StrategicResponsetoInstitutionalInfluencesonInformationSystemsOutsourcing SoonAng;LarryL.Cummings OrganizationScience,Vol.8,No.3.(May- Jun.,1997),pp.235-256. StableURL: http://links.jstor.org/sici?sici=1047-7039%28199705%2F06%298%3A3%3C235%3ASRTIIO%3E2.0.CO%3B2-N OrganizationScienceiscurrentlypublishedbyINFORMS. YouruseoftheJSTORarchiveindicatesyouracceptanceofJSTOR'sTermsandConditionsofUse,availableat http://www.jstor.org/about/terms.html.JSTOR'sTermsandConditionsofUseprovides,inpart,thatunlessyouhaveobtained priorpermission,youmaynotdownloadanentireissueofajournalormultiplecopiesofarticles,andyoumayusecontentin theJSTORarchiveonlyforyourpersonal,non-commercialuse. Pleasecontactthepublisherregardinganyfurtheruseofthiswork.Publishercontactinformationmaybeobtainedat http://www.jstor.org/journals/informs.html. EachcopyofanypartofaJSTORtransmissionmustcontainthesamecopyrightnoticethatappearsonthescreenorprinted pageofsuchtransmission. TheJSTORArchiveisatrusteddigitalrepositoryprovidingforlong-termpreservationandaccesstoleadingacademic journalsandscholarlyliteraturefromaroundtheworld.TheArchiveissupportedbylibraries,scholarlysocieties,publishers, andfoundations.ItisaninitiativeofJSTOR,anot-for-profitorganizationwithamissiontohelpthescholarlycommunitytake advantageofadvancesintechnology.FormoreinformationregardingJSTOR,[email protected]. http://www.jstor.org TueJan808:44:382008 Strategic Response to Institutional Influences on Information Systems Outsourcing Soon Ang Larry L. Cummings Information Management Research Center (IMRC), Nanyang Business School, Nanyang Technological University, NanyangAvenue 2263, Republic of Singapore Professor in Strategic Management and Organization, University of Minnesota, 271 19th Avenue South, Minneapolis, Minnesota 55455 I nstitutional theory tends to discount the ways by which organizations can be strategically proactive in their adaptation to environmental and institutional influences. The central contribution of this paper to organization theory is an empirical investigation of individual firm strategic responses to institutional influences at a time when hypercompetition has altered the competitive dynamics of the industry. The focus of the paper is on banking industry, which is both hypercompetitive and highly institutionalized, and on the strategic appropriation of information responses of individual banks to institutional influences on outsourcing of information technology capabilities. The paper has impor- tant implications for the understanding organization change in terms of institutional strategic adaptation theories. Arie Y. Lewin Abstract assume organizations to be passive players. In fact, This paper underscores the importance of examining strate- veneration of institutional norms and rules for orga- gic response to institutional influences in light of hypercom- nizational behavior is so deeply ingrained that re- petition. Focusing on the banking industry, which is hyper- searchers often disregard individual organizations' competitive and highly institutionalized, affords a unique ability to respond proactively, creatively, and strategi- opportunity to understand how individual corporations in cally to institutional influences. We investigate one such an industry respond strategically to institutional pres- such response, the strategic response of individual sures. banks to institutional influences on information system We examine critical contingencies arising from hypercom- (IS) outsourcing. We argue that in hypercompetitive petition that moderate institutional influences on informa- environments, institutional theory must accommodate tion systems outsourcing in commercial banks. Using data strategic responses of individual organizations. from 226 banks and hierarchical moderated regression analy- ses, we show that the propensity of banks to conform to or resist institutional pressures depends on the nature of institu- tional pressures, perceived gain in production economies, The Institutional Cage of Banking financial capacity to resist institutional influences, and trans- Commercial banks in the United States operate in action cost considerations. highly institutionalized environments. Historically, (Institutional Influences; Outsourcing; Managerial Dis- stringent banking legislation1 restricted operations and cretion; Strategic Managerial Action) suppressed competition. First, rules on chartering and branching limited banks' ability to compete with one another geographically. No new bank could set up Introduction business without acquiring a national or state charter. According to Oliver (1991), Perrow (1985), DiMaggio State law often prohibited intrastate branching by (1988), and Scott (1995), institutional theorists often state-chartered banks, and federal law required na- 1047-7039/97/0SO3/0235/$05.00 Copyright 01997. Institute for Operations Research and the Management Sciences ORGANIZATSICOINENCE/VO8~, .NO.3, May-June 1997 235 SOON ANG AND LARRY L. CUMMINGS Strategzc Response to Irzstltlitlonal I~lfZuenceso n Infolmatzolz Sj>sternsO utso~~rcfng tional banks to follow whatever rules individual states enhancing strategies (Office of Technology Assessment established (Baer and Mote 1992, Burns 1988). 1984, 1987). Continuing technological advances in com- Second, restrictions on the services banks could offer puters and communications steadily reduced transac- limited competition between banks and other financial tion costs of banking. Technological innovations in institutions. Commercial banks monopolized demand financial products also transformed financial services. deposits. However, they were not allowed to deal in Sophisticated cash management, securitization of mort- corporate securities, underwrite new corporate issues, gage loans, massive trading in government securities, or engage in commercial activities. Periodically, federal and money market mutual funds for ordinary citizens regulators would visit banks to ensure they conformed became possible with IT handling the speed and enor- to rules and regulations. Thus, incumbents in the bank- mous volume of financial transactions (Steiner and ing industry thrived in market niches carved out by Teixeira 1990) while telecommunications spread infor- bank regulation (OECD 1992). mation instantaneously, linking formerly separate fi- nancial markets into an integrated world market. Hypercompetitive Pressures in Banking In light of increasing hypercompetitive pressures un- In recent years, regulation that formerly had fended off der which banks are operating, we contend that banks competition from other financial institutions weak- may not simply acquiesce to institutional influences. ened. Banks became immersed in hypercompetition: Rather, whether they conform to or resist such influ- rapidly escalating competition based on new and con- ences depends on their responses toward an increas- tinually shifting product or geographic markets, fre- ingly deregulated and competitive environment. quent entry of unexpected competitors, radical re- We report a study on bank's response to institutional definition of market boundaries, rapidly changing influences on IS sourcing. First we discuss IS sourcing technologies, and short product life cycles (D'Aveni in the banking industry, and describe institutional 1994). norms that affect IS sourcing strategy. We then de- In banking, novel substitutes escalated competition. velop hypotheses on how the relationship between Securitization enabled businesses to use marketable institutional influences and IS sourcing is moderated securities to bypass bank loans as sources of funds. by strategic contingencies arising from hypercompeti- Competition intensified when nonbank corporations tive pressures. After the theoretical development, we were permitted to acquire banks. By spinning off either describe the study sample, variables, and method, and deposit-taking or commercial loan operations from ac- discuss the estimation process and results. Finally, we quired banks, firms could carry on the remaining activi- summarize the results and corresponding theoretical ties free of regulatory restraints. By the late 1980s, and managerial implications. such nonbank banks had breached the wall that for- merly protected banks from outside competition. Information System Outsourcing in Banking At the same time, banks found ways of expanding As corporations searched for ways to grow and main- traditional business domains. With the relaxation of tain their competitive edge, outsourcing emerged as a regulation of bank holding companies, banks were dominant organizational strategy for achieving those legally permitted to compete with nonbanks in mort- goals. In outsourcing, firms7 orientations toward inter- gage banking, discount brokerage services, financial nal action gives way to greater dependence on external counseling, and data processing services (Burns 1988). service providers (Kanter 1989, Quinn 1992). Unlike Hypercompetitive pressures also arose from redefini- the old model of organization characterized by hierar- tion of geographic boundaries of the industry. Large chical ownership and avoidance of external depen- money-center banks increasingly explored foreign dence, new models of organizations are characterized countries for funds and loans, thereby avoiding restric- by networks of lateral and vertical interlinkages across tive domestic regulation. Those banks in turn intro- firms (Nohria and Eccles 1992). Outsourcing thus epit- duced smaller banks to global banking by selling them omizes a more open and networked form of organizing participation in their foreign loans. At the same time, for resources. large banks began to securitize on a global scale by One central value-chain activity that companies have shifting from domestic banks or securities markets to outsourced is the information services function (Quinn foreign banks and international securities markets 1992). A Frost and Sullivan Market Intelligence survey (White 1993). found that more than 50% of all companies with IT Information technology (IT) transformed banking by budgets in excess of $5 million were outsourcing or opening up new cost-saving, risk-reducing, and profit- actively considering outsourcing IS. Within the banking 236 ORGANIZATSIOCNIE NCE/VO8I,. NO.3, May-June 1997 SOON ANG AND LARRY L. CUMMINGS Strategic Response to Institutional Influences on Information Systenzs Outsou~.cing industry, the figure jumped to 85% (Fortune, Decem- Bank Administration Institute's Automation Alterna- ber 12, 1994). tive publication (Bank Administration Institute 1980) and American Bankers Association National Opera- tions Surveys (American Bankers Association 1981, Institutional Pressures and Information 1986, 1990) and share innovative bank practices, in- Systems Outsourcing cluding IS sourcing. At its core, institutional theory seeks to explain homo- Federal regulators exert substantial influence on geneity of organizational forms and practices. As dis- bank practices and operations. The Federal Reserve cussed by DiMaggio (1988), the role of institutional Bank and Office of the Comptroller of Currency con- influence is particularly powerful in explaining organi- duct periodic examinations to ensure that banks con- zational phenomena in regulated industries where well form to myriad rules and regulations. Included in the laid-out rules, structures, external regulation and prac- bank examination is a bi-annual audit of banks' elec- tices govern organizational forms and operations. tronic data processing (EDP) or IS practices. Among According to DiMaggio and Powell (19831, organiza- other concerns, the audit assesses IS sourcing and tional fields that exert institutional influences on orga- recommends alternatives if current arrangements do nizations evolve through a process of institutional defi- not provide banks with relevant and pertinent informa- nition or "structuration." The structuration process tion for decision making, or are inadequate to secure consists of four parts: an increase in the extent of financial assets of bank customers (Federal Financial interaction among organizations in the field, the emer- Institutions Examination Council 1992). gence of sharply defined interorganizational structures Institutional theory has become an important theo- of domination and patterns of coalition, an increase in retical framework for investigating diffusion of organi- the information load with which organizations in the zational practices (Galaskiewicz and Wasserman 1989, field must contend, and the development of a mutual Tolbert and Zucker 1983). However, firms may not awareness among participants in a set of organiza- simply conform to institutional pressures. In a critique tions that they are involved in a common enterprise of institutional research, Oliver (1991) stressed the (DiMaggio 1983, cited by Powell and DiMaggio 1991, inadequate attention given to strategic responses. In p. 65). For example, in a study of the diffusion of IS another essay, Perrow (1985) argued that instead of outsourcing among Fortune 500 firms, Loh and Venka- emphasizing only the "taken-for-granted character traman (1992) found that the much-publicized Kodak- of institutional rules, myths and beliefs as shared social IBM outsourcing arrangement legitimated the practice reality," institutional theory must accommodate the of IS outsourcing among Fort~lne5 00 firms and stimu- role of organizational self-interests and active agency. lated mimicking of outsourcing practices among large In line with this argument, we contend that despite organizations. Thus, according to institutional theory, the strong influences from peer banks and the federal the major impetus toward homogeneity of such organi- examiners, banks may enact different strategic re- zations is the structuration process of institutional in- sponses to institutional pressures for IS sourcing. fluences arising from external constituents such as key Specifically, given hypercompetitive pressures to im- suppliers, resource and product consumers, regulatory prove performance, we contend that conformity to agencies, and other organizations that produce similar institutional influence for IS outsourcing will be contin- services or products (DiMaggio and Powell 1983). gent on economic factors such as perceived economic Within the banking industry, influences on IS out- gain from conformity, financial capacity to resist insti- sourcing come from both peer banks and federal regu- tutional influence, and transaction costs implied by lators. Peer banks exert considerable influence on each acquiescence to institutional pressures. other because of tight professional networks formal- ized by memberships in regional and national bank associations. At the regional level, banks affiliate with Moderators of Institutional state associations such as the Michigan Bankers Asso- ciation, the Minnesota Bankers Association, and the Conformity Independent Community Bankers of South Dakota. At Figure 1 is the research model for our study. Institu- a national level, banks belong to the American Bankers tional influence is shown as affecting a bank's IS sourc- Association, and Independent Bankers Association of ing strategy. This relationship is strengthened (positive America. Associations hold regular meetings, conduct sign) or weakened (negative sign) by individual banks' IT usage surveys, publish IT guidelines such as the perceptions of gain in production economies; financial ORGANIZATISOCNIE NCE/VO8I,. NO. 3, May-June 1997 237 SOON ANG AND LARRY L. CUMMINGS Strategic Response to I~zstit~~tioInnafllu ences on Infotmation Systenzs 0utsour.cing Figure 1 Research Model duced operating expenses by $150 million in 1991 Financial (American Banker 1991). We therefore expect Perceived Gain Capacity to in Production Resist Economies Conformity (PRODCOST) (SUCK) HI. The greater the external production cost advan- tage, the stronger the relationship between instit~ltional IRTORRIATION in$?uencea nd IS outsourcing. IRTLUENCE SYSTEMS .Peer 'Federal Regulator Financial Capacity to Resist Conformity Asset Specificity Tecnological (SPECIFIC) U~lcertainity Organizations may not conform to institutional influ- ences if they have the financial capacity or working Functional Supplier Complexity Presence capital to resist external pressures. When external (COMPLEX) (SUPPLIER) pressures advocate outsourcing, managers may resist by TRANSACTION COSTS deploying slack resources to amass IT resources inter- *H7: Size is hypothesized to affect the interaction of institutional nally. That reaction is consistent with Jensen's (1989) influence and each of the moderators on I/ S outsourcing. observation that managers prefer to deploy slack to- ward asset capitalization rather than distribute it as dividends to shareholders because increased asset capi- capacity to resist institutional influence, and transac- talization enhances the social prominence and political tion costs associated with a particular sourcing strategy. power of senior executives (Baumol 1959). Investments in IT represent major asset capitalization. Deemed "crown jewels" in banks (Huber 1993), IT symbolizes Perceived Gain in Production Economies firm growth, advancement, and progress. Because in- According to Oliver (1991), when an organization an- vestments in IT can promote social prominence, they ticipates that conformance will enhance economic fit- make managers more likely to resist outsourcing pres- ness, acquiescence will be the most probable response sures. to institutional influence. In IS sourcing, we expect In contrast, when slack resources are low, managers banks to adopt IS outsourcing when it affords substan- are likely to conform to institutional outsourcing pres- tial gains in production economies or operational cost sure (Sutton and D'Aunno 1989). Anxiety is provoked savings. In other words, if external service providers during low slack because financial distress is often offer advantages in production cost economies over attributed to managerial incompetence and organiza- internal IS services, we expect banks to conform more tional ineffectiveness (Meindl et al. 1985, Pfeffer and readily to institutional influences for outsourcing. Salancik 1978, Whetten 19801, and firms react by The emphasis on production costs is exacerbated by downsizing to reduce costs and recoup losses (Rubin hypercompetitive and economic forces facing the bank- 1977, Tomasko 1987, Warren 1984). ing industry that threaten corporate survival. Accord- In 1992, the General Accounting Office found that ing to Steiner and Teixeira (19901, a significant 50% of poorly performing banks outsourced to generate a bank's noninterest expense is for IT operations that short-term financial slack (GAO 1992). Outsourcing support funds movements, such as transaction process- contracts were drawn whereby the service providers ing, check processing, cash management, and data purchased IT assets at substantially higher prices than networks management. As heavy consumers of IT re- the market value in return for higher servicing fee sources, banks are struggling with spiraling costs of amortized over an 8- to 10-year period. Outsourcing maintaining a progressive IT infrastructure. Recent enabled banks to maintain capital, defer losses on asset surveys on IT management have shown "cost contain- disposal, and show an instantaneous increase in finan- ment of IS" to be the dominant concern in financial cial value on the balance sheet. Accordingly, when institutions (American Banker 1988, 1990, 1991, 1992). slack resources are low, we expect firms to conform to Outsourcing has been extolled as a means of cutting IT external pressures to outsource. costs (Lacity and Hirschheim 1993). Banks purportedly reap an average of 1520% operational cost savings from outsourcing, thus substantially reducing financial H2. The lower the level of slack resources in a bank, outlays on IT (Gillis and Biafore 1993). For example, the stronger the relationship between instit~ltionailn fluence First Fidelity Bankcorp with $29 billion in assets re- and IS outsourcing. Conversely, the higher the level of 238 ORGANIZATSICOINE NCE/V8O, ~N.O .3, May-June 1997 SOON ANG AND LARRY L. CUMMINGS Stmteg~cR esponse to Inst~tutlonaIln fluences on Infotmatzon Systems Outsourcing slaclc resources in a bank, the weaker the relationship H3. The greater the specificity of IS assets, the weaker between institutional influence and IS outso~ircing. the relationship between instit~~tionainl fluence and IS outsourcing. Transaction Costs Transaction costs incurred by conforming to institu- tional pressures also can affect organizational resis- Functional Complexity. Complexity is the degree to tance. From the transaction cost perspective, outsourc- which activities are diversified within the IS function. ing creates a market-contracting, interorganizational A highly complex organization is characterized by many relationship between a firm and its external service occupational roles, subunits (divisions and depart- provider, and requires the firm to incur substantial ments), levels of authority, and operating sites. IS costs of negotiating, monitoring, and supervising exter- activities are many and varied (Price and Mueller 1986). nal contractual parties. Transaction costs are exacer- They range from long-term strategic activities such as bated by the level of specific assets, technological developing an information systems strategy and IT uncertainty, functional complexity, and supplier pres- planning, to tactical decision making on IS human ence (Joskow 1988, Walker and Weber 1984, resource management, to operational control over Williamson 1985). computer capacity, production scheduling, and secu- Specific Assets. Specific assets are physical and hu- rity. As personal computers and telecommunication man assets that are specialized and unique to the networks supplement mainframe operations, the scope extent that they generate less value outside the con- of IS services expands to PC and network management. tractual relationship (Joskow 1988, Williamson 1985). As banks disperse geographically, they must confront IS assets can be divided into IT infrastructure compris- the challenges of spatial complexity (Blau and Schoen- ing hardware and software and IT skills and special- herr 1971) in distributed IS services. The wave of bank ized know-how. In IS services of banks, specific IT mergers and acquisitions in the late 1980s left many infrastructure will include specialized equipment, oper- banks with not only distributed IS data centers, but ating procedures, and systems tailored for use in a also the onerous task of integrating disparate informa- single organization. Specialized IT skills and knowl- tion resources residing on incompatible IT infrastruc- edge consist of expertise acquired only through several tures. years of specialized IT training and knowledge that is According to transaction cost analysis, functional useful in only a narrow range of business applications. complexity precipitates difficult contracting which According to transaction cost analysis, assuming that drives up transaction costs (Mahoney 1992, Masten parties to an exchange will perform efficiently and 1984, Ulrich and Barney 1984). Indeed, a typical out- forgo opportunistic behavior is imprudent (Anderson sourcing contract for IS services is highly complex. For and Coughlan 1987). As a result, specific assets cause example, Brandon and Segelstein (1984) prescribed a problems because a firm's continued use of such assets checklist of more than 170 essential contracting ele- depends on the good-faith behavior or forbearance of ments for use in negotiating an outsourcing contract. the service provider. Firms are exposed to the possibil- The elements include technical and monetary details ity of "opportunistic expropriation" if the service such as price adjustments when volume changes, qual- provider chooses to capitalize on the transaction (Klein ity standards, response time for online work, et al. 1978). For example, opportunistic expropriation turnaround time for batch work, hardware configura- may occur ex post when service providers commodify tion, site access, staff expenses, costs of termination/ and standardize IS services to the extent that no unique cancellation, and dispute resolution mechanisms. Con- needs of any customers are met, thereby reducing the tractual items increase significantly when IS services quality and service to any single client. are geographically dispersed because contractural par- The safeguard against opportunistic appropriation is ties must cope with additional issues such as file and to rely on additional monitoring and enforcing mea- data transfer rights, software access, vendor access sures to ensure coalignment of interests between the procedures, and data and program compatibility. firm and the external service providers. Accordingly, Hence, we expect: specific IS assets would lead to increased transaction costs and influence the firm toward an internal IS service, as ownership reduces the motivation to expro- H4. The greater the functional complexity, the weaker priate the value of specific investments (Heide and the relationship between institutional influence and IS John 1990). outsourcing. ORGANIZATSIOCNIE NCE/VO8~, N. O. 3, May-June 1997 239 SOON ANG AND LARRY L. CUMMINGS Strategic Response to It~st~tut~oInnafllu ences otz Itzformation Systems Outsourcing Technological Uncertainty. Organizational decision tions, whereas large banks concentrate on wholesale makers have strong preferences for certainty, stability, and international bank services. Large banks have and predictability in organizational life (DiMaggio 1988, greater resources and power to influence their environ- DiMaggio and Powell 1983, Pfeffer and Salancik 1978, ments and are less dependent on other constituents in Zucker 1977). When the environmental context is highly the organizational field (DiMaggio and Powell 1983, uncertain and unpredictable, the organization will ex- Oliver 1991, Pfeffer and Salancik 1978, Thompson ert great effort to reestablish control and ensure stabil- 1967). Consequently, we expect size to be an important ity of future organizational outcomes. One stabilizing moderator variable in banks' strategic responses to strategy is to mimic or imitate legitimate actions and institutional pressures arising from their organizational responses of successful organizations (DiMaggio and fields. Specifically, we expect the relationship between Powell 1983, Galaskiewicz and Wasserman 1989). institutional influence and IS outsourcing to be weaker With a high degree of uncertainty in technological for large banks if moderator factors mitigate the attrac- change and the corresponding risks of technological tiveness of conforming to institutional norms. obsolescence, banks would be more likely to conform to institutional pressures to outsource. Banks would H7. Size of firm will affect significantly the interaction shift their IS services to external IT service providers to of institutional influence and moderations of IS outsourc- relinquish the financial and administrative burden of ing. rapidly depreciating IT equipment or stagnating skills of IS veterans. As technological uncertainty increases, Research Method internal economies of specialization deteriorate in rela- tion to external economies of specialization of service To test our hypotheses in the banking industry, we providers, because service providers can spread their drew a sample from banks affiliated with the American innovation risks over a large clientele. Accordingly, we Bankers A~sociation.S~tratified sampling based on predict that acquiescence is most likely when banks bank size was used as we expected differential effects perceive technological uncertainty to be high. of institutional influence and moderating economic factors on IS outsourcing between large and small H5. The greater the technological uncertainty, the banks. To minimize standard error in the difference, stronger the relationship between institutional influence we stratified by bank size to ensure roughly equal and IS outsourcing. proportions of responses from large and small banks - - - (Ross et al. 1983). Supplier Presence. Supplier presence is the availabil- Three hundred eighty-five banks were sampled, 85 ity of reputable and trustworthy service providers with more than $5 billion in asset^,^ and a random (Walker and Weber 1984). Opportunistic inclinations selection of 100 banks in each of three other size by any party in a contractual arrangement pose little strata: large ($1 billion to $5 billion in assets), medium risk if competitive exchange relations are characterized ($0.3 billion to $1 billion in assets), and small (less than by a large number of potential suppliers (Pisano 1990). $0.3 billion in assets). The four-way classification of In fact, firms may be constrained in their outsourcing very large, large, medium, and small banks reflects the choices if a full array of IS services is not available scheme adopted by the American Bankers Association from another supplier. The presence of suppliers pro- in their membership database. For the purpose of our motes a bank's conformity to institutional pressures to study, banks with more than $1 billion in assets were outsource because greater supplier presence reduces reclassified as large banks and those with less than $1 small-numbers bargaining problems and dampens op- billion as small banks. portunism (Williamson 1985). A questionnaire was mailed in the second half of 1992 to bank officers with corporate responsibility in "6. ~h~greater the supylierp resence, the stronger the IS. Each respondent acted as the informant for his or relationship between institLltional influence and IS out- her bank's sourcing arrangement and practices pertain- sourcing. ing to information systems service^.^ Impact of Organizational Size Items in the questionnaire measured concepts in the Commercial banks in the U.S. are not homogeneous in research model. Other than bank size and sourcing their nature of business, bank strategy, use of IT re- mode, items were measured on 7-point Likert scales. A sources, or the customer base they serve (Markus and draft instrument was pretested qualitatively and quan- Soh 1993). Small banks tend to focus on retail opera- titatively to ensure that the final version was valid for 240 ORGANIZATISOCNIE NCE/VO8I, .N O.3, May-June 1997 SOON ANG AND LARRY L. CUMMINGS Strntegrc Resporzse to Instltutlorznl Influerzces on Irzfornzntron Systems Olrtsou~clrzg use in a large sample. The questionnaire was pretested All but the first two sourcing alternatives (in-house with the chief information officers in 21 banks within operations and IS subsidiary at parent bank) were the Minneapolis-St. Paul, Minnesota metropolitan area. classified as IS outsourcing. Use of in-house IS services Banks used in the pilot study were omitted from the was considered insourcing. For banks that relied on main study. The pilot study ensured clarity of the their parent banks for IS services, decisions about questionnaire and ascertained that theory-based items policy, management, and operation of IS resided with tapped issues of concern in sourcing decisions. the parent bank and not the subsidiary. Consequently, In the main study, telephone interviews and mail informants at the subsidiary banks often had little questionnaires were the primary means of collecting knowledge of or control over the administrative choice data. Each bank in the random sample was contacted of IS source. Because informants of subsidiary banks by telephone to identify the person who held corporate lacked first-hand knowledge of policies on IS sourcing, responsibility for IS. A letter was sent stating the they either returned incomplete questionnaires or wrote purpose of the study and requesting participation. "don't know" across the survey instrument. We there- About a week or 10 days after the letter was sent, the fore discarded responses from banks that relied on potential respondents were contacted by telephone. their parent banks or bank holding companies for IS From those who verbally agreed to participate, prelimi- services and eliminated them from the subsequent nary information was elicited about the bank, including analyses6 the type of IS sourcing arrangement and the bank's Institutional irzfluences are external social pressures affiliation (if any) with a parent company with respect toward conformity. Institutional influences on bank to IT. Reasons were elicited from those who declined operations and practices arise from peer banks (PEER) participation. A packet of materials containing a cover and federal examiners (FEDERAL). letter, survey questionnaire, and a self-addressed, Extenzal production cost advantage (PRODCOST) is stamped envelope was sent to participants. the degree to which an external service provider is Returned questionnaires were examined for com- perceived to have an advantage in production cost pleteness of information. Respondents who skipped economies over internal bureaucratic management of items on the questionnaires were contacted again by IS services. Production cost was operationalized as telephone to obtain the missing information. Executive hardware costs, software costs, and personnel costs. summaries of the preliminary and final analyses of the Slack resources (SLACK) is the level of working study were sent to each respondent. capital, measured as the amount of financial resources The total number in the final sample was 243, yield- available for IS relative to past years' resources and ing a 63.1% response rate. Chi-square analysis was peer banks' resources allocated for IS. conducted to determine whether there was a difference Asset specificity (SPECIFIC) is the degree to which in the distribution of banks that participated and those investments in IS yielded unique value to any single that did not. In terms of bank size, banks participating firm. The construct was operationalized as (1) the level in the study were representative of the original sample of investment in specialized equipment, (2) the level of (x2 = 3.62,-p > 0.05). specialized technical skills specific to the needs of a particular bank, and (3) the level of specific business skills and knowledge pertaining to a particular bank. Variables Technological uncertainty (UNCERTAIN) is the rapid Appendix A contains questionnaire items for each con- and unexpected change in IT developments. It was struct. Outsourcing, the primary dependent variable, operationalized as the degree to which a bank can refers to the source of IS services a bank adopts. Banks forecast accurately its technical requirements as well as usually choose one of six major sources of IS services: the degree to which a bank can anticipate IT obsoles- (1) in-house computer operations, (2) an information cence. systems subsidiary at the parent bank,5 (3) other banks F~inctionalc omplexity (COMPLEX) is the degree of that provide IS services, (4) service bureaus, (5) facili- formal structural differentiation within an organization ties management, (6) joint venture, cooperative com- (Price and Mueller 1986). In the context of the IS puter service arrangement (Federal Financial Institu- function, the construct was operationalized as (1) an tions Examination Council 1992, p. 109). A seventh overall assessment of the degree of complexity in man- category, an "other" arrangement category, was cre- aging IS operations, (2) the number of hardware plat- ated to capture any additional IS source that does not forms and systems configurations, and (3) the degree of fit appropriately into any of the first six categories. sophistication of the software portfolio. ORGANIZATSIOCNIE NCE/VO8I,. NO. 3, May-June 1997 241 SOON ANG AND LARRY L. CUMMINGS Strategic Response to Institutional Influences on Infornzation Systems Outsourcing Supplier presence (SUPPLIER) is the availability of suggesting a construct that is relatively broad in con- reputable and trustworthy external IT service providers ceptual scope. in the market. It was operationalized by the perceived According to Cronbach and Meehl (19551, a scale presence of adequate service providers and the ability can have construct validity even when individual items to find comparable providers to replace the services of have low intercorrelations as long as those items sam- a current one. ple the same conceptual domain. Because items in Size (SIZE) is represented by a dichotomous vari- slack resources and asset specificity covered multiple able, zero, for small banks and one for large banks. dimensions of the respective conceptual domains, there Small banks are those with assets less than $1 billion is some assurance of construct validity despite modest and large banks are those with assets of at least $1 reliability. billion. Statistical Analysis We tested our hypotheses by using hierarchical moder- ated logistic regression models. That procedure in- Construct Validation volves forming multiplicative terms as moderator vari- For each multiple-item construct, the items were sub- ables, and using a series of logistic regression analyses jected to factor analysis to ensure a single-factor struc- to determine the relative contribution of the moderator ture. A common rule of thumb is that for unidimen- terms to the explanation of variance in the dependent sional construct measures, all or most of the items variable. Accordingly, the probability of a firm choos- should load more heavily on the first factor than on ing to outsource its information system services in any other factor (Carmines and Zeller 1979, p. 60; preference to insourcing can be modeled as a function Spector 1992, p. 55). Only one factor emerged from a of the main effects and the interaction terms as factor analysis conducted on each multi-item construct, thus confirming the unidimensional nature of the items. Items therefore were summed into a total-score mea- Probability of choosing sourcing mode sure for each construct. Subsequently, inter-item reliability analysis using Cronbach's alpha was applied to measures based on multi-item scales. Nunnally (1967) suggested reliability where: in the 0.5 to 0.6 range in the early stages of research and 0.95 as being the desirable standard for research in applied educational settings. In business settings, there are no generally accepted guidelines (Peter 19791, al- though Van de Ven and Ferry (1980) suggested a range and zl>x2>X. .a-r.e, >ex-plpan atory variables; and co- of 0.55 to 0.90 for constructs of narrow to moderately efficients, b,, _b,, . . . , b are the corresponding coeffi- broad conceptual scope. We used that range to assess cients with b, as the intercept term. The explanatory the reliability of the measures. Alpha coefficients for variables are (1) the main effects of institutional influ- the constructs are reported in Appendix A. ence (PEER or FEDERAL), the hypothesized variable Slack resources (a = 0.43) and specific assets (a = (PRODCOST, SLACK, SPECIFIC, etc.), and the con- 0.56) yielded relatively low reliability. For slack re- trol variable, bank size (SIZE), (2) two-way interaction sources, informants were asked to assess slack on the effects of institutional influence with the hypothesized basis of two norm references: peer bank's slack and variable, institutional influence with size, and the hy- past year's slack. Confirmatory factor analysis showed pothesized variable with size, and (3) a three-way inter- good fit for a single factor ( ~ <2 0.0 5, p > 0.98); but action effect of institutional influence, hypothesized poor fit for a two-factor structure (x2 = 5.78,p < variable, and size. 0.02), suggesting that despite low inter-item reliability, To assess the explanatory power of the three-way the items represented one factor. interaction term, we applied a hierarchical moderated For specific assets, informants were asked to assess regression strategy using an improvement likelihood three different types of specific investments: human ratio chi-square (Hauck and Donner 1977, Hosmer and assets in the form of IS technical skills and specific Lemeshow 1989). Analogous to the hierarchical _F-test business skills, and non-human assets in the form of in multiple regression (Aiken and West 1991, Ch. 6), computer equipment. The intercorrelations among the the improvement chi-square statistic computes the three forms of investments ranged from 0.10 to 0.39,7 change in the -2 log likelihood (- 2 LL) between suc- 242 ORGANIZATISOCNIE NCE/VO8I,. NO.3, May-June 1997 SOON ANG AND LARRY L. CUMMINGS Strategzc Respot~seto Instztut~orzaIlt ~fluetzceso r? Infommatzon Systems Outsou~c~fzg cessive steps of building a model. Accordingly, we containing all of the lower-order terms, formulated and tested nested models of variables with and without the focal interaction term using the im- Log(OUTSOURCING) provement chi-square as the statistic to assess the + + + relative contribution of that interaction term. = ,PEER ,PRODCOST ,SIZE When large numbers of interaction terms are in- + + cluded in a model, serious multicollinearity is likely. _b,(PEER'*PRODCOST) _b, (PEER"'S1ZE) We rescaled the original variables using procedures + recommended by Aiken and West (1991). All continu- b,(PRODCOST*SIZE) , ous variables were "centered" by subtracting the corre- sponding variable mean from each value. Such rescal- computed the 2LL from a model containing these - ing does not affect the substantive interpretation of the lower-order terms plus the three-way interaction term, coefficients (hken and West 1991). Effects of each hypothesized variable on the Log(0UTSOURCING) relationship of institutional influence (PEER or FEDERAL) to OUTSOURCING were analyzed sepa- = -n + ,PEER + ,PRODCOST + ,SIZE rately rather than including all of them in the same logistic regression model. That strategy seemed appro- + _b,(PEER'*PRODCOST) + , (PEER*SIZE) priate given that the greater the number of indepen- dent and moderator variables included in a model, the + ,(PRODCOST*SIZE) lower to power of the test8 The following series of hierarchical logistic equations + _b, (PEER"PR0DCOST"SIZE) , illustrates how hierarchical moderated logistic regres- sion assesses the effect of a moderator, externalpro- and tested the incremental -2LL improvement with a duction cost advantage (PRODCOST), on the relation- hierarchical chi-square test. If an improvement in chi- ship between PEER influence and OUTSOURCING. square value was statistically significant, a three-way First, the three-way interaction term (PEER* interaction effect was present. PRODCOST*SIZE) determined whether there was a A significant three-way interaction term would mean difference in the PEER'TRODCOST interaction be- that the effect of the hypothesized moderator variable tween large and small banks. The logistic regression (in this case, PRODCOST) on the relationship of peer model included all main effects, 2-way and 3-way inter- influence to outsourcing is significantly different be- action effects of peer influence (PEER), the moderator tween large and small banks. Subsequently, coefficients variable (PRODCOST), and the control variable of the two-way interaction of (Peer*Prodcost) for large (SIZE). and small banks would be examined to determine whether the significant impact of the hypothesized variable resides in large or small banks. If the hierarchical chi-square test revealed that the three-way interaction term of PEER:kPRODCOST* SIZE was not significant, it would mean that the effect of the hypothesized variable (PRODCOST) on the relationship of PEER to outsourcing is independent of size. The coefficient of the two-way interaction term (PEER"PRODC0ST) would then be examined for all banks. +~,(PEER*PRODCOST*SIZE) Results Table 1 reports the means, standard deviations, and intercorrelations among the major constructs in the To determine the significance of the (PEER' study. The intercorrelations show a significant positive PRODCOST*SIZE) interaction term, we computed bivariate association between peer influence and IS -2 log likelihood (-2LL) improvement from a model outsourcing (c = 0.51,-p < 0.01) and a significant posi- ORGANIZATSIOCNIE NCE/VO8I, .N O. 3, May-June 1997 243
Description: