chang-2015-human resource management journal

23
A multilevel examination of high-performance work systems and unit-level organisational ambidexterity Yi-Ying Chang, Department of Business Administration, National Taiwan University of Science and Technology Human Resource Management Journal, Vol 25, no 1, 2015, pages 79–101 This paper examines the process linking high-performance work systems (HPWS) and organisational ambidexterity both at the unit and firm level of analyses by integrating strategic HRM, human capital and social capital perspectives. Multisource and multilevel data from 2,887 employees and 536 managers of 58 banks was collected. Results revealed that firm-level HPWS were positively related to unit-level employee human capital. Unit-level employee human capital partially mediated the relationship between firm-level HPWS and unit organisational ambidexterity. Furthermore, firm-level social climate moderated the effect of firm-level HPWS on unit organisational ambidexterity through unit-level employee human capital. This paper contributes to HPWS and ambidexterity research by revealing the impacts of firm-level HPWS and mediating mechanisms, as well as identifying boundary conditions for pursuing unit-level organisational ambidexterity. Contact: Dr Yi-Ying Chang, Department of Business Administration, National Taiwan University of Science and Technology, 43, Keelung Road, Section 4, Taipei 106, Taiwan. Email: [email protected] Keywords: firm-level high-performance work systems; organisational ambidexterity; firm-level social climate; human capital INTRODUCTION R esearch on ambidexterity indicates that the use of high-performance work systems (HPWS) is an important antecedent to facilitate ambidexterity, which results in better firm performance (Patel et al., 2013). Extant research focused on HRM and ambidexterity tends to treat both HPWS and ambidexterity as overarching organisational-level phenomenon (Junni et al., 2013). In line with this, Patel et al. (2013) investigated the direct relationship between HPWS and organisational ambidexterity. Organisational ambidexterity refers to the capacity of an organisation to simultaneously utilise existing market opportunities efficiently, and to initiate creative and innovative solutions to meet future market demands (Lubatkin et al., 2006). Moreover, the literature on HPWS tends to study outcomes at the organisational level, such as organisational performance. More recently, studies on HPWS have investigated the mechanisms through which HPWS affects individual-level employee attitudes and behaviours (e.g. Takeuchi et al., 2009). Few attempts have been made to explore the mechanisms and boundary conditions, such as HR practices, both at the higher organisational level and the lower organisational level to support the occurrence of organisational ambidexterity (Junni et al., 2013; Turner et al., 2013). There are several important reasons to uncover the mechanisms and boundary conditions through which HPWS affects unit organisational ambidexterity. First, HRM studies have investigated the role of HRM practices, such as HPWS, in the creation of a context conducive to ambidexterity (Kang and Snell, 2009; Patel et al., 2013). HPWS literature has consistently doi: 10.1111/1748-8583.12061 HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 25 NO 1, 2015 79 © 2014 John Wiley & Sons Ltd. Please cite this article in press as: Chang, Y.-Y. (2015) ‘A multilevel examination of high-performance work systems and unit-level organisational ambidexterity’. Human Resource Management Journal 25: 1, 79–101.

Upload: kashif-munir-idreesi

Post on 17-Aug-2015

26 views

Category:

Documents


18 download

DESCRIPTION

fkdnfkldf

TRANSCRIPT

A multilevel examination of high-performance worksystems and unit-level organisationalambidexterityYi-YingChang, DepartmentofBusiness Administration, NationalTaiwanUniversityofScienceandTechnologyHumanResourceManagementJournal, Vol25, no1, 2015, pages79101Thispaperexaminestheprocesslinkinghigh-performanceworksystems(HPWS) andorganisationalambidexterity both at the unit and rm level of analyses by integrating strategic HRM, human capitaland social capital perspectives. Multisource and multilevel data from 2,887 employees and 536 managersof 58bankswascollected. Resultsrevealedthatrm-level HPWSwerepositivelyrelatedtounit-levelemployee human capital. Unit-level employee human capital partially mediated the relationship betweenrm-level HPWS and unit organisational ambidexterity. Furthermore, rm-level social climatemoderated the effect of rm-level HPWS on unit organisational ambidexterity through unit-levelemployee human capital. This paper contributes to HPWS and ambidexterity research by revealing theimpacts of rm-level HPWS and mediating mechanisms, as well as identifying boundary conditions forpursuingunit-levelorganisationalambidexterity.Contact: Dr Yi-Ying Chang, Department of Business Administration, National TaiwanUniversity of Science and Technology, 43, Keelung Road, Section 4, Taipei 106, Taiwan. Email:[email protected]: rm-level high-performance work systems; organisational ambidexterity; rm-levelsocialclimate; humancapitalINTRODUCTIONResearchonambidexterityindicates that the use of high-performance worksystems(HPWS)isanimportantantecedenttofacilitateambidexterity, whichresultsinbetterrm performance (Patel et al., 2013). Extant research focused on HRM and ambidexteritytends to treat both HPWS and ambidexterity as overarching organisational-level phenomenon(Junni et al., 2013). Inline withthis, Patel et al. (2013) investigatedthe direct relationshipbetweenHPWSandorganisational ambidexterity. Organisational ambidexterityreferstothecapacity of an organisation to simultaneously utilise existing market opportunities efficiently, andtoinitiatecreativeandinnovativesolutionstomeet futuremarket demands(Lubatkinet al.,2006). Moreover, theliteratureonHPWStendstostudyoutcomesattheorganisationallevel,suchasorganisational performance. Morerecently, studiesonHPWShaveinvestigatedthemechanisms through which HPWS affects individual-level employee attitudes and behaviours(e.g. Takeuchi et al., 2009). Fewattempts have beenmade toexplore the mechanisms andboundaryconditions, suchasHRpractices, bothat thehigherorganisational level andthelower organisational level tosupport theoccurrenceof organisational ambidexterity(Junniet al., 2013; Turneret al., 2013).Thereareseveralimportantreasonstouncoverthemechanismsandboundaryconditionsthrough which HPWS affects unit organisational ambidexterity. First, HRMstudies haveinvestigated the role of HRM practices, such as HPWS, in the creation of a context conducivetoambidexterity(KangandSnell, 2009; Patel et al., 2013). HPWSliteraturehasconsistentlybs_bs_bannerdoi: 10.1111/1748-8583.12061HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 25 NO 1, 2015 79 2014 John Wiley & Sons Ltd.Please cite this article in press as: Chang, Y.-Y. (2015) A multilevel examination of high-performance work systems and unit-level organisationalambidexterity. Human Resource Management Journal 25: 1, 79101.argued that the practices themselves do not generate a competitive advantage; rather,performance stems from the human resources that are developed by the system (Huselid, 1995;Wright et al., 2001). Accordingly, behavioural ambidexterityislikelytocomefromthermsdistinctiveHRbase(Barney, 1991). Thatis, organisationscanachieveambidexteritybybeingexible with time allocation and focusing the attention of human resources towards explorationand exploitation (Lepak et al., 2003; Gibson and Birkinshaw, 2004; Patel et al., 2013). Byunderstanding the differences among people of the same organisation in terms ofambidexterity, rather than viewing the workforce as one entity, a rm may realise that it needstoemployadifferent set of HPWSinordertofosterunit organisational ambidexterity. Forinstance, differentpositions(verticallyandhorizontally)withinarmmaydemanddifferentlevelsof ambidexterityinorder tobeat betweenthespecicdemandsof ajobandthebehaviouroftheperson.Second, it is important to understand why HPWS may be more or less effective in fosteringambidexterity. PreviousstudiesofHPWShaveemphasisedaneedtodiscoverthemediatingandboundarymechanismsthat explaintheperformanceimplicationsofHPWS(BeckerandHuselid, 2006). More recently, scholars have called for more research to investigate how HPWSaffects ambidexterity through various mediators such as human capital (Kang and Snell, 2009;TurnerandLee-Kelly, 2013; Turneret al., 2013)andacrossmultiplelevels(Jansenet al., 2009;BirkinshawandGupta, 2013; Junniet al., 2013). Whileorganisationalambidexterityresidesatallhierarchicallevelsoftheorganisation,itcanalsooccuratallorganisationallevels.Infact,as the multi-unit rmhas become a pervasive organisational formin the contemporarybusiness landscape (Usher, 1999), organisational ambidexterity at the business unit level is vitalto such organisations. Lower level business units often simultaneously execute exploration andexploitation activities, in addition to undertaking paradoxical thinking and managingconicting demands, and they do this with more direct contact with important organisationalstakeholders, such as employees and customers. As a result, unit-level ambidexterity (i.e. doingtwo different things equally well, such as efficiency and exibility, adaptability and alignment,integration and responsiveness, and exploration and exploitation) could help the rm succeed(BirkinshawandGupta, 2013).Other studies have usedthe social capital perspective toexamine the impact of HRMpractices onambidexterityonly at those organisational levels that are mediatedbysocialclimate (e.g. PrietoandSantana, 2012). We expect that the relationshipbetweenrm-levelHPWSandunit organisational ambidexteritywouldbemediatedbythehumancapital ofindividual employees. Human capital refers to the knowledge, skills and abilities ofindividual employees that are valuable to the rm(e.g. SubramaniamandYoundt, 2005;Turner et al., 2013). HPWScanbuy andmake desirableindividual employeeknowledge,skillsandabilities, whichinturncanbeusedtocreatevaluefortherm(Snell andDean,1992; BeckerandGerhart, 1996). Arecent meta-analytical HRMstudyrevealedthat humancapital andmotivationactedasmediatorsbetweenHPWSandoperational outcomes(Jianget al., 2012). HPWScanbeusedtopromoteorimpedetheefficiencyofindividualemployeeswhenhigh-quality, valuable, unequalledhumanresourcesact inwaysthat arerequiredforimplementingandachievingorganisational outcomes bothat rmandunit levels (Barney,1991; Wrightet al., 1994). TheuseofHPWScancreateandmaintainvaluablehumancapital,includinggenericandorganisational-specichumancapital, whichinturnleads toahighlevel of unit organisational ambidexterity (Kang and Snell, 2009). Going forward, it isimportant to identify the HPWS that is critical for promoting human capital at the unit level,andthemediatingroleof thehumancapital of individual employeesbetweenHPWSandunit organisational ambidexterity.HPWS and organisational ambidexterityHUMAN RESOURCE MANAGEMENT JOURNAL, VOL 25 NO 1, 2015 80 2014 John Wiley & Sons Ltd.Furthermore, in this research, we employ a knowledge and learning perspective to examinethe role of social climate as a moderator at the rm level in the relationship between HPWS andunit organisational ambidexterity. The use of HPWS can enable andmotivate individualemployees to deliver high levels of performance. The high performance implication here can beseen as the completion of organisational goals (e.g. Becker and Huselid, 1998; Delery and Shaw,2001) orthecompletionof individual employees tasks. Thismeansit couldbearguedthatHPWShasatendencytofocusmoreontheexploitationoftheformal, theinducedandtheknown, but less on the exploration of the informal, the uncertain and the unknown. Hence, weargue that HPWS needs to be complemented with a social climate (i.e. trust, cooperation, andsharedcodes andlanguages) inorder tofoster a context that promotes or impedes bothexplorationandexploitationconcurrently. Inother words, theeffectiveness of HPWSmaydepend on the boundary conditions of a rms social context that are used to facilitate the unitsorganisational ambidexterity. Empirical ndings revealed that a rms context, characterised byitssocial climate, isconducivetoknowledgecombinationandexchange(CollinsandSmith,2006) andambidexterity(GibsonandBirkinshaw, 2004). Goingforward, wearguethat theeffectiveness of a context characterised by HPWS depends on a rm-level social climate in orderto provide for the variety and safety needed for employees to engage in ambidextrous learning.Finally, while there is much research on single-level outcomes, research on the effect of HPWSon unit-level ambidexterity at multiple levels is scarce (Raisch and Birkinshaw, 2008; Junni et al.,2013). Therefore, scholars (e.g. Guptaet al., 2006; Junni et al., 2013; Turner et al., 2013) havesuggested cross-level analysis of ambidexterity as a promising direction for future research. Inthis study, we examine the relationshipbetweenrm-level HPWSandunit organisationalambidexteritythat crossesrmandunit levels(cf. TushmanandOReilly, 1996; GibsonandBirkinshaw, 2004; OReilly and Tushman, 2004). No previous research has examined this crucialquestion: Howdoes rm-level HPWS affect unit-level organisational ambidexterity acrossmultiple levels, through the unit-level human capital of employees and rm-level social climate?We endeavour to make several theoretical contributions to the literature. First, we go beyondthefocusonasinglelevel ofanalysisofambidexteritybyinvestigatingrm-level HPWSinpromoting organisational ambidexterity across unit and rmlevels, thereby addressingprevious calls for anunderstandingof organisational ambidexterityacross multiple levels(OReillyandTushman, 2011; BirkinshawandGupta, 2013; Junniet al., 2013).Second, this study examines whether the relationship between rm-level HPWS andorganisational ambidexterity at the unit level would be similar to the relationship found at thermlevel by Patel et al. (2013) (i.e. homology; Morgeson and Hofmann, 1999). As Chen et al. (2005:376) stated, [i]f researchers nd that relationships are homologous across levels of analysis, itadds to the parsimony and breadth of theories. In contrast, should relationships prove not to behomologous across levels, it signals a boundary condition and a need to rene theories.Third, ourfocusonunit-levelhumancapitaladdsamissingpiecetoresearchonstrategicHRM and organisational ambidexterity. We not only test the idea that unit-level human capitalmediates the effects of HPWSonunit organisational ambidexterity, but alsoextendit byshowingthattheeffectmaydependonrm-level social climateasaboundaryconditiontopromotetheunitorganisationalambidexterity. Inaddition, wenotonlytesttheideathatanindividual employees human capital stimulates the occurrence of unit organisationalambidexterity, but also extend it by revealing the effect of rm-level HPWS on unitorganisational ambidexterity, through the mediating effect of an individual employees humancapital. Overall, wecontributetothebroader literatureonambidexterityandHRM, whichstates that animportant issue is to identify the conditions anindividual needs to haveto. . .excel at both exploration and exploitation (Gupta et al., 2006: 696). We identifyYi-Ying ChangHUMAN RESOURCE MANAGEMENT JOURNAL, VOL 25 NO 1, 2015 81 2014 John Wiley & Sons Ltd.organisational systems i.e. HPWS, humancapital andboundaryconditions (rms socialclimate) that facilitate the occurrence of unit organisational ambidexterity. More specically,wecontributetotheunderstandingofhowHRpracticesimpact lowerlevel unitsofarmrather than focusing on macro-level outcomes. We make these contributions by using amultisourceandmultilevelresearchdesignandasampleofbankingservicermsinTaiwan.Studies into ambidexterity have revealed that bank branches achieve service and salesambidexterity by creating a work context that assists their judgement about how to divide theirtime, attention and efforts between requests for service excellence and sales targets (Gibson andBirkinshaw,2004).Inordertoachievethealignmentofservingandsellinginthebanks,itisimportanttodesignpeersupportandprocesses, aswell assystemssuchasHRsystems, toencouragemoresocialinteractionamongteammembers(Yuet al., 2013). Thiswillshapetheoccurrence of ambidexteritybothat the bankbranches andamongindividual employees(Jansenet al., 2008; Peters, 2011; Yuet al., 2013). Thisstudyisamongtherst toempiricallyexamine HPWS and managers ambidexterity in a developing economy. Previous research hasfocused predominantly on high-technology rms in advanced economies (e.g. Wang and Raq,2014), and on optometry and telecom industries in advanced economies (e.g. Cagarra-NavarroandDewhurst, 2007).THEORY AND HYPOTHESISTheoretical backgroundBowen and Ostroff (2004: 206) stated that the content of work systems should be largely drivenby the strategic goals and values of the organisation, and that HR practices must be designedaround a particular strategic focus, such as service or innovation. The core of this statement isthat work systems must reect how employees add value, and this is accomplished by linkingthepracticeswithinasystemtowardssomestrategicobjective. Therefore, allpartsofHPWSmust be chosen and designed to achieve a specic strategic objective. This strategic perspectiveofHPWSissimilartotheargumentthatworksystemsshouldutilisebothhorizontallyandvertically aligned HR practices in order to affect both the ability and motivation of employeesandtoachievetheorganisationsgoals(BeckerandGerhart, 1996; Takeuchi et al., 2007). AllHPWS have goals to attract, retain and motivate human resources towards the achievement oforganisational objectives, by producing a t between the knowledge, skills and abilities of theperson and the duties and responsibilities required by the job (Patel et al., 2013). Scholars haveindicatedthat humanresources, ratherthanthepracticesdevelopedbythesystems, canbeused to generate better performance (Huselid, 1995; Wright et al., 2001). Similarly, theoccurrence of unit organisational ambidexterity is likely to arise from the rms unique HR baserather than froma set of practices (Barney, 1991). More explicitly, unit organisationalambidexteritycanbecreatedbybeingexiblewithtimeallocationandfocusingattentiononhumanresources(Lepaket al., 2003; GibsonandBirkinshaw, 2004).Similarly, scholars have stated that versatile individuals have the motivation and ability topursue a range of apparently conicting opportunities, tackle conict and engage inparadoxical thinking(GibsonandBirkinshaw, 2004; SmithandTushman, 2005). Moreover,individual employees withvaluableknowledge, skills andexperiences areabletoperformmultiple functions and tasks, so that they act more like generalists than specialists (Leana andBarry, 2000). Importantly, GibsonandBirkinshaw(2004: 211) statedthat ambidexterity iscreated by encouraging individuals to make their own judgments as to how best divide theirtimebetweentheconictingdemandsforalignment andadaptability. Inotherwords, unitorganisational ambidexteritycanbepromotedwhenorganisationsusepracticestodevelopHPWS and organisational ambidexterityHUMAN RESOURCE MANAGEMENT JOURNAL, VOL 25 NO 1, 2015 82 2014 John Wiley & Sons Ltd.resourceexibilityintheiremployment, suchthathumanresourceshavethediscretionandmotivationtodevotetheir effortstoactivitiesassociatedwithexplorationandexploitation(Patel et al., 2013). Previous studies haverevealedthat rms needtoinvest inthehumanresourcespracticesthatboostemployeemotivation, skill, andadaptability(Adleret al., 2009:109) in order to help their employees achieve the diverse goals of the organisation (Wright andSnell, 1998). AlthoughresearchershaveexaminedtherelationshipbetweentheuseofHPWSand ambidexterity, the focus has been on contextual ambidexterity, knowledge combination andexchange, or adaptive capability(Wei andLau, 2010; Patel et al., 2013). While this line ofresearch is valuable in revealing the potential impact of HPWS on organisational ambidexterity,it offers limited insight into the use of rm-level HPWS to facilitate organisationalambidexterityat the lower levels of rms. Furthermore, it has beenassumedthat HPWSrepresentsaconduit forthedevelopment ofunit organisational ambidexterity; however, theactual mechanisms linking HPWS andunit organisational ambidexterity across multilevelanalysishaverarelybeenexamineddirectly.Building on HPWS perspective, knowledge and learning perspective, and the organisationalambidexterity approach, we identify that rm-level HPWS promotes the development ofunit-level employee human capital, and this in turn inuences the increase in unit organisationalambidexterity. Furthermore, across levels, rm-level social climate moderates the effect ofrm-level HPWSonunit organisational ambidexteritythroughunit-level employeehumancapital. The discussion leads to an integrative model presented schematically in Figure 1.Firm-level HPWS, unit-level employee human capital and unit-levelorganisational ambidexterityPrevious studies have focused mainly on macro-level HPWS, especially the macro-level impactsof HPWS on organisational ambidexterity (Prieto and Santana, 2012; Patel et al., 2013). Scholarshave theoretically and empirically suggested that macro-level HR practices are not useful to theFigure 1Anintegrativemultilevelmodelofhigh-performanceworksystems, humancapitalandunitorganisationalambidexterityFirm-Level Use of High-Performance Work Systems Unit-Level Employees Human Capital Unit Organisational Ambidexterity Unit Level Firm Level Firm-Level Social Climate Notes: The dashed line separates firm-level constructs and unit-level constructs. Arrows crossing the dashed line representcross-level relationships with the outcome variables.Yi-Ying ChangHUMAN RESOURCE MANAGEMENT JOURNAL, VOL 25 NO 1, 2015 83 2014 John Wiley & Sons Ltd.samedegree(orinthesameway)inallemployeegroups(Lepaket al.,2007).Correspondingtothese arguments, rm-level HPWSoffers a contextual cue for managers toformtheirperceptions and experience of the work system (Liao et al., 2009: 375). Research has found thatanindividuals strategicvalues of humancapital wereaffectedbymacro-level HPWS(e.g.DeleryandDoty, 1996; LepakandSnell, 1999). Moreover, anindividual employeeshumancapitalisassessedbythevalueofthatemployeesknowledge, skillsandabilitiestotherm(e.g. Wright et al., 2001; Subramaniam and Youndt, 2005). These valuable human resources notonlytendtobemoreeffectiveforacquiringandassimilatingnewknowledgefromvarioussources(BrownandDuguid, 1991), butalsohavethepotentialabilitytolearn, combineandapply new knowledge within the organisation (Wright and Snell, 1998; Taylor and Greve, 2006).Strategic HRM research has argued that HRM can be used to buy and make required employeeknowledge, skillsandabilities, whichcancreatevaluefortheorganisation(e.g. BeckerandHuselid, 1998; Lado and Wilson, 1994). Previous studies (e.g. Subramaniam and Youndt, 2005)have argued that the role of human capital is rarely examined explicitly as a mediator betweenHPWS and organisational ambidexterity at the unit level. In the context of the service industry,individualemployeesneedtolearntotackleconictingdemandsinordertosearchfornewknowledgeandskillsandreneexistingservicesandproductstomeetcustomersneeds. Atthesametime, individualemployeesneedtofullvariousrolesandcarryoutvarioustaskswithina certainperiodof time (FloydandLane, 2000; GibsonandBirkinshaw, 2004) asrequestedbycustomers. One of the keyfunctions of rm-level HPWS is toimprove anemployees knowledge and skills and to make managers more adaptive to deal withcontradictory dual demands (Lengnick-Hall and Lengnick-Hall, 2003). Firm-level HPWS is ableto assist various functional employees, such as customer service experts, to cooperate togetherat the unit level and subsequently engage in disperse exploratory and exploitative activities byintegratingtheiremployeecapabilities (Kaupplia, 2010). Forinstance, rm-level HPWScanhelpselectanddevelopservicetalentsattheunitlevel throughservicequality-focusedandexibility-focusedrecruiting, training, information sharing, andincentive systems. In thisinstance, information-sharing practices can assist unit employees to acquire and assimilate new,in-depthknowledgewithinanarrowrangeof restrictions (BrownandDuguid, 1991) andsituated in diverse knowledge domains to both pursue ambidextrous activities (Kang and Snell,2009).Firmsalsouserecruiting/staffingpracticestohelpunitemployeesupgradetheirskillsandabilities tobeambidextrous, throughtheselectionof employees withapotential andopenness to learning new skills and the selection of employees with current competency to tthe current job requirements (Lepak and Snell, 1999). Firms can develop ambidextrousemployees at the unit level through intensive training to improve current job-related skills (BaeandLawler, 2000; Guthrie, 2001) andthroughskill/knowledge-basedincentivesystemsthatencourage employees at the unit level to gain new knowledge and skills beyond their currentjobs (Guthrie, 2001). The above-mentioned rm-level HPWS, in turn, boosts individualmanagershumancapitalbyhelpingmanagershostcontradictorythinking,fullcurrentandnew job roles, rene existing knowledge, and acquire new knowledge. As rm-level HPWS ismorelikelytoaffecttheunit-level employeehumancapital (asdiscussedearlier), weexpectthat unit-level employee human capital will mediate the inuence of rm-level HPWS on unitorganisationalambidexterity.Hypothesis1: Firm-levelHPWSrelatestounit-levelemployeehumancapital.Hypothesis 2: Unit-level employee human capital mediates the relationship betweenrm-levelHPWSandunitorganisationalambidexterity.HPWS and organisational ambidexterityHUMAN RESOURCE MANAGEMENT JOURNAL, VOL 25 NO 1, 2015 84 2014 John Wiley & Sons Ltd.Firm-level HPWS and unit organisational ambidexterity: moderating effect of rm-levelsocial climateIntegrating social capital perspective (e.g. Nahapiet and Ghoshal, 1998) and HPWS perspective(Wright andMcMahan, 1992; Huselid, 1995), wearguethat rm-level HPWSmaycreateapositivesocialclimateatthermlevel, whichthenexertsapositivetop-downeffectonunitorganisational ambidexterity. As the social capital perspective suggests, social climate within arm is vital for increasing interaction and information exchanges (Mayer et al., 1995; Nahapietand Ghoshal, 1998). Trust also enhances the possibility of an exchange of information and ideasbecause trustworthy social conditions improve actors beliefs that an existing exchange will leadto later reciprocities (Coleman, 1990). Asocial climate of trust supports the exchange of valuableideas among employees that will, in turn, lead to greater innovation and rm growth (CollinsandSmith, 2006).Astrongrm-level social climateof trust, cooperation, andsharedcodes andlanguagemayalsoact asasituational enhancer (Howell et al., 1996), andcanfurther strengthenthepositive inuence of rm-level HPWS on unit organisational ambidexterity through unit-levelemployeehumancapital. Thereareat least tworeasons for this statement. First, astrongrm-level social climate provides a specic context for rm-level HPWS to promote anindividual employees awareness of creativesolutions andinnovativethinking, whichwillupgradetheunit-level employeehumancapital intheworksetting. Hofmannet al. (2003)demonstrated that the social climate within a rmemphasises or de-emphasises certaincontent-specic role expectations for employees as theyrespondtothe inuence of theirleaders. Wearguethatwhenanemployeeisembeddedinarm-widecontextcharacterisedbyahighlevel of trust, cooperationandsharedunderstanding, theemployeeis likelytoaccessandbecomeexposedtovariousnewknowledgedomainsforambidextrousactivities(SubramaniamandYoundt, 2005)becausethecontentofrm-levelHPWSresonateswiththerms social climate. As a result, individual employees are more likely to upgrade theirknowledge, skills and abilities as requested by the rm-level HPWS, and thus are more readytoparticipateincreatingasocialclimatethatcontributestothermsgoalsandvalues. Theresult is aboost tointra-unit knowledge exchange andcreative thinking, whichcreates apositiveenvironmentforunitorganisationalambidexterity. Ontheotherhand, whenthereisaweaksocialclimateoftrust, cooperation, andsharedcodesandlanguageatthermlevel,individualemployeesmaybelesseffective, asthesocialcontextfosteredinthatrmbytheHPWSmaygiverisetocognitivedissonanceamongthermsemployees. Thus, knowledgeexchangewithinthermmaybestied, whichwouldnegativelyaffect unit organisationalambidexterity at the unit level. HPWS may shape a managers motivation to pursue a varietyofdifferentgoalsandtoengageinambidextrousbehaviours. Yettheabilityoforganisationstoleveragetheir HPWSeffectivelyandtomakeuseof theunit-level humanresources byimprovingunit employees knowledge, skillsandabilitiesisconstrainedbytheuncertaintyandequivocalityassociatedwiththewaysinwhichexplorationandexploitationneedtobepursued (Floyd and Lane, 2000). A rms social climate may help unit employees reduce suchuncertainty and equivocality by fostering the exchange of information, opinions andjudgements with other colleagues (Mayer et al., 1995). It stimulates a comprehensiveassessment of divergent goals as well as debate about expected outcomes (Maurer et al., 2011)thatcontributestoimprovedtransparencyandcommonexpectationsaboutthefullmentofexploratoryandexploitative roles andbehaviours. Giventhese benets froma rm-levelsocialclimate, wearguethatarmssocialclimatecanfurtherreinforcesuchclimateeffectsonunit organisational ambidexterity.Yi-Ying ChangHUMAN RESOURCE MANAGEMENT JOURNAL, VOL 25 NO 1, 2015 85 2014 John Wiley & Sons Ltd.Asocial climateimproves mutual learningprocesses amongmanagers andmayfurtherimprovethecompetencyaswellaswillingnessofunitemployeestoengageinambidextrousbehaviours. Without a shared code and language, individual managers may nd it difficult toapply informational resources and knowledge within their own activities because they will eachtend to develop specic languages, world views and thought worlds (Cox, 1993). However, ina social climate that is characterised by heightened trust, collaboration and shared languages,it is easier toassimilate andapplydivergent ideas andknowledge fromother colleagues(Nahapiet and Ghoshal, 1998). By providing a useful context for combining knowledge acrossinternal organisational boundaries, asocial climateimproves theeffectiveness of HPWSincontributing to unit organisational ambidexterity, through the advancement of employeehumancapital.A strongsocialclimatemeansalowvulnerabilityonthepartofrmcontext(BowenandOstroff, 2004). Such a climate offers a safe context for unit employees with valuable knowledge,skills and abilities to perform various roles and tasks. The unit employees are likely to perceivethe rm context as facilitating their efforts to achieve unit ambidextrous behaviour and to feelmore comfortable directing their efforts towards fostering unit ambidextrous behaviours, suchas engaging in contradictory thinking and undertaking various roles (Dirks and Ferrin, 2001).Second, as noted earlier, unit-level employee human capital promotes the intra-unitexchangeofinformationandknowledgebycreatingaconstructivesocial climatewithintherm. Wereasonthat arm-level social climateof trust, cooperation, andsharedcodesandlanguage, which facilitates an inter-unit ow of information and knowledge, complements theintra-unit exchange by enhancing employees human capital in order to promote unitorganisational ambidexterity. Whenthereis ahighlevel of intra-unit exchange, individualemployees are likely to be exposed to and have access to a range of internal knowledge. In sucha case, individual employees at the entry point of knowledge from other parts of the rm aremore capable of interpreting and understanding the acquired knowledge either using their ownenhanced expertise or using that of other employees that they can easily locate within the rm.Moreover, as newlyacquiredknowledgeis quicklysharedacross therm, it canbemoreefficiently incorporated into routine activities, and understood and applied in more depth andbreadth within the unit in the context of organisational ambidexterity. Thus, fromtheknowledge and learning perspective (Kang and Snell, 2009), a strong rm-level social climatecreatesaboundaryconditionthat ampliesthepositiveeffect of rm-level HPWSonunitorganisationalambidexteritythroughunit-levelemployeehumancapital. Incombination, theaboveargumentssuggest:Hypothesis 3: Firm-level social climate moderates the positive relationship betweenrm-level HPWS and unit organisational ambidexterity through the indirect effect ofunit-level employee human capital, such that the effect is stronger when there is a positivesocialclimateatthermlevel.METHODSample and procedureWetestedour hypotheses inasampleof bankingrms inTaiwan. Thebankingsector isespeciallysuitablefor our studybecauseunit-level organisational ambidexterityis vital inassistingbankingrmstoserveexistingproducts/servicesandrenewproducts/servicesorprocesses in order to survive in the changing environment. We chose the branches because theyare geographically diverse, and are independent decision units in the areas of types of productsHPWS and organisational ambidexterityHUMAN RESOURCE MANAGEMENT JOURNAL, VOL 25 NO 1, 2015 86 2014 John Wiley & Sons Ltd.and services provided, markets and customer base. Each branch has (a) its own seniormanagement team, (b) its own costs and revenues, and (c) its own responsibilities with respectto various traits of their operations, such as following exploration and exploitative innovation.Each branch offers a variety of products, services andsales, including enterprise andindividual loans, mortgages, savings and insurance, corporate banking, corporate leasing, etc.Inaddition, eachbranchoperates inmarkets withvarious environmental dynamismandcompetitiveness which condition needs branches pursuing diverse innovations (Han et al., 1998;Jansenet al., 2008). Studiesintoambidexterityinthebankingsector(e.g. Yuet al., 2013)haverevealed that banks, both at branch and unit levels, can achieve service and sales ambidexterityby using HR practices, such as incentive systems, to motivate branch managers and employees.In early 2010, we sent surveys to selected participants together with a supporting letter fromthechief executiveofficer (CEO) of therm. Thesurveys weredevelopedinEnglishandtranslatedintoChinese usingthe back-translationmethod(Brislin, 1980). Altogether, 1,490managersand4,000employeesweresurveyedacrossatotal of 808branchesof 58bankingrms. After 4 weeks, with three rounds of reminders, we received responses from 616 managersand2,998employees.Our nal sampleincluded808branchesin58rms, withresponsesfromatotal of 536managers(35.9per centresponserate)and2,887employees(72.1per centresponserate). Wecompared rms included in our nal sample with rms that had been excluded, and did notndsignicant differencesbetweenthemintermsof thenumberof full-timeemployeesornumber of units of banks. Following Armstrong and Overton (1977), we also compared early(rst10per cent)andlate(last10per cent)respondentstoassessnon-responsebiasoneachdimension of management innovation. No signicant differences emerged across thesedimensions. Ofthemanagersinoursample, 63per centwerefemale. Theiraverageagewas35.1years.Onaverage,themanagershadworkedinthermfor12.3yearsandhadbeenintheir current job for 4.24 years. Of the employees in our sample, 59 per cent were female. Theiraverage age was 37 years, with an average organisational tenure of 18 years. Of the managersinoursample, 63per centwerefemale.Toalleviatethecommonmethodbias, weobtaineddatafrommultiplesources. First, 536managers from 808 branches of 58 banking rms were chosen randomly from a list providedby the rm. These participants were told the studys purpose and were asked to rate unit-levelambidexterity. Second, the 2,887 employee respondents ratedtheir ownhumancapital ingeneral. Third, twoseniormanagersfromeachrmsheadquartersratedrm-level HPWS.Fourth, threeemployeesfromeachrmsheadquartersratedtherm-levelsocialclimate.Weconductedaseriesofconrmatoryfactoranalyses(CFA)toexaminethediscriminantvalidity of these constructs. Like Liao et al.s (2009) one-factor solution of rm-level HPWS, wefound that the one-factor model of rm-level HPWS t the data better [2(72) = 735.38,non-normedt index(NNFI) = 0.92, comparativet index(CFI) = 0.92, incremental t index(IFI) = 0.92, and root mean square error of approximation (RMSEA) = 0.05]. In addition, the onehigher-factor of unit-level organisational ambidexterity represents the construct well [2(65) = 537.91, NNFI = 0.91, CFI = 0.91, IFI = 0.91, andRMSEA = 0.05]. Taken together, theseresultsprovidedevidenceofdiscriminantvalidity.MeasuresUnit organisational ambidexterityUnit organisational ambidexterity was measured using a12-itemscale adapted fromLubatkin et al. (2006). The 12 items showed good reliability( = 0.90).Becauseofthehighlevelofagreementbetweenraterswithinthesameunit[meanrwg = 0.92, intraclass correlationcoefficient (ICC)(1) = 0.23, ICC(2) = 0.74], we averagedtheYi-Ying ChangHUMAN RESOURCE MANAGEMENT JOURNAL, VOL 25 NO 1, 2015 87 2014 John Wiley & Sons Ltd.responses of managers within each unit to create an aggregated measure of unit-levelorganisational ambidexterity. Weusedthemainstudydata536managers ratingsof unitorganisational ambidexterity to conduct a CFA. The CFA results showed that the one-factormodel t thedatawell [2= 57.35, df = 9, p < 0.01, RMSEA = 0.04, CFI = 0.92, goodness-of-tindex (GFI) = 0.92, Tucker-Lewis index (TLI) = 0.90]. The results indicated that the measure wasvalidandreliable. Inthis study, unit organisational ambidexterityshowedgoodreliability( = 0.81). Following previous studies (e.g. He and Wong, 2004; Lubatkin et al., 2006), this studycalculated the unit organisational ambidexterity as the multiplication score of exploration andexploitation, andtheabsolutedifferencescoreofexplorationandexploitation.Firm-level HPWSWeusedthe37-itemHPWSthat Liaoet al. (2009) developedfromextantliterature especially for banking rms (Zacharatos et al., 2005). The HPWS was adaptedspecicallytohelpambidextrous managers improveserviceperformanceonaseven-pointscale. The HPWS scale has eight unique dimensions for HR systems. Two senior managers fromeach rms headquarters rated the items that were used to direct individual managers in eachbanking rm. We followed previous research and used the eight dimensions to create an indexof HPWS (Sun et al., 2007; Takeuchi et al., 2007; Chuang and Liao, 2010) ( = 0.92). Tests showedthatmanagersfromthesamermhadahighlevelofagreementregardingrm-levelHPWS[meanrwg = 0.90, ICC(1) = 0.24, ICC(2) = 0.72].Firm-level social climateFollowingpreviousresearch, weviewedarmssocial climateasconsisting of trust, cooperation, and shared codes and language (Mayer et al., 1995; Mayer andDavis, 1999; Chatman and Flynn, 2001; Collins and Smith, 2006). Trust was measured using a12-itemscaleadaptedfromMayerandDavis(1999). Firm-level social climatewasratedbythree employees at each rms headquarters. The measures showed good reliability ( = 0.93 fortrust; = 0.85 for cooperation; and = 0.91 for shared codes and language). Tests revealed thatthethreeemployeesfromeachrmsheadquartershadahighlevel ofagreementregardingrm-level social climate: trust [mean rwg = 0.91, ICC(1) = 0.22, ICC(2) = 0.71], cooperation[mean rwg = 0.91, ICC(1) = 0.22, ICC(2) = 0.71], and shared codes and language [meanrwg = 0.90, ICC(1) = 0.23, ICC(2) = 0.72]. We, therefore, averaged the responses of the employeeswithin each rms headquarters to create aggregated measures of rm-level trust, cooperation,andsharedcodesandlanguage.Unit-levelemployeehumancapital Weusedave-itemhumancapitalscaledevelopedbySubramaniamandYoundt(2005)onaseven-pointscale. Theitemswereadaptedtodescribeservice-relatedknowledge, skillsandabilities( = 0.89). Testsshowedthatdirectorsfromthesame rm had a high level of agreement regarding unit-level employee human capital [meanrwg = 0.90, ICC(1) = 0.23, ICC(2) = 0.73].Control variablesFirst, we controlled for manager age and tenure within the rm, which areexpected to positively relate to organisational ambidexterity (Tushman and OReilly, 1996: 27).Second, we controlled for a managers tenure in his or her current function, which is related toanincreasinglevel of specialisationandtherefore is expectedtonegativelyrelate tounitorganisational ambidexterity (cf. Gibson and Birkinshaw, 2004: 211). Third, we controlled for levelof education because increasing levels of education are linked with an increasing cognitive abilitytoprocessinformationandtolearn(Papadakiset al.,1998).Thismaypositivelyrelatetounitorganisationalambidexterity(Adleret al., 1999: 51). Educationaleffectsincludedtwodummyvariables: onereectingmanagerswithamastersdegreeor higher, andanother reectingHPWS and organisational ambidexterityHUMAN RESOURCE MANAGEMENT JOURNAL, VOL 25 NO 1, 2015 88 2014 John Wiley & Sons Ltd.managers with bachelors degrees. Managers with degrees below the bachelors level were thereferencegroup(Momet al., 2009). Fourth, wecontrolledsizeeffect(numberofsubordinatesunder a manager) (Lewin et al., 1999) and environmental uncertainty because these are linked toa rms motivation to adjust to changing resource conditions (Lubatkin et al., 2006).RESULTSTable 1presentsdescriptivestatisticsandcorrelations.Wetestedthehypothesesusinghierarchical linearmodelling(HLM) analyses. Wegrandmean-centredtheinterpretationfor theHLMresults, whichensurestheLevel 1affectsarecontrolledforduringtestingoftheincremental effectsoftheLevel 2variables, andreducesmulticollinearity in Level 2 estimation by lessening the correlation between the Level 2 interceptandslopeestimates(HofmannandGavin, 1998).BeforeconductingHLManalyses, weexaminedthedegreeof between-groupvarianceinindividual-level humancapital andunitorganisational ambidexterity. Resultsofnull modelsrevealedthat26per centofthevarianceinindividual-levelhumancapital, 32per centofthevarianceinunitorganisational ambidexterity(multiplicationofexplorationandexploitation)and24per cent of thevarianceinunit organisational ambidexterity(absolutedifferenceofexploration and exploitation) reside between individuals (the grouping variable), respectively.Thechi-squaretestsrevealedthat thebetween-individual variancesweresignicant; i.e. theintercepttermssignicantlyvariedacrossindividuals.Hypothesis1predictsthatrm-level HPWSrelatestounit-level employeehumancapital;Hypothesis 2 predicts that unit-level employee human capital mediates the relationshipbetweenrm-levelHPWSandunitorganisationalambidexterity. TheresultsinModel1andModel 2 revealed that rm-level HPWS was signicantly related to unit-level employee humancapital ( = 0.42, p < 0.01), unit-level employee human capital mediated the relationship betweenrm-level HPWSandunit organisational ambidexterity( = 0.53, p < 0.01, Model 2), andtheeffect of rm-level HPWS remained signicant but was reduced in magnitude ( = 0.22, p < 0.01,Model2). Thus, Hypotheses1and2weresupported.Hypothesis 3 proposes a positive cross-level interaction between the rm-level social climateandunit-levelemployeehumancapitalinpredictionsofunitorganisationalambidexterity.InModel 5, we regressed the slope estimates for unit-level employee human capital obtained fromLevel 1onsocial climateat Level 2totest this interaction(BrykandRaudenbush, 1992).Moreover, as spurious cross-level interaction may be found if between-groups interactions arenot controlledfor (HofmannandGavin, 1998), we includedthe interactions of rm-levelHPWS rm-level social climate at Level 2. The results revealed that the interaction ofrm-level HPWS rm-level social climate was not signicant, whereas the cross-levelinteractionwas signicant ( = 0.42, p < 0.01, Model 5). These results provide support forHypothesis3andsuggest that apositiverm-level social climateenhancedtheinuenceofunit-levelemployeehumancapitalonunitorganisationalambidexterity.Toexaminewhether theinteractioneffect was mediatedbytheindividual-level humancapital speciedinthisstudy, wefollowedEdwardsandLamberts(2007)procedure, whichintegrates moderated regression analysis and path analysis to analyse comprehensivelysimultaneous moderation and mediation. In Table 2, for the unit-level employee human capitalpath, the indirect effect was 0.20 (p < 0.05) at high rm-level social climate and 0.10 (p < 0.05)at low rm-level social climate, and the difference was signicant [0.20 (0.10) = 0.30, p < 0.05].Theresults, thus, suggest that theindirect effect of rm-level HPWSonunit organisationalambidexterity, through unit-level employee human capital, was stronger at a higher rm-levelYi-Ying ChangHUMAN RESOURCE MANAGEMENT JOURNAL, VOL 25 NO 1, 2015 89 2014 John Wiley & Sons Ltd.TABLE1Means,standarddeviationsandcorrelationsMeanSD1234567891011121.Managerage36.195.672.Managertenureintherm11.972.270.103.Managertenureinthecurrentjob3.450.530.21***0.16***4.Size(logofmanagerssubordinates)1.220.150.15**0.13**0.035.Education1(masterabove)0.300.450.000.000.19***0.12**6.Education2(bachelor)1.000.000.21***0.010.030.010.007.Environmentaluncertainty5.571.150.23***0.08**0.030.15***0.12**0.038.Unit-levelemployeehumancapital4.011.190.030.13**0.01*0.15***0.13**0.010.10**9.Firm-levelHPWS4.531.030.010.07*0.05*0.16***0.14**0.000.040.08*10.Firm-levelsocialclimate4.080.590.020.000.020.010.010.000.08**0.11**0.49***11.Unitorganisationalambidexterity(multiplicationofexplorationandexploitation)16.754.180.010.010.000.010.000.010.06**0.29***0.07*0.28***12.Unitorganisationalambidexterity(absolutedifferenceofexplorationandexploitation)0.800.630.010.020.000.000.030.000.05**0.010.010.16**0.17***p