999) can influence. In the following four hypotheses, we examine how such predictive expectancies, which influence the degree to which a given occurrence will lead to an expectancy violation, affect the use of impression offsetting. Within our context, although markets often view acquisitions negatively, there is significant variation in perceived expectancy violations depending on characteristics of the firm and on context associated with the deal (Haleblian et al., 2009). Accordingly, acquisitions allow for the examination of conditions under which organizations are more or less likely to engage in impression offsetting as a given acquisition may trigger stronger or weaker negative expectancy violations. We restrict our analyses to acquirer and target characteristics previously examined within the acquisitions literature. Specifically, we examine acquirer and target related variables that increase the risk associated with an acquisition, which make a perceived expectancy violation more likely. Along these lines, we chose the variables of acquirer prior acquisition activity (e.g., Haleblian Morck, Shleifer, & Vishny, 1990;Seth, 1990) and target pre-merger profitability (Capron & Shen, 2007). As we develop below, as each of these acquirer and target related variables increase acquisition risk, they also increase the likelihood of impression offsetting.
Acquirer prior acquisition activity. As a firm's prior actions inform its predictive expectancies (e.g., Burgoon, 1993), we suggest one particularly salient source of such expectancies: firm acquisition history. This contention is consistent with expectancy violation research that suggests as norm deviating behavior accumulates through behavioral repetition, a larger negative expectancy violation reaction is generated as compared to a single behavior (e.g., Bachman & Guerrero, 2006). Repeated violations of investor expectations are thought to reduce stakeholder trust (Fehr & Gachter, 2000) in which stakeholder skepticism of subsequent violations increases with behavioral repetition (Pfarrer, DeCelles, Smith, & Taylor, 2008;Simpson, 2002). Together, these ideas suggest that if a given actor repeatedly violates a given expectation, observers will perceive each subsequent violation more severely. Consistent with this tenet of expectancy violation theory, prior researchers found that as the level of a firm's acquisition activity increases, market participants perceive the firm as carrying excessive growth related risk, and as such the market's response to their subsequent acquisition activity becomes more negative (Laamanen & Keil, 2008).
Since markets see the acquisitions of serial acquirers more negatively than less frequent acquirers, their announced acquisitions should be more likely to stimulate negative expectancy violations as compared to other acquirers. Moreover, from the perspective of the serial acquirer, as it increases its acquisition activity, and thus has a history of repeatedly violating the expectation of engaging in activities that promote organizational performance, it may learn that markets typically perceive acquisitions negatively (Haleblian & Finkelstein, 1999 Q:9 ), especially those of experienced acquirers (Laamanen & Keil, 2008). Hence, based on awareness of the negative perception of markets that derives from direct experiential learning, serial acquirers will be more prone to initiate impression offsetting than less experienced acquirers in order to counteract negative perceptions associated with acquisition activity. Thus, relative to less experienced acquirers, serial acquirers generally have more experience with negative acquisition outcomes, which results in an added motivation to positively influence stakeholders to defend, and potentially build, its identity as a competent serial acquirer. Combined, these ideas suggest: Hypothesis 3. A firm's recent acquisition activity will be positively associated with releasing unrelated positive information around acquisition announcements.
Acquirer reputation. We also argue that investors hold predictive expectancies for high reputation firms. While some research suggests high-reputation actors are held to higher standards and are more strongly repudiated for missteps than low-reputation firms (e.g., Milbourn, 2003;Wade, Porac, Pollock, & Graffin, 2006), other studies suggest reputation protects firms and reduces the negative outcomes experienced by such firms (e.g., Pfarrer, Pollock, & Rindova, 2010). We expect, however, that organizational leaders at high reputation firms may be more concerned about the potential downside associated with a negative expectancy violation and are thus more likely to engage in impression offsetting for two reasons.
First, because firm reputation is based upon the ability to consistently deliver quality (Lange, Lee, & Dai, 2011), shareholders will likely hold the predictive expectancy that high-reputation firms will continue to perform at a high level. Engaging in acquisitions, which are fraught with risk and often destroy shareholder value, is inconsistent with this predictive expectancy. Indeed, as an actor's prior actions influence predictive expectancies (e.g., Floyd et al., 1999), shareholders expect higher returns from high reputation actors (Fombrun, 1996). Along these lines, prior research has shown that high-reputation firms engage in riskier investments than other firms in order to meet the high expectations they faced (Petkova, Wadhwa, Yao, & Jain, 2014 Q:10 ). Therefore, shareholders will more likely perceive risky acquisition actions from these high reputation actors as expectancy violations than they would from firms held to lower expectations.
Second, based on the quality of their abilities and their outputs, high-reputation organizations benefit from substantial accumulated levels of public recognition (see Barnett & Pollock, 2012 for a review of this literature). Given these benefits, high-reputation organizations may have an enhanced motivation to manage impressions of their acquisitions to protect their reputations by avoiding the perception of a negative expectancy violation. Accordingly, Burgoon and Hale (1988) argued that positively evaluated actors have the potential to commit a more grievous violation because the gap between actual and expected behavior is likely greater for such actors. Consistent with this idea, Rhee and Haunschild (2006) found that high-reputation automakers experienced larger losses in market share from negative impressions than low-reputation organizations. Accordingly, high-reputation organizations may be more likely to attempt to mitigate an anticipated negative market reaction-such as impression offsetting-given the potential for stakeholders to demonstrate strong reactions. Thus, we hypothesize: Hypothesis 4. High-reputation organizations will be more likely than other organizations to release unrelated positive information around acquisition announcements.
Target relatedness. Specific situation characteristics also inform predictive expectancies (Burgoon, 1993). Based on this tenet of expectancy violation theory, we argue that managers will perceive a higher likelihood that investors will view an acquisition as a negative expectancy violation as the risk associated with the target increases. Prior research suggests, within the context of acquisitions, there is significant variation in the likelihood of a positive outcome (e.g., Pablo, 1994). As acquisitions increase in target risk, managers expect investors to recognize this difficulty and react negatively to the announcement. Although all acquisitions carry significant risk (Eisenmann, 2002), unrelated acquisitions are generally considered riskier (Morck et al., 1990) and are thus more likely to lead to a negative expectancy violation. Moreover, acquirers that have knowledge of their own industry or industries may be less able to assimilate and use knowledge from a target organization in a different, unrelated industry (Cohen & Levinthal, 1990;Vermeulen & Barkema, 2001). Consistent with this notion, prior research shows lessrelated acquisitions tend to produce more negative market reactions than more-related acquisitions (Morck et al., 1990;Seth, 1990).
As such undertaking an unrelated acquisition represents a stronger potential predictive expectancy violation than the case of a related acquisition. Expectancy violation theory recognizes the valence of the violation is an important determinant in which the stronger the negative behavior, the more likely observers will perceive it as a negative expectancy violation (Burgoon & Hale, 1998 Q:11 ). Moreover, shareholders will likely associate undertaking such a risky acquisition as intentional by organizational leaders in which intentional violations are typically associated with increased evaluations of responsibility and negative evaluations (e.g., Bachman & Guerrero, 2006: 946;Fincham & Bradbury, 1987 Q:12 ). Thus, as the likelihood for a predictive negative expectancy violation increases for an acquisition, organizational leaders will become more motivated to act preemptively in order to influence the market's reaction through the use of impression offsetting. Along these lines we hypothesize: Hypothesis 5. Target unrelatedness will be positively associated with releasing unrelated positive information around acquisition announcements. Target pre-merger profitability. In a similar vein, as with unrelated targets, unprofitable targets also represent a high degree of acquisition risk, because it is difficult to turn around the performance of a firm experiencing poor performance (Capron & Shen, 2007). More specifically, if a target organization is currently profitable, an acquirer may be effectively acquiring a revenue stream and some opportunity to increase efficiency by consolidating support functions such as marketing and accounting. A target's profitability may also signal that it has valuable resources that the acquirer can use (see Capron & Shen, 2007). In this instance, it might be relatively easy for organizational leaders to highlight the ongoing profitability of the organization as part of the justification for entering into a relationship with a target firm. In contrast, acquiring an unprofitable target combines the challenge of integrating the two companies as well as transforming an acquired unprofitable organization into a profitable component of the acquiring firm. Accordingly, organizations may anticipate acquiring an unprofitable target, which inherently involves greater risk than acquiring a profitable one. As such, it is more likely to lead to a negative expectancy violation than a profitable target. Indeed, a lack of profitability suggests that the burden for generating potential synergies with the acquiring organization increases, in that leaving an unprofitable target to operate independently is likely to be value-destroying for the acquiring firm. In addition, the challenge involved in creating a relationship with an unprofitable acquired organization may lead markets to react more negatively to targets with higher risk profiles. These factors also lead to an enhanced likelihood of a perceived negative expectancy violation; hence we expect acquirers will be more motivated to use impression offsetting when they acquire an unprofitable organization.
Hypothesis 6. Unprofitable targets will be more likely than profitable ones to be associated with releasing unrelated positive information around acquisition announcements.

Sample and Data Collection
Our sample consists of acquisition announcements for deals in which both the acquirer and target were public companies. In addition, these deals met the following criteria and were: (a) greater than $100 million in value; (b) announced between 1995 and 2009; and (c) subsequently completed. Our sample consists of 770 acquisition announcements fitting our criteria. Missing data reduces our sample to 758 for models predicting impression offsetting and 701 for models predicting cumulative abnormal returns.
We collected data from Thomson Reuters SDC Platinum for each acquisition announcement including the date, acquirer, target, acquisition price, form of consideration offered, acquirer and target industries, and premium paid. Using Compustat, we also collected data on the acquirers' organization size, profitability, and level of diversification. From Risk Metrics, we collected data on the acquirers' board characteristics. We employed Thomson Reuters I/B/E/S to capture whether earnings announcements indicate negative, zero, or positive earnings and whether those earnings miss, meet, or beat consensus estimates. We also collected and analyzed press releases to examine impression offsetting behaviors and baseline counts of positive announcements by acquirers in our sample. We report variable means, standard deviations, and correlations in Table 3.
Press release collection. We collected press releases in two waves. The first wave covered the threeday event window, covered the period from one day before to one day after the event for each acquisition, and the second wave, which we collected for our baseline positive announcements measure, covered the period from four months prior to the event window through one month prior to the event window.
Event window press releases. For our event window collection, we collected and analyzed 4,969 press releases to isolate and code those from the acquirers in our sample. Our primary sources of data were PR Newswire and Business Wire, the two dominant press release distributors. In the event window collection, we accessed PR Newswire and Business Wire using both Factiva and LexisNexis. After substantial investigation, we found that Business Wire was substantially identical between Factiva and LexisNexis, but PR Newswire in LexisNexis covered more organizations. Accordingly, we gathered data for 666 acquisitions from Factiva databases and an additional 77 acquisitions from LexisNexis databases (i.e., a total 743 acquisitions out of 770 or 96.5%). As an additional robustness check, we later compared our Factiva data (supplemented with LexisNexis data for 77 acquisitions) to LexisNexis data for the full sample covering the same three-day window and found them to be substantially similar across the entire sample (r 5 .98; p , .001). We also used organization websites when we could not find press releases in PR Newswire or Business Wire. We gathered data for 24 acquisitions directly from organization press websites and an additional three acquisitions using Google searches to find press release pages on organization websites that were not listed on their press websites.
Based on the content of the collected press releases, we developed content categories. Though most press releases cover one news announcement, we also include those that have multiple types of news in one press release (e.g., an earnings release that also includes an executive retirement and a revision of previously issued guidance for a future period). We provide a list of all categories and the number of occurrences of each category in Appendix A.
Baseline press releases. For our baseline collection, we gathered over 300,000 press releases and, we used a subset of 53,200 press releases that fit a threemonth date range for our Baseline positive announcements control that we describe below. Like our event window collection, our primary sources of data are PR Newswire and Business Wire. Consistent with our event window collection, we accessed PR Newswire and Business Wire using LexisNexis, having learned in our event window collection that the PR Newswire database in LexisNexis covers more organizations. We gathered baseline data for 765 organizations from LexisNexis databases (i.e., 765 acquisitions out of 770 or 99.4%). Like our event window collection, we also used organization press websites to gather data for the remaining five acquisitions.
We used computer-assisted technology to facilitate processing the massive number of press releases in the baseline collection. In particular, we developed custom software, written in the Python 3.3 programming language, to identify and isolate press releases released by focal organizations in our sample. Our software uses custom per-organization patterns that we developed by manually examining organization press releases, coding custom rules, and iteratively checking them against our actual baseline data. This process eliminated press releases that organizations did not directly issue in the  sample and reduced our sample to 15,028 press releases for the baseline positive announcements measure we describe below and two robustness measures. 2 Once processed, we exported the data in spreadsheet form for human coding of our measures described below.

Dependent Variables
Impression offsetting. We measure impression offsetting as the count of positive, material press releases issued by the acquiring firm within the event window (i.e., day 21 to day 11). Employing this three-day window is consistent with prior research on AIM (e.g., Graffin et al., 2011) and recognizes that, even if organizational leaders issue a press release one day after a focal event, they prepared that release in advance of the event. Our results and conclusions are substantively unchanged when we restrict the window to day 21 and day 0.
For clearly positive events (e.g., new product releases, social good efforts like donations) we coded press releases as impression offsetting. For event types that may or may not be positive, we looked for additional information. Specifically, for earnings releases, we compared announced earnings to consensus expectations to categorize them as positive, and thus as impression offsetting, or neutral or negative, which are not coded as impression offsetting. For other events like divestitures, we coded press releases as impression offsetting if the organization made an unambiguously positive characterization of the event in the text of the press release. In total, we categorized 415 events as positive, 49 as neutral, and 56 as negative. We provide a full list of press release categories in Appendix A. The most frequent categories were new product announcements (197), customer wins (79), earnings releases (51), social good events (41), and other acquisition announcements (34).
We used a primary coder to code all 4,969 press releases into one or more of the categories. In addition, we assessed the reliability of our primary coder by using a secondary coder for 100 randomly chosen press releases excluding those that we coded as either being from other parties or announcing the focal acquisition. To assess inter-rater reliability we employed Cohen's k (Cohen, 1968). For the sample of 100 press releases the k was 0.99 and values above 0.80 are considered excellent (Fleiss, 1981;Landis & Koch, 1977). Given this high level of inter-rater agreement, we coded the remaining press releases with one coder.
Cumulative abnormal returns. Abnormal returns represent the part of the return on a security that is unanticipated by an economic model of expected returns for the same security. Cumulative abnormal returns (CAR) are the sum of daily abnormal returns for a security over a period that captures the influence of the event during that period. Specifically, we assessed returns of the security in our sample against the return of the value-weighted market portfolio using the following formula: where R it 5 the return on stock i for day t, R mt 5 the return on the value-weighted market portfolio for day t; a i 5 a constant, b i 5 b of stock i, and T 1 and T 2 are the lower and upper limits of the event window, respectively.
We calculate the estimates of a and b during a 250day window that falls between 295 and 45 days before the focal acquisition. In line with McWilliams and Siegel's (1997) guidance on short event windows, we report a three-day event window centered on the acquisition (i.e., day 21 to day 11). Our CAR variable is the average of the abnormal returns for the event window. Our results are substantively unchanged to the alternative event windows using the date of the acquisition (i.e., day 0 only) and a two-day window covering the day of the announcement and the next day (i.e., day 0 to day 11).

Independent Variables
Prior acquisition activity. Consistent with prior studies (e.g., Laamanen & Keil, 2008;Vermeulen & Barkema, 2002), we measure prior acquisition activity as the count of acquisitions fitting our sample 2 Although, we gathered press releases for an entire year prior to the acquisition (-13 months through 21 month), our initial collection including this time window resulted in over 300,000 press releases. Since this was a massive amount of data to process and manually code, we investigated whether it would be possible to scope the data analysis to a more reasonable size. First, we manually coded our three-month measure. Then, we coded a 12-month measure for a 10% random subsample as a robustness check. Our analyses that showed the number of positive, material press releases issued by firms in the 12-month window was highly correlated with the number issued within a three-month window (r 5 0.87). Given the likely high level of similarity between these time windows as well as the highly laborintensive nature of the coding, we opted to use our threemonth measure. specification the acquirer made in the three years immediately prior to the announcement date (i.e., day -[365 * 3] to day 21).
High reputation. Following prior research, we code acquirer high reputation using annual lists of Fortune's Most Admired and the Wall Street Journal and Harris Interactive's corporate reputation rankings (Pfarrer et al., 2010). We code this variable as 1 when the acquiring organization was within the top 25 of either list in a particular year and as 0 otherwise. We lagged this variable in all analyses. As a robustness check, we also created a variable that counts the number of times an organization appeared on the list in the prior five years. Our results and conclusions are substantively unchanged when we employ this alternate variable.
Unprofitable target. We code our unprofitable target variable as 1 if the target organization in an acquisition has a zero or negative net income in their annual report in the year prior to acquisition announcement and as 0 otherwise. To create this variable we flagged all acquisitions where the offer price-to-EPS Q:13 ratio was missing in the Thomson Reuters SDC Platinum database we used to collect acquisition information. Once we manually verified that this missing data indicated negative earnings, we coded it as unprofitable.
Target unrelatedness. We code our unrelated variable as 1 for acquisitions in which the acquirer and target have different two-digit SIC Q:14 codes. Our results and conclusions are substantively unchanged using three-and four-digit SIC codes.
Impression offsetting. We use our impression offsetting variable described above as an independent variable in our test of Hypothesis 2.

Control Variables
Baseline positive announcements. We measure baseline positive announcements as the three-day average count of positive material announcements by the focal firm for a three-month period prior to the acquisition announcement (i.e., day 2121 to day 230). We coded using a process identical to the one we used for impression offsetting. In our sample, the overall average three-day count is 0.0897 for a threeday event window ([1.883 3 3]/63, where 1.883 is the overall average count for entire baseline periods in the sample, 3 is the number of days in the event window, and 63 is the number of business days in a three-month period).
As a robustness check, we also coded a 10% random subsample for similar six-month (i.e., day 2213 to day 230) and 12-month (i.e., day 2395 to day 230) measurement periods. The three-month measure is highly correlated with both the six-month measure (.91) and the 12-month measure (.87), suggesting strong convergent validity between the three-month count and other longer interval counts, which suggests the use of the three-month measure as an accurate baseline measure. In total, we coded 15,028 press releases, which included the three-month measure and the two robustness measures discussed above.
Governance controls. We include four acquirer corporate governance controls in our analyses. We control for CEO pay using Compustat's total CEO compensation (TDC1) variable-which we log transformed-and for CEO tenure as research suggests that each may generally influence a CEO's strategic decisions (Mueller, 1987), as well as the specific type of acquisitions a CEO pursues (e.g., Sanders, 2001). We also capture outside director percentage and total directors to control for the relative strength of monitoring by the board of directors as such governance characteristics have been associated with firm acquisition activity (Hoskisson & Hitt, 1990). We lag each of these variables in all of our analyses.
Acquirer controls. We also include seven additional acquirer controls in our analyses. Acquiring firm performance may be associated with acquisition success (Morck et al., 1990); hence, we control for lagged ROA to capture performance in the year prior to the acquisition announcement. Prior work on acquisitions has shown free cash flow influences acquisition likelihood (e.g., Lang, Stulz, & Walkling, 1991), so we control for the percentage of free cash flow, which we measure as (operating incometaxesinterest expensedepreciationcommon and preferred stock dividends)/common equity. In addition, it may be that stock price influences acquisition likelihood (Harford, 2005); hence, we control for stock returns, which we measure as an organization's annual stock returns, assuming reinvestment of dividends. There is also evidence that firm size influences acquisition behavior (e.g., Moeller et al., 2004). We control for acquirer size by using the log of the acquiring organization's sales. We transformed this variable due to its skewed distribution, and we lagged it in all analyses. Because acquisitions in more concentrated industries may be perceived differently by markets (see Hou & Robinson, 2006 Q:15 ), we control for acquirer industry concentration using a Herfindahl index of the concentration (of sales) of the acquirer's three-digit SIC industry code using data from Compustat, following Hou and Robinson (2006) Q:16 . Since environmental factors have also been shown to influence acquisition activity (Harford, 2005), we control for environmental dynamism to control for environmental influences on strategic decision making by examining the degree of sales instability in the markets in which the acquirer is a participant. Consistent with prior studies (e.g., Dess & Beard, 1984;Tosi, Aldag, & Story, 1973), we measure it as dispersion of sales over the prior five years and lagged it one year in all analyses. Also, in our test of Hypothesis 2, we use our high reputation and prior acquisition activity variables as controls.
Acquisition controls. We include four acquisition level controls in our analyses. Percentage stock used in acquisition deals have been shown to influence market reactions (e.g., Travlos, 1987); as such we control for percent stock which is the percentage of the total cost of the acquisition that the acquirer pays using its own stock. In addition, the attitude of an acquisition transaction has been asserted to influence acquisition performance, so we control for the attitude of the acquisition classified as whether the acquirer's bid is friendly or hostile (Browne & Rosengren, 1987). Also, in our test of Hypothesis 2, we use our unrelatedness and unprofitable target variables as controls.
Announcement controls. We include three announcement controls in our analyses. We measure Friday announcement as one when an acquisition is announced on a Friday and zero otherwise. We control for Friday announcement to help rule out the alternative explanation that managers will tend to announce an acquisition on a Friday when they expect a negative reaction (i.e., "taking out the trash"), which may act as a substitute impression management tactic (e.g., Dellavigna & Pollet, 2009). We also control for neutral announcements and negative announcements, each measured as a count variable and coded alongsidebut not included in the count of-our impression offsetting variable described above. We control for these announcements to rule out the alternative explanation that other AIM techniques are being used in the acquisition context and influence our results.
Year dummies. We include year dummy variables in all of our regression analyses to control for timevarying macroeconomic effects. For aesthetic reasons, we omit these year dummy variables from our tables of regression results.

Analysis
We test Hypothesis 1 by comparing our observed count of positive announcements around acquisitions with baseline three-day average counts (a) from our three-month baseline measure using a paired t-test and (b) from a prior study using a two-sample z-test for proportions. We test Hypothesis 2 using ordinary least squares (OLS) regression with robust standard errors. We test Hypotheses 3 through 6 using negative binomial regression with robust standard errors because these models employ count dependent variables with over-dispersion.

RESULTS
In Hypothesis 1, we predicted that impression offsetting around acquisitions will occur more often than is predicted by organizations' baseline count of positive announcements. We compare our observed count of positive event announcements to a baseline three-day average count of announcements from our own baseline measure and from a prior study. For the three-month period beginning four months prior and ending one month prior to the focal acquisition event, we constructed a baseline count of positive event announcements. We observed that in this baseline period, firms averaged 0.0897 positive announcements every three days. In contrast to the baseline period, during the impression offsetting window, which is also three days, we observed that 205 out of 770 acquisitions, or 26.6% of firms, in our sample have at least one positive event announcement within the 21 day to 11 day event window. In addition, we found an overall average count of 0.5221 positive announcements and our observed count of positive press releases around acquisitions is statistically significantly more positive than our baseline positive press release count (t 5 10.0619; p , 0.001) and suggests that firms release 482% more positive unrelated news surrounding acquisitions than suggested by our baseline. Finally, we also compared the number of negative press releases in the event window to a baseline negative press release three-day average count using a paired t-test. We found that negative announcements during the event window (mean 5 .0377) actually decrease relative to the baseline three-day average count (mean 5 .0627) of negative announcements (t 5 3.1198; p 5 0.000). This result suggests organizations release negative press releases contemporaneously with acquisitions in an amount 39.94% lower than suggested by our negative announcements baseline measure. Thus, offsetting consists of both increased positive announcements and decreased negative announcements relative to baseline levels. Combined, these results suggest that impression offsetting differs from strategic noise, which involves the release of both positive and negative announcements and provides support for Hypothesis 1.
We predicted in Hypothesis 2 that impression offsetting will be positively associated with abnormal returns around acquisition announcements. The models that test Hypothesis 2 are presented in Table 4. Model 1 in Table 4 presents the control model, while Model 2 adds our measure of impression offsetting. The coefficient for impression offsetting in Model 2 is positive and significant (p , 0.05), supporting Hypothesis 2. For instance, when an organization engages in impression offsetting at the mean level (two positive announcements for those organizations using any level of impression offsetting) with an acquisition announcement, the mean market reaction is reduced from 21.4% to 20.8%. At the mean of our sample this represents a positive change of $246 million in market capitalization and that impression offsetting reduces the size of the negative market reaction by 44%. As a robustness check, we also examined Hypothesis 2 using a two-stage least squares (2SLS) model, and our results and conclusions were substantively unchanged when we used this alternative model. In our 2SLS robustness model, we used shares outstanding and average top management team total compensation as instruments for our impression offsetting variable. 3 Following Semadeni, Withers, and Certo (2014), we conducted two post-estimation tests regarding the validity of our instruments, both of which supported their validity. First, we tested whether our instruments are weak using an F test (with Stata's "estat firststage" post-estimation command). The test was highly significant (F 5 14.96; p , 0.001), rejecting the null hypothesis of weak instruments. Second, we tested the over-identifying restrictions of the twostage model using Wooldridge's x 2 test (with Stata's "estat overid" post-estimation command). The test was not significant (x 2 5 1.21; p 5 0.271), failing to reject the null hypothesis that the instruments are uncorrelated with the structural error term and the model is properly specified, and thereby supporting the validity of our two instruments. Table 5 presents the models that test Hypotheses 3 through 6. In Hypothesis 3, we predicted that a firm's recent acquisition activity will be positively associated with impression offsetting. Model 1 presents the control model, while Model 2 includes all independent variables predicting strategic offsetting. The coefficient for prior acquisition activity in Model 2 is positive and statistically significant (p , 0.01), supporting Hypothesis 3. Accordingly, when acquisitions in the prior three years increase from the mean in our sample (roughly one acquisition) to one standard deviation above the mean (roughly three acquisitions), the number of offsetting announcements increases by approximately 0.4 announcements.
In Hypothesis 4, we predicted that high-reputation organizations will be more likely to engage in impression offsetting around acquisition announcements. The coefficient for high reputation in Model 2 is positive and statistically significant (p , 0.01), supporting Hypothesis 4. This finding suggests the number of offsetting announcements increases by roughly one announcement for high reputation organizations relative to non-high reputation firms. We predicted in Hypothesis 5 that acquisition target unrelatedness will positively influence the likelihood of the acquiring organization engaging in impression offsetting around acquisition announcements. The coefficient for unrelatedness in Model 2 is positive but not statistically significant, failing to support Hypothesis 5. In Hypothesis 6, we predicted that, when acquisition targets are not profitable, organizations will be more likely to engage in impression offsetting around acquisition announcements. The coefficient for unprofitable target in Model 2 is positive and statistically significant (p , 0.05), supporting Hypothesis 6. This result suggests that when an acquisition target is unprofitable, the number of a firm's positive unrelated news announcements increases by approximately 0.3 releases. We also conducted a series of robustness checks that specified impression offsetting as binary (tested with a legit model) or rightcensored at 2, 3, 4, and 5. Across these models, we continued to see a broad pattern of support for Hypotheses 3 through 6. 4 We also conducted robustness checks in which we controlled for the premium paid by the acquirer (as a proportion of the target's market value), and our robustness results are substantially similar to our main models.
Our results suggest consistent support for our argument that organizational leaders release positive information content to offset potential negative expectancy violations like acquisitions. In addition, our results suggest the use of impression offsetting increases as characteristics of the acquirer or the acquisition increase the likelihood of an anticipated negative expectancy violation. Finally, impression offsetting appears to be an effective AIM tactic because it positively influences the market reaction to acquisition announcements.

DISCUSSION
We introduced a new form of AIM-impression offsetting: an AIM tactic organizational leaders employ when they expect a focal event will negatively violate the expectations of external stakeholders. Our results suggest when firm executives anticipate that observers may view an upcoming announcement as a deviation from expected behavior (i.e., a negative expectancy violation), they are prone to employing impression offsetting in which they contemporaneously issue positive, but unrelated, press releases with an acquisition announcement.
Ours is the first study to quantify the benefits of any impression management technique. We found evidence that impression offsetting limits negative abnormal market returns to acquisition announcements. Specifically, we found that impression offsetting inhibited perceptions of a negative expectancy violation such that when firms issued one positive release, the negative market reaction associated with acquisition announcements decreased by roughly 44%. In addition, we theorized and found that organizational leaders are more or less likely to engage in impression offsetting depending upon the anticipated severity of an expectancy violation. These findings speak directly to the theoretical mechanisms underlying the use of this previously unexplored impression management tactic. Since it is the anticipation of a potential expectancy violation that drives the use of this tactic, it logically follows that the more severe the violation, the more likely organizational leaders will engage in impression offsetting. In our context, we found that the higher the riskiness of the acquisition, the more likely organizational leaders were to engage in impression offsetting. Specifically we found that serial acquirers make greater use of impression offsetting because markets often respond negatively to acquisition programs (e.g., Laamanen & Keil, 2008). Hence, employing impression offsetting to a current acquisition may positively influence shareholders' perceptions of the firm's ability to execute subsequent acquisitions, which may improve the reaction to future acquisitions. Moreover, we found that high-reputation organizations that derive value from their reputation as an intangible asset and likely perceive greater risk to their social evaluations, act to protect it using impression offsetting. We also found a lack of profitability led to a greater use of impression offsetting, which may suggest a relative lack of valuable resources or a more difficult and/or riskier integration process contributed to managers' perceptions of a higher risk of negative expectancy violation.

Implications for Impression Management Research
Our study contributes to the literature on AIM in a number of ways. First, we organized and extended the limited and disparate AIM literature. Only two AIM tactics have been explored previously in the accounting and strategic management literatures: big bath accounting and strategic noise. Strategic noise involves organizational leaders releasing of both positive and negative unrelated information when the anticipated response is unclear. Big bath accounting occurs when organizational leaders release additional negative information when the expected reaction is negative. Impression offsetting represents a previously unexplored AIM tactic employed by organizational leaders. Second, the results of the current study, when combined with other recent research on AIM (e.g., Elsbach et al., 1998;Graffin et al., 2011), suggest that firms engage in a range of impression management activities in anticipation of a potential expectancy violation.
In the case of impression offsetting, future research could explore prescriptive and predictive characteristics that trigger this AIM technique. Since managers are often aware of investors' negative reactions to acquisitions and likely wish to avoid market repudiations of their actions, future researchers may wish to examine how managerial characteristics, such as a CEO's reputation or tenure, influence the likelihood they engage in impression offsetting. In addition, although we only focused on negative expectancy violations, positive expectancy violations may also occur when firm behavior exceeds expectations. We thus encourage future researchers to explore conditions under which firms exceed expectations, and the resulting reactions by markets and stakeholders to this new information.
Our results also have implications for the broader impression management literature in two respects. First, although we discussed anticipatory and reactive impression management as distinct and separate forms of influencing markets, it may be the case that impression management has both reactive and anticipatory elements. For instance, organizational leaders may delay announcements for those events for which they control the timing of the announcement if an unanticipated expectancy violation occurs. Moreover, firms might delay announcing acquisitions if an unanticipated expectancy violation occurs, such as a scandal or environmental accident. Hence, the anticipated timing of the announcement of an event that is under the control of organizations could actually be a form of impression management, even if the event itself were unanticipated. Along these lines, future researchers may wish to explore the joint relationships between anticipatory and reactive impression management in more detail.
Second, our supplementary analyses suggest that firms are not only more likely to release positive unrelated information around acquisition announcements, but also that the release of negative information decreases in a statistically significant manner. This finding suggests there is agency behind the timing of the release of a firm's significant negative occurrences. How and when firms elect to release significant bad news has received little direct attention. Therefore, a better understanding of when firms are more or less likely to release negative information would seem to represent fertile ground for future impression management research.

Implications for Acquisitions Research
Our findings also have implications for research examining acquisitions. To the extent that prior studies of acquisition announcements did not control for confounding events, the results of these studies may have been positively biased in light of organizations employing impression offsetting. This suggests the stock market reaction for a significant percentage of acquisition announcements-roughly 27% in our sample-in prior studies may have been positively biased. In other words, despite the fact the prior research finds a generally negative market reaction to acquisitions, the size of the observed negative reaction may be understated. Our results also suggest this bias occurs in systematic ways as we found more active and higher reputation acquirers were more apt to engage in impression offsetting. This finding, combined with our results on the effectiveness of impression offsetting, suggests prior studies may have reported biased abnormal returns particularly for these types of acquisitions. Future research may wish to re-examine findings relating to the market reaction for active acquirers and highreputation firms. Indeed, it may be the case that, due to their higher levels of acquisition experience, serial acquirers may have developed an impression offsetting capability, which upwardly biases the market reaction of their acquisitions.

Practical Implications
Our results also have practical implications for managers. We found that impression offsetting successfully reduced the negative market reaction to a potential negative expectancy violation in our sample. Hence, our results may encourage organizational leaders to employ this tactic. For investors and other market participants, however, impression offsetting-and its effectiveness-suggests that they should be aware of how an organization is managing the event, and whether this needs to be taken into consideration when valuing the organization's actions. This exploration may be particularly important when organizations take on projects with high likelihoods of being perceived as negative expectancy violations. Thus, investors should interpret press releases around potentially negative events with caution.
Our research also has implications for managers in terms of the timing of announcements, which are under their control. Indeed, our results combined with other studies of AIM (e.g., Elliott & Shaw, 1988;Graffin et al., 2011) suggest that organizational leaders will actively attempt to influence stakeholder reaction to significant organizational events when these leaders have advanced knowledge of the event becoming known to stakeholders. While the current study focuses on how organizational leaders proactively manage impressions associated with a potential negative expectancy violation, managers may wish to engage in this sort of impression management in other contexts. Indeed, organizational leaders can anticipate numerous occurrences, such as earnings announcements, alliance announcements and new product announcements, and may involve positive and/or negative potential expectancy violations.

Limitations and Future Research
One limitation of our study is that we were not able to directly observe organizational leaders intentionally releasing information to offset the market's negative reaction to acquisition announcements. We rather inferred this motivation based upon an observed empirical pattern relative to a baseline behavior. Specifically, we calculated a baseline count of confounding press releases around any announcement and observed a significant higher count of offsetting announcements around acquisitions. Based on the deviation from the baseline, we inferred agency on the part of acquiring firms. Our approach, however, is consistent with recent work in finance that inferred managerial malfeasance when CEOs repeatedly experienced "lucky stock grants," defined as receiving an option at the lower stock price that month (Bebchuk, Grinstein, & Peter, 2010). Future researchers may wish to leverage this tactic to better understand potential motives for other patterns of behavior within firms such as executive or board members' actions.
Given this AIM tactic is both inexpensive and effective, it is not clear why more firms do not use it. In addition, though, we explored the frequency to which organizational leaders engage in impression management in other contexts. Porac et al. (1999) found that organizations selected 31% of its peer group from firms outside of its industry, which they took to represent a form of impression management. Similarly, Graffin et al. (2011) found that 20% of firms engaged in impression management surrounding CEO announcements. While our results seem to be consistent with other studies that have measured the percentage of firms that do or do not engage in impression management, it is unclear why so many firms do not engage more actively in impression management. Might the high and increasing rate of turnover in the executive suite be limiting the development of such impression management capabilities? Could the strengthening of governance and regulations regarding the release of information be limiting executives' discretion to engage in such tactics? Future research may wish to explore the differences between these groups of firms and the rate of diffusion for such tactics.

Conclusion
We introduced a new AIM tactic-impression offsetting. Our results suggest when firm executives anticipate that observers may view an upcoming announcement as a negative expectancy violation, they contemporaneously issue positive, but unrelated, press releases with an acquisition announcement. We also found that impression offsetting is effective, as it significantly decreases the negative abnormal return around an acquisition announcement. Finally, we found that riskier acquisitions were more associated with this AIM technique. Stated simply, our findings indicate that firms can successfully influence market reactions to organizational announcements. To build on our findings, we encourage future work on AIM in order to better understand how organizational leaders attempt to influence market participants.