Explaining Coordination Quality in Public Service Delivery

Abstract

Governments manage coordination problems with different levels of success. The existing literature suggests that organizational features are important for explaining coordination quality, that is, how successful governments are in managing coordination problems. However, few studies have systematically compared coordination quality in multiple political systems. This article examines how governments’ organizational features explain variations in the coordination quality of public services in 259 local political systems in Norway. Three organizational features are considered, namely governments’ use of coordination instruments, their structural specialization, and task interdependence between their services. For the analysis, the article combines existing and original survey data on the coordination of public service delivery. Multilevel regression analysis indicates that public managers perceive coordination quality to be better when local governments introduce more coordination instruments and when task interdependence is higher. However, local governments’ structural specialization does not correlate with perceived coordination quality. These findings are important because they suggest that local governments have the capability to facilitate coordination. Moreover, local governments’ coordination efforts inform how public managers perceive their own abilities to address coordination problems and, by extension, foster more coherent service delivery.

Keywords:

• coordination quality
• local government
• public service delivery
• structural specialization
• task interdependence

Introduction

Governments’ success in managing coordination problems varies. Coordination, a persistent challenge in all governments, involves getting different, and often interdependent, organizations to contribute toward a common goal (Koop & Lodge, 2014, p. 1313). Coordination is required because the activities of one organizational unit often depend on those of another. The effectiveness of public service delivery is, therefore, related to the quality of coordination between the organizational units involved in service provision. Consequently, coordination has practical implications for public problem solving (Kapucu & Hu, 2022; Lodge & Wegrich, 2014; Satheesh et al., 2023). Understanding why and under what conditions coordination quality varies between governments can, therefore, improve our general understanding of public service delivery.

Governments’ organizational features are important factors in explaining variations in coordination processes and outcomes (Egeberg, 2012). A similar analytic distinction applies to coordination quality, which can be considered a process or an outcome (Christensen & Lægreid, 2020). This article adopts the latter perspective. As an outcome, coordination quality refers to the extent to which organizations manage, mitigate, and resolve coordination problems. Building on the work of Christensen and Lægreid (2020), this article aims to answer the following research question: To what extent do governments’ organizational features explain variations in their coordination quality?

In the extensive literature on public sector coordination (see, for example Gulick, 1937; Scharpf, 1994), there is a strong focus on explaining how and why governments engage in coordination (Bach & Wegrich, 2019; Bouckaert et al., 2010; Senninger et al., 2021; Trein & Ansell, 2021). A subset of this literature focuses on coordination as an outcome (for a discussion, see Peters, 1998). Existing studies have considered several potential explanations for why coordination outcomes vary, such as how relational preconditions, for example trust, inform coordination outcomes in regional transport organizations (Davoudi & Johnson, 2022), how formal and informal coordination mechanisms impact coordination effectiveness in health crises (Kapucu & Hu, 2022; Soujaa et al., 2021), and how governments’ organizational and cultural features explain coordination quality in 17 European countries (Lægreid et al., 2016). However, these studies tend to be either very broad or very narrow in their empirical focus, thus providing limited systematic explanations of coordination outcomes. Cross-country comparisons, such as those of Lægreid et al. (2016), face challenges in controlling for contextual variations and excluding competing explanations. At the same time, studies of single political systems, such as Christensen and Lægreid (2020) and Senninger et al. (2021), offer no means for comparing systems.

This study contributes to the current literature by systematically comparing the coordination quality of public services in 259 local political systems in Norway. All public services provided by local governments were included in the analysis. Although all Norwegian local governments have the same legal status and provide the same services, they differ in their organizational features. Thus, the research design leverages variations across local governments to evaluate how their organizational features explain differences in coordination quality between their services.

This article considers three organizational-level explanations that may account for differences in coordination quality: the structural specialization of governments, the scope of governments’ coordination instruments, and the level of task interdependence between organizational units. Christensen and Lægreid (2020) explicitly linked the first two features to coordination quality but not the third. Task interdependence is one reason why organizational units require coordination (Elston et al., 2018). Consequently, governments may seek to limit task interdependence through structural design. However, because policy problems often cross organizational boundaries governments vary in how much they limit task interdependence between their services (Wegrich & Stimac, 2014). To use structural specialization as a proxy for task interdependence is, therefore, not uncommon in the coordination literature (see, however, Kopp & Lodge, 2014). To uphold the important analytical distinction between structural specialization and task interdependence, this article includes task interdependence as a separate determinant of coordination quality.

Empirically, this study examined public managers’ perceived coordination quality in delivering local public services as a dependent variable. The focus was on coordination between organizational units within the same government rather than coordination between different governments. The analysis combined existing and original survey data, with a total sample of 1,417 public managers responsible for public service delivery in 259 Norwegian local governments. The results from the cross-sectional, multilevel regression analysis suggested that structural specialization is unrelated to perceived coordination quality, whereas a government’s use of coordination instruments and level of task interdependence are associated with better perceived coordination quality. This study highlights the potential benefits of coordination instruments as a means for local governments to ensure more coherent service delivery.

The rest of the article is structured as follows: The first section introduces the concept of coordination quality. The second section presents the analytical framework and discusses the three organizational features and their relationship with coordination quality. The third section presents the research approach, the data, and the variables. The fourth section highlights the findings and discusses them in light of the theoretical expectations. The concluding section discusses the implications of the findings for theory and practice and considers the limits of the study as well as avenues for future research.

Coordination quality

Coordination quality has no broadly established definition in the public administration (PA) literature. In their study, Christensen and Lægreid (2020) argued that coordination quality can reflect both the quality of coordination processes and the quality of coordination outcomes. Process-related quality concerns coordination across internal and external, or vertical and horizontal, divisions of government. Outcome-related quality, which is the focus of the present study, has to do with what coordination achieves in terms of policy coherence. Although policy coherence is a pertinent coordination problem and thus an important part of coordination quality, I contend that coordination quality is more than policy coherence. A more nuanced definition is needed to enhance the analytical usefulness of coordination quality as a concept. Therefore, this study builds on Christensen and Lægreid (2020) reasoning by considering how governments manage a broader set of coordination problems.

Conceptually, it can be difficult to distinguish between process and outcome. However, approaching coordination quality as an outcome enables distinguishing formal features, such as coordination instruments, from their outcomes. Analytically, formal features and outcomes are different phenomena (Christensen et al., 2016; Soujaa et al., 2021). A similar argument is provided by Kapucu and Hu (2022, p. 776), who argue that “it is challenging to evaluate coordination as an outcome without carefully examining coordination activities, structures and processes.”

The following section conceptualizes coordination quality. The argument rests on the assumption that, for public managers, coordination involves problems that must be overcome (Koop & Lodge, 2014). To illustrate this argument, the article focuses on five generic coordination problems derived from the PA literature. I do not claim that the list is exhaustive; rather, it represents the coordination issues typically addressed in the literature. Following this line of reasoning, coordination quality is defined as the extent to which organizations manage, mitigate, and resolve coordination problems.

From coordination problems to dimensions of coordination quality

The coordination problems that have received the most attention in the PA literature are redundancy, lacunae, and incoherence (Peters, 1998), as well as informational and motivational problems (Scharpf, 1994). Table 1 shows the relationships between these coordination problems and the various dimensions of coordination quality. This does not mean that each problem has only one “solution”; rather, the table demonstrates my general argument. Importantly, my emphasis is not on coordination problems per se but on how well they are managed, mitigated, and resolved as perceived by public managers. For example, the problem of policy incoherence, as emphasized by Christensen and Lægreid (2020), can be managed through policy consultation. In this case, the evaluation of coordination quality is based on how “effective” policy consultation is in mitigating policy incoherence or ensuring policy coherence. This outcome-oriented logic is applied to all dimensions of coordination quality.

Table 1. Conceptualizing Coordination Quality.

Generic problem	Problem manifestation	Dimensions of coordination quality
Redundancy	Overlap	Assignment of responsibilities
Lacunae	Underlap	Procedural requirements to work together
Incoherence	Policy inconsistency	Policy consultation and direction
Informational problems	Information mismatch	Information exchange
Motivational problems	Silo politics Opportunistic behavior Shirking	Developing joint identities (joint problem awareness) Developing joint identities (problem-solving mindset)

Note. Based on Peters (1998), Koop and Lodge (2014), and Wegrich and Stimac (2014).

The aforementioned coordination problems manifest in different ways. For example, redundancy refers to the overlap between public services—that is, situations in which policies and actors address the same issue, which sometimes results in redundant uses of governments’ resources (Koop & Lodge, 2014). Overlap can be managed by assigning responsibilities to clarify which service or policy is handled by which actors. Lacunae refer to situations in which no policies and actors are responsible for a certain issue (Wegrich & Stimac, 2014). This phenomenon, known as underlap, often causes actors to neglect specific tasks and can be managed by introducing procedural requirements to work together. Incoherence involves policy inconsistency, a pervasive issue in bureaucratic policymaking (Senninger et al., 2021). Policy inconsistency can be managed by implementing policy consultations and direction to avoid that contradictory decisions by different actors effectively negate one another (Koop & Lodge, 2014). Informational problems often entail information mismatches and a reluctance to share information, which hamper coordination because information is a requirement for effective coordination (Koop & Lodge, 2014; Scharpf, 1994). Information mismatch can be managed by facilitating information exchange between services. Finally, motivational problems involve “silo politics,” which are often accompanied by shirking and opportunistic behavior, indicating a general lack of motivation to engage in coordination (Bach & Wegrich, 2019). The result is typically that organizations act without regard for other organizations (Scharpf, 1994). As a structural issue, silo politics cannot be solved, but it can be managed by developing joint problem awareness or joint identities between services.

Explaining coordination quality: Analytical framework

One important tenet in organization theory is that formal organizational features inform administrative behavior, such as coordination quality (Egeberg, 2012). For example, Gulick (1937) argued that coordination is qualitatively different depending on the type of structural specialization. According to organization theory, an organization’s formal features produce systematic attention biases that influence the probability of public managers acting or perceiving parts of their work in specific ways (Bach & Wegrich, 2019; Christensen & Lægreid, 2020, pp. 149–151). Consequently, organizational features, such as the three studied in this article, likely influence managers’ perceived coordination quality.

This study considered the following three organizational features: (1) the scope of coordination instruments, (2) the specialization of administrative structures, and (3) the level of task interdependence between organizational units. Figure 1 illustrates how coordination quality is studied in this article. Task interdependence is considered an important mediator in the relationship of coordination instruments and structural specialization with coordination quality. Given that all three features constitute key variables in the study of coordination, it is important to scrutinize their relationship with coordination quality separately and together.

Figure 1. Analytical framework.

Scope of coordination instruments

Governments use multiple instruments to coordinate their services (Bouckaert et al., 2010). The adoption of coordination instruments signifies governments’ purposeful attempts to foster coordination. The scope of coordination refers to the number and breadth of coordination instruments that a government uses to facilitate coordination between services. The literature provides examples of coordination instruments that are positively and significantly correlated with coordination quality, such as systems for sharing information and/or reporting results (Christensen & Lægreid, 2022), and cross-cutting policy programs and working groups (Lægreid et al., 2016). Moreover, studies have demonstrated that governments’ coordination practices are characterized by substantial “hybridity,” whereby “coordination often plays out as a combination of hierarchy, negotiations/networks and markets” (Christensen & Lægreid, 2020, p. 147; Kapucu & Hu, 2022). According to Bouckaert et al. (2010, pp. 268–269), there is potential for “synergies” when combining different coordination instruments. The actual occurrence of synergies is not a given and is essentially an empirical question. However, the notion of synergies suggests that multiple coordination instruments do not necessarily negate each other; instead, they can collectively improve governments’ capacities for coordination (Bouckaert et al., 2010, pp. 268–269). Consequently, coordination quality is expected to be higher when the scope of coordination is greater:

• H1: The greater a government’s scope of coordination, the better the coordination quality as perceived by public managers.

Specialization of governments’ administrative structures

Specialization involves the structuring of work between different parts of the government apparatus (Egeberg, 2012). Governments are specialized horizontally, that is, divided into different but equal ministries (Senninger et al., 2021). Horizontal specialization reflects the internal division of work between organizational units at the same administrative level (Hall et al., 1967). Another type of structural specialization involves establishing units “outside” the traditional ministerial bureaucracy. Such organizational units, which are established at an arm’s-length distance from political control, are often called agencies and the process of creating them agencification (Van Thiel & Smullen, 2021). Agencies are characterized by their relatively autonomous position in relation to executive leadership and their structural separation from parent ministries.

The specialization of a government’s administrative structure (either horizontal or agency based) provides the basis for coordination. Managing a large number of organizational units, which occurs when structural specialization is high, increases the challenge of ensuring coordination (Bouckaert et al., 2010). The greater the number of units involved in managing a coordination problem, the greater the heterogeneity of the interests to be coordinated, which diminishes the probability that a coordination problem will be managed effectively (Koop & Lodge, 2014). Consequently, the potential for successful coordination is lower when structural specialization is higher:

• H2: The more specialized a government’s structure, the lower the coordination quality as perceived by public managers.

Task interdependence

The concept of task interdependence describes situations in which organizational units require resources, such as expertise or information, held by others (Elston et al., 2018; Lundin, 2006, p. 653; Park, 2022). Because interdependent tasks require coordination, organizations attempt to limit the level of task interdependence (Koop & Lodge, 2014; Wegrich & Stimac, 2014, p. 45). One solution is to try and assign all interdependent tasks to the same organizational unit. Consequently, task interdependence manifests itself as the extent to which managers from different but interdependent units coordinate with one another. Although such contact patterns constitute “networks” of interactions between services, they are not necessarily networks or collaborations in the formal and conceptual sense (Griffin et al., 2023).

The above hypotheses regarding the impact of structural specialization and the scope of coordination on coordination quality do not consider the fact that interdependence may vary between units in terms of tasks. As Lundin (2006) argued, the level of task interdependence may condition the impact of organizational features on interorganizational coordination. First, the scope of coordination is less likely to positively influence coordination quality when task interdependence is low. If task interdependence is low, a greater scope of coordination may not be required to ensure good coordination quality because the demand for coordination is already limited:

• H3a: The positive relationship between the scope of coordination and coordination quality is contingent on the level of task interdependence.

Second, structural specialization is less likely to negatively influence coordination quality when the level of task interdependence is low. When task interdependence is low, the demand for coordination is limited, and greater structural specialization is less likely to hinder good coordination quality:

• H3b: The negative relationship between structural specialization and coordination quality is contingent on the level of task interdependence.

Data and methods

Research context

This study examined Norwegian local governments, one specific type of public organization. All Norwegian local governments, 356 as of 2020, have the same legal status and are legally obliged to provide the same set of services. Local governments, which are financed through a mix of funding from the central government and their own revenues, accounted for 32.8% of total government expenditure in 2021 (OECD Fiscal Decentralization Database).

Many important public services are delivered by local governments, including primary education, sewage, health care, nursing, and care services. Thus, Norwegian municipalities have a sizable task portfolio, and their overall local autonomy (relative to the central government) is among the highest in Europe (Ladner et al., 2019). This autonomy includes making decisions about internal organization, such as how to organize service delivery. Local governments perform executive functions only. Moreover, local governments decide how to address the coordination demands between their services. Whereas some local governments formalize specific “coordination goals” for policy areas that require considerable service coordination, others do not. In the field of public health, for example, local governments vary considerably in their uses of goal setting. Moreover, some local governments have permanent “coordinator” positions to ensure service coordination for specific client groups, such as people with dementia or cancer, and family-oriented services. This study examined the coordination between local governments’ own services and did not address coordination with private organizations, other local governments, and the central government. Because some local governments provide services through intermunicipal arrangements, a small share of local governments’ coordination practices involve services in different local governments. However, these situations are exceptions rather than the rule. The initial decision to rely on intermunicipal service delivery lies with local governments, as does the responsibility of coordinating these services.

Data sources

In this study, I combined existing and original survey data. First, I used existing data on local governments’ internal administrative structures from the local government survey. The local government survey contained data on administrative structures, including the organization and provision of 22 executive tasks. These executive functions covered the range of services provided by Norwegian local governments (Jones et al., 2021). The local government survey was conducted on behalf of the Ministry of Local Government and Regional Development in 2020–2021. A total of 260 local governments answered the questions on their administrative structures, which amounted to approximately 70% of the Norwegian local governments at the time of data collection.

Second, I collected data using an original survey on local governments’ service delivery coordination (henceforth the coordination survey). Invitations to answer the coordination survey were sent to operational managers in the 260 local governments that had answered the local government survey. Operational managers head organizational units responsible for providing one or several public services. All local public services were included, from preschool and health care to road maintenance and waste collection. Based on the 260 local governments, I identified an initial sample of 5,548 operational managers. The survey was completed by 1,475 operational managers, yielding a response rate of approximately 27%. The response rate was modest, which is common in executive surveys of this kind (Hammerschmid et al., 2013). The survey was representative of the population of local governments and of the population of operational managers. The full list of survey items is provided in the supplementary materials.

Although the study combined different data sources, it employed several variables from the coordination survey, making common source bias (CSB) a potential methodological issue (Jakobsen & Jensen, 2015). Through the survey design, I implemented several measures to limit the potential for CSB. The first measure concerned question types. Importantly, factual questions, such as those involving observable behaviors, are less affected by CSB than other types of questions, especially “evaluative” questions (Meier & O’Toole, 2013, p. 447). Therefore, the empirical analysis included factual independent variables, such as coordination instruments. The only evaluative variable in this study was the dependent variable (for similar strategy, see Bjurstrøm, 2020). Second, the survey items appeared in the following order: the evaluative dependent variable, followed by the factual independent variables. By administering the evaluative question first, I avoided the possibility of respondents presenting themselves as consistent by adjusting their evaluations of coordination quality to those of previous questions (Meier & O’Toole, 2013), such as regarding the number of coordination instruments used. Presenting the factual questions last made this possibility less likely. Furthermore, the two questions were separated by 30 survey items, which further reduced the potential for biased responses. Third, to limit the potential for social desirability responses, the questions were worded to explicitly avoid asking for respondents’ self-evaluations; instead, the respondents were asked to evaluate the joint coordination initiatives with which they had been involved. Moreover, indicators were kept specific rather than general, and I avoided using similar scales for the dependent and independent variables (Meier & O’Toole, 2013; Podsakoff et al., 2012). Finally, despite the methodological challenges discussed above, the survey method remains a commonly used method in the literature (Christensen et al., 2016; Soujaa et al., 2021).

Estimation strategy

The coordination survey was designed to be combined with the local government survey. This coupling of data sources resulted in an inherently hierarchical data structure because operational managers are nested in local governments. Multilevel regression is a suitable method for analyzing hierarchical data. However, the intraclass correlation coefficient was low (0.048), which suggests that multilevel regression techniques were not strictly necessary. Nonetheless, due to the clearly hierarchical nature of the data, I opted for multilevel regression.

Dependent variable: Perceptions of coordination quality

In this study, the dependent variable was operational managers’ perceptions of coordination quality along different dimensions. The measure of coordination quality was based on data from the coordination survey, which included eight items measuring respondents’ perceptions of coordination quality. The question was worded as follows: “Consider your cooperation with other municipal services. Over the last five years, to what extent has cooperation in which you have been involved accomplished the following: …” This question was followed by a list of coordination quality dimensions (Table 2). Importantly, this question asked public managers to evaluate the cooperation in which they had been involved rather than their own contributions and/or performance. Table 2 provides an overview of the coordination quality dimensions. The responses were measured on a 7-point Likert scale. “I do not know” was not offered as a response category, and the respondents were not required to answer all questions to complete the survey. The sample contained 1,417 respondents from 259 different local governments after all missing responses were removed.

Table 2. Operationalization of Coordination Quality.

Dimensions of coordination quality	Operationalization in the coordination survey
Assignment of responsibilities	Clarifying the assignment of responsibilities between services
Procedural requirements to work together	Uncovering blind spots or gaps in service delivery
Policy consultation and direction	Coordinated service delivery (“end-product”)
	Holistic/integrated case management
Information exchange	Maintaining ongoing information exchange between services
Developing joint identities (joint problem awareness) Developing joint identities (problem-solving mindset)	Resolving disagreements between services
	Sustaining/fostering collaborative relationships
	Shared understanding of goals across services

Note. Based on Peters (1998), Koop and Lodge (2014), and Wegrich and Stimac (2014).

Given the difficulty of measuring coordination quality objectively, the perceptions of those involved in coordination are likely to yield the best available information on coordination quality. As the few existing studies indicate, coordination quality is a matter of perspective because no single public manager in a government can be expected to have a full overview of all problems and ideal solutions (Christensen & Lægreid, 2020). Thus, each public manager offers a partial perspective on the coordination quality of a government. Operational managers are uniquely situated in the coordination process and can provide unique perspectives on what works well (or not) in the coordination of public services.

To test the usefulness of coordination quality as an analytical concept, I performed an exploratory factor analysis of the eight survey items measuring coordination quality (see supplementary materials). The analysis indicated that a one-factor solution was most appropriate for the data. Only one factor qualified according to Kaiser’s criterion (eigenvalue > = 1). This result was in line with my initial theoretical expectations. The factor analysis indicated that it was meaningful to discuss coordination quality as a construct. Factor loadings for all eight items measuring coordination quality exceeded the 0.60 threshold. The internal reliability of the factor was very high (Cronbach’s alpha = 0.91). Factor scores were extracted and used as an indicator of coordination quality in the multilevel regressions (for a similar strategy, see Jacobsen, 2022). However, initial analyses with each item of coordination quality as a dependent variable yielded the same results.

Explanatory variables

Scope of coordination

I used data from the coordination survey to construct an index for the scope of coordination. The index captured the number and breadth of coordination instruments used to facilitate coordination between public services. The index was based on a modified version of Bouckaert et al. (2010) typology of coordination instruments. The coordination survey employed this typology, including 10 categories of coordination instruments, with a total of 28 instruments (for a full list, see supplementary materials). Each category of coordination instruments included a set number of coordination instruments, each of which was initially measured on a three-point ordinal scale: “not used,” “used to some extent,” and “used to a large extent.” To limit the potential for CSB between the independent and dependent variables, the responses for each coordination instrument were dichotomized as either “used” or “not used.” This involved removing ex post the evaluative aspect that could have contributed to CSB. This approach ensured that the indicators were factual. Moreover, this strategy did not force the respondents to answer either yes or no, which could have resulted in acquiescence, or “yes-saying” (Meier & O’Toole, 2013).

To construct a measure for each coordination instrument category, I divided the number of coordination instruments used by the total potential number of instruments in that category, typically three. The index for the scope of coordination was calculated by summing the score for each of the 10 coordination instrument categories, yielding a measure that ranged from 0 to 10. A score of 0 indicated that no coordination instrument from any category was used, whereas 10 indicated that all coordination instruments from all categories were used.

Specialization of governments’ administrative structures

In this study, the ways in which local governments specialized their administrative structures were important explanatory variables. Two types of structural specialization were included in the analysis. First, horizontal specialization, which indicates the number of units within the same administrative tier, was measured as the number of service delivery units within a local government (Hall et al., 1967). In a local government, the number of service units equals the number of operational managers. The variable was constructed as an ordinal scale ranging between 0 and 10, with higher scores indicating greater horizontal specialization of the administrative structure. Each value corresponded to a set interval, with 0 indicating a number of service delivery units between 0 and 9, 1 indicating a number of units between 10 and 19, and so forth. A value of 10 indicated a number of units equal to or greater than 100.

To measure agencification, I used data from the local government survey. The degree of agencification indicated the extent to which local governments relied on agencies for their service delivery (for a similar strategy, see Overman & van Thiel, 2016). The local government survey included data on the organization of 22 executive tasks delivered by local governments (for the classification of agencies, see supplementary materials). These 22 tasks covered virtually the entire range of public services provided by Norwegian local governments. Therefore, the measure effectively captured the level of structural disaggregation of a local government’s service delivery. The degree of agencification was calculated by dividing the number of tasks provided by an agency in a given government by the total number of tasks. The score indicated the share of total tasks delivered by agencies in a local government. The measure ranged from 0 to 1, with 1 indicating that all 22 tasks were delivered by an agency.

Task interdependence

The data for measuring task interdependence were collected from the coordination survey. Instead of asking all operational managers about their dependence on other units, which may have reflected the extent of their actual coordination with others rather than their dependence (Lundin, 2006, pp. 659–660), the survey asked operational managers to indicate all services with which they coordinated at least monthly. The number of services ranged from 0 to 20. If an operational manager indicated one or more of their own services, these services were excluded from the count. The level of task interdependence was calculated by dividing the actual number of services with which operational managers coordinated monthly by the total potential number of services. The measure ranged from 0 to 1. A score of 0 indicated that there was no regular coordination, meaning that task interdependence was negligible. A score of 1 indicated that a unit coordinated with 20 services monthly, which suggested that task interdependence was substantial. This measure captured the demand that each unit had for coordination with other services provided by the same government and was thus a suitable proxy for task interdependence (Griffin et al., 2023, p. 470). Moreover, the measure did not use respondents’ self-evaluations and relied exclusively on observable, factual data, which limited the potential for CSB (Meier & O’Toole, 2013, p. 447).

Control variables

The analysis contained three organizational-level controls. First, I included size as a control variable. Size can be measured as organizational size, which is indicated by either population or personnel size. Both measures are likely related to how much coordination is required and how much organizational capacity is available for facilitating coordination (Andrews et al., 2016). Moreover, size is sometimes used as a proxy for the structural complexity of an organization (Bjurstrøm, 2020). Local government size was measured as personnel size using data on full-time equivalents from Statistics Norway. I used a logged version of size due to skewness. Size correlates strongly with horizontal specialization and was thus included in separate models to avoid multicollinearity (see the Supplementary Material).

Second, I used data from the local government survey to control for the number of administrative levels. Although all public managers included in this study had the same formal position in the administrative hierarchy, the number of administrative levels varied among local governments (Hall et al., 1967). The administrative level was measured as the number of administrative levels between the operational managers and the chief administrative officer. The number of administrative levels indicates how close operational managers are to the chief administrative officer in a local government, which can affect their perceptions of coordination quality (Christensen & Lægreid, 2020).

Tasks, or sectoral differences, are sources of variation in coordination quality. Differences in coordination quality are likely between services that vary in terms of their complexity, saliency, or technicality (Christensen et al., 2019; Pollitt, 2006). In the analysis, tasks were included as a set of 11 dummy variables indicating operational managers’ sectoral affiliations. The classification of sectoral affiliation followed the standards of Statistics Norway.

Similar to other studies of coordination quality, I controlled for two individual-level features (see, for example Christensen & Lægreid, 2020): tenure, measured as the number of years spent in a position, and education, measured as the level of the highest completed education. Gender was not significant in the initial analyses and was thus removed completely from the analysis.

Descriptive statistics

Descriptive statistics for the variables are available in the supplementary materials. On average, local governments had between 30 and 40 service delivery units and delivered approximately half of the 22 tasks through agencies (a score of 0.44). This implies that a high number of operational managers worked in local governments with administrative structures whose specialization involved a mix of internal and semi-autonomous service delivery units. Moreover, the mean scope of coordination was practically in the middle of the scale (5.04), which means that, on average, governments used a fair number of coordination instruments to facilitate coordination between their services. Lastly, task interdependence had a mean of 0.44, which indicated that, on average, operational managers were in contact with approximately 10 services monthly for coordination purposes. This suggests that many services depend on coordination with other services to perform their own tasks adequately. The level of task interdependence varied substantially among operational managers (SD = 0.24) in different municipalities and among sectors (see supplementary materials).

Findings and discussion

The results from the multilevel regression analysis are shown in Table 3 and indicate three main findings. First, the analysis demonstrated that a greater scope of coordination correlated with a higher perceived coordination quality (Model 1). This finding is in line with the theoretical expectation that when the scope of coordination instruments used by local governments is higher, public managers perceive coordination quality to be better. Thus, the analysis indicates that adopting coordination instruments can improve governments’ management of coordination problems. The positive correlation between the scope of coordination and perceived coordination quality supported H1.

Table 3. The Impact of Organizational Features on Perceived Coordination Quality (Multilevel Regressions).

	Dependent variable: coordination quality
	(1)	(2)	(3)	(4)	(5)	(6)
Scope of coordination (index)	0.159*** (0.012)		0.152*** (0.013)		0.152*** (0.025)	0.154*** (0.025)
Horizontal specialization		−0.002 (0.014)		−0.003 (0.013)		−0.012 (0.013)
Degree of agencification		0.042 (0.218)		0.035 (0.214)		−0.008 (0.203)
Task interdependence			0.291*** (0.104)	0.534*** (0.108)	0.293 (0.277)	0.310 (0.278)
Sector: Reference category is primary and lower secondary school
Sector: Preschool /childcare	0.328*** (0.091)	0.406*** (0.096)	0.336*** (0.091)	0.414*** (0.095)	0.336*** (0.091)	0.337*** (0.091)
Sector: Internal administration	0.393*** (0.106)	0.400*** (0.112)	0.348*** (0.107)	0.316*** (0.112)	0.348*** (0.107)	0.342*** (0.107)
Sector: Child welfare and social services	0.333*** (0.098)	0.433*** (0.103)	0.329*** (0.098)	0.416*** (0.102)	0.329*** (0.098)	0.328*** (0.098)
Sector: Culture	−0.025 (0.110)	−0.036 (0.116)	−0.012 (0.109)	−0.013 (0.115)	−0.012 (0.110)	−0.017 (0.110)
Sector: Health care, nursing, and care services	0.276*** (0.079)	0.368*** (0.083)	0.280*** (0.078)	0.369*** (0.082)	0.280*** (0.078)	0.279*** (0.078)
Sector: Technical services	0.312*** (0.105)	0.273** (0.111)	0.337*** (0.105)	0.323*** (0.110)	0.337*** (0.105)	0.341*** (0.105)
Sector: Planning, building permits, maps, and surveying	0.359*** (0.105)	0.337*** (0.111)	0.365*** (0.105)	0.350*** (0.111)	0.365*** (0.105)	0.364*** (0.105)
Sector: Municipal property management	0.301 (0.254)	0.284 (0.269)	0.255 (0.254)	0.202 (0.267)	0.255 (0.255)	0.257 (0.255)
Sector: Municipal enterprise	0.626*** (0.184)	0.616*** (0.195)	0.680*** (0.185)	0.714*** (0.194)	0.680*** (0.185)	0.676*** (0.185)
Sector: Miscellaneous	0.346*** (0.100)	0.424*** (0.106)	0.337*** (0.100)	0.400*** (0.105)	0.337*** (0.100)	0.337*** (0.100)
Number of administrative levels	0.006 (0.025)	0.022 (0.028)	0.006 (0.025)	0.020 (0.028)	0.006 (0.025)	0.013 (0.027)
Tenure (years)	0.002 (0.006)	−0.003 (0.007)	0.0005 (0.006)	−0.005 (0.007)	0.0005 (0.006)	0.0005 (0.006)
Education level	0.098*** (0.038)	0.108*** (0.040)	0.094** (0.037)	0.100** (0.040)	0.094** (0.037)	0.098*** (0.038)
Task interdependence × scope of coordination (interaction term)					−0.0004 (0.050)	−0.004 (0.050)
Constant	−1.313*** (0.145)	−0.592*** (0.168)	−1.388*** (0.147)	−0.784*** (0.170)	−1.389*** (0.180)	−1.390*** (0.198)
Number of municipalities	259	259	259	259	259	259
Random intercept (municipality)	0.04	0.05	0.04	0.04	0.04	0.04
Observations	1,417	1,417	1,417	1,417	1,417	1,417
Log likelihood	−1,865.807	−1,943.332	−1,863.278	−1,932.468	−1,865.353	−1,869.064
Akaike inf. crit.	3,765.613	3,922.663	3,762.555	3,902.936	3,768.707	3,780.128
Bayesian inf. crit.	3,854.971	4,017.277	3,857.169	4,002.806	3,868.576	3,890.511

Notes. Unstandardized coefficients, standard errors in parentheses. Estimation method is maximum likelihood. Reference category for sector is primary and lower secondary school.

Significance levels: *p < .1; **p < .05; ***p < .01.

Second, the analysis did not find that greater structural specialization correlated with lower perceived coordination quality. Instead, the analysis revealed a non-significant, non-directional correlation between the level of specialization (that is, horizontal specialization and agencification) and perceived coordination quality (Model 2). The strong assumption that structural specialization biases public managers in their perceptions of coordination quality does not hold under empirical scrutiny. Consequently, H2 was not supported.

Third, I hypothesized that the relationships between the scope of coordination and coordination quality (H3a) and structural specialization and coordination quality (H3b) are contingent on the level of task interdependence. On the one hand, the analysis found that task interdependence correlated positively and significantly with perceived coordination quality. On the other hand, the inclusion of task interdependence did not alter the substantial findings regarding the scope of coordination, which remained positive and significant (compare Models 1 and 3), and structural specialization, which remained non-significant (compare Models 2 and 4). Thus, this finding does not support the assumption that task interdependence conditions the impact of either coordination instruments (H3a) or structural specialization (H3b) on perceived coordination quality. Interestingly, the analysis indicated that task interdependence had a more direct relationship with coordination quality.

All results were robust when controlling for sectoral affiliation, the number of administrative tiers in the municipality, organizational size, and respondents’ tenure and education level. Organizational size was non-significant and did not alter the substantial results in any model (see the supplementary material). Several dummies for sectoral affiliation correlated significantly with perceived coordination quality, indicating that the complexity, technicality, and saliency of the sectors biased public managers’ perceptions of coordination quality (Christensen et al., 2019). Moreover, education level was positively and significantly correlated with perceived coordination quality, which suggests that not all individual features may be as unimportant for coordination quality as the current literature suggest (Christensen & Lægreid, 2020; Soujaa et al., 2021).

Finally, task interdependence may result from local governments’ efforts to coordinate different organizational units, known as “ex-post interdependence” (Elston et al., 2018, pp. 1819–1820). For example, local governments’ coordination practices may influence the extent to which their organizational units can deliver services unilaterally. To control for a potential interaction between the scope of coordination and task interdependence, two additional models were estimated with an interaction term between these variables. The interaction term was not significant in either of the models (Models 5 and 6), which suggests that there is likely no considerable “ex-post interdependence.” Moreover, the coefficient for the scope of coordination was robust when the interaction term was included, but the coefficient for task interdependence was not.

The empirical analysis indicates that although local governments’ formal organizational features explain variations in coordination quality, the features considered in this article matter differently (Egeberg, 2012). Only one of the four hypotheses was supported. First, the analysis suggests that the adoption of formal coordination instruments is conducive to public managers’ perceived success in managing coordination problems. According to the most optimistic view, expanding the scope of coordination was perceived as a valuable investment in more coherent service delivery, which is in line with existing studies of coordination quality (Christensen & Lægreid, 2022; Lægreid et al., 2016; Soujaa et al., 2021) as well as studies of cross-sectoral collaborations (Satheesh et al., 2023). This finding provides tentative evidence regarding the potential of synergies between coordination instruments, as suggested by Bouckaert et al. (2010, pp. 267–269). Admittedly, the analysis cannot demonstrate the actual presence of synergies. Instead, the analysis showed that local governments benefit from adopting hybrid “mixes” of coordination arrangements (Kapucu & Hu, 2020). The combination of coordination instruments into different mixes may, therefore, be of some value in practice, at least according to those responsible for coordinating services (Bouckaert et al., 2010, pp. 267–269). This finding paints an optimistic picture of the potential to achieve coordination between different services in a government.

Second, the analysis indicates that one should be careful in assuming that perceived coordination quality mainly reflects the specialization of a government’s administrative structures. The assumption that specialization influences coordination quality is likely the clearest and most discussed theoretical assumption in the literature (Bouckaert et al., 2010; Dan, 2014; Egeberg, 2012). Surprisingly, this analysis did not find any evidence in favor of this assumption for either horizontal specialization or agencification. This finding is especially surprising in the case of agencification, which adds to the complexity of managing public service delivery (Van Thiel & Smullen, 2021). While greater levels of specialization indicate a greater heterogeneity of interests to be coordinated, this does not necessarily translate into poorer perceptions of coordination quality. Although this finding contradicts conventional wisdom, it is not uncommon to find a non-directional correlation between specialization and coordination in the literature (Overman, 2016, pp. 1252–1253). When one relates this finding to the literature, typically studies of national governments, some interesting similarities emerge. Even at the national level, where the potential for “siloization” is likely much stronger due to extensive sectoral specialization, Christensen et al. (2016) found that structural-organizational features affected perceived coordination quality less profoundly than did cultural-institutional features (see also Lægreid et al., 2016). Therefore, when interpreting the current findings in the context of local service delivery, where the potential for siloization is lower, it may not be so surprising that structural specialization does not correlate significantly with coordination quality. Moreover, local public service delivery has low political sensitivity compared to tasks such as policymaking and regulation at the national or even supra-national levels (Senninger et al., 2021), which also supports this interpretation of the findings. In addition, organizations specialized by territory, as exemplified by local governments, are generally expected to produce better coordination within their territories compared to other forms of specialization (Christensen & Lægreid, 2020; Gulick, 1937). The findings of this study support this expectation. In this study, public managers were only asked to assess how they managed internal coordination problems—that is, problems with other services belonging to the same local government. In other words, the coordination of services at the local level may benefit from being under the shadow of the same hierarchy (Scharpf, 1994). This contextual feature may also help explain why, in this study, structural specialization was unrelated to perceived coordination quality—the formal distance between the studied units was generally short.

Third, the analysis suggests that the level of task interdependence between services is related to perceived coordination quality. This is unsurprising because task interdependence is one reason why organizational units coordinate (Elston et al., 2018; Wegrich & Stimac, 2014). However, the analysis does not provide any evidence that task interdependence conditions the impact of either the scope of coordination instruments and/or structural specialization on perceived coordination quality. Although the findings for task interdependence contradict the initial expectations, they highlight the importance of considering task interdependence when studying coordination, as previously demonstrated by Lundin (2006) and Park (2022). In addition, this analysis provides empirical support regarding the significance of studying the influence of task interdependence separately from that of structural specialization (Wegrich & Stimac, 2014). Furthermore, the positive correlation between task interdependence and coordination quality is surprising because task interdependence involves coordination problems (Koop & Lodge, 2014). However, Park (2022, pp. 374–375) also found that task interdependence in organizations can facilitate improved “cooperative behavior,” in line with the present analysis. Moreover, studies have shown that a “history of interactions” leads to shared values, trust, and norms of reciprocity (Davoudi & Johnson, 2020, p. 19; Soujaa et al., 2021). The benefits of such relational dynamics have also been stressed in the literature on cross-sectoral collaborations (Christensen, 2023; Satheesh et al., 2023, p. 421). In the context of local service delivery, where all services are embedded in the same organizational structure, task interdependence may simply be conducive to achieving these relational dynamics. This would help explain the finding that public managers responsible for more interdependent services, rather than less, perceived themselves as more successful in managing coordination problems.

Conclusions

This study demonstrates that organizational features explain variations in governments’ coordination quality, supporting a common assumption in organization theory. Although the findings are in line with existing research, they indicate that organizational features vary in their importance. Out of the three organizational features studied in this article, that is structural specialization, the scope of coordination instrument, and task interdependence, only the scope of coordination instruments supported the initial theoretical expectations. Thus, this study demonstrates the need for more theory testing in contemporary scholarship on public sector coordination (see also Trein & Ansell, 2021).

This study contributes to the literature by providing a more empirically comprehensive and systematic comparison of governments’ coordination quality than what is currently available. More specifically, I compared coordination practices across all services provided by Norwegian local governments (instead of considering only a selection of services in a single government), which is rare in studies of coordination (for a notable exception, see Senninger et al., 2021). By using a combination of original and existing survey data, the analysis empirically identified two types of structural specialization, horizontal specialization and agencification, and distinguished task interdependence from structural specialization. Although the distinction between task interdependence and structural specialization is theoretically and empirically significant, it has been scarcely investigated in the literature. Moreover, the analysis considered a broader range of local governments’ coordination instruments and developed a more nuanced conceptualization of coordination quality compared to the literature (Davoudi & Johnson, 2022; Souja et al., 2021).

This study has several implications. In terms of theory-building, the study further develops coordination quality as an analytical concept and demonstrates how it can be used to study coordination outcomes empirically. Moreover, the analysis empirically demonstrates that task interdependence matters for coordination (Elston et al., 2018). This finding reaffirms the importance of distinguishing structural specialization from task interdependence when studying coordination. This study demonstrates once again that structural specialization does not influence coordination quality, or coordination as an outcome. Thus, it is possible that structural specialization may be more important when considering coordination processes than coordination outcomes (Bouckaert et al., 2010).

In practical terms, this study shows that in the context of local governments in which all services fall under the shadow of the same hierarchy, it is possible to ensure high-quality coordination. The study suggests that governments can facilitate coordination between services by providing more coordination instruments; at least, this is the perception held by public managers. Thus, in addition to centrally decided coordination instruments, it may be helpful to give operational managers some discretion in how they facilitate “local” coordination between services. Furthermore, the study suggests that public managers should be less concerned regarding the detrimental effects of structural specialization; instead, they should facilitate relationships between services, as suggested by the finding on task interdependence. In the context of this study, task interdependence is likely conducive to building trust. Consequently, local governments and their public managers should try and limit “external” coordination to situations in which it is absolutely needed (Heims, 2019).

This study has certain limitations. One is the lack of causal evidence because the analysis only used cross-sectional data. The findings must be interpreted accordingly. However, the theoretical arguments presented in this article are derived from decades of PA scholarship, which makes strong causal assumptions about the relationships between the variables included in this study (for a discussion, see Bouckaert et al., 2010). In the end, more research, both longitudinal and involving other contexts, is needed to validate the extent to which my findings constitute causal patterns. Second, the measure of coordination quality was based on public managers’ perceptions. Although this is a common strategy in the literature, it nonetheless means that the analysis did not measure coordination quality objectively. This limits the generalizability of the findings to contexts in which coordination is “internal” to organizations, such as local governments in other countries. Furthermore, generalizability depends on the tasks being coordinated, which, in this study, included everyday service delivery. Service delivery differs considerably from more politically salient tasks, such as regulation and policymaking, which are common at the national level. The more different a task is from service delivery, the lesser the potential for generalization. In the future, scholars should consider how to systematically assess and compare governments’ coordination outcomes, both within and across different types of public organizations.

Acknowledgements

I would like to thank Signy Irene Vabo and Tobias Bach for their valuable comments on this paper and the participants in the research group on Policy, Bureaucracy and Organization at the Department of Political Science, University of Oslo, for their feedback on an earlier version of this paper. I also wish to thank the anonymous reviewers for their very helpful suggestions.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data that support the findings of this study are not available due to national ethical guidelines.

Additional information

Funding

This article is part of a PhD-project funded with grant from the Norwegian Ministry of Local Government and Regional Development.

Notes on contributors

Leif E. Kårtvedt

Leif E. Kårtvedt is a PhD candidate at the Department of Political Science, University of Oslo. His research focuses on public service delivery in local government.