Critical success factors for is outsourcing implementation from an interorganizational...

INTRODUCTION

The enormous growth in the size of information systems (IS) outsourcing arrangements has been staggering as evidenced by numerous high-profile multibillion dollar outsourcing deals involving such familiar names as Boeing, Bank One and Xerox. This upward trend in IS outsourcing is set to continue. International Data Corporation (IDC) predicts that the worldwide outsourcing market would grow from $100 billion in 1998 to $151 billion in the year 2003 (31). The U.S. market alone is expected to grow from $51.1 billion in 1998 to $81 billion in 2003 (31). The reasons for this phenomenon vary, but the overriding belief is that IS outsourcing will result in many benefits including cost savings, improved quality of IS services, access to up-to-date technology, flexibility in IS operations, and focus on core competencies (43, 44). Despite such benefits, IS outsourcing does not always achieve desired results. Lacity and Hirschheim (23), in their case study of 14 U.S. firms, found that many of the initially proclaimed benefits of IS outsourcing were not realized. Further, Willcocks and Lacity (48) studied 116 outsourcing decisions, assessing objective against outcome, cost savings expected versus those achieved, and "satisfaction" levels outcome for the clients. They found that 38% of outsourcing arrangements were successful, 35% were failures, and 27% resulted in mixed results. In light of these less than desired outcomes, there is a need to gain more insight into the critical success factors in outsourcing relationships (42, 49).

The primary objective of this study is to empirically investigate the factors contributing to the success or failure of IS outsourcing relationships between client firms and vendors. This study is intended to examine problems from previous studies related to the success factors of implementation of IS outsourcing and to augment knowledge for successful implementation of IS outsourcing. For the purpose of this study, IS outsourcing is defined broadly as "the practice of turning over part or all of an organization's IS functions to one or more external service provider(s)" (15, p. 91). First, the unit of analysis in this study will be the relationship between a client firm and a vendor in a specific outsourcing task. Utilizing a relationship as a unit of analysis allows an in-depth and precise analysis of an IS outsourcing relationship. The vast majority of studies in interorganizational relationship success also adopted a relationship rather than an organization as a unit of analysis. Second, this study will employ a more comprehensive model that includes factors not addressed in the previous studies, such as relational exchange characteristics, communication behavior, and task characteristics. Knowledge acquired in this study is expected to provide a framework for helping IS managers in the on-going management of the relationships as well as in the selection of vendors.

THEORETICAL BACKGROUND

A relationship developed between a client firm and a vendor in IS outsourcing can be characterized as an interorganizational relationship (IOR) since two parties from different organizations work together. Various theories and frameworks of IORs have been utilized in organizational economics, organizational theory, strategic alliance, and relationship marketing literature. In this section, two major approaches in IORs, transaction cost analysis and relational exchange theory, are briefly reviewed to understanding IS sourcing from IOR perspective. The research model in this study is primarily based on these two approaches.

Transaction Cost Analysis

The focus of analysis in transaction cost analysis (TCA) is to identify efficient governance structures that match transaction characteristics (17). In TCA, governance structure problems are studied in transaction cost minimizing terms. TCA postulates that three primary dimensions of the transaction - uncertainty, frequency of exchange, and asset specificity - determine appropriate governance structures. When uncertainty is high, transactions are recurrent and/or when asset specificity is high in transactions, organizations are expected to adopt in-sourcing to avoid high transaction costs generated in transacting with outside parties when transaction characteristics opposite to those described above exist, organizations are expected to adopt outsourcing (50).

Although TCA has been widely applied in studying organizational governance, recently it has been under severe criticism. For instance, Ring and Van de Ven (40) pointed out three major limitations when TCA is applied to analyze interorganizational relationships. First, TCA assumes that managers will be motivated solely by efficiency considerations. Other factors, such as equity and flexibility, are largely ignored. Second, in TCA, opportunism is the primary behavioral principle. However, cooperative interorganizational behavior may occur frequently because of long-term efficiency resulting from cooperation and deletion of actors whose behaviors are habitually opportunistic (20). Third, TCA explores only two kinds of governance mechanisms - markets and hierarchies. It does not adequately explore other available interorganizational governance structures between the extremes of market and hierarchy, such as joint ventures, partnerships, and outsourcing arrangements. In addition, TCA overemphasizes the structural features of the interorganizational exchange (53). It views IORs solely in terms of structural properties, but fails to recognize developmental processes of IORs. Hence, processual/behavioral aspects of interorganizational exchange are largely ignored in TCA.

Although TCA has been mainly used in the investigation of governance structure, it also provides normative prescriptions regarding the organization of governance structure. It is suggested that an appropriate match between governance structure and task characteristics in TCA enhances performance. For instance, tasks with high uncertainty and high asset specificity will cause high transaction costs when transacting with outside vendors and those high potential transaction costs will eventually lead to decreased relationship performance. On the other hand, tasks with low uncertainty and low asset specificity will cause relatively low transaction costs, eventually leading to good relationship performance.

Research in IS outsourcing has largely neglected the relationship aspects of the client-vendor behavior while tending to study outsourcing transactions as discrete events. This tendency is reflected in the dominant use of the TCA framework in the empirical research in IS outsourcing. However, some researchers (13, 29) view IS outsourcing agreements as strategic alliances or partnerships and emphasize the importance of managing outsourcing relationships. In IS outsourcing, the average length of contract is estimated to be ten years (29). As the outsourcing contract term becomes longer and the scope of outsourcing expands to more core, strategic areas of IS functions, it is essential to view IS outsourcing from the partnership perspective rather than from a discrete transaction perspective.

Relational Exchange Theory

Relational exchange theory, as Macneil's (27) neoclassical contractual framework is often called, expanded Williamson's (50) initial description of market versus hierarchy of the interorganizational governance structure. The theory suggests that the governance structure can be arranged on a continuum of relationalism anchored by market (discrete exchange) and hierarchy (relational exchange) at the polar extremes (34). Macneil (27) differentiated discrete exchange from relational exchange along several key dimensions. Discrete exchange is relatively short-term, and the relationships between highly autonomous buyers and sellers are designed to facilitate economically efficient transfer of goods or services (40). Communication between parties is very limited and the contents are very narrow. Since virtually no social exchange is engaged, the identity of parties can be completely ignored. An appropriate example in a pure form might be a one-time purchase of unbranded gasoline out-of-town at an independent station paid for with cash (12). Whereas, in relational exchange, each transaction must be viewed in terms of its history and its anticipated future, the participants are expected to derive complex, personal, non-economic satisfaction and engage in social exchange (12). Its pure form occurs in a form of hierarchy within an organization.

The concept of interorganizational structure or relational structure presented in the relational exchange theory provides a significant opportunity to study hybrid forms of interorganizational cooperative arrangements, which are neither markets nor hierarchies. Powell (39) argued that simultaneous pressures toward efficiency, flexibility, and speed are pushing more and more firms to form hybrid arrangements. An IS outsourcing arrangement is a kind of hybrid interorganizational arrangement since it involves a long-term commitment with about ten years as the average length of contract. Relational exchange theory, to be discussed later in the attributes of relational structure, also addresses the behavioral/processual aspects of relationships. Hence, in this research, relational exchange theory will be used as the backbone of studying IS outsourcing relationship.

RESEARCH MODEL AND HYPOTHESES

The model in this study is primarily based on the relational exchange theory. Two TCA factors, uncertainty and asset specificity, are also included in a form of task characteristics to provide an economic efficiency perspective. Based on the literature review presented in the previous section, literature in IS outsourcing, and relationship marketing and other IOR literature, factors affecting success or failure of implementation of IS outsourcing are identified and presented in Figure 1. Factors identified are relational exchange characteristics and task characteristics.

Relational Exchange Characteristics

The first component of the model identifies relational exchange characteristics. Researchers in relationship marketing and strategic alliances have used a variety of sets of dimensions to measure the relational structure based on the situational factors of the research settings. Examples of dimensions include solidarity, continuity expectation, role integrity, mutuality, monitoring of the vendor, joint actions, flexibility, vendor assistance, information exchange, restraint in the use of power, and harmonization of conflict. Among these dimensions, the present study utilizes solidarity, continuity expectation, role integrity, flexibility, and monitoring of the vendor as the characteristics of relational structure. These dimensions are chosen to capture the essence of characteristics of the relational structure in IS outsourcing relationship. The first four items, solidarity, continuity expectation, role integrity, and flexibility, are included since these items are among the most frequently utilized. The remaining item, monitoring of the vendor, is included since it is considered a very important success factor in an IS outsourcing relationship by the researchers.

Solidarity

The norm of solidarity refers to the extent to which an ongoing relationship (as distinct from a series of discrete transactions) is created and sustained (21). It represents the norm of holding exchanges together (27). In relational exchange, the preservation of a unique and continuing relationship is internalized by exchange parties as being important in and of itself, thus exhibiting a partnership mentality between parties. In comparison, discrete exchange displays little solidarity between parties due to its tendency to be adversarial in relationships (52).

Continuity Expectation

The norm of continuity expectation refers to the expectation of future exchange between parties (34). Note that the definition involves anticipated duration into the future rather than the historical duration of the relationship. In discrete exchange, the parties expect a low probability of future interactions. As transactions become more relational, the parties expect that the current relationship will last for the given contract period and renewal of relationship is also expected. When the parties expect continuity of the relationship, each party is expected to perform its activities more faithfully because "the shadow of the future" has been enlarged and thus future interactions between parties provide an opportunity to reward good behavior and punish opportunism (5).

Role Integrity

Role integrity refers to the extent to which the parties maintain highly complex and multi-dimensional roles (21). In relational exchange, expectations of ongoing transactions necessitate complex roles involving a variety of business and non-business issues (10). In contrast, the roles to be maintained in discrete exchange are simplistic, just requiring simple buy-sell interactions (52). In IS outsourcing, the parties are expected to exhibit highly complex and multidimensional roles in order to deal with the complexity of information systems.

Flexibility

Flexibility refers to smooth alterations in practices and policies in the event of unexpected or changing conditions (6). As the transaction becomes more relational, the terms of trade becomes more open-ended. Thus, planning and adjustment are essential to cope with uncertain environments. In comparison, a discrete exchange relationship requires little flexibility due to "binding and specific" terms of trade (34). In IS outsourcing relationship, maintaining flexibility is one of the important goals as well as a critical success factor (24, 29). Due to the long-term nature of IS outsourcing arrangements, it is nearly impossible to precisely pre-specify everything in detail. Flexibility is required to cope with evolving technology and changes in the organization's business posture, market, and climate (29).

Monitoring of the Vendor

Monitoring of the vendor involves the monitoring or supervisory actions that the client firm undertakes to secure satisfactory vendor performance in the execution of the agreement (34). The monitoring activities in IS outsourcing include developing performance standards, measuring results, and then interpreting them continuously (29). In a typical discrete transaction where well-specified, simple products are delivered, enforcement of the contract is a simple task. When the contract becomes more relational, active supervision by the client firm is necessary to ensure satisfactory performance. In the IS area, contracts tend to be more relational since the contract is typically long-term and precisely pre-specifying everything in detail is very difficult. According to a survey on telecommunications outsourcing performed by Thobe (47), the respondents generally did not feel that they have clear performance measures and indicated that clear definitions of responsibilities and performance measure as the most important factor for successful outsourcing arrangements.

Relational Exchange Characteristics and Success

As today's IS outsourcing contract term becomes longer and the scope expands to more core, strategic areas of IS functions, it is essential to view the IS outsourcing relationship from a partnership or relational exchange perspective rather than from a discrete exchange perspective (29). Cheon (7) found that the quality of partnership has a positive effect on the performance of IS outsourcing. Empirical studies in IORs have shown that a higher level of relational exchange characteristics has a positive effect on both polity performance (trust and satisfaction) and economic performance in interorganizational arrangements (2, 41). Thus, the following hypothesis is proposed regarding relational exchange characteristics.

H1: Success of an IS outsourcing relationship is positively associated with the degree of:

A. solidarity

B. continuity expectation

C. role integrity

D. flexibility

E. monitoring of the vendor

Transaction Cost Analysis and Task Characteristics

Transaction cost analysis (TCA) has been the predominant framework employed to investigate the determinants of IS outsourcing. TCA provides explicit normative prescriptions regarding the organization of the governance structure. That is, an appropriate match between governance structure and task characteristics in TCA is expected to enhance performance. However, little empirical testing has been conducted to investigate the performance implications of the TCA framework (19). Following the tradition of recent empirical studies (19, 33, 34) that tested performance implications of TCA factors, this paper will test the link between TCA factors and performance. Among three major TCA factors, only uncertainty and asset specificity will be considered in the model. The remaining factor, frequency of exchange, is excluded since the transactions are performed on a continuous basis in IS outsourcing (4).

Uncertainty

Uncertainty refers to the degree to which future states of the world cannot be anticipated and accurately predicted (38). Uncertainty is created when insufficient information is available to make precise contract specifications so that "future contingencies for which adaptations may be required cannot be anticipated at the outset" (51, p. 237).

Three areas of uncertainty heavily studied in interorganizational literature are technological, measurement, and demand (or volume) uncertainty. Technological uncertainty in a potential client-vendor relationship may come from adoption of new standards, introduction of new functionalities, and obsolescence of hardware and software (25). Measurement uncertainty may come from the difficulties of evaluating and monitoring (1) the quality of products/services delivered, (2) the cost-performance trends, and (3) the quality of IS staff (4, 25). Finally, demand uncertainty may come from the fluctuations in demand of hardware, software, IS personnel, and telecommunications/network requirements.

TCA predicts that high uncertainty will cause high transaction costs due to costs associated with writing, negotiating, monitoring, and enforcing contracts with outside vendors and that high potential transaction costs will eventually lead to decreased relationship performance. In an empirical study of software development projects, Nidumolu (33) identified a negative relationship between project uncertainty and project performance. In an empirical study of a marketing channel relationship, Heide and Stump (19) found a negative relationship between volume uncertainty and relationship performance. In terms of uncertainty and success of IS outsourcing relationship, the following hypothesis is proposed:

H2A: Success of an IS outsourcing relationship is negatively associated with the degree of uncertainty in the outsourcing task.

Asset Specificity

Asset specificity refers the degree to which the transaction utilizes specialized assets that cannot be transferred to any other transaction. Unlike assets that can be used to serve a variety of situations, assets with high asset specificity have primary value only in the particular relationship between the client firm and the vendor. Loh (25) identified three sources of asset specificity in the IS outsourcing context. First, technical resource specificity arises when client firms have their hardware, software, and communications architecture or platforms uniquely developed for customized usage. Second, human resource specificity arises when IS staff are trained to operate only customized applications that are distinctive to the organization. Finally, technical procedure specificity arises when client firms have unique technical procedures in the process of systems design, operations, and maintenance.

TCA postulates that high asset specificity will cause high transaction costs and that high potential transaction costs will eventually lead to decreased relationship performance. In an empirical study of a marketing channel relationship, Heide and Stump (19) found a negative relationship between asset specificity and relationship performance. In terms of uncertainty and success of IS outsourcing relationship, the following hypothesis is proposed:

H2B: Success of an IS outsourcing relationship is negatively associated with the degree of asset specificity in the outsourcing task.

Control Variable: Vendor Capability

Vendor capability is probably the most critical factor for the successful implementation of IS outsourcing (22, 29). Many studies in interorganizational and interpersonal relationships studies have also related ability or capability as a critical characteristic of trust, which is a strong predictor of relationship success. In view of strategic resource theories, such as the resource-based theory and resource dependence theory, organizations engage in outsourcing arrangements when they fail to generate necessary resources or capabilities internally (45). Hence, organizations are likely to engage in the IS outsourcing arrangements when organizations' IS capabilities fall short of expectation. Empirical studies in the determinants of IS outsourcing (26, 32, 45) have confirmed the role of IS capability in IS outsourcing decision.

The fact that IS capability is a major determinant of IS outsourcing implies that organizations need competent vendors to satisfy their outsourcing needs. In addition, IS outsourcing is usually contracted for a long term and it is very difficult to switch to another vendor during the contract term or to insource again. The vendor should have the capability to keep up with ever-changing technology and its financial structure should be stable to maintain undisrupted service. Important factors determining vendor capability include experience and track record, technical competence, and financial status (9, 14, 16, 22).

In this study, vendor capability is treated as a control variable since the researcher is primarily interested in the effect of independent variables discussed above. The treatment of vendor capability as a control variable will contribute to eliminating the confounding effects and focusing on the effect of research variables. Further control of industry sector and organizational size was not considered since there is evidence that industry type and organizational size do not affect the context of research on IS outsourcing (26).

Success Measures of Implementation of IS Outsourcing

In MIS research, having well-defined success measure(s) or dependent variable(s) is very important since MIS research is intended to make a contribution to the world of practice (11). In the IS arena, research is interorganizational relationship (IOR) is sparse and most success measures have been utilized to measure performances of specific IS activities in non-IOR contests. There is no consensus on the appropriate measures of success of IS outsourcing (36) due to lack of study in the investigation of success of IS outsourcing.

Areas of outsourcing surveyed in this study are not restricted to any specific area, but include all areas of IS outsourcing. Due to difficulties in constructing success measures in all areas of IS outsourcing and the space limitation of the questionnaire, quantitative or task-specific measures, such as the system response time in mainframe operation outsourcing and network availability in telecommunications outsourcing, are not utilized in this study. Based on the literature of IORs in marketing channels, strategic alliances, and studies in IS outsourcing, this study employs two indicators of outsourcing success: satisfaction and perceived benefits.

Satisfaction is an affective measure that represents the degree of a client firm's satisfaction with the vendor. Satisfaction is defined as "a positive affective state resulting from the appraisal of all aspects of a firm's working relationship with another firm" (2, p. 66). Since the measurement of satisfaction involves the evaluation of all aspects of the relationship, satisfaction is considered as a close proxy for perceived effectiveness of the relationship (2). Satisfaction has been widely utilized as a measure of relationship success in IOR studies (1, 30).

Perceived benefits are a client firm's perception of benefits gained from a specific outsourcing relationship (7). Since benefits of IS outsourcing are also underlying reasons for or expectations from outsourcing arrangements, perceived benefits measure the degree of accomplishment of expectations from the client firm's perspective. Hence, they are accepted as good measures of IS outsourcing success. Three major types of benefits identified in the literature are strategic, economic, and technological benefits (7, 9, 16, 36). First, strategic benefits refer to the achievement of focusing on core competence, enhancing strategic use of IT and enhancing flexibility. Second, economic benefits refer to the capability to produce IS services at lower costs. The degree of economic benefits depends not only upon achieving economies of scale and scope in the areas of human and technological resources such as hardware and software, but also upon controlling cost structure. Third, technological benefits refer to the achievement of gaining access to leading edge IT and avoiding the risk of IT obsolescence that results from accelerating changes in the nature of IT infrastructure.

RESEARCH METHOD

This study involved a cross-sectional field study via a questionnaire-based mail survey. This study focuses on the relationship in IS outsourcing. Hence, the unit of analysis in this study is the relationship between a client firm and one of its vendors. Researchers have recommended dyadic analysis to comprehensively investigate the interorganizational relationship (18). However, considering the immense difficulty of locating exact parties in IS outsourcing relationships, this study focuses on one perspective of the dyadic relationship, the client firm's view of the relationship with its referent vendor.

The sampling frame for this research consisted of the large U.S. firms listed in the Directory of Top Computer Executives (West Edition) published by Applied Computer Research, Inc. all listed organizations have their own IS departments. Among the organizations listed in the directory, not-for-profit organizations and government organizations were excluded from the survey. Further control of the industry sector and organizational size were not considered since there is evidence that the industry type and organizational size do not affect the context of research on IS outsourcing (26). The questionnaire was distributed to a sample of 2200 firms in 12 Midwestern and central states on the list of the directory. The questionnaire was addressed to the top executive in charge of information systems. Approximately three weeks after the initial mailing, a follow-up mailing to 1000 randomly selected companies was conducted with a copy of the questionnaire enclosed.

The questionnaire was basically composed of two portions. The first portion asked about the respondent's perceptions about IS outsourcing in general and the company's overall outsourcing practice without consideration of any specific outsourcing arrangement. The second portion of the questionnaire asked about the nature of the relationship in regard to a specific outsourcing arrangement. The respondents were asked to identify a specific vendor with which they are familiar. The relationship with this selected vendor then served as the referent for all further questions. Companies that were not engaged in any outsourcing arrangement did not complete the second portion of the questionnaire.

Of the 2200 questionnaires mailed, 368 were returned. Upon further evaluation, 13 incomplete questionnaires were determined unusable. Thus, 355 responses were usable, resulting in a usable response rate of 16.1%. Possible nonresponse bias was evaluated by comparing early respondents (67.3% within three weeks) with late respondents, due to unavailability of nonrespondent firms' data. The underlying assumption here is that late respondents are in some way more like nonrespondents (3). The comparison indicated no significant differences between early and late respondents in characteristics such as total sales, IS department budget, number of total employees, and number of IS employees, at the significance level of 0.05.

Profile of the Respondents

The respondents had an average length of 21.8 years of experience in the IS field. A large number are Directors/Assistant Vice Presidents (VPs) (36%), CIOs/VPs (34%), and Managers (28%). The responding organizations had an average number of 5051 total employees. They represented a broad spectrum of industries, with the largest segment coming from manufacturing (36%).

Measurement

Whenever possible, measures that have been utilized and validated are adopted for this study. All the variables were measured with multiple-item scales. Each item was measured according to a seven-point Likert-type scale. The measures of relational exchange characteristics were adopted from the relationship marketing literature. For the two TCA factors, uncertainty and asset specificity, this study utilized the measures developed by Nam (32). Measures of vendor capability were based on the items used by Loh (25).

The present study utilizes two measures of success in the implementation of IS outsourcing: satisfaction and perceived benefits. Satisfaction was measured by three items adapted from Park (37), who in turn based his scale on the work of Anderson and Narus (1). Though the measure of satisfaction was adopted from the relationship marketing literature, it is deemed acceptable to be utilized in this study since it represents overall satisfaction without referring to any industry-specific terms. Perceived benefits were measured by nine items based on Cheon (7). Each item of perceived benefits was measured by a seven-point Likert-type scale, anchored from "much worse" to "much better" in comparison to the client firm's expectation. This type of measurement approach represents the concept of "outcomes given comparison level (Outcomes\CL)" proposed by Thibaut and Kelley (46) from the perspective of social exchange theory. The comparison level in the present context can be defined as a standard representing the quality of outcomes the client firm has come to expect from a given type of relationship, based upon present and past experience with similar relationships, and knowledge of other IS outsourcing relationships (2). Thus, perceived benefits in this study are conceptualized as outcomes obtained from a relationship, against the comparison level defined above. The comparison level is introduced as an anchor for assessing perceived benefits to control different expectations by different IS activities outsourced.

Measurement Assessment

Each set of multiple-item scales was initially subjected to an examination of item-to-total correlations to identify items that did not belong to the specific scale. An item with low item-to-total correlation indicates that the item is not drawn from the same domain and should be deleted to reduce error and unreliability (35). Items were deleted if their item-to-total correlation was below 0.35. In order to assess unidimensionality, principal components factor analysis was conducted on subsets of variables. Varimax was the rotation method for all analysis. Factor loadings of less than .50 are dropped from further analysis. Table 1 lists summary scale statistics.

The table lists the extracted final constructs. All items except two cases cleanly loaded to the intended construct. First, for the relational exchange characteristics, all items of solidarity and continuity expectation were merged into one factor. Upon inspection of the items, the new factor was subsequently labeled as "partnership" to reflect the fact that the items depict closeness and long-term orientation, which are the essence of partnership in the working relationship. It is not uncommon that items from different scales are combined into one factor in relational exchange scales. Second, for perceived benefits, two factors emerged from nine items. The split of perceived benefits into two factors was not intended, but it is not unexpected since the items represent a diverse set of benefits including strategic, technological, and economic dimensions. Upon an inspection of the items, the two factors were termed as "perceived non-economic benefits" and "perceived economic benefits," respectively. The items of non-economic benefits generally reflect strategic and technological benefits. As a result, three dependent variables will be utilized for hypothesis testing.

Table 1 also lists the result of the reliability test. All Cronbach's alphas except that of asset specificity exceeded the generally accepted minimum value of 0.70, demonstrating satisfactory evidence of internal consistency. The Cronbach's alpha for asset specificity was 0.54. Nunnally (35) suggested that a coefficient value of between 0.5 and 0.6 is sufficient for early basic research. Considering the exploratory nature of this study, the researcher decided to retain the measures of asset specificity for further analysis.

RESULTS

Characteristics of Outsourcing Relationships

For the hypothesis testing purpose, only the second portion of the questionnaire was utilized. Of the 355 responses of the survey, 148 responses did not complete the second portion of the questionnaire. Ninety of them were non-outsourcing firms. A total of 207 outsourcing relationships were utilized for hypothesis testing. The descriptive statistics of outsourcing relationships are summarized in Table 2.

The table includes a breakdown of outsourcing arrangements by IS activity, length of the contract, and contract amount. It indicates that the sample of outsourcing relationships is heavily concentrated on the application development/maintenance area (46.4%). A majority of the relationships had a contract period of two years or longer (51.3%). However, the average contract length of 2.7 years was shorter than the contract period of five to ten years typically discussed in the literature. Less than 24% of the contracts had an amount of less than $100,000. The contract amount also represents a wide spectrum of contract size.

Tests of Hypotheses

Multiple regression analysis was employed to test the hypotheses. For each group of hypotheses (three groups: H1A to H1E, H2A to H2B), multiple regressions were run separately for each of the dependent variables: satisfaction, non-economic benefits, and economic variables. Following guidelines suggested by Cohen and Cohen (8), the control variable (vendor capability) was added to each regression run before adding the independent variables. In this way, the effect of a control variable can be partialed out prior to hypothesis testing.

The results of hypothesis testing are shown in Table 3. The results of regression runs for three dependent variables were consolidated to one table for easy comparison.

Hypothesis 1 posited that relational exchange characteristics are positively associated with successful implementation of IS outsourcing. As Table 3 shows, partnership (i.e., the combined measure of solidarity and continuity expectation) is found to be positively associated with satisfaction (p<.01), providing support for H1A and H1B. Flexibility is significantly associated with both satisfaction and non-economic benefits (p<.01), providing support for H1D. Monitoring of vendor is significantly associated with both non-economic and economic benefits (p<.01), providing support for H1E. Interestingly, role integrity is negatively associated with satisfaction. Since role integrity is proposed to be positively associated with dependent variables, H1C is rejected. Overall, relational exchange characteristics, except the case of role integrity, are significantly, positively associated with at least one of three success measures.

Hypothesis 2 is concerned with the effect of task characteristics on the success of IS outsourcing. It posited that uncertainty (H2A) and asset specificity (H2B) are negatively associated with success of IS outsourcing. Table 3 indicates that asset specificity is negatively associated with satisfaction (p<.01), providing support for H2B. However, uncertainty is found to be not associated with any of the success measures. Thus, H2A is rejected.

In all six regression runs (three dependent variables by two sets of hypotheses) discussed above, the control variable (vendor capability) was found to be positively related to dependent variables (all p's<.01), indicating that vendor capability is a prime predictor of IS outsourcing success.

As discussed above, hypothesis testing has been conducted for each group of hypotheses. In order to check the overall soundness of hypothesis testing, one overall regression run, which addresses all seven independent variables and the control variable, was performed for each dependent variable. The findings were quite similar to those of separate runs presented in Table 3.

DISCUSSION

The following variables were found to be significantly, positively related to predicting the success of IS outsourcing (either satisfaction or benefits): partnership, flexibility, and monitoring of the vendor. These findings suggest that as these variables are present in a high degree, there is a greater likelihood of outsourcing success. On the other hand, the following variables were found to have negative influence on the success of IS outsourcing: role integrity and asset specificity. A great degree of presence of these variables is expected to reduce the chance of the success in IS outsourcing.

Except for role integrity, all dimensions of relational exchange were positively and strongly related to at least one of the three success measures. These findings generally suggest that a strong formation of relational exchange attributes, as opposed to discrete exchange, is essential to successfully implement IS outsourcing. That is, a traditional discrete governance structure, characterized by adversarial arm's-length relationships, should be supplanted by cooperative relational exchange relationships.

Partnership was the strongest predictor (Beta=.29) of satisfaction among relational exchange characteristics. Since partnership, measured by solidarity and continuity expectation, largely represents the feeling of closeness or cohesiveness of the relationship, it may psychologically influence satisfaction, the affective measure. Flexibility appears to be a very strong predictor of success, as it is significantly related to two success measures, satisfaction and non-economic benefits. The finding of flexibility is consistent with the IS literature which suggests that flexibility is a key for outsourcing success to cope with evolving technology and changes in business needs. Monitoring of the vendor was not significantly related to satisfaction, but it is the only variable in relational exchange significantly associated with both non-economic and economic benefits. Hence, when it comes to actual outcomes represented by benefits, monitoring seems to be a critical element. The finding is consistent with IS literature that puts a heavy emphasis on monitoring to ensure successful performance in IS outsourcing. Firms need to put more effort to monitoring activities in such areas as developing performance standards, measuring results, and interpreting them.

The negative association between role integrity and satisfaction is both surprising and inconsistent with the prediction. It is possible, however, that the greater level of role integrity may generate unattainable high expectations among parties and this in turn may result in a lower satisfaction rating.

Asset specificity was found to be negatively associated with satisfaction. The finding is consistent with the prediction. Task characteristics including asset specificity are largely not controllable after an outsourcing relationship begins. Hence, firms need to avoid outsourcing tasks with high asset specificity. Uncertainty was not found to be significantly associated with any of the success measures. A possible explanation may lie in the nature of uncertainty in IS outsourcing. In the IS arena, uncertainty is largely influenced by technological elements since IS tasks are very sensitive to technological changes. Mahoney (28) argues that a higher level of technological uncertainty leads organizations to utilize less firm-specific assets. Thus, the resulting reduction in asset specificity may confound the effect of technological uncertainty. Overall, task characteristics do not seem to be good predictors of success of IS outsourcing.

Finally, the importance of vendor capability, which was utilized as a control variable, needs to be mentioned. For all regression runs, vendor capability was found to be strongly and significantly related to success with beta values ranging from .28 to .67. Hence, the single most important thing a client can do for outsourcing success can be to select capable vendors.