MANAGING THE RISKS OF PROACTIVITY: A MULTILEVEL STUDY … · fective in executing their...

r Academy of Management Journal2016, Vol. 59, No. 4, 1339–1360.http://dx.doi.org/10.5465/amj.2014.0177

MANAGING THE RISKS OF PROACTIVITY: A MULTILEVELSTUDY OF INITIATIVE AND PERFORMANCE IN THE

MIDDLE MANAGEMENT CONTEXT

LOTTE GLASER*Erasmus University Rotterdam

WOUTER STAM*VU University Amsterdam

RIKI TAKEUCHIHong Kong University of Science and Technology

Drawing on theories of behavioral decision making and situational strength, we de-veloped and tested a multilevel model that explains how the performance outcomes ofpersonal initiative tendency depend on the extent of alignment between organizationalcontrol mechanisms and proactive individuals’ risk propensities. Results from a sampleof 383 middle managers operating in 34 business units of a large multinational corpora-tion indicated that risk propensity weakens the positive relationship between personalinitiative tendency and job performance. This negative moderating effect was furtheramplified when middle managers receive high job autonomy but was attenuated inbusiness units with a strong performance management context. We discuss the implica-tions of these findings for research on proactivity, risk taking, and organizational control.

Within our organization, we want everyone to be in-novative; we want them to challenge the status quo.But to do that productively we need to deeply engagethem in our strategy and culture so they understandhow far we want them to go—and how far is too far.

(RobertE.Moritz,ChairmanofPricewaterhouseCoopers)

An important and recurring theme in the organiza-tional literature has been the design of structures andprocesses that increase predictability of employeebehavior (Ouchi, 1980). Yet as today’s business envi-ronments are becoming ever more dynamic, bothscholars and practitioners have recognized the grow-ing need for flexible work roles in which employeesmust exercise initiative rather than just “do their jobs”(Frese & Fay, 2001; Griffin, Neal, & Parker, 2007). Thisshift in focus has been guided by the expectation thathigh-initiative employees engage in self-starting, pro-active behaviors that contribute to individual andorganizational effectiveness (Crant, 2000; Grant &Ashford, 2008). These potential benefits are not al-ways realized, however, as evidenced by the mixedfindings on the performance outcomes of personalinitiative (Grant, Nurmohamed, Ashford, & Dekas,

2011; Tornau & Frese, 2013). Indeed, because initia-tive extends beyond formal job descriptions, proactiveemployees may often take inappropriate actions thatultimately fail to create value. Organizations thus facethe dilemma of how to exercise control over proactiveemployees without overly constraining them.

Unfortunately, understanding about how this ten-sion can be resolved remains incomplete. Althoughprior research has alluded to the risks of proactivity,most theoretical and empirical attention has focusedon the factors that promote initiative at work (e.g.,Frese & Fay, 2001; Parker, Williams, & Turner, 2006).This research has argued that providing freedom andsupport is essential for encouraging initiative, whichmight suggest that organizational controls are un-desirable. However, conditions that elicit personalinitiative in the workplace may not be the same con-ditions that enhance its performance benefits. An im-portant question therefore is how the risks of personalinitiative can be adequately controlled and at whatlevel in the organization this is best done. In answeringthis question, some have recognized that top manage-ment may guide employee initiative by designingappropriate control systems (e.g., Marginson, 2002;Shimizu, 2012). An assumption in this emerging lit-erature is that all proactive employees require the same* Both authorsmadeanequal contribution to themanuscript.

1339

Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder’s expresswritten permission. Users may print, download, or email articles for individual use only.

http://dx.doi.org/10.5465/amj.2014.0177

type of control. Yet individuals may actually varygreatly in their capacity to manage risks related toundertaking new initiatives (Chan, 2006), implyingthat substantial heterogeneity may exist in proactiveemployees’ idiosyncratic needs for control. It followsthat researchers must better understand these indi-vidual differences in order to determine when par-ticular control mechanisms strengthen or weaken theperformance benefits of personal initiative.

One purpose of this study is to advance knowledgeof the individual differences that influence employees’abilities to manage the risks of proactivity and ob-tain higher job performance from personal initiative.Scholars have argued that individuals often try to as-sess the potential risks before taking initiative but fewhave examined the determinants and performanceoutcomes of such expectancies. Drawing on behav-ioral decision theory (March&Shapira, 1987; Sitkin&Pablo, 1992), we propose that risk propensity consti-tutes an important individual trait thatcanclarifywhysome employees obtain better job performance frompersonal initiative than others. In contrast to priorassertions that proactive individuals must be willingto take risk (e.g., Fay & Frese, 2000), we suggest thatindividuals with higher risk propensities in fact areless effective in controlling the risks associated withinitiative. Accordingly, our study not only highlightsthe significance of risk propensity for understandingthe contingent performance benefits of personal ini-tiative, but also provides a theoretical explanation forwhy proactive employees may have different needsfor organizational control.

Another purpose of this study is to elucidatewhatforms of control enable employeeswith varying risk

propensities to translatepersonal initiative intohigherjob performance. To date, the control literature haspredominantly focused on formal control strategiesthat evaluate behavior or outcomes by means ofrules and procedures (Cardinal, 2001; O’Reilly &Chatman, 1996). But given that uncertainty makes itdifficult to specify appropriate initiative behaviors exante, there is a need for research that considers theeffectiveness of a wider range of controls. We developand test a multilevel model that explains how the re-lationship between personal initiative and job per-formancedepends on the extent of alignment betweenemployees’ risk propensities and organizational con-trolmechanisms. At the job level, we examine the roleof autonomy, which is a core facet of formal controlthat has been shown to influence proactivity (Parkeret al., 2006). At the business unit level, we considerperformance management context as a form of in-formal control that captures whether employees arebounded by a shared ambition and collective orien-tation toward discipline (Gibson &Birkinshaw, 2004).Drawing on situational strength theory (Mischel, 1977),wepropose that thenegativemoderating effect of riskpropensity on the initiative–performance link is am-plified by autonomy but attenuated by performancemanagement context. Our study thus improves the-oretical understandings of when particular controlmechanisms enhance the performance benefits of per-sonal initiative, delineatinghow formal and informalcontrols distinctly influence the job performance ofproactive individuals with low and high risk pro-pensities. Figure 1 depicts our multilevel model.

We tested this model using a sample of middlemanagers in a large multinational corporation. This

FIGURE 1Multilevel Research Framework

1340 AugustAcademy of Management Journal

research context was compelling because initiativeis a key part of middle managers’ job description.Given their unique position as organizational linkingpins, personal initiative may help middle managersto successfully reconcile top management’s strategicdirections with implementation issues surfacing atlower levels (Kanter, 1982; Wooldridge, Schmid, &Floyd, 2008). However, it is also at the middle man-agement level where the risks of initiative are mostsalient. Compared to top- or operating-levelmanagers,middlemanagersmust navigate amore complex set ofinteractions by resolving divergent expectations froma larger number of stakeholders (Floyd & Lane, 2000).Middle managers’ initiatives are therefore highly in-terdependent with, and visible throughout, the rest ofthe organization, rendering their future impact highlyunpredictable. Thus, our study extends pastwork thathas mostly examined initiative in low-level jobs byexamining a research setting where initiative is notonly essential for job performance but also a riskyendeavor that needs careful management.

THEORETICAL FRAMEWORK ANDHYPOTHESES

The burgeoning literature on proactivity has in-troduced multiple constructs that all aim to capture“anticipatory action that employees take to impactthemselves and/or their environments” (Grant &Ashford, 2008: 8). Here, we focus on the concept ofpersonal initiative tendency, defined as an indi-vidual’s propensity to engage in work behaviors thatare self-starting, proactive, and persistent in over-coming barriers (Frese & Fay, 2001). Consistent withthe idea that these three facets often reinforce oneanother (Frese, Fay, Hilburger, Leng, & Tag, 1997),we view personal initiative tendency as a unidimen-sional construct capturing one’s overall disposi-tion to be self-starting, proactive, and persistent. Theconcept is similar to other constructs that captureone’s propensity to show initiative such as proactivepersonality (Bateman & Crant, 1993), but is broaderin scope than various context-specific concepts likefeedback seeking, issue selling, or individual in-novation (Crant, 2000; Parker & Collins, 2010). Per-sonal initiative tendency thus signifies whetherindividuals are generally inclined to pursue self-setgoals instead of assigned ones, proactively respondto threats or opportunities rather thanwait for othersto follow, and try to overcome resistance to changeinstead of giving up easily.

We acknowledge that conceptualizing personalinitiative as a relatively stable individual attribute is

certainly not uncontroversial. Proponents of thedispositional perspective argue that enduring in-dividual differences in personal initiative do existand explain considerable variance in proactive be-haviors beyond other personality traits (Bateman &Crant, 1993; Seibert, Crant, & Kraimer, 1999). Yetcritics have noted that an individual’s initiative maynot necessarily be consistent across situations, suchthat a focus on actual behavior may be more appro-priate (Parker & Collins, 2010). In this study, we as-sume a middle ground between these opposingviews by proposing that a person’s general tendencyto show initiative may not always activate the samepattern of behaviors, but rather influences the generallikelihood thatan individualbehavesproactively.Ourmodel thus recognizes that both individual differ-ences and situational influences regulate the specificbehavioral manifestations and associated perfor-mance outcomes of an individual’s personal initiativetendency.

Personal Initiative and Job Performance of MiddleManagers

Although prior research has mostly examined theperformanceconsequencesof initiative in lower-leveljobs, personal initiative is particularly important atthe middle management level. Middle managersserve as organizational linking pins who are oftenexpected to proactively identify new opportuni-ties emerging at lower levels and overcome obsta-cles by mobilizing support for initiatives from topmanagers (Kanter, 1982; Wooldridge et al., 2008).Juggling multiple roles creates substantial uncer-tainty for middle managers about how to satisfyconflicting demands from different stakeholders(Floyd & Lane, 2000). To successfully navigate suchambiguous situations, middle managers may bene-fit from taking initiative by engaging in proactivebehaviors, such as seeking feedback and buildingnetworks, which help to reduce uncertainty andincrease control at work (Frese, Garst, & Fay, 2007;Seibert et al., 1999).

The foregoing suggests that middle managers withhigh personal initiative tendencies will be more ef-fective in executing their boundary-spanning roleand thus achieve better job performance. However,prior findings on the performance benefits of per-sonal initiative remain inconclusive. Crant (1995)found a positive link between proactive personalityand job performance of real estate managers. Seibertet al. (1999) and Thompson (2005), using samplesof college graduates, also discovered that proactive

2016 1341Glaser, Stam, and Takeuchi

personality was positively related to career successand supervisor-rated job performance, respectively.In contrast, Chan (2006) and Erdogan and Bauer(2005) found that proactive personality was un-related to either subjective or objective measures ofjob performance. Studies byGrant et al. (2011) revealthat personal initiative behavior improved the pro-ductivity of call center employees, whereas self-reported personal initiative tendency was unrelatedto the job search success of students. A recent meta-analysis by Tornau and Frese (2013) confirmed thatprior research results have been inconsistent. A keyinsight from their meta-analysis is that effect sizesof personality measures of personal initiative onsupervisor-rated job performance are, on average,moderately positive (rwc 5 .15 to .17), but exhibitsubstantial heterogeneity. It thus appears that per-sonal initiative tendency does not always improvejob performance and that moderators likely exist ininfluencing this relationship.

One interpretation of these mixed findings is thattaking initiative is a risky endeavor that often hasunpredictable consequences. Personal initiative en-tails risk because initiating change without beingtold implies that proactive individuals may takeactions that are poorly timed, use inappropriatemethods, or encounter resistance from coworkers(Ashford, Rothbard, Piderit, & Dutton, 1998; Grant,Parker, & Collins, 2009). These risks are particularlyacute in the middle management context becauseinitiatives developedat this intermediate levelmighteasily trigger a “ripple effect” of unexpected re-sponses across the organization. Top managers in-deed often struggle to evaluate how new initiativeswill advance the organization’s goals (Burgelman,1983), creating the risk that middle managers’ ini-tiatives are either rejected or create little value. Theserisks are exacerbated in large organizations where itis difficult to closely monitor the numerous dis-persed initiatives emerging at the middle manage-ment level (Shimizu, 2012).

In this study, we seek to clarify the role of organi-zational control in enabling proactive middle man-agers with varying risk propensities to translatepersonal initiative into higher job performance. De-fined as an individual’s tendency to take or avoidrisk, risk propensity has been shown to regulatewhether an individual’s attention is primarily fo-cused on information related to opportunities orthreats (Lopes, 1987; Sitkin & Pablo, 1992). Based onthis insight, we argue that middle managers withhigher risk propensities perceive fewer risks associ-ated with proactivity, expend less effort to carefully

evaluate and mitigate these risks, and thus obtainfewer performance benefits from personal initiative.

The proposedmoderating role of risk propensity isbased on three assumptions that need to be clarified.First, we view personal initiative tendency and riskpropensity as theoretically distinct constructs thatare not causally related. This view is consistent withfindings in the entrepreneurship literature indicat-ing that proactiveness and risk taking representtwo key dimensions of the entrepreneurial orien-tation construct that vary independently of oneanother (Covin & Lumpkin, 2011). Second, our ar-guments emphasize that risk propensity modifiesthebehavioral expressionof an individual’spersonalinitiative tendency rather than that it influences thesebehaviors directly. Although risk propensity mightalso directly promote proactive behavior, Ashfordet al. (1998) and Grant and Rothbard (2013) found nosupport for such a link. We suggest it is useful to dis-tinguish between the likelihood that one behavesproactively and the quality of the behavior in terms ofspecific actions that areundertaken. Inour theorizing,we assert that the former is predicted by managers’personal initiative tendency, whereas the latter isinfluenced by their risk propensity.1 Third, we treatrisk propensity as a relatively stable personality trait,which is consistent with findings showing that in-dividuals have fairly consistent risk preferences (Das& Teng, 2001). Yet we do acknowledge that managerscan be risk-seeking in one situation, but risk-averse inanother situation by focusing on their risk propensitywithin the job domain, excluding risk preferencesrelated to personal decisions (cf. MacCrimmon &Wehrung, 1990). Furthermore, we explicitly considerhow the outcomes of risk propensity depend on con-text by examining the cross-level moderating influ-ences of situational strength. Situational strengthinvokes the notion that features of the context inwhich individuals operate can limit the behavioralexpression of personality (Meyer, Dalal, & Hermida,

1 Our theoretical arguments and hypotheses are con-sistent with an alternative model focusing on actual ini-tiative behaviors instead of one’s generalized tendencyto take initiative. In both cases, we expect that risk pro-pensity influences the precise nature of the proactivebehavior in terms of its form, intended impact, timing,and tactics instead of just the frequency of the behavior(Grant & Ashford, 2008). This means that, unless thesedimensions are explicitly captured in an empirical study,risk propensity can be expected to also moderate the linkbetween traditional measures of personal initiative be-havior (which typically capture its frequency) and jobperformance.


2010). Strong situations generate uniform expecta-tions of suitable behaviors, whereas weak situationslack clear and consistent cues on appropriate actions(Mischel, 1977). Grounded in the interactionist per-spective, our model recognizes that while risk pro-pensity tends to focus middle managers’ attention onpotential gains or losses associated with personal ini-tiative, situational strength can either correct or exac-erbate these biases in risk perception.

Risk Propensity as a Moderator

Middle managers with proactive tendencies mayfrequently expose themselves to substantial risksthat can harm their job performance. Behavioral de-cision theory indicates that the way people perceiveand handle these risks is governed by their risk pro-pensity (Das & Teng, 2001; Sitkin & Pablo, 1992).Individuals with high risk propensities dispropor-tionately focus their attention on potential opportu-nities and therefore tend to regulate their behaviorsthrough eagerness strategies that emphasize gainsand advancement. Conversely, individuals with lowrisk propensities focus primarily on potential threatsand regulate their behaviors through vigilance strat-egies aimed at preventingnegative outcomes (Crowe& Higgins, 1997). In support of these distinctions,Grant andAshford (2008) argued that proactivity caneither be promotion-focused or prevention-focusedand proposed that the two forms may produce dis-tinct performance outcomes.

Research findings indicate that the behavioralmanifestations of risk propensity occur through itsimpact on risk perception. Risk-seekers often per-ceive the same objective situation as less risky thanrisk-averters, which induces greater risk taking thantheir willingness to bear risk would predict (Sitkin &Weingart, 1995). Middle managers with high riskpropensities mainly focus on information aboutpotential opportunities, leading to a reduced aware-ness of downside risks (March & Shapira, 1987). Dis-counting negative outcomes may create a false senseof optimism that limits amanager’s capacity tohandlerisky situations (Hmieleski & Baron, 2009). Accord-ingly, we argue that proactive middle managers withhigher risk propensities are less likely to carefullyevaluate and mitigate relevant risks and thus obtainfewer performance benefits from personal initiative.Several mechanisms account for this hypothesizedmoderating effect.

First, showing initiative requires that managersactively develop self-set goals that go beyond formaljob requirements. Since risk-seekers tend to focus on

attaining “hits” and avoiding missed opportunities(Baron, 2004), they will more easily conclude thata particular initiative is desirable. In turn, settinga lower threshold prompts risk-seekers to take actionmore quickly (Wally & Baum, 1994) and, as a conse-quence, develop initiatives that turn out to be poorlytimed or of little value. Mishra and Lalumiere (2011)indeed found that risk-seekers engage in more impul-sive behaviors and often neglect the long-term impactof their actions. Ironically, being overly optimistichelps risk-seekers to get attention from top managers(Ashford et al., 1998), but also leads to greater dis-content when initiatives fail to deliver promised re-sults. In contrast, risk-averse individuals require ahigh probability of success to tolerate exposure to po-tential failure (Brockhaus, 1980) and thus focus pri-marily on avoiding initiatives that could fail. Thisconcernwith preventing “false alarms” (Baron, 2004)makes proactive middle managers with low riskpropensities more likely to only take initiative whendoing so has a high probability of success. Interestingly,their initial pessimism about the potential impact oftheir initiatives helps risk-averters to often exceed topmanagers’ expectations, leading to better performanceevaluations.

Second, personal initiative involves anticipatingfuture developments andproactively taking action toaddress them. Doing so often ignites resistance fromothers who oppose proposed changes and preventinitiatives from having impact (Ashford et al., 1998).Given that threats are less salient to risk-seekers(Lopes, 1987), they may underestimate oppositionagainst their initiatives and thus fail to performactivities that prevent setbacks later on, such asrequesting feedback, developing a plan, and moni-toring progress. Evidence indicates that risk-seekersindeed undertake fewer risk adjustment actionsto gain control over outcomes and limit exposureto possible losses (Wehrung, Lee, Tse, & Vertinsky,1989). In contrast, proactive middle managers withlow risk propensities focus on cues signaling possi-ble resistance to their initiative. Increased salienceof threats, in turn, induces risk-averters to morecarefully evaluate the merits and feasibility of theirinitiative. Research by Dowling and Staelin (1994)confirmed that risk-averse individuals engage inmore risk-reduction behaviors (i.e., informationsearch) to handle risky situations. Brockner,Higgins,and Low (2004) argue that conducting such due dil-igence is critical for the success of entrepreneurialinitiatives because it highlights potential problemareas and helps to identify actions that can miti-gate these risks. Doing so may align initiatives with


organizational objectives, suggesting that top man-agement is more likely to appreciate the per-sonal initiative of middle managers with lower riskpropensities.

Third, personal initiative demands strong persis-tence in overcoming barriers that could limit aninitiative’s impact. Yet removing obstacles may bedifficult, if not impossible, suggesting that middlemanagers must judge when to give up (Staw, 1981).In this regard, risk-seeking managers tend to believethey have greater control over outcomes than theyactually do (March & Shapira, 1987), which makesthem more likely to escalate commitment to failinginitiatives (Keil, Tan, Wei, Saarinen, Tuunainen, &Wassenaar, 2000). At the same time, the opportunitycosts of such persistence can be substantial becauserisk propensity increases an individual’swillingnessto challenge the status quo (Ashford et al., 1998).Risk-seeking middle managers must therefore ex-pend relativelymore effort to overcome resistance totheir initiatives, such that the contextual perfor-mancebenefits of personal initiativemay come at theexpense of lower task performance, resulting in fewbenefits for overall job performance.

In sum, the preceding arguments suggest that riskpropensity regulates the behavioral manifestationsof middle managers’ personal initiative tendency.Compared to risk-averse individuals, risk-seekerstend to develop expectancies about the benefits oftaking initiative that are biased upward, which re-duces their efforts to carefully evaluate and mitigaterelevant risks. Failure to perform such risk adjust-ments increases risk-seekers’ tendencies to escalatetheir commitment toward potentially inappropriateinitiatives, resulting in reduced performance bene-fits associated with personal initiative. Thus, we of-fer the following hypothesis:

Hypothesis 1. The positive relationship betweenpersonal initiative tendency and job perfor-mance is weaker for middlemanagers with highrisk propensities than formiddlemanagers withlow risk propensities.

The Amplifying Role of Job Autonomy

Although risk propensity generally reduces theperformance benefits of personal initiative tendency,we argue that high job autonomy constitutes a keyboundary condition for this moderating effect. Au-tonomy indicateswhethermiddlemanagers receivediscretion to structure their own work and makeindependent decisions (Hackman&Oldham, 1976).

According to theories of situational strength (Mischel,1977), low autonomy constitutes a strong situationthat directs individuals’ attention to those behaviorsthat are deemed appropriate and restricts opportuni-ties for acting on one’s own dispositions. Conversely,high autonomy represents a weak situation in whichthe attention focus and behavior of managers arelargely guided by individual differences instead ofexternal forces (Barrick & Mount, 1993). Based onthese insights, we argue that risk propensity only re-duces the performance benefits of personal initiativein situations of high autonomy.

At low levels of job autonomy, risk-seekers andrisk-averters will gain few, but comparable, perfor-mance benefits from personal initiative. Limitedautonomy constitutes a strong situation (Barrick &Mount, 1993), where the need to follow prescribedprocedures restricts risk-seekers from taking initia-tive without addressing the associated risks. For ex-ample, top managers may not give permission toproceed with undesirable initiatives, or may pro-vide feedback that can prevent unexpected out-comes (Campbell, 2000). However, low autonomyalso constrains one’s ability to take actions thatsupport new initiatives, suggesting that middlemanagers with high risk propensities obtain fewperformance benefits from initiative when job au-tonomy is low. Meanwhile, for risk-averse man-agers, personal initiative tendency is also unlikely toincrease job performance when autonomy is low.Having little autonomy signals that taking initiativemay not be feasible or rewarded (Den Hartog &Belschak,2012),which further increases risk-averters’concern for avoiding mistakes. Thus, in low auton-omy situations, proactivemiddlemanagerswith lowriskpropensitieswill either fail to take initiativewhenit is required or spend unnecessary time mitigatingtrivial risks, both of which contribute little to jobperformance.

In contrast, when autonomy is high, the link be-tween personal initiative tendency and job perfor-mance strongly varies as a function of risk propensitysuch that it ismore positive for risk-averters than forrisk-seekers. Prior research shows that autonomyincreases employees’ confidence in their ability toinfluence work outcomes by taking initiative (Freseet al., 2007; Parker et al., 2006). Accordingly, insituations of high autonomy, risk-averse middlemanagers are more likely to pursue value-creatinginitiatives theywould otherwise have forgone out offear of failure. Yet autonomy also comes at the costof increased accountability and ambiguity aboutappropriate behaviors. Since risk-averse middle


managers focusmore on avoiding negative outcomes(Crowe&Higgins, 1997), theywillmake greater effortto ensure that initiatives are well aligned with topmanagers’ expectations. High-autonomy situationsthus motivate and enable proactive middle man-agers with low risk propensities to perform necessaryrisk adjustments, such as acquiring information anddeveloping plans, which enhance the performancebenefits of personal initiative (Frese & Fay, 2001).

For middle managers with high risk propensity,autonomy weakens the relationship between per-sonal initiative and job performance. Individualswho are risk-seeking primarily consider the gainsassociated with initiative and tend to discount thepossibility of failure (Sitkin & Pablo, 1992). Auton-omous situations further magnify these perceptualbiases by inducing feelings of control and enactivemastery (Parker et al., 2006). This means that whenrisk-seekers are given autonomy, they are evenmorelikely to falsely believe that initiatives will succeedand be appreciated. Meanwhile, autonomy also en-ablesmiddlemanagerswith high risk propensities toact upon these biases because top managers are lesslikely to intervene. Autonomous risk-seekers maythus become isolated from the rest of the organiza-tion and be free to develop self-serving initiativesthat fail to create value (Shimizu, 2012). Accord-ingly, proactive middle managers with high riskpropensities are particularly likely to develop ini-tiatives that contribute little to job performance insituations of high autonomy.

Taken together, the preceding suggests that riskpropensity only reduces the performance benefits ofpersonal initiative tendency when middle managersare given high autonomy. Since low-autonomy sit-uations counter the overconfidence of risk-seekersbut exacerbate the pessimism of risk-averters, bothwill gain few, but comparable, performance benefitsfrom taking initiative. On the other hand, high-autonomy situations furthermagnify the tendency ofrisk-seekers to overestimate the potential gains ofinitiative, while they curb the excessive concernof risk-averters with potential losses. Risk-seekerstherefore obtain fewer performance benefits frompersonal initiative than risk-averters when auton-omy is high. Hence we hypothesize:

Hypothesis 2. Job autonomy moderates the neg-ative interactive effect of personal initiative ten-dency and risk propensity on job performance insuchaway that this interaction only occurswhenautonomy is high.

The Attenuating Role of Business Unit PerformanceManagement Context

So far we have argued that risk propensity reducesthe performance benefits of personal initiative ten-dency, particularly under conditions of high auton-omy. It may be difficult, however, to overcome thissituation by curtailing the job autonomy of middlemanagers. Doing so requires that top managementspecifies desired behaviors ex ante and continuouslymonitors performance, which may be impossible incontextswhere future jobdemands areuncertain andmiddle managers value independence (O’Reilly &Chatman, 1996). Accordingly, some have arguedthat to minimize undesirable outcomes of proac-tivity, top managers may develop an organizationalcontext that creates shared expectations and com-mon frameworks (Campbell, 2000). Such work en-vironments entail a strong situation that offers clearand consistent cues about what proactive behaviorsare deemed appropriate (Meyer et al., 2010). Thus,rather than purely restricting individual discretion,a strong organizational context provides middlemanagers with adequate information and incentivesto ensure that their initiative supports organizationalobjectives.

In this study, we focus on a business unit’sperformance management context, which captureswhether work environments emphasize both disci-pline and stretch (Gibson & Birkinshaw, 2004). Dis-cipline induces individuals to voluntarily strive tomeet all expectations generated by their explicit orimplicit commitments. Establishing clear standardsof performance and behavior, having a system ofcandid and fast-cycle feedback, and applying sanc-tions consistently help to instill discipline (Ghoshal& Bartlett, 1994). Stretch, on the other hand, refersto a context in which employees voluntarily strivefor more, rather than less, ambitious objectives. De-veloping a shared ambition and collective identity,along with promoting a sense of personal accom-plishment, contributes to the establishment of stretch(Ghoshal & Bartlett, 1994). Together, discipline andstretch define the strength of a business unit’s per-formancemanagement context (Gibson & Birkinshaw,2004).

We expect that a high personal initiative tendencyis less rewarding for risk-seekers than for risk-avertersin a weak performance management context. Whendiscipline is lacking, middle managers experienceambiguity about what behaviors are appropriate andmay perceive this as either a threat or an opportunityfor taking initiative (Grant&Rothbard, 2013). Because


threats are more salient to individuals with low riskpropensities, their initiative tendency is likely toelicitproactive behaviors aimed at reducing ambiguityresulting from unclear performance standards andinconsistent feedback (Grant & Ashford, 2008). Thus,in weak performance management contexts, risk-averters’ natural concern for avoiding mistakes in-creases the likelihood that their personal initiative isappreciated and rewarded. Risk-seekers, on the otherhand, mainly focus on the opportunities afforded bya lack of discipline such that they tend to overlook theimportance of performing adequate risk adjustments(Wehrung et al., 1989). Hence, in the absence of ade-quate information and incentives, proactive middlemanagers with high risk propensities are more likelyto develop initiatives that contribute little to their jobperformance.

Moreover, weak performance management con-texts lack a shared ambition such that initiatives thatchallenge the status quo are less appreciated. Here,risk-seeking middle managers are more willing totolerate exposure to failure by promoting novel ideasthat deviate from prevailing practices (March &Shapira, 1987). Doing so will bring few performancebenefits, however, when managers are not boundedby a shared commitment to strive for excellence. Inunits with limited stretch, middle managers withhigh risk propensities thus obtain few performancebenefits from personal initiative because top man-agers resist innovation and risk-taking (Burgelman,1983). Risk-averters, in contrast, are primarily con-cerned with avoiding failure such that their initia-tive tendencies mostly elicit proactive behaviorsthat reinforce the organization’s existing activities.Such initiative may be particularly appreciated andrewarded in business units with little stretch as itsupports the status quo (Burris, 2012).

Conversely, in strong performance managementcontexts, we expect that personal initiative gener-ates greater performance benefits for middle man-agerswithhigher riskpropensities.Disciplinecreatesshared expectations about appropriate behaviorsand ensures timely feedback that can prevent mis-alignment of initiativeswith organizational objectives(Campbell, 2000). As a result, risk-seeking middlemanagers are less likely to escalate their commitmentto inappropriate initiatives. Discipline also impliesthat top managers frequently monitor results andapply consistent sanctions to enforce accountability(Ghoshal & Bartlett, 1994). Such fast-cycle feedbackmotivates and enables risk-seekers to seek input, de-velop plans, and perform other risk adjustments thathelp to align initiativeswithorganizational objectives.

In contrast, risk-averters seek to avoid failure and thusmay become overly cautious in a strong performancemanagement context. Indeed, when top managersstress the importance of delivering results and applysanctions consistently, risk-averse managers mayonly develop incremental initiatives that contributelittle to their job performance.

In addition, in strong performance managementcontexts that induce organizationmembers to stretchtheir goals, top managers are more supportive ofinitiative that challenges the status quo. Shared as-pirations to strive for excellence render managerialrisk-taking more desirable because top managersrecognize that simply continuing existing activitiesis no longer adequate (Burgelman, 1983). As a result,they may come to favor the initiative of middlemanagers with high risk propensities, who typicallypromote more novel ideas and solutions than risk-averters. Initiative is thus more likely to enhance thejob performance of middle managers with high riskpropensities in business units with strong perfor-mance management contexts.

In sum, we argue that risk propensity becomes lessharmful for the performance benefits of initiativewhen middle managers operate in a strong perfor-mance management context. The shared ambitionand consistent sanctions that prevail in these workenvironments make it more likely that risk-seekingmiddle managers undertake initiatives that are ap-preciated by top managers. In contrast, in weak per-formance management contexts, clear and consistentcues about acceptable forms of initiative are lacking,such that middle managers with high risk pro-pensities are more likely to act on their tendency totake initiative without performing adequate risk ad-justments. Thus, we hypothesize:

Hypothesis 3. Business unit performance man-agement context moderates the negative in-teractive effect of personal initiative tendencyand risk propensity on job performance in sucha way that this interaction only occurs in weakperformance management contexts.

METHODS

Research Setting

The empirical context of this study is a globaltransport and logistics service company with ap-proximately 160,000 employees worldwide in 2010.The company had grown considerably in the yearsprior to our study with annual revenues of approxi-mately $17billion in 2010. It serves as an appropriate


research setting for testing our hypotheses for severalreasons. First, the company faced substantial com-petitive pressures in its domestic market due toslowing growth rates and increasing deregulation.To remain competitive, it consolidated its marketshare in the home market but also explored newopportunities in emerging markets. Hence, middlemanagers were increasingly expected to show per-sonal initiative and support the company’s strategicchanges. Second, the company’s global activitieswere organized into 34 business units. These unitswere geographically dispersed, had relatively au-tonomous operations and decision rights, and eachhad their own senior management team. Accord-ingly, we expected to observe sufficient individual-and unit-level variance in the predictor and criterionvariables in our model.

Research Design and Data Collection

To minimize single-informant bias, we collectedprimary and secondary data from multiple datasources. The dataset was part of a larger researcheffort to understand middle managers’ role in stra-tegic renewal (see Glaser, Fourne, & Elfring, 2015).Primary data were collected mid-2010 by surveyingboth themiddlemanagement and the top executivesof each business unit. At the time of the survey,human resource representatives provided us withthe contact details of all 687 individuals represent-ing the senior executives and middle managers ofthe company’s 34 business units. We started bycontacting the middle managers who reported di-rectly to the executive team of each unit, requestingthem to complete several scales capturing theirpersonal initiative tendency, risk propensity, andjob autonomy. Next, another survey was distributedto unit top executives that assessed their unit’s per-formance management context. Out of a total of 687surveys sent, we received usable responses from383middle managers (69% response rate) and 72 exec-utives (56% response rate) across 34 business units.We tested for nonresponse bias by comparing keyattributes of respondents and non-respondents. Lo-gistic regression analyses indicated no significantdifference on gender (g 5 .03, p 5 .91), tenure(g52.04, p5 .46), or job grade (g52.11, p5 .19).

The secondary dataset was collected through in-ternal company records at the end of 2010 (sevenmonths after collecting the primary survey data).Wecollected year-end job performance appraisals ofthe middle managers including their job grades. Weenhanced the validity of our measures and reduced

common method variance by gathering data frommultiple sources.

Measures and Validation

Surveys were administered in English, which wastheworking language in all businessunits of the focalorganization. Unless otherwise noted, the measureswere rated on a scale ranging from 1 (“strongly dis-agree”) to 7 (“strongly agree”).

Personal initiative tendency. Middle managers’tendency to take personal initiative was measuredusing a seven-item scale that captures the extent towhich middle managers take an active, self-startingapproach to work and go beyond formal job re-quirements (Frese et al., 1997). For example, middlemanagers responded to the item: “Whenever there isa chance to get actively involved, I take it.” Themeasure has been validated as a unidimensionalconstruct capturing one’s overall disposition to beself-starting, proactive, and persistent in prior re-search (Frese et al., 2007). We factor-analyzed theitems and found that all of them loaded above .70.Allitems loaded on a single factor having an eigenvalueof 3.86 and accounting for 55% of the variance (a 5.86). For each middle manager, we averaged theseven items into a single overall personal initiativetendency score.

Risk propensity. Risk propensity was measuredusing a four-item scale based on the work of Gomez-Mejia and Balkin (1989) and Zhao, Seibert, and Hills(2005). The measure captures a middle manager’sgeneralized tendency to take or avoid risk and hasbeen validated in prior research on proactivity atwork (e.g., Ashford et al., 1998). A sample item was:“I often take risks inmy job.”We factor-analyzed theitems and found that all of them loaded above .76.Allitems loaded on a single factor having an eigenvalueof 2.19 and accounting for 72% of the variance (a 5.81). Consequently, for each middle manager, weaveraged the items into a single overall risk pro-pensity score.

Job autonomy. Job autonomy was assessed usinga nine-item scale adapted from Hornsby, Kuratko,and Zahra (2002). For example, middle managerswere asked to indicate the extent towhich they agreewith the following statement: “I have the freedom todecide what I do on my job.” We factor-analyzedthe items and found that all of them loaded above.69. All items loaded on a single factor having aneigenvalue of 5.12 and accounting for 57% of thevariance (a 5 .90). We considered job autonomy asan individual level variable rather than a unit level


variable because autonomy was largely individuallynegotiated within the units. The ICC1 score for au-tonomy indeed showed that less than half of thevariance in autonomy is between units, indicatingconsiderable within-unit variability in job auton-omy.We averaged all nine items into a single overalljob autonomy score for each middle manager.

Performancemanagement context.Wemeasuredperformance management context by adopting aseven-item scale fromGibson and Birkinshaw (2004).Top executives were asked to indicate the degreeto which their business unit encouraged stretch anddiscipline. For example, executives rated to whatextent their unit encouraged individuals at their levelto “Issue creative challenges to their employees, in-stead of narrowly defining tasks” and “Hold em-ployees accountable for their job performance.” Allitems loadedonasingle factorhavinganeigenvalueof3.11 and accounting for 45%of the variance (a5 .78).For each business unit, we averaged all seven itemsinto a single performancemanagement context score.

Job performance. The focal organization useda common job performance appraisal instrument forall middle managers. Specifically, it had developeda competency framework that was used to assessmiddle managers’ job performance against severalobjectives. These objectives were set at the begin-ning of the year and supervisors evaluated eachsubordinate’s performance on these objectives bythe end of the year. The framework focused onmiddle managers’ functional, value-based, and lead-ership competencies. The functional competenciesare those that pertain to a particular job function.These competencies were defined at the departmentlevel since they often incorporate specific task-relatedskills (e.g., “databasemanagement”). The value-basedcompetencies reflected the type of people and be-haviors that were valued by the organization. For in-stance, two values that were considered importantwere integrity and social responsibility. The leader-ship competencies assessed middle managers’ abilityto lead others and be adaptive to changes in theworkplace. For instance, strategic orientation, com-munication, and change management were part ofmiddle managers’ leadership competencies. Overall,the competency framework assessed middle man-agers’ job performance on 11 competencies: strategicawareness, people management, communication,managing results, delighting customers, teamwork,continuous improvement, integrity, change manage-ment, business development, and social responsibil-ity. Each competency was evaluated on a five-pointscale (E 5 “significantly below expectations” to

A5 “significantly above expectations”). Integratingthese ratings, supervisors constructed an overall jobperformance appraisal using the same five-pointscale (i.e., E5 “significantly below expectations” toA 5 “significantly above expectations”) and pro-vided written comments justifying the appraisal.

The company provided us with the overall job per-formance appraisal rating for each middle manager atthe end of this evaluation process (i.e., seven monthsafter measuring the predictor variables). For thisstudy, we recoded this summative job performancescore (E, D, C, B, A) into numeric scores (i.e., 1 to 5).

Control variables. We controlled for possible al-ternative explanations by including relevant controlvariables. At the individual level, we controlled forfour important variables. First, we controlled forgender as research findings suggest a mixed picturewith regard to the influence of gender on proactivebehaviors (Kanfer, Wanberg, & Kantrowitz, 2001).Gender was dummy coded, with 15 female and 05male. In our sample, 18% of the respondents werefemales. Second, we included job grade as a controlvariable becausemanagers at higher levels often havegreater responsibility for undertaking initiative andmay experience higher job complexity (Frese et al.,2007). Job grade was dummy coded as the companyapplied two different job grades that reflect the stra-tegic, hierarchical position of each middle manager.Third,we includedyears in job asa control variable asprior work experiences may indicate differences inknowledge and abilities that influence middle man-agers’ initiative and job performance (Frese & Fay,2001). Last, we controlled for functional area of thejob because functions may differ in the extent towhich they demand or reward personal initiative.Wecreated four dummy-variables for each functionalarea: Marketing & Sales; IT & Operations; Finance &Accounting; and Other (reference category).

At the business unit level, we controlled for threedifferent variables. First, we controlled for size andincluded the natural logarithm of the number of full-time employees in each business unit as this mayaffect resource availability to support new initia-tives and add complexity when initiatives must beimplemented. Second, we controlled for client focusbecause business units may specialize in differentmarkets.We differentiated betweenunits that servedbusiness clients (coded as 1) and those that servedconsumer clients (coded as 0). Third, we controlledfor the geographic location of the business unit toaccount for possible geographical differences inmiddle managers’motivation or opportunity to takerisk and initiative. Herewe created a dummy variable


for the main geographic location in which the com-pany operated (05Western Europe and 15 OutsideWestern Europe).

Aggregating Business Unit Data

Since business unit performance managementcontext represents a “shared unit level construct”(Kozlowski & Klein, 2000), we conducted severalanalyses to ensure that our data exhibited sufficientwithin-unit agreement and between-unit variation.ICC1 and ICC2 were calculated for the performancemanagement context measure to assess whether itmet the statistical criteria for aggregating individualresponses from each unit’s top executives. ICC1 andICC2 for this scale were 0.32 and 0.77, respectively.Furthermore, we calculated an interrater agreementscore (rwg; LeBreton & Senter, 2008) for the perfor-mance management context variable. We used theinterrater reliability to assess agreement between twoor more top executives on the assignment of the unitvariable. The interrater reliability is Cohen’s k, whichranges from 0 to 1 (although negative numbers arepossible), where larger numbers mean better reliabil-ity.Most researchers prefer k values to be at least 0.60and most often higher than 0.70 before claiming agood level of agreement (Glick, 1985). The medianinterrater agreement value for performance manage-ment context was 0.76, suggesting adequate agree-mentforaggregation.Together, theseanalysessupportedour decision to aggregate performance managementcontext to the unit level by taking the mean of topexecutives’ individual responses for each unit, asrecommended by Kozlowski and Klein (2000).

RESULTS

Confirmatory Factor Analyses

Before testing our hypotheses, we conducted con-firmatory factor analysis (CFA) using Lisrel 8.72

(Joreskog & Sorbom, 2005) to examine the factor struc-ture of the four study variables. A 4-factor model(personal initiative tendency, risk propensity, jobautonomy, and performance management context)reached good fit [x25 461.60, df = 266, p, .00; rootmean square error of approximation (RMSEA) 50.05; non-normed fit index (NNFI) 5 0.97; good-ness of fit index (GFI)5 0.91; comparative fit index(CFI) 5 0.97]. The fit of the 4-factor model was signif-icantly better than an alternative 3-factormodel wherepersonal initiative tendency and risk propensity wereconstrained to load on one factor (Δx25 395.22, Δdf53, p , .00). Furthermore, we tested discriminant val-iditybycomparinganunconstrainedwithaconstrainedmodel. The constrained model sets the correlation be-tween twoconstructs equal to one.AsTable 1 indicates,the four multi-item constructs all exhibit satisfactorydiscriminant validity.We examined potential commonmethod variance by testing whether adding a single la-tent method factor would significantly improve modelfit. Adding the additional method factor did not signif-icantly improvemodel fit [Δx25 (Δdf5 60, n5 455)5142.69, n.s.], indicating that common method bias isunlikely to be severe.

We used hierarchical linear modeling (HLM)(Raudenbush, Byrk, & Congdon, 2004) to test ourhypotheses because HLM is particularly well suitedfor nested data. HLM not only estimates model co-efficients at each level, but also predicts the randomeffectsassociatedwitheachsamplingunitatevery level.We first ran a null model for the dependent variablewith no predictor to ensure that there was sufficientvariance between the business units. The test revealedthat a significant amount of the variance in middlemanagers’ job performance resided betweenunits (x2550.55,p, .05). The ICC1value for jobperformancewas0.12, indicating that the variability between units waslarge and using HLMwas appropriate.

We first tested Hypothesis 1 by regressing the indi-vidual (level 1) variancecomponenton the individual-level predictor. We then tested Hypotheses 2 and 3

TABLE 1Results of Scale Analyses using Confirmatory Factor Analysis

Model x2 df Δx2 Δdf RMSEA NNFI GFI CFI

4-factor 461.60*** 266 Baseline model .05 .97 .91 .973-factor 856.82*** 269 395.22*** 3 .08 .91 .85 .922-factor 1478.80*** 271 621.98*** 2 .12 .81 .73 .841-factor 1905.83*** 272 427.03*** 1 .14 .75 .69 .79

Note: 4-factor: risk propensity, performancemanagement context, autonomy, personal initiative tendency; 3-factor: risk propensity, performancemanagement context, autonomy; 2-factor: risk propensity, performance management context; 1-factor: all items loading on one factor.

***p , .001


where the independent variables weremean centeredprior to the formation of interaction terms, as recom-mended by Aiken andWest (1991). Finally, we testedcross-level interactions by regressing level 1 slopes(i.e., relationships between level 1 predictors andoutcomes) onto unit level (level 2) predictors. Wegroup mean centered the level 1 predictors, as this isthe recommended centering approach when cross-level interactions are involved (Hofmann, Griffin, &Gavin, 2000). Finally, although it is difficult to esti-mate precise effect sizes in cross-level models, we re-port Snijders and Bosker’s (1999) overall pseudo R2

(;R2), indicating the proportional reduction of level 1and level 2 errors owing to thepredictors of themodel.

Descriptive Statistics

Table 2 presents the means, standard deviations,and correlations among the variables under studywhere the individual variables are measured at level1 and performance management context at level 2.Personal initiative tendency correlated positivelywith risk propensity (r 5 .21, p , .01) and job au-tonomy (r 5 .24, p , .01). Risk propensity and jobautonomy were also positively correlated (r 5 .27,p , .01). Furthermore, job performance correlatedpositively with personal initiative tendency (r5 .09,p, .05), butwasneither significantly correlatedwithrisk propensity (r 5 .00, n.s.) nor job autonomy (r 5.02, n.s.). Finally, performance management contextwas not significantly correlated with any of the in-dividual level constructs.

Hypothesis Testing

Table 3 summarizes the results of the HLM ana-lyses for Hypotheses 123. Control variables (in-cluding gender, job grade, years in job, functionalarea, unit size, unit client focus and unit geographicallocation) were included first (Model 1). We thentested whether middle managers’ personal initiativetendency was positively related to job performance.As shown in Model 2a, middle managers’ personalinitiative tendency is significantly related to their jobperformance (g 5 .13, p , .01), replicating priorfindings of research on personal initiative (Frese &Fay, 2001). This step accounted for 4% additionalvariance in job performance over the first step withcontrols (total pseudo R2 5 .11).

Next, we tested our baseline hypothesis stipulatingthat risk propensity has a negative moderating influ-ence on the relation between middle managers’ ini-tiative and job performance (Hypothesis 1). As shown

in Model 3a, the individual-level interaction termbetween personal initiative tendency and risk pro-pensitywas significantly negative (g52.16,p, .01),indicating that the positive relation between personalinitiative tendency and job performance becomesweaker when risk propensity is high than when it islow. We performed a simple slope analysis (Aiken &West, 1991) to examine whether the slope is signifi-cantly different from zero. Results indicated that thesimple slope was negative but not significant whenrisk propensity is high (t52.19, n.s.), whereas it waspositive and statistically significant (t5 4.49, p, .00)when risk propensity was low. The addition of two-way interaction terms accounted for 7% additionalvariance in job performance (total pseudo R2 5 .18).These results provide support for our Hypothesis 1.

We then entered the cross-level interaction termsto test Hypotheses 2 and 3, which took into accountthe contingent moderating roles of job autonomyand performancemanagement context, respectively.The addition of two, three-way interaction termsaccounted for 6% additional variance in job perfor-mance (total pseudo R2 5 .24). As shown in Model4a, the interactive effect of personal initiative ten-dency, risk propensity, and job autonomy on jobperformancewas negative and significant (g52.10,p , .01). To further probe this result, we plotted theinteraction effect using Aiken and West’s (1991)procedures. In support of Hypothesis 2, Figure 2shows that the negative interaction effect of personalinitiative tendency and risk propensity on perfor-mance mainly occurs when job autonomy is high.We performed simple slope analyses for each re-gression line to test whether its slope was signifi-cantly different from zero (Aiken &West, 1991). Theresults show that the relationship between personalinitiative tendency and job performance was signif-icantly positive when risk propensity was low andjob autonomywas high (t5 3.02, p, .00) but neutralwhen risk propensity and job autonomy were bothhigh (t 5 2.23, n.s.). Under conditions of low au-tonomy, personal initiative tendency was unrelatedto job performance for both middle managers withlow risk propensities (t 5 1.10, n.s.) and high riskpropensities (t 5 2.23, n.s). Slope difference testsrevealed that the slope ofmiddlemanagerswith highrisk propensities and high job autonomy differedsignificantly from the slope ofmiddlemanagerswithlow risk propensities and high job autonomy (t 522.66, p , .01). For middle managers with low jobautonomy, the slope differencewas not significant (t52.78, n.s.). Combined, these results provide supportfor Hypothesis 2.


TABLE2

Mea

ns,StandardDev

iation

,andCorrelation

s

Variable

Mea

nSD

12

34

56

78

910

1112

13

1Gen

der

0.18

0.39

2Jobgrad

e0.25

0.43

–.12*

3Yea

rsin

job

2.52

1.42

.00

.03

4Marke

ting&salesfunction

0.28

0.45

.04

–.01

.08

5IT

&op

erationsfunction

0.20

0.40

–.05

–.07

–.01

–.32*

*6

Finan

ce&ac

counting

function

0.26

0.44

.05

.07

–.00

–.37*

*–.30*

*

7BU

size

49.0

29.5

.01

–.16*

*.13*

*.38*

*–.14*

*–.14*

*8

BU

clientfoc

us

0.64

0.49

–.05

–.01

.19*

*.22*

*–.11*

–.18*

*.37*

*9

BU

geog

raphic

loca

tion

0.40

0.49

–.05

.00

.08

–.10*

–.04

–.08

.04

.54*

*10

Personal

initiative

tenden

cy4.81

0.73

.04

.04

–.06

.03

–.01

–.10*

–.05

–.01

.13*

*(.86

)

11Riskpropen

sity

4.17

1.24

–.13*

–.03

–.00

.13*

.01

–.17*

*–.06

–.04

–.05

.21*

*(.81

)12

Jobau

tonom

y5.14

0.98

.01

.05

–.05

.09

–.08

–.01

–.03

–.10

–.10

.24*

*.27*

*(.90

)13

Perform

ance

man

agem

ent

context

4.93

0.65

.03

.01

.07

.10

–.02

–.03

.15*

*.30*

*.25*

*–.04

–.04

.00

(.78

)

14Jobperform

ance

3.77

1.10

.04

–.31*

*–.02

–.09

–.00

.02

.02

–.09

–.03

.09*

.00

.02

.02

Note:

n5

383middle

man

agers(lev

el1)

in34

businessunits(lev

el2),internal

consisten

cyreliab

ility(a)e

stim

ates

areon

thediago

nal.

**p

,.01

*p,

.05;

two-tailed

tests.


In testing Hypothesis 3, Model 4a shows that theinteraction effect of personal initiative tendency, riskpropensity, andperformancemanagement context onjob performancewas positive and significant (g5 .18,p , .01). We again plotted the interaction to betterunderstand the contingent relationships. In supportof Hypothesis 3, Figure 3 shows that the negativeinteraction effect of personal initiative tendencyand risk propensity on job performance only occursin weak performance management contexts. Again,simple slope analyses were performed, indicatingthat in weak performance management contexts, the

relationship between personal initiative tendencyand job performance was significantly positive formanagerswith low riskpropensities (t53.73,p, .00)but neutral for managers with high risk propensities(t 5 21.09, n.s.). In strong performance managementcontexts, however, personal initiative tendency waspositively related to job performance (t 5 2.09, p ,.05) among middle managers with low risk pro-pensities, but unrelated to performance for thosewithhigh risk propensities (t5 1.08, n.s.). Slope differencetests revealed that the slope of middle managers withhigh risk propensities differed significantly from the

TABLE 3Results of HLM Analyses Predicting Middle Managers’ Job Performance

Variables Model 1 Model 2a Model 2b Model 3a Model 3b Model 4a Model 4b

Control variablesGender .05(.11) –.00(.12) –.02(.13) –.02(.13)Job grade –.01(.09) –.03(.09) –.01(.09) –.03(.08)Years in job –.01(.02) –.01(.02) –.01(.02) –.02(.02)Marketing & sales function –.41(.16)* –.40(.16)* –.42(.15)** –.44(.15)**IT & operations function –.12(.21) –.11(.19) –.08(.20) –.10(.20)Finance & accounting function –.08(.15) –.07(.14) –.09(.13) –.09(.13)BU size –.00(.00) –.00(.00) –.00(.00) –.00(.00)BU client focus –.52(.17)** –.58(.18)** –.58(.18)** –.58(.18)**BU geographic location .30(.20) .29(.20) .29(.20) .29(.20)

Main effectsPersonal initiative tendency .13(.05)** .10(.05)* .17(.05)** .17(.05)** .21(.05)** .20(.05)**Risk propensity –.08(.04)† –.07(.04) –.05(.04) –.06(.03)† –.04(.03) –.05(.03)Job autonomy .01(.06) .01(.07) –.01(.06) .01(.07) –.00(.07) –.01(.07)Performance management context .13(.10) .03(.10) .13(.10) .03(.10) .13(.10) .03(.10)

Two-way interactionPersonal initiative tendency x risk

propensity (Hypothesis 1)–.16(.06)** –.18(.06)** –.17(.06)** –.16(.05)**

Personal initiative tendency x jobautonomy

.14(.04)** .10(.04)** .07(.06)* .07(.05)*

Risk propensity x job autonomy –.05(.03)† –.03(.02)† –.02(.03) –.03(.03)Performance management context x

personal initiative tendency.00(.07) .05(.10) .04(.11) .06(.10)

Performancemanagement context x riskpropensity

.12(.04)** .11(.04)** .10(.04)** .09(.04)*

Performance management context x jobautonomy

–.13(.09) –.13(.11) –.11(.11) –.13(.11)

Three-way interactionPersonal initiative tendency x risk

propensity x job autonomy(Hypothesis 2)

–.10(.04)** –.08(.04)*

Personal initiative tendency x risk propensity x performancemanagement context (Hypothesis 3)

.18(.07)** .17(.06)**

Pseudo R2 .07 .11 .06 .18 .12 .24 .21

Note: n5 383 middle managers (level 1) in 34 business units (level 2). Coefficients (based on grand centering) are reported with standarderrors in parentheses.

Pseudo R2 values estimate the amount of total variance in the dependent variable captured by predictors in the model.** p , .01* p , .05† p , .10


slope of middle managers with low risk propensities(t523.40,p, .01) for unitswith aweakperformancemanagement context. In contrast, the slope differencewas not statistically significant (t 5 2.58, n.s.) forunitswitha strongperformancemanagement context.Hypothesis 3 was thus supported.

Supplementary Analyses

We performed several supplementary analyses toexamine the robustness of the findings. First, weconducted the same set of analysis without anycontrols (as shown in Model 2b, 3b, and 4b inTable 3). These results are highly comparable to the

ones we found with various controls. Second, wecreated interaction terms among middle managers’prior job performance, personal initiative tendency,risk propensity, and job autonomy to examine po-tential reverse causality—i.e., prior job performancex personal initiative tendency x risk propensity onautonomy (g 5 2.01, n.s.), prior job performancex personal initiative tendency x job autonomy onrisk propensity (g 5 .03, n.s.), and prior job perfor-mance x risk propensity x job autonomy on per-sonal initiative tendency (g 5 2.03, n.s.) were allnon-significant. Third and related, supplementaryanalyses indicated that prior job performance hadno statistically significant main effect on personal

FIGURE 2Interaction of Personal Initiative Tendency, Risk Propensity, and Job Autonomy


initiative tendency (g 5 .00, n.s.), risk propensity(g 5 2.03, n.s.), and job autonomy (g 5 .04, n.s.).These two results combined indicate that priorperformance is unlikely to be driving our results,addressing potential reverse causality concerns.Fourth, we redid our main analyses with middlemanagers’ prior job performance included as anadditional control variable. Results from these an-alyses were largely identical to the results reportedearlier. Of course, controlling for past performancechanges the interpretation of our results becausepersonal initiative tendency then predicts changesin job performance instead of absolute performance

levels. Given that our hypotheses focus on the latter,wedecidednot to control for prior job performance inour final analyses (results from our supplementaryanalyses are available on request).

DISCUSSION

To date, research on proactivity at work has em-phasized that providing freedom and support is es-sential for promoting employee initiative. Our studycomplements this work by developing a multilevelmodel that clarifies the potential enabling role oforganizational control for the job performance of

FIGURE 3Interaction of Personal Initiative Tendency, Risk Propensity, and Business Unit Performance

Management Context


proactive individuals with varying levels of riskpropensity. Findings from a sample of middle man-agers indicated that risk propensity reduces theperformance benefits of personal initiative, but onlywhen individuals receive high job autonomy oroperate in business units with weak performancemanagement contexts. We also found that whileautonomy improved the job performance of pro-active individuals with low risk propensities, per-formance management context increased the jobperformance of proactive individuals with high riskpropensities. Thus, a key insight of this study is thatthe performance outcomes of personal initiativedepend on the extent of alignment between indi-vidual differences in risk propensity and organiza-tional control mechanisms. Below we discuss theimplications of our findings for research on proac-tivity, risk-taking, and control.

Theoretical Implications

By revealing that an individual’s risk propensityreduces the performance benefits of personal initia-tive, our study advances recent efforts aimed at un-derstanding when proactivity leads to better jobperformance (e.g., Grant et al., 2011; Li, Liang, &Crant, 2010). This negative moderating influence ofrisk propensity is a noteworthy finding because itcalls into question the common assumption thatproactive individuals must be willing to take risks(Grant & Rothbard, 2013). It also answers calls formore research on why different employees formdifferent perceptions of the potential risks associatedwith proactivity (Crant, 2000) by demonstrating thesignificant role of individual differences in riskpropensity. Our findings suggest that high levels ofrisk propensity may bias middle managers’ risk as-sessments, leading them to overestimate potentialopportunities afforded by initiative and to disregardthe corresponding risks. Ironically, it is exactly thoseindividuals who are most tolerant of the risks asso-ciated with proactivity that gain the fewest perfor-mance benefits from personal initiative.

More broadly, our study attests to the value of in-tegrating behavioral decision theory with researchon proactivity. Decision theorists have producedmany insights into the determinants and conse-quences of risk behavior (e.g., Das & Teng, 2001;Sitkin & Pablo, 1992), but these have not yet beenlinked to the proactivity literature. In this study, wetook a first step toward merging these researchstreams by proposing that when individuals havehigher risk propensities, their personal initiative

tendencywill elicit fewer risk adjustment efforts andthus provide fewer performance benefits. Our re-sults offer indirect support for this logic, but moreresearch is needed to verify the proposed cognitiveand behavioral mechanisms. For instance, scholarsmay directly capture individuals’ risk perceptionsregarding different forms of initiative and examinethe efficacy of different risk reduction strategies (seeWehrung et al., 1989). Futurework could also extendour focus on risk propensity to include other indi-vidual differences that might influence risk behav-ior, such asdomain familiarity (Sitkin&Pablo, 1992)and regulatory focus (Crowe &Higgins, 1997). As faras these individual differences decrease employees’risk adjustment efforts, we expect them to also re-duce the performance benefits of personal initiative.

Our study also helps to reconcile competing viewson the performance consequences of risk propensity.On one hand, risk propensity may increase perfor-mance by enhancing managers’ alertness to oppor-tunities, persistence in the face of adversity, andpaceof decisionmaking (Brockhaus, 1980;Wally&Baum,1994). On the other hand, risk propensity may re-duce performance by biasing managers’ judgmentand decision making (Hmieleski & Baron, 2009).Our multilevel framework contributes to resolvingthis debate by introducing situational strength asa key contingency factor. Specifically, we found thatrisk propensity only reduces the performance bene-fits of initiative when middle managers receive highjob autonomy or operate in a weak performancemanagement context. This finding is insightful be-cause it clarifies how individual differences inrisk propensity interact with contextual factors toinfluence a manager’s performance. Although strat-egy scholars have recognized that the value of risk-taking depends on environmental conditions (Palmer&Wiseman, 1999), this idea has not yet been appliedto the job performance literature where few studieshave examined the career benefits of risk propen-sity (for an exception, see MacCrimmon & Wehrung,1990).

Our study demonstrates that risk propensity onlyreduces performance when middle managers havehigh personal initiative tendencies and operate inweak situations. Interestingly, this result appearsinconsistent with the person-job fit literature, whichhas argued that proactive individuals must be com-fortable with taking risks because showing initiativeis risky (Fay & Frese, 2000). It also contrasts withempirical evidence indicating that managerial risk-taking is particularly beneficial in ambiguous envi-ronments (Gupta & Govindarajan, 1984). We believe


this apparent contradiction can be reconciled byrecognizing that risk propensity might increasemanagers’willingness to take risk when it is needed,but decrease their post-decision ability to managerisk. This contention raises new questions regardingthe particular risk adjustment actions used by risk-averse individuals to control the risks of personalinitiative. For instance, future studies could examinethe efficacy of various risk-reduction tactics, suchas gathering information, developing new decisionalternatives, hedging exposure to potential losses,and changing the timing or sequence of actions (seeWehrung et al., 1989). Scholars may also considerwhen these strategies are most effective. For exam-ple, the success of proactive individuals’ risk miti-gation efforts might depend on their political skills,social network ties, and knowledge resources. Riskreduction may also not always be possible, espe-cially when individuals face severe time pres-sures. Further research is needed to examine thesepossibilities.

Our study also has implications for research onorganizational control. Based on the assumption thatcontrol undermines personal initiative, past re-search has mostly focused on the “soft” elements ofan organization context that provide freedom, en-courage experimentation, and support entrepre-neurship (e.g., Ashford et al., 1998; Li et al., 2010).Yet our results demonstrate that high job autonomyand a weak performance management context—factors that are often believed to facilitate proac-tivity (e.g., Parker et al., 2006)—were ineffective inenhancing the performance benefits of personalinitiative for managers with high risk propensities.The present study thus complements past researchby suggesting that the “hard” elements of an orga-nization context, as captured by its control systems,can actually enhance the effectiveness of employeeinitiative and thus deserve more attention in futureresearch.

Importantly, our study not only indicates thatcontrol can be beneficial for initiative, but alsoreveals that the efficacy of certain control mecha-nisms depends on individual differences in riskpropensity. We found that proactive middle man-agers with low risk propensities performed betterin high-autonomy situations, while their per-formance was little influenced by business unitperformance management context. In contrast, theperformance of proactive middle managers withhigh risk propensities was hardly affected by au-tonomy, but was significantly influenced by perfor-mancemanagement context. These divergent results

clarify that it is not the level of control per se thataffects whether initiative improves performance,but rather the extent to which control mechanismsare aligned with employees’ individual needs forcontrol. This finding is interesting because theidea that proactive individuals vary in their re-sponses to organizational control systems remainslargely unexplored. Our study thus makes a con-tribution by advancing a contingency-based per-spective on the enabling role of control for proactiveemployees, which relates the relative effectivenessof different control mechanisms to individual dif-ferences in risk propensity. Clearly, additionalresearch is needed to clarify how the personal-ities, motivations, and abilities of proactive in-dividuals affect their responses to different forms ofcontrol.

A somewhat unexpected finding is that neither ofthe two control mechanisms examined in this studyevoked a significant positive link between initiativeand performance for risk-seeking middle managers.Yet we found that, unlike job autonomy, perfor-mance management context did significantly im-prove the job performance of proactive managerswith high risk propensities. Overall, these resultshighlight the need for further research on the con-tingent benefits of different control mechanisms foremployee initiative. One direction for future studieswould be to compare performance evaluation sys-tems that employ strategic versus financial controls.Strategic controls might be more effective in guidingthe initiative of risk-seeking individuals becausethey emphasize long-term and strategically relevantperformance assessment criteria (Hitt, Hoskisson,Johnson, & Moesel, 1996). In exploring this possi-bility, scholars could draw from behavioral agencytheory (Wiseman & Gomez-Mejia, 1998) to examinehow reward and incentive systems interact withamanager’s risk propensity in shaping the outcomesof personal initiative. Another research opportunitywould be to more closely examine the potential en-abling role of informal control. Informal controlsmight be more appropriate for guiding employeebehaviors like personal initiative that involve hightask uncertainty (O’Reilly & Chatman, 1996). Ourresults on the benefits of performance manage-ment context for risk-seeking individuals appear tosupport this logic, but more work is needed thatexamines a wider range of informal controls. Thus,research examining how peer control by coworkersand supervisors affects the risks and performanceoutcomes of personal initiative would be a fruitfulextension of our study.


Limitations and Future Research Directions

This study has several limitations. First, usingmostly cross-sectional data prevented us from fullydisentangling the causal relationships in our model.We did implement a seven-month time lag betweenmeasuring the predictor and criterion variables, butthis does not completely rule out the possibility ofreverse causality. Longitudinal studies are thus re-quired to clarify the interplay between personalinitiative, risk propensity, and job performance overtime. Prospect theory suggests that whether indi-viduals engage in risk-taking depends on whethertheir current performance exceeds or falls behindtheir past performance (Kahneman & Tversky, 1979).Hence an alternative explanation for our findings isthat poorly performing middle managers becomemore proactive and risk-seeking. Although our sup-plementary analyses did not support this possibility,we invite future research to more closely investigatethe role of aspiration levels in explaining the emer-gence and outcomes of proactivity at work.

Second, this study did not directly examine thebehavioral mechanisms underlying the hypothe-sized relationships. By explicitly capturing thesebehaviors, future studies may clarify how risk pro-pensity affects an individual’s engagement in dif-ferent forms of proactivity. This work could drawfromregulatory focus theory (Crowe&Higgins, 1997)by comparing how the eagerness strategies ofrisk-seekers and vigilance strategies of risk-avertersshape the performance outcomes of personal ini-tiative. Moreover, there is reason to believe that amanager’s risk propensity can be domain-specific(March & Shapira, 1987). It is therefore critical toevaluatewhether the findings from this study,whichfocused on people’s generalized tendency to takeinitiative, are applicable to other, more context-specific proactive behaviors.

Third, we tested our model using only one overalljob performance measure. It is possible that consid-ering distinct performance dimensions would leadto different results. For instance, risk propensitymight reduce the task performance of proactive in-dividuals, but increase their innovative job per-formance. Future research could employ multipleperformance measures to more fully capture thecontrasting influences of personal initiative and riskpropensity on managers’ job performance. One in-teresting possibility that warrants further study isthat risk propensity not only influences the meanperformance benefits of personal initiative, but alsothe variance ofmanagers’ job performance over time.

Considering performance variability would enablefuture studies to more directly capture the risks as-sociated with personal initiative.

Fourth, we followed priorwork (e.g., DenHartog &Belschak, 2012; Parker et al., 2006) by using a self-reported measure of job autonomy. Our approachis consistent with evidence attesting the validityof employees’ perceptions of work characteris-tics (Frese et al., 2007). Nonetheless, individualdifferences in risk propensity might have causedmanagers in our sample to consistently underesti-mate or overestimate their actual autonomy. Whilesuch misperceptions can be costly, they may bedifficult to avoid because managers often encoun-ter situations of “mixed discretion” (Hambrick &Finkelstein, 1987) in which they have high auton-omy in some domains but low autonomy in others.Research incorporating both perceptual and objec-tive measures of autonomy is needed to examinewhen managers develop biased perceptions of theirdiscretion and how such biases affect the perfor-mance outcomes of personal initiative.

Fifth, we only examined middle managers oper-ating in a single multinational corporation. This ap-proach helped us to probe the generalizability offindings from prior research that has mostly exam-ined proactivity in low-level jobs, while controllingfor context-specific conditions thatmight impact theoutcomes of initiative. Arguably, middle managersare generally expected tobemoreproactive andoftenhave greater latitude in taking initiative than lower-level employees. Hierarchical level, functional area,and specific organizational factors may all affectwhether individuals are able and expected to showpersonal initiative. Thus, research replicating ourstudy across awider range of jobs, organizations, andindustries is warranted.

CONCLUSION

Organizations increasingly promote personal ini-tiative in the workplace, yet doing so creates signifi-cant risks in that proactive employees may takeinappropriate actions that fail to create value. Thepresent research improves understanding of howthese risks can be managed in the middle manage-ment context by combining insights from behavioraldecision theory and situational strength theory intoa multilevel framework. A key insight of this studyis that the organizational control mechanisms thatincrease the performance benefits of personal initia-tive are distinctly different for individuals with lowand high risk propensities. The theory and findings


presented in this article thus not only refine the com-monbelief that control isharmful for initiative, but alsounderscore the importance of aligning control systemswith proactive individuals’ idiosyncratic needs forcontrol. Clearly, more research is needed to furtherunravel how organizations can manage the risks ofpersonal initiative while leveraging its benefits.

REFERENCES

Aiken, L. S., & West, S. G. 1991.Multiple regression: Test-ingand interpreting interactions. ThousandOaks,CA:Sage.

Ashford, S., Rothbard, N., Piderit, S., &Dutton, J. 1998. Outon a limb: The role of context and impression man-agement in selling gender-equity issues. Administra-tive Science Quarterly, 43: 23–57.

Baron, R. A. 2004. The cognitive perspective: A valuabletool for answering entrepreneurship’s basic “why”questions. Journal ofBusinessVenturing, 19: 221–239.

Barrick, M. R., & Mount, M. K. 1993. Autonomy as a mod-erator of the relationships between the Big Fivepersonality dimensions and job performance. The Jour-nal of Applied Psychology, 78: 111–118.

Bateman, T., & Crant, J. M. 1993. The proactive componentof organizational behavior: A measure and correlates.Journal of Organizational Behavior, 14: 103–118.

Brockhaus, R. H. 1980. Risk taking propensity of entrepre-neurs.AcademyofManagement Journal, 23: 509–520.

Brockner, J., Higgins, E. T., & Low, M. B. 2004. Regulatoryfocus theory and the entrepreneurial process. Journalof Business Venturing, 19: 203–220.

Burgelman, R. A. 1983. A process model of internal cor-porate venturing in the diversified major firm. Ad-ministrative Science Quarterly, 28: 223–244.

Burris, E. R. 2012. The risks and rewards of speaking up:Managerial responses to employee voice.Academy ofManagement Journal, 55: 851–875.

Campbell, D. J. 2000. The proactive employee: Managingworkplace initiative. The Academy of ManagementPerspectives, 14: 52–66.

Cardinal, L. 2001. Technological innovation in the phar-maceutical industry: The use of organizational controlin managing research and development. OrganizationScience, 12: 19–36.

Chan, D. 2006. Interactive effects of situational judgmenteffectiveness and proactive personality on work per-ceptions andwork outcomes. The Journal of AppliedPsychology, 91: 475–481.

Covin, J., & Lumpkin, G. 2011. Entrepreneurial orientationtheory and research: Reflections ona needed construct.Entrepreneurship Theory and Practice, 35: 855–872.

Crant, J. M. 1995. The proactive personality scale and ob-jective job performance among real estate agents. TheJournal of Applied Psychology, 80: 532–537.

Crant, J. M. 2000. Proactive behavior in organizations.Journal of Management, 26: 435–462.

Crowe, E., &Higgins, E. 1997.Regulatory focus and strategicinclinations: Promotion and prevention in decision-making. Organizational Behavior and Human De-cision Processes, 69: 117–132.

Das, T., & Teng, B. 2001. Strategic risk behaviour and itstemporalities: Between risk propensity and de-cision context. Journal of Management Studies,38: 515–534.

Den Hartog, D. N., & Belschak, F. D. 2012. When doestransformational leadership enhance employee pro-active behavior? The role of autonomy and rolebreadth self-efficacy. The Journal of Applied Psy-chology, 97: 194–202.

Dowling, G., & Staelin, R. 1994. A model of perceived riskand intended risk-handling activity. The Journal ofConsumer Research, 21: 119–134.

Erdogan, B., & Bauer, T. 2005. Enhancing career benefitsof employee proactive personality: The role of fitwith jobs and organizations. Personnel Psychology,58: 859–891.

Fay,D., &Frese,M. 2000.Conservatives’ approach towork:Less prepared for future work demands? Journal ofApplied Social Psychology, 30: 171–195.

Floyd, S. W., & Lane, P. J. 2000. Strategizing throughoutthe organization: Managing role conflict in strategicrenewal. Academy of Management Review, 25:154–177.

Frese, M., & Fay, D. 2001. Personal initiative: An activeperformance concept for work in the 21st century.Research in Organizational Behavior, 23: 133–187.

Frese, M., Fay, D., Hilburger, T., Leng, K., & Tag, A. 1997.The concept of personal initiative: Operationaliza-tion, reliability and validity in two German samples.Journal of Occupational and Organizational Psy-chology, 70: 139–161.

Frese,M., Garst, H., & Fay, D. 2007.Making things happen:Reciprocal relationships between work characteris-tics andpersonal initiative in a four-wave longitudinalstructural equation model. The Journal of AppliedPsychology, 92: 1084–1102.

Ghoshal, S., & Bartlett, C. A. 1994. Linking organizationalcontext and managerial action: The dimensions ofquality of management. Strategic Management Jour-nal, 15: 91–112.

Gibson, C. B., & Birkinshaw, J. 2004. The antecedents,consequences, and mediating role of organizationalambidexterity. Academy of Management Journal,47: 209–226.


Glaser, L., Fourne, S. P. L., & Elfring, T. 2015. Achievingstrategic renewal: The multi-level influences of topand middle managers’ boundary-spanning. SmallBusiness Economics, 45: 305–327.

Glick, W. H. 1985. Conceptualizing and measuring orga-nizational and psychological climate: Pitfalls in mul-tilevel research. Academy of Management Review,10: 601–616.

Gomez-Mejia, L. R., & Balkin, D. B. 1989. Effectiveness ofindividual and aggregate compensation strategies.Industrial Relations, 28: 431–445.

Grant,A.,&Ashford,S. J. 2008.Thedynamicsofproactivityatwork. Research in Organizational Behavior, 28: 3–34.

Grant, A., Nurmohamed, S., Ashford, S. J., & Dekas, K.2011. The performance implications of ambivalentinitiative: The interplay of autonomous and con-trolled motivations. Organizational Behavior andHuman Decision Processes, 116: 241–251.

Grant, A.M., Parker, S. K., &Collins, C. 2009. Getting creditfor proactive behavior: Supervisor reactions dependon what you value and how you feel. Personnel Psy-chology, 62: 31–55.

Grant,A.M., &Rothbard,N. 2013.When indoubt, seize theday?Security values, prosocial values, andproactivityunder ambiguity. The Journal of Applied Psychol-ogy, 98: 810–819.

Griffin, M., Neal, A., & Parker, S. K. 2007. A new model ofwork role performance: Positive behavior in uncertainand interdependent contexts. Academy of Manage-ment Journal, 50: 327–347.

Gupta, A., & Govindarajan, V. 1984. Business unit strategy,managerial characteristics, and business unit effec-tiveness at strategy implementation. Academy ofManagement Journal, 27: 25–41.

Hackman, J., & Oldham, G. 1976. Motivation through thedesign of work: Test of a theory. Organizational Be-havior and HumanDecision Processes, 16: 250–279.

Hambrick,D.,&Finkelstein,S.1987.Managerialdiscretion:Abridge between polar views of organizational outcomes.Research in Organizational Behavior, 9: 369–406.

Hitt,M.A., Hoskisson, R. E., Johnson, R. A., &Moesel, D. D.1996. The market for corporate control and firminnovation. Academy of Management Journal, 39:1084–1119.

Hmieleski, K. M., & Baron, R. A. 2009. Entrepreneurs’ op-timism and new venture performance: A social cogni-tive perspective. Academy of Management Journal,52: 473–488.

Hofmann, D. A., Griffin, M. A., & Gavin, M. B. 2000. Theapplication of hierarchical linear modeling to organi-zational research. In K. J. Klein & S. W. J. Kozlowski(Eds.), Multilevel theory, research, and methods inorganizations: 467–511. San Francisco: Jossey-Bass.

Hornsby, J. S., Kuratko, D. F., & Zahra, S. A. 2002. Middlemanagers’ perception of the internal environment forcorporate entrepreneurship: Assessing a measurementscale. Journal of Business Venturing, 17: 253–273.

Joreskog, K., & Sorbom, D. 2005. Lisrel 8.72. Chicago, IL:Scientific Software International.

Kahneman, D., & Tversky, A. 1979. Prospect theory: Ananalysis of decision under risk. Econometrica, 47:263–292.

Kanfer, R., Wanberg, C. R., & Kantrowitz, T. M. 2001. Jobsearch and employment: A personality-motivationalanalysis and meta-analytic review. The Journal ofApplied Psychology, 86: 837–855.

Kanter, R. 1982. The middle manager as innovator. Har-vard Business Review, 60: 95–106.

Keil,M., Tan, B. C. Y.,Wei, K., Saarinen, T., Tuunainen,V.,& Wassenaar, A. 2000. A cross- cultural study on es-calation of commitment in behavior software projects.Management Information Systems Quarterly, 24:299–325.

Kozlowski, S. W. J., & Klein, K. J. 2000. A multilevel ap-proach to theory and research in organizations: Con-textual, temporal, and emergent processes. In K. J.Klein & S. W. J. Kozlowski (Eds.), Multilevel theory,research, and methods in organizations: 3–90. SanFrancisco, CA: Jossey-Bass.

LeBreton, J.M., & Senter, J. L. 2008. Answers to 20 questionsabout interrater reliability and interrater agreement.Organizational Research Methods, 11: 815–852.

Li, N., Liang, J., & Crant, J. M. 2010. The role of proactivepersonality in job satisfaction and organizational citi-zenship behavior: A relational perspective. The Jour-nal of Applied Psychology, 95: 395–404.

Lopes, L. 1987. Between hope and fear: The psychology ofrisk. In L. Berkowitz (Ed.),Advances in experimentalsocial psychology, vol. 20: 255–295. New York, NY:Academic Press.

MacCrimmon,K., &Wehrung,D. 1990.Characteristicsof risktaking executives.Management Science, 36: 422–435.

March, J., &Shapira,Z. 1987.Managerialperspectiveson riskand risk taking.Management Science, 33: 1404–1418.

Marginson, D. E. W. 2002. Management control systemsand their effects on strategy formation at middle-management levels: Evidence from a U.K. organiza-tion. Strategic Management Journal, 23: 1019–1031.

Meyer, R. D., Dalal, R. S., &Hermida, R. 2010.A review andsynthesis of situational strength in the organizationalsciences. Journal of Management, 36: 121–140.

Mischel, W. 1977. The interaction of person and situation.In D. Magnusson & N. S. Endler (Eds.), Personality atthe crossroads: Current issues in interactional psy-chology: 333–352. Hillsdale, NJ: Erlbaum.


Mishra, S., & Lalumiere,M. L. 2011. Individual differencesin risk-propensity: Associations between personalityand behavioral measures of risk. Personality and In-dividual Differences, 50: 869–873.

O’Reilly, C., & Chatman, J. 1996. Culture as social control:Corporations, cults, and commitment. Research inOrganizational Behavior, 18: 157–200.

Ouchi, W. 1980. Markets, bureaucracies, and clans. Ad-ministrative Science Quarterly, 25: 129–141.

Palmer, T., & Wiseman, R. 1999. Decoupling risk takingfrom income stream uncertainty: A holistic model ofrisk. Strategic Management Journal, 20: 1037–1062.

Parker, S. K., & Collins, C. 2010. Taking stock: Integratingand differentiating multiple proactive behaviors. Jour-nal of Management, 36: 633–662.

Parker, S. K.,Williams, H.M., & Turner, N. 2006.Modelingthe antecedents of proactive behavior at work. TheJournal of Applied Psychology, 91: 636–652.

Raudenbush, S. W., Byrk, A. S., & Congdon, R. 2004.HLM6:Hierarchical linear and nonlinearmodeling.Lincolnwood, IL: Scientific Software International.

Seibert, S., Crant, J., & Kraimer, M. 1999. Proactive per-sonality and career success. The Journal of AppliedPsychology, 84: 416–427.

Shimizu, K. 2012. Risks of corporate entrepreneurship:Autonomy and agency issues.Organization Science,23: 194–206.

Sitkin, S., & Pablo, A. 1992. Reconceptualizing the de-terminants of risk behavior.AcademyofManagementReview, 17: 9–38.

Sitkin, S., & Weingart, L. 1995. Determinants of riskydecision-making behavior: A test of themediating roleof risk perceptions and propensity. Academy of Man-agement Journal, 38: 1573–1592.

Snijders, T. A. B., &Bosker, R. J. 1999.Multilevel analysis:An introduction to basic and advanced multilevelmodeling. London, UK: Sage.

Staw, B. 1981. The escalation of commitment to a course ofaction.AcademyofManagementReview, 6: 577–587.

Thompson, J. A. 2005. Proactive personality and job per-formance: A social capital perspective.The Journal ofApplied Psychology, 90: 1011–1017.

Tornau, K., & Frese, M. 2013. Construct clean-up in pro-activity research: A meta-analysis on the nomologicalnet of work-related proactivity concepts and their in-cremental validities. Applied Psychology, 62: 44–96.

Wally, S., & Baum, J. 1994. Personal and structural de-terminants of the pace of strategic decision making.Academy of Management Journal, 37: 932–956.

Wehrung, D. A., Lee, K. H., Tse, D. K., & Vertinsky, I. B.1989. Adjusting risky situations: A theoretical frame-work and empirical test. Journal of Risk and Un-certainty, 2: 189–212.

Wiseman, R. M., & Gomez-Mejia, L. R. 1998. A behavioralagency model of managerial risk taking. Academy ofManagement Review, 23: 133–153.

Wooldridge, B., Schmid, T., & Floyd, S. W. 2008. Themiddle management perspective on strategy process:Contributions, synthesis, and future research. Journalof Management, 34: 1190–1221.

Zhao, H., Seibert, S. E., & Hills, G. E. 2005. The mediatingrole of self-efficacy in the development of entrepre-neurial intentions. The Journal of Applied Psychol-ogy, 90: 1265–1272.

Lotte Glaser ([email protected]) is an assistant professor at theRotterdam School of Management, Erasmus UniversityRotterdam. She received her PhD from the VU Universityin Amsterdam. During her PhD, Lotte continued workingfor a multinational, first at the management developmentdepartment and then at the strategy department of thecompany’s global headquarters. Her research focuses onmulti-level antecedents and consequences of intra- andinter-organizational networks, corporate entrepreneurship,and organizational paradoxes.

Wouter Stam ([email protected]) is a professor of entrepre-neurship at VU University Amsterdam. He has previouslybeen a facultymember at Hong KongUniversity of Scienceand Technology and a visiting scholar at the WhartonSchool, University of Pennsylvania. His research interestsincludenewventure creation, corporate entrepreneurship,networks, and managerial decision making.

Riki Takeuchi ([email protected]) is a professor in the De-partment of Management at the School of Business andManagement, Hong Kong University of Science & Tech-nology. He received his PhD from the Robert H SmithSchool of Business, University of Maryland at CollegePark. His research interests revolve around various socialexchange relationships and includecross-cultural/expatriateadjustment, strategic human resource management, and or-ganizational citizenship behaviors.


mailto:[email protected]



MANAGING THE RISKS OF PROACTIVITY: A MULTILEVEL STUDY … · fective in executing their...

Documents

Transcript of MANAGING THE RISKS OF PROACTIVITY: A MULTILEVEL STUDY … · fective in executing their...