Implementing Successful Public–Private IT Outsourcing Relationships: Relational View to Fostering Public Value

Abstract

This article proposes a novel theoretical model, via integrating the relational view and boundary object theory, to explore the determinants of public value in public–private IT outsourcing (ITO). This study emphasizes the influence of relational factors, such as benevolence, technical capability, ambidexterity, and semantic and pragmatic knowledge interfaces, on public value creation in public–private ITO. This model is examined in an empirical fashion through a study of public–private outsourcing projects in Mexico’s state-level public administration, contributing to a nuanced and integrated understanding of relational factors in public–private ITO. The article concludes with managerial implications, limitations, and avenues for future research.

1. Introduction

Outsourcing has gained importance in the public sector, shifting the focus of public administration from service provision to the management of contractual relations [1,2]. Public–private IT outsourcing (ITO) is defined in our study as an organizational form whereby external providers offer a variety of IT activities to develop and execute public services on the basis of a medium- to long-term contractual agreement with a public entity [3]. Outsourcing frequently responds to pressures to reduce costs [4]. Meanwhile, collaborative processes with external service providers have also become a policy imperative in public administration to spur organizational change and innovation in the presence of limited resources and policy constraints [5,6,7].

In spite of the potential benefits they can bring, IT outsourcing relationships face challenges to their success in creating public value because of potential misalignment in initial conditions, objectives, structures, knowledge management processes, outcomes, and accountabilities, especially between public and private partners [4,8,9].

The present literature emphasizes the importance of factors such as organizational commitment, organizational learning, trust, sense of mission orientated toward public value, and bureaucratic coordination to foster IT development in public services, especially in developing countries, where lack of coordination and adequate partnerships prevails [10]. Efficient public–private IT outsourcing relations are viewed as essential partnerships in that sense [11,12,13]. However, more often than not, lack of trust and knowledge sharing hinder interorganizational collaborative efforts in the public sector [14].

Still, the extant literature offers insufficient guidance in finding the relational factors that enable public organizations to create public value from interorganizational relations [4,15]. There is still much to be learned about the ways to develop shared understandings for productive relationships to create and maintain IT services in public administration and, ultimately, create value for citizens through such IT services [16]. Therefore, the main research question that motivates this article is: what is the impact of relational factors between service providers and clients on public value creation in public–private IT outsourcing?

This article argues that the lack of guidance from the literature referred to above stems from an insufficient specification of knowledge interfaces at the organizational boundaries between contracting members in their relations. Most contributions focus on knowledge-sharing or allude to the importance of knowledge management without elaborating this notion in detail. However, a refined view of knowledge management at interorganizational boundaries is needed to lead to performance [4,17,18].

In this article, a specific model based on institutional theory [19,20] and boundary object theory [21] was designed and validated to extend the relational view of interorganizational collaborations [22,23]. This research operationalizes and validates conceptual models previously proposed in the literature to explore in greater depth the role of knowledge interfaces across organizational boundaries to enhance public value creation in public–private IT outsourcing [24,25].

Thus, this article contributes to the existing literature by proposing an integrated theoretical model, refining the notion of knowledge interfaces in this sense. The model was validated through a quantitative study of IT outsourcing projects in Mexico’s public administration at the state level. Our model is original, as it explores not only the immediate outputs of the relationship (i.e., the quality of the information service developed) but the impact of those outputs on the creation of public value. Only a few contributions have validated or formulated integrated models based on the relational view, and even fewer have done so in the public sector, in spite of the relevance of such a framework in this context [15,26]. Notably, we show that knowledge-sharing cannot be reduced to a single dimension to account for the complexity of underlying processes at the interorganizational interfaces in public–private ITO relations. As analyzed here, pragmatic boundaries must be distinguished from semantic boundaries to support value creation. This study’s results also show that the relations between trust and performance are mediated by the knowledge interfaces and by the quality of IT artifacts and services produced, thanks to efficient knowledge-sharing interfaces. Consistently with the relational and boundary objects approaches, a distinct relational factor, e.g., ambidexterity in the mobilization of formal and informal communication channels to organize the exchange of knowledge under different forms, was added to the model.

After the introduction, a review of the literature analyzes the types of structures that public administrations develop in public IT outsourcing relations and puts into evidence significant research gaps in this respect. In Section 3, a theoretical framework, a research model and related hypotheses, linking processes, structures, and outcomes are presented according to a relational view of interorganizational relationships and boundary object theory. In Section 4, the research methodology is introduced. In Section 5, the results of the questionnaire applied in accordance with our model and variables are analyzed. In the discussion section, the results are compared with the existing literature to highlight the importance of factors to organize the interface between clients and service providers, from the public and private sector, respectively. Finally, in the conclusion section, research limitations and suggestions for future research are formulated.

2. Literature Review

In this literature review, the determinants of public value in public–private IT outsourcing relations are examined. Public–private IT outsourcing relations tend to adhere to the principles of the new public governance, which emphasize the importance of cooperation, partnerships, and interorganizational collaboration with various stakeholders in the implementation of public policies [27,28], taking inspiration from theories of collaborative governance [29,30]. Outsourcing relationships do require the adoption of a collaborative approach to public management. Such a collaborative approach can be defined as the organized and structured process through which various groups, both public and private, develop, implement, and evaluate collective strategies to achieve public value [6]. Interorganizational collaboration means that governments must increasingly share resources, information, and skills with external stakeholders, particularly from the private sector, to implement actions on a partnership basis to collectively achieve results that no organization could have obtained on a separate basis [11,31,32].

2.1. ITO in the Public Sector

Public–private ITO is different from outsourcing within the private sector. Public–private ITO arrangements often involve complex contracts and agreements due to the involvement of government regulations such as public procurement rules, specific audit requirements, specific dispute resolution processes, and provisions for transparency and accountability. Public–private ITO relations also represent a specific type of relations, as they suppose the existence of a relational contract requiring specific adaptive mechanisms, such as joint decision-making structures based on coordination, negotiation, and communication for mutual adjustment [33], as well as specific administrative controls stemming from bilateral dependency to deal with increased monitoring requirements and costs in comparison with in-house administration, such as long-term contracts supported by contractual safeguards and relational norms for information disclosure and dispute-settlement [34].

One other major distinctive ITO characteristic lies in the integration of public interest concerns into contractual and organizational features to develop projects in order to balance the needs of the public and diverse stakeholders with the goals of efficiency and cost-effectiveness [35]. Hence, this research proposes using public value as a measure of the impact of the information systems and services (IS) produced through ITO [36,37,38]. Information system quality measures the desired characteristics of an information system, including usability, availability, reliability, adaptability, and response time from the user perspective [39].

The concept of public value has proven useful to understand the wider impacts of IS in public management and digital government research [40]. The public management literature also offers many ways to understand the concept of public value [41]. Public value is fluid and continuously shaped by aggregate but diverse expectations and perceptions of value by a variety of stakeholders, including public managers [42,43,44,45]. Public value is defined through the notions of operational efficiency, transparency, equity, cooperation and trust between government and citizens, citizen participation, and the advancement of democratic values [46,47,48,49,50,51]. The measurement of public value frequently divides public value between internal public value, i.e., efficiency and effectiveness, on the one hand, and external public value, i.e., citizen-centric outcomes such as user satisfaction, fairness, equality, transparency, accountability, and increased participation, on the other hand [8,49,52,53]. Public administrators are supposed to maintain and enhance these different dimensions of public value while considering the methods to develop IS projects under stringent requirements and specific rules, routines, and values [54,55].

The success of public–private IT outsourcing relies on the private sector’s supposed advantages over the public sector to achieve public value. The advantages of the private sector are assumed to be superior cost-efficiency and effectiveness, stronger accountability, and superior technical skills and resources due to specialization [2,51] to compensate public administrations’ shortages in this area [56]. Such advantages are also related to knowledge complementarities, which can be created between the public and private sectors in those settings [30].

Meanwhile, public–private ITO presents specific risks and hazards related to cultural differences between the private and public sectors and possible “hold-up” risks from service providers from the private sector [57]. In addition, outsourcing may increase administrative complexity, further degrading public value in the provision of IT-related public services [52]. Outsourcing may also increase environmental uncertainty and therefore risks, thus undermining accountability [58].

Actually, few of these partnerships seem to lead to improved public service delivery [59]. IT service providers often fail to incorporate essential organizational processes into public administration systems to serve governmental goals [7]. Thus, it becomes critical to maintain productive client–vendor relationships to reach the benefits that IT outsourcing is supposed to bring and avoid the risks it carries [12,60,61]. To do so, we propose to look at those relations through the lens of the relational view of interorganizational collaborations.

2.2. Relational View of Interorganizational Collaborations

Dyer et al.’s [22] relational view focuses on the implementation of interorganizational relations. In this view, the quality of relationships depends on four main determinants: complementary resources and capabilities; relation-specific assets; knowledge-sharing routines; and effective governance [22]. In this perspective, scholars should pay attention to collaborative mechanisms and instruments, organizational forms or governance structures, and related management and leadership styles supporting the operation of interorganizational collaborative forms [62,63]. Agger and Sørensen [64] show that partners in collaborations must create and maintain a mutual understanding with each other, given diversity and differences in knowledge sets and experiences, thanks to formal and informal agreements. Those formal and informal agreements need to be implemented simultaneously, which calls for ambidexterity, referring, in the context of this study, to the ability to combine formal and informal communication and control procedures in a flexible and pragmatic way [65]. Ambidexterity is expected to play an important role in favoring cooperative behavior [66,67].

The relational view also emphasizes the importance of a shared sense of identity and belonging, along with a common repertoire developed through collaborative processes such as conversations, joint activities, and problem-solving, to create goodwill that fosters cooperative behavior [68].

The relational approach is well-suited to the study of IT development in public organizations, as it intertwines organizational forms and institutions in the design, development, implementation, and use of technology [6,62,69]. In this context, a relational approach points to the importance for public managers to use horizontal governance based on dialogue and negotiation in order to coordinate and make the necessary adjustments [5,6]. Coordination within these arrangements is based mainly on relational mechanisms such as trust, shared values, standards, and mutual consultation, away from rational bureaucratic mechanisms based on control, hierarchy, and vertical chains of command.

In the relational approach, trust reflects the extent to which negotiations are fair and commitments are upheld. It also reflects one party’s belief that its requirements will be fulfilled through future actions undertaken by the other party [23]. However, in spite of its potential relevance to study interorganizational relations between public and private actors, this approach has been seldom used in the public management literature [15,26].

2.3. Importance of Knowledge Interfaces in Public-Private ITO

Collaborative innovation in the public sector depends on the facilitation of the exchange of knowledge and ideas between actors in the public and private sectors to stimulate processes of mutual learning that may improve the understanding of problems and challenges at hand and extend the range of creative ideas about how to solve them [66]. Public managers should strive to establish close relationships with their innovation partners thanks to mutual trust and empowerment, thus creating a common vision instead of using monitoring and control and thereby focusing on public value. Still, more knowledge is needed about how collaborative relations between governments and external partners should be designed and governed to ensure that the focus is put on creating public value [70].

The creation and maintenance of organizational interfaces are important to facilitate the exchange of knowledge, as organizational interfaces in interorganizational relationships shape representations, interpretations, and capacity of cocreation of processes [71], particularly in public sector settings [72]. Cooperation is defined as organizations working together to achieve mutual benefits [12]. Cooperation involves a process of “working together to identify and implement the best possible solutions” [73] more than just adhering to requirements.

What are interorganizational interfaces and why are they important? Collaborative interfaces include objects, communication, and work routines, as well as governance structures that mediate relations in sociotechnical systems [21], in the public sector [74]. Interfaces draw boundaries, work routines, and work dependencies [75]. In interfacing processes, managers use “boundary objects” such as codified manuals of procedures, timeline charts, standardized report forms, contracts, and organizational charts in order to coordinate operations between organizational members or external partners [4,21].

According to Ansell and Gash [29], institutional design conditions collaborative processes, such as face-to-face meetings, steering committees, intranet forums, government resource planning systems (GRPs), joint training, joint reports, business manuals, shared methodologies, or even conversations, for participants to understand each other’s business objectives and processes [76]. People working in these interfaces are called “boundary spanners”: they are in charge of diagnosing and solving problems, evaluating systems, and controlling performance [77].

Karaba and their colleagues suggest that outsourcing interfaces such as contracts may be considered as boundary objects that facilitate alignment in goals and knowledge-sharing in an outsourcing relationship [4]. As stated by Carlile [17,18], boundary objects are characterized by three dimensions, representativeness, concreteness, and transformability, which influence the effectiveness of knowledge sharing in the relation. “Representativeness” means that partners must have a common language or lexicon to align with each other [18]. Still, cocontracting partners need not only share the same terminology but must translate each other’s knowledge. To do so, “concrete” boundary objects must be created and developed as a basis for partners to share know-how, know-where, and know-whom for decisions to be taken in a coordinated way. “Concreteness” means that partners create effective interdependencies in the work achieved between them in a “semantic” way [17,18]. However, the mere existence of those boundary objects at the semantic level is not sufficient. They must be used in practice to transform the relationship in a “pragmatic” way for partners to achieve a sufficient degree of flexibility to adapt rules and practices to adjust to each other [17]. Nonetheless, research on boundary objects suggests that they can effectively build bridges across organizational boundaries; they can also become sources of conflict and reinforce power when the objects are imposed by a group [78,79]. Research on boundary objects also suggests that they are effective only when they are effectively used by individuals to jointly share and transform their knowledge [80]. Moreover, the ability to use the object effectively cannot be taken for granted [81].

In conclusion of the literature review, most contributions in the literature keep an oversimplified view of knowledge-sharing [82]. Thus, it is important to establish a relation between collaborative governance and public value to offer a more realistic view of knowledge interfaces in this respect. The second gap identified in the literature is a lack of empirically validated integrated frameworks, which relate the antecedents of knowledge creation, the quality of IT artifacts and services produced in a collaborative way, and public value in public–private IT outsourcing relationships.

3. Theoretical Framework, Research Model, and Hypotheses

In this article, the notion of knowledge interfaces is used and operationalized as a key aspect to develop complementary resources and capabilities, relation-specific assets, knowledge-sharing routines, and effective governance consistently with the relational approach. At the same time, the antecedents of such interfaces, and their role in creating successful IT projects and public value in public administration, are examined to design an integrated original model.

Consequently, our model of public value creation in public–private ITO, based on the relational view of interorganizational collaborations and the theory of boundary objects, integrates the notions of trust [23,83] and knowledge interfaces [84] as key variables to understand success in public–private IT outsourcing. Our model shows the importance of the quality of organizational interfaces between collaborative partners as an antecedent of the quality of the information system produced and, hence, public value. Trust, under benevolence and technical capability dimensions, appears as a main organizational factor explaining the quality of the knowledge interface.

From the model displayed in Figure 1, the hypotheses explained in the following section were derived. The first relations proposed in this model concern the organizational factors explaining the quality of knowledge interfaces among contracting partners, i.e., technical capability, benevolence, and ambidexterity. The following hypotheses establish relations between knowledge interfaces and value attributes, i.e., the quality of the technical artifacts produced, internal public value, and external public value.

3.1. Relation Between Technical Capability and the Quality of Knowledge Interfaces

Technical capabilities, i.e., competence or expertise, refer to a dimension of trust that captures the skills that enable a party to have influence within a specific domain or technical area to effectively implement tasks related to that area [83]. It highlights and enhances the task- and situation-specific nature of the knowledge to be mobilized between both parts [81]. Technical capability has been identified as a key contribution of the private sector partners within the ITO relationship [2,52,85]. Thus, the first hypothesis was formulated:

Hypothesis H1a. 

Public administration clients’ perception of service providers’ technical capabilities has a positive influence on the quality of knowledge interfaces between them.

3.2. Relation Between Benevolence and the Quality of Knowledge Interfaces

Benevolence constitutes another dimension of trust that has long been considered as an important characteristic in terms of development and success of interorganizational relationships in general [60]. Benevolence corresponds to the expectation that a party will behave fairly, even when the possibility for opportunism is present in interorganizational relationships [85]. Benevolence appears as an antecedent of knowledge-sharing [86,87]. More specifically, thanks to repeated exchanges and reciprocity, trust becomes embedded in a social context, favoring routines to share knowledge [88]. When benevolence is present, collaborating partners are more willing to engage in the cooperative interactions required for knowledge sharing to occur [68]. Researchers have included the notion of benevolence as a basis for trust, such as negotiating in good faith [29], goodwill between partners [11], reliability and honesty [89], and integrity and sincerity between outsourcing partners [12] to favor knowledge-sharing. Thus, the second hypothesis was formulated:

Hypothesis H1b. 

Public administration clients’ perception of service providers’ benevolence has a positive influence on the quality of knowledge interfaces between them.

Service providers’ technical capability, in that sense, is an essential element of trust in outsourcing relations. Expertise from service providers is a factor that leads to trust [60,84,90]. This tends to indicate that the relation between service providers’ technical capability and knowledge interfaces may be mediated by trust in the service provider. Hence, Hypothesis H1c was formulated:

Hypothesis H1c. 

Service providers’ technical capability has a positive influence on the perception of service providers’ benevolence by the client in public administration.

3.3. Relation Between Ambidexterity and the Quality of Knowledge Interfaces

Different IT outsourcing governance mechanisms help organizations achieve different goals in managing their IT services. Prior research has classified these mechanisms into two major categories: contractual (based on formal control and modes of communication) and relational governance (based on informal controls and modes of communication) [91,92]. A balance between relational and contractual governance has been deemed critical for the success of such collaborations [65]. This type of balance is captured under the notion of ambidexterity, as the ability to simultaneously manage formal and informal control and modes of communication in interorganizational relationships [67]. Such ambidexterity responds to tensions resulting from cocontracting partners’ behavior uncertainty, goal misalignment, and information misappropriation or distortion [65,67,92,93,94]. Formal and informal communication channels are crucial for coordination to occur in complex projects involving an intensive exchange of knowledge under different forms [93]. They can support alignment and adaptability for the purpose of project development and effective use of boundary objects [75]. According to Caccamo et al. [95], the role of formal and informal mechanisms to enhance knowledge integration and boundary objects for the sake of innovation in interorganizational projects must still be further explored. However, the contributions of formal and informal mechanisms in this context are still being debated. Hence,

Hypothesis H1d. 

The simultaneous use of formal and informal communication channels between clients in public administration and service providers in IT outsourcing projects has a positive influence on clients’ perception of the quality of knowledge interfaces between both partners.

3.4. Relation Between the Quality of Knowledge Interfaces and the Quality of Information Systems

Boundary objects are essential to manage the interfaces between contracting partners. Boundary objects can take many forms, such as diagrams, maps, prototypes, databases, manuals, models, methods, or contracts. In public–private outsourcing relationships, the contract plays an important role, specifying timelines, communication and coordination mechanisms, and definitions of roles and responsibilities [4]. In this way, boundary objects at the “semantic boundary” in the relationship, provide a medium to share knowledge, producing common grounds for the collaboration and enhancing communication [17]. The presence of such objects appears essential to develop new information services in a collaborative way. Thus,

Hypothesis H2a. 

The quality of semantic knowledge interfaces between clients and service providers has a positive influence on the quality of the information system produced.

However, if a good quality of semantic knowledge interfaces seems necessary to achieve good quality for the information system produced, it may not be sufficient. As mentioned, having the possibility to mobilize boundary objects to manage the relationship is one thing, but using them effectively is another [81]. The presence of those objects may not guarantee that they are going to be exploited to transform the relation and enhance the creation and maintenance of new shared knowledge to solve conflicts or develop new applications [74].

The effectiveness of boundary objects at the “pragmatic boundary” corresponds to the ability to transform the relation between cocontracting partners to find joint solutions to the problems they face [17]. The common knowledge existing between contracting partners may need to be transformed for partners to effectively share and assess knowledge at the boundary, against the path-dependent tendency of actors to emphasize existing knowledge and entrenched customs and interests. Hence, Hypothesis H2b was:

Hypothesis H2b. 

The quality of pragmatic knowledge interfaces between clients and service providers has a positive influence on the quality of the information system produced.

3.5. Information System Quality to Public Value

Outsourcing success is defined as the degree to which predefined objectives are realized [96]. The potential success of government IT projects depends on the quality of the service produced for citizens. The quality of the information system produced is expected to have a positive influence on the organization’s productivity, service quality, and efficiency, among other potential benefits and results, thus connecting to public value, which integrates internal aspects such as efficiency and effectiveness, and external aspects such as user satisfaction and democratic values such as fairness, equality, transparency, accountability, and increased citizen participation [53,54]. Therefore, it can be argued that the quality of the relationship does not result directly in public value but has to be mediated by the IS outsourcing projects’ performance. The quality of IS systems produced may impact both types of public value. Thus, Hypotheses 3a and 3b were formulated:

Hypothesis H3a. 

The quality of information systems produced in public–private ITO projects has a positive influence on internal public value.

Hypothesis H3b. 

The quality of information systems produced in public–private ITO projects has a positive influence on external public value.

4. Materials and Methods

4.1. Sample

This study was part of a research project dealing with the outsourcing of information technology processes in public administration in Mexico. For the purpose of this survey, we indicated to participants that outsourcing, i.e., outsourced services or “outsourcing” of information technology, consisted of the transfer of activities related to information technology (IT) services managed through collaboration with third parties (private entity or public sector agency) on the basis of an annual or multiyear contract.

ITO has been widely adopted by the Mexican federal government since the beginning of the current century, most notably during the period of President Fox [55]. Although outsourcing has played a role in digital innovation at the state level, that role has been much more limited than it has been among federal agencies [97]. Our exploration of outsourcing at the state level in this project suggests that Mexican states tend to outsource mostly application development services and data center management.

In terms of relevant features of the institutional context, the field of action in the exercise of state governments in Mexico depends on the electoral calendar. General elections at the state level are held every six years. Periods of stability in the government team occur between electoral periods, opening up favorable spaces for the improvement of IT projects. When governments change, continuity of projects and interorganizational relations may be jeopardized because of the tendency of new teams to discredit or abandon the work and achievements of their predecessors.

To build the sample in this study, the authors of this article created a directory of 500 key informants in Mexico’s IT departments (CIOs or high-level executives in charge of the implementation and coordination of IT projects at the organizational level) in state public administration in different ministries. The information was extracted from the 32 state government transparency portals in Mexico and through direct phone calls to check and update the information to ensure a reliable listing. To gather data for this study, the Qualtrics survey platform was used. The survey was carried out for three months, between February and April 2018. Participants were invited to respond to the survey, with their anonymity guaranteed. To improve the response rate, two reminders were sent to nonrespondents at one-month intervals.

As governments do have multiple projects at once, public administrators in our sample were asked to answer the survey questions about one specific project in order to control the information received. Seventy-one percent of the projects were completed less than one year before the key informant answered the questionnaire, and twenty-nine percent of the projects were nearing completion when the key informant responded to the questionnaire.

Table 1. Types of IT projects in the sample size.

IT Project	Number	%
Software development	29	47%
Data center operations	20	32%
Digitalization of documents and files	9	15%
Cloud services	4	6%
Total	62	100%

reveals that those projects corresponded to the development of outsourced services usually requested by public administrators. A variety of services were observed, though a large majority, in the study’s context, concerned software development and data center operations, including maintenance.

Sixty-two participants responded to the survey in its entirety. This represents a rate of usable response of 12.4%, which is within the bounds of the average response rate for online surveys in academic research [98] and in the range of response rates of other articles using the same kind of sample survey target (local public officials in charge of projects) [99]. Further, the sample can be considered as representative of the population involved (decision-makers), all the more so as a large majority, over 80%, of Mexican states were represented in the sample. Twenty-six out of thirty-two states were represented. The six remaining states were Campeche, Chiapas, Nayarit, Guerrero, Tamaulipas, and Oaxaca. In this way, most major states in Mexico were actually represented. As noted by Cook et al. [100] (p. 81), the representativeness of responses is a more crucial issue than the response rate itself. With a coverage of 80% of Mexican states in the sample, there was a satisfactory level of response representativeness such that the sample adequately represented the target population.

4.2. Measurement

The variable measurement items are presented in

Table 2. Measurement items.

Category	Item
Benevolence
BENE 1	The service provider considers the welfare of my organization, as well as its own, when making important decisions in this project.
BENE 2	The service provider considers the interests of my organization and their organization when problems arise.
BENE 3	The service provider makes beneficial decisions to us under any circumstances.
Technical capability
TECH 1	My service provider is very capable of performing their job in this project.
TECH 2	My service provider is known to be successful at the things they do.
TECH 3	My service provider has specific technological capabilities that can increase our performance.
Ambidexterity	When coordinating activities with the service provider, we use formal channels (official reports and briefings, formal planned meetings, logs and other official communications through standard procedures, etc. …) and informal channels (face-to-face conversations, phone conversations, and email or text messages through unplanned interactions or meetings) …
AMB 1	… to exchange knowledge.
AMB 2	… to resolve conflicts.
AMB 3	… to adapt or change processes.
Semantic interface
SEM 1	We and our service provider share manuals, models, and methodologies with each other.
SEM 2	We and our service provider share know-how from work experience with each other.
SEM 3	We and our service provider share each other’s know-where and know-whom if decisions need to be taken.
Pragmatic interface
PRAGMA 1	My service provider and I have different understandings of our project goals.
PRAGMA 2	Often, during the collaboration with the service provider, I initially think I understand them, but afterwards, this turns out not to be correct.
PRAGMA 3	I have a different perception of solutions to problems than my service provider.
PRAGMA 4	It is hard to come to a joint solution with my service provider.
Information system quality
INF 1	Info systems produced through the collaboration are easy to use.
INF 2	Info systems produced through the collaboration are useful.
INF 3	Info systems produced through the collaboration can be customized to our needs.
Internal public value
PV 1	During the course of the collaboration, we have been able to reduce the cost of operations in our organization in this project.
PV 2	During the course of the collaboration, we have been able to access top-class, state-of-the-art IT capabilities in this project.
External public value
PV 3	During the course of the collaboration, we have been able to create a more transparent government thanks to this project.
PV 4	During the course of the collaboration, we have been able to increase systems’ timeliness for the citizen thanks to this project.
PV 5	During the course of the collaboration, we have been able to increase systems’ user friendliness for the citizen thanks to this project.
Project complexity
COMPLEX_1	The project entailed new technical requirements for our organization.
COMPLEX_2	The project included new technologies.
COMPLEX_3	The project involved the coordination of multiple units.
Experience
EXP_1	Number of years of tenure in the present position.

. Likert scales were used to measure responses to each question or statement of the variables under study. To analyze the data, the code ranged from 1 = strongly disagree to 5 = strongly agree. For trust-related dimensions (technical capability and benevolence), existing scales from Lee and Kim [11] were used. For ambidexterity between formal and informal communications, the authors of this article built their own scales, as none were available in the existing literature. These scales were elaborated to measure the quality of semantic and pragmatic knowledge interfaces. To do so, Carlile’s definitions [17,18] were used to design two distinctive scales. One scale, about semantic knowledge interfaces, measured via three items the presence of boundary objects, such as manuals, methodologies, and models, by which the know-how, know-whom, and know-where for coordination needs and potential conflicts may be addressed. The other scale, on pragmatic knowledge interfaces, assessed boundary objects in the way they were effectively, pragmatically used to transform relations, examining the ability of actors to reach joint solutions together, aligning their views. In the four items, we used reverse coding to reduce response bias [101].

The quality of the information system produced was measured by four items taken from the enacted technology scale of Picazo-Vela et al. [54]. To measure internal and external public value, this article considered the public value operationalization developed by Picazo-Vela et al. [54]. Their operationalization integrates efficiency, effectiveness, user satisfaction, and democratic values as the main outputs of public value. We adopted those dimensions to differentiate and measure internal and external public value. As shown in Table 2, two items were used to assess internal public value, while three items captured external public value.

4.3. Data Analysis Method

In this study, we used partial least-square structural equation modeling (PLS-SEM). This approach is now widely applied in many social science disciplines, particularly in management information systems [102]. It is a causal-predictive approach to SEM that emphasizes prediction in estimating statistical models [103]. PLS-SEM estimates partial model structures by combining principal component analysis with ordinary least squares regressions [103,104,105]. In contrast with CB SEM, which uses the covariance matrix of the data and estimates the model parameters by only considering common variance, PLS-SEM accounts for the total variance and uses the total variance to estimate parameters [102]. According to Goodhue et al. [106] and Hair et al. [102], researchers should select PLS-SEM when the analysis is concerned with testing a theoretical framework from a prediction perspective; when the structural model is complex and includes many constructs, indicators, and/or model relationships; or when a relatively small population restricts the sample size, which is the case here.

Our SEM analysis adhered to established guidelines from the literature on general research methods [102,107] and the specific field of digital government research [105,108]. Per Hair et al. [102], for small sample sizes, PLS-SEM is generally preferred over other SEM techniques because of its superior statistical power. Moreover, PLS-SEM stands out as the best option because of its flexibility and robustness [107,108]. To check the sufficiency of the sample size, an alternative to simplifying the model that does not involve collecting more data is to adopt the inverse square root method, which uses the inverse square root of the size of the sample in estimating standard error [109]. In this approach, we discarded a path coefficient that was too low in the context of a given sample size, as it was nonsignificant regardless of the corresponding p-value [109]. At a 10% significance level, with a sample size of 62, the minimum path coefficient for indicators should be at least 0.269, according to the formula

$$n_{min}>{\left.\left({\displaystyle \frac{2.123}{\mid P_{min}\mid }}\right.\right)}^2$$

Thus, the relation between experience and information systems quality (path coefficient = 0.241), even though the p-value was above 2, was not validated. Other relations kept the same validity level.

Furthermore, Hair et al. [107] and Gil-Garcia [108] recommend a sample size at least 10 times greater than the number of indicators used to define the formative construct with the most indicators. Similarly, the sample size should be 10 times larger than the number of structural paths directed at the latent variable with the most connections in the structural model. Given these considerations and the characteristics of our research model, a sample size of 40 could be deemed sufficient to conduct PLS-SEM analysis.

5. Results

Items were adequately related to their constructs, as reported in Figure 2. The results of construct validity are reported in

Table 3. Construct validity.

Variable	Cronbach’s Alpha	Composite Reliability	AVE
BENE	0.894	0.897	0.826
TECH	0.869	0.87	0.792
AMBI	0.881	0.881	0.808
SEM	0.801	0.807	0.717
PRAGMA	0.917	0.918	0.801
ISQ	0.907	0.907	0.843
IPV	0.841	0.841	0.863
EPV	0.954	0.954	0.915
COMPLEX	0/827	0/897	0.744

and

Table 4. Discriminant validity (HTMT matrix).

	1	2	3	4	5	6	7	8
Technical capability
Benevolence	0.411
Ambidexterity	0.243	0.149
Pragmatic knwl. interface	0.585	0.399	0.292
Semantic knwl. interface	0.478	0.517	0.483	0.339
Information system qual.	0.695	0.537	0.209	0.567	0.595
Internal public value	0.878	0.501	0.367	0.615	0.452	0.756
External public value	0.783	0.538	0.210	0.452	0.394	0.703	0.944

, which show Cronbach’s alpha, composite reliability (CR), and average variance extracted (AVE) values for each construct. Those values can be considered as satisfactory given the usual thresholds for such indicators in exploratory research [102]. The results reported in Table 4 showed satisfactory discriminant validity between constructs. The values of AVE were greater than the nondiagonal values, which confirmed the discriminant validity of the model.

Table 5. Outer model using bootstrapping method [104].

Dependent Construct	Independent Construct	Path Coefficients and p Values
Benevolence R2 (adj) = 0.118	Technical capability	0.364 *
Semantic interface quality R2 (adj) = 0.275	Benevolence	0.306 *
	Technical capability	0.230
Pragmatic interface quality R2 (adj) = 0.289	Benevolence	0.184
	Technical capability	0.421 ***
Information system quality R2 (adj) = 0.571	Semantic interface quality	0.396 ***
	Pragmatic interface quality	0.341 **
Internal public value R2 (adj) = 0.395	Information system quality	0.583 ***
External public value	Information system quality	0.585 ***
R2 (adj) = 0.377
Complexity	Information system quality	0.399 ***
	Internal public value	−0.003
	External public value	−0.001
Experience	Information system quality	0.241
	Internal public value	−0.112
	External public value	0.035

Sig. (* p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001).

shows the results of the inner model. Out of the seven hypotheses formulated, four hypotheses were validated, and three were partially validated. H1b was partially validated, as benevolence was significantly and positively related to semantic knowledge interfaces, but it was not significantly related to pragmatic knowledge interfaces. It was expected that benevolence could offer the conditions to elaborate solutions together between service providers and clients, based on a common understanding between both. H1a was also partially validated, as technical capability, i.e., the other dimension of trust apart from benevolence, was not directly and significantly related to semantic knowledge interfaces, but it was significantly and positively related to pragmatic knowledge interfaces. However, it can be argued that technical capability was still positively and significantly related to semantic knowledge interfaces through the mediation of benevolence, as H1c indicated. H1d was partially validated, as ambidexterity in the use of formal and informal channels of communication was significantly and positively related to semantic knowledge interfaces but not to pragmatic knowledge interfaces.

Meanwhile, H2a and H2b were validated, as both knowledge interfaces had significant impacts on the quality of the information systems. H3 was validated as well for both dimensions of public value, showing the relevance of operating distinctions between two different dimensions of knowledge interfaces (semantic and pragmatic) and between two distinct dimensions of public value (internal and external public value).

There was no direct significant relation between semantic knowledge interfaces and internal or external public value and no significant relation between pragmatic knowledge interfaces and internal or external public value. The mediation was partial, and weakly so (p-value: 0.027), only in the relation between pragmatic knowledge interfaces and internal public value. Thus, mostly, public value is not directly determined by the quality of knowledge interfaces, as the relation between knowledge interfaces and public value is mediated by the quality of the IT service or artifacts produced.

Two control variables, i.e., experience and project complexity, were integrated. Interestingly, complexity was significantly related to the quality of the information system produced but was not related to any dimension of public value directly. The relations between experience and quality of information systems and public value were not validated in our sample.

6. Discussion

The model proposed in this article deals with the factors determining the performance of interorganizational relationships in the context of the public sector, which is an important question to address, as public–private ITO does not necessarily lead to good performance, given that parties frequently experience problems in coordination and shared knowledge. In doing so, this approach leads to improvements over existing models.

Ansell and Gash’s article stands as a common reference in the study of “collaborative governance” [29] to drive decision-making processes in interorganizational relationships. Several authors in the literature [11,61] agree with Ansell and Gash [29] to acknowledge that the quality of partnership positively influences IT outsourcing success, reducing information asymmetry between cocontracting partners. Swar et al. [12] proposed a model where partnership depends on the development of key capabilities (i.e., communication, cultural compatibility, flexibility, information sharing, confidentiality maintenance, and conflict handling).

The efficacy of knowledge management has been frequently cited as an important dimension of partnership quality to perform outsourcing arrangements between clients and vendors from the public and private sectors, respectively [96]. The authors recognize the importance of knowledge management recommend leaders to play the role of facilitators; however, “facilitating leadership” remains fuzzy and vague in their perspective. Facilitating leadership is associated with building trust, maintaining ground rules, guaranteeing the conditions for dialogue to take place, exploring mutual gains, etc. [29]. The notion of “shared understandings” remains underspecified in the literature [12,29].

Levina and Vaast [71] made substantial progress by showing that organizations use devices such as codified manuals of procedures, timeline charts, standardized report forms, contracts, and organizational charts as boundary objects to manage interorganizational relations. Carlile also revealed that knowledge sharing is a multidimensional notion that cannot be reduced to a single dimension [17,18]. Moreover, research has shown that the effectiveness of a collaboration at the boundary depends not only on the characteristics of the objects but on the ability to use them to effectively translate and transform knowledge and align goals [80,81].

Those models and perspectives have their merits, identifying key variables. To improve them, however, it is important to establish several key insights that can be validated by empirical study, as was done in this article.

First, trust is a multidimensional notion that should encompass two distinctive facets: technical capability and benevolence. Second, the relation between trust and performance is mediated by the capabilities to manage interorganizational interfaces that trust affords. Third, in addition to trust-related factors, the ability to simultaneously manage formal and informal communication networks between service providers and clients is a crucial factor to enhance knowledge interfaces. Fourth, knowledge interfaces are consequences of trust rather than antecedents, in the sense that trust favors “shared understandings”, e.g., common problem definition and identification of common values in joint action, negotiation, and conflict resolution, consistently with Zaheer et al. [85]. Fifth, knowledge-sharing cannot be reduced to a single latent variable to account for the complexity of underlying processes at the interorganizational interfaces in public–private ITO relations. Knowledge-sharing does not occur spontaneously; it is the product of purposive actions depending on the types of knowledge interfaces developed.

Our study suggests that only technical capability, an important antecedent of trust, matters for pragmatic knowledge interfaces. This result was related to the nature of the outsourcing projects included in the sample, which included outsourcing infrastructure-related projects, such as data centers or cloud services. In this kind of project, knowledge transformation may be less relevant, as public managers do not necessarily need to change their practices as they move their infrastructure outside their organizational boundaries. Another potential explanation may involve some other specific aspects of public administration that may prevent the effective use of the knowledge interfaces to produce transformation [81]. Among these reasons, quick turnover of key staff, legal constraints related to inflexible laws on purchasing services in the public sector in Mexico, short-term budgets, and a lack of a well-functioning career planning system can be found, thus hampering the development of experience and knowledge in this context.

We expected ambidexterity to be a strong predictor for the ability of partners to develop joint solutions together [67,93,95]. While ambidexterity, per our definition, appeared to be useful for sharing knowledge in the form of manuals and protocols in a smooth and efficient way, it did not seem to contribute to pragmatic knowledge interfaces. Therefore, it is neither necessary nor sufficient in that sense, as benevolence or formal and informal communication channels by themselves cannot compensate the relative absence of technical capability to solve problems arising from collaborations between service providers from the private sector and public administrators.

Finally, the relation between knowledge interfaces and public value was mediated by the quality of the IT service or artifacts produced. This result was consistent with the sociotechnical approach frequently found in the literature on technology-enabled change, such as structuration and institutional theories [19,20]. In these approaches, particularly in Fountain’s enactment theory, organizational and institutional factors shape the selection of technological features or “enactments” of technology. These enactments mediate the production of value [54]. In the model proposed in this paper, particular organizational arrangements embedded in the semantic and pragmatic interfaces enable (or make it harder) to produce a specific technology enactment, which in turn produces public value. This perspective is also consistent with the more general literature on system success found in the information systems literature [36,37], which looks for causal connections between technology artifacts and the creation of value.

Measuring and empirically validating relevant knowledge interfaces is important and necessary to avoid the oversimplification mentioned earlier when following a relational view of interorganizational relationships complemented by the theory of boundary objects applied to IT outsourcing relationships in the public sector. Overall, it is important to account for the differences between public and private organizations in terms of goals, core values, and relational mechanisms.

Recent studies [110] have shown that, while outsourcing may help mitigate the effects of financial constraints, budget cuts, in times of crisis such as pandemics, often target outsourced services, potentially leading to disruptions in essential operations. This could be dangerous for public service continuity. Thus, essential sectors, such as health, education, sanitation, and public security, often retain outsourced services because of their critical nature, e.g., in the context of public services in Brazil [110]. Thus, we do not expect major changes in the robustness of the results as of today in comparison with the time when our survey was implemented.

7. Conclusions

In this article, the quality of interface as the main determinant for public value in the delivery of IT-based public services was examined. As a theoretical contribution, an integrated theoretical model based on the relational view is proposed, complemented by the theory of boundary objects, to refine the notion of knowledge interfaces. Knowledge interfaces were examined in a novel way in this context, significantly contrasting semantic knowledge interfaces and pragmatic knowledge interfaces, while most of the existing literature has tended to limit itself to a unidimensional view of knowledge in interorganizational relationships [89]. Our model also has the originality to integrate the notion of ambidexterity in the simultaneous use of formal and informal communication channels to support knowledge interfaces. The existence of a mediation between knowledge interfaces and two distinct types of public value was established through the use of the notion of quality of the IT services/applications produced. The research model was validated through a quantitative study of IT outsourcing projects in Mexico’s public administration at the state level.

In terms of practical implications, this research shows that public managers need to actively manage the relationship in outsourcing configurations as opposed to separating themselves from the execution of the activity or the task. Managing the relationship can be translated into several formal and informal activities.

• Public managers need to cultivate trusting relationships with their service providers. As suggested in the literature, building trust in a relationship can be most effectively developed through continuous effort and repeated interactions [83,87].
• Building trusting relationships requires attention to the technical competency and benevolence of ITO partners. Public managers involved in the outsourcing process should include an evaluation of the potential for service providers to become trusted partners in a long-term relationship as a criterion in their selection. Beyond the selection process, cultivating trust will continue to be a success factor [85].
• Public managers should redefine their role, keeping strong operational interfaces with the providers’ systems, supported by well-written relational contracts. To do so, there must be some degree of preparation from service providers and a strong capacity of absorption from clients. However, public sector organizations must prepare themselves to redefine their activities as orchestrators of interorganizational relationships.
• Building and maintaining strong operational interfaces involves considering how trust and knowledge interfaces can be developed and nurtured. In this way, besides strategies to build an effective and trusted relationship, public organizations should not abandon the maintenance of associated knowledge and skills related to the project, actively investing in acquiring understanding of technical tools and solutions.
• As our model suggests, relational contracts should include a renewed sense of ambidexterity between formal and informal channels to be able to negotiate changes and adjustments in the contract in an effective manner and to increase chances of success. As projects evolve, understanding of the goals and requirements also evolve. Therefore, it becomes necessary to continuously adjust project components to better respond to evolving requirements.
• In addition, it appears that sharing documentation and methods provides only a precondition for effective alignment between service provider and government agency, given the distinction between the pragmatic and semantic dimensions of knowledge interfaces. Clear procedures and responsibilities, coordination, and arbitration of disagreements are important as well, given that those elements facilitate effective conflict resolution and adjustment in the contractual relationship at the pragmatic level.
• In the particular context of Mexico and other developing countries, understanding the difficulties of sustaining relationships over time should encourage public managers to promote contracts with lengths not limited to the period of current administration and electoral deadlines, as is still too often the case.
• Finally, proper attention should be given to the internal and external dimensions of public value in orientating collaborative efforts with private service providers, who may lack sufficient cultural awareness of their clients’ necessities in that perspective.

This study faces certain limitations. It would be interesting to examine interactions between the variables under study due to system dynamics representations, e.g., as in Luna–Reyes [111]. Future studies that would include larger samples from diverse contexts could increase the generalizability of the findings. Future research may consider the inclusion of multiple stakeholders’ perspectives on one specific project. Future research could also examine complementary external factors potentially affecting public value under its different forms in ITO relations, such as communication methods, cultural differences, regulatory environments, project management practices, the technical capacity of public administrators, technology resistance from the citizen, or budget constraints. This research could be extended by exploring more facets of public value among the seven categories of public value identified by Jørgensen and Bozeman [53].