Microsoft word - collabservices_wp.doc

Collaborative Service Arrangements:
Patterns, Bases, and Perceived Consequences
Peter J. May
Center for American Politics and Public Policy Søren C. Winter
Danish National Centre for Social Research This paper has later been published in Public Management Review 9 (4): 479-502. Acknowledgements
Mads Stigaard, Mette F. Sørensen, Ina R. Bøge, Nina Friisberg, Helle N. Jensen, SFI- Survey and UNI-C provided substantial research and other assistance with this project. The authors are grateful to anonymous reviewers, Bodil Damgaard, Carolyn Hill, Laurence Lynn, Kenneth Meier, Vibeke Lehmann Nielsen, and Laurence O’Toole for their advice, to the municipal managers who participated in the study, and to AKF and ECO-analyses for provision of some of the data. The research was supported by grants from the Research Program of the Danish Ministry of Employment and the Danish National Institute of Social Research for a project directed by Søren C. Winter. The findings are not necessarily endorsed by the sponsoring organizations or the survey Collaborative Service Arrangements:
Patterns, Bases, and Perceived Consequences
Abstract
While much of prior research on collaboration addresses the service delivery network as a whole, we address collaborative relationships between one type of organization— municipal employment services—and a range of governmental and non-governmental partners for employment services in Denmark. Municipalities differ in the type, degree, and character of collaboration with these partners. As others have found in prior research, we find that organizational benefits, trust, and a variety of contextual factors help shape the extent of collaboration. But, the relevance of these and problem-solving benefits in particular differs among collaborators. Our modeling of the influence of collaboration on perceived employment outcomes suggests that these impacts are relatively minor. They are greater when there is active involvement of municipal employment managers in fostering cooperative relationships with collaborators. In short, collaboration requires a healthy and active relationship to foster improved outcomes. These findings have implications for future research about collaborative service delivery concerning the measurement of collaboration, different bases for it, and potential impacts.
Collaborative Service Arrangements:
Patterns, Bases, and Perceived Consequences
The literature about the provision of social services highlights the importance of collaborative service arrangements. Two key presumptions underlie collaborative arrangements. One is that they enhance service provision by some combination of reducing costs, increasing efficiency, fostering innovation, and enhancing flexibility. A second is that on balance collaboration leads to better service outcomes. These are typically taken as truisms in the more popular treatment of the subject (e.g. Goldsmith and Eggers 2004: 25-38; Osborne and Plastrik 1998: 203-240). But the validity of these presumptions clearly has important implications that are the heart of current research about governance and service delivery (see reviews by Forbes and Lynn 2005, Hill and Any assessment of the implications of collaborative arrangements requires a fuller consideration of what they entail. This research adds to the understanding of collaborative arrangements and of the bases for inter-organizational cooperation. We address collaboration for employment services in Denmark. This setting provides a useful contrast to consideration of collaborative arrangements in the United States. Like the U.S., employment services in Denmark are delivered through a multi-tiered organizational structure of governmental and non-governmental entities. But, the Danish context also engages to a much greater degree unions, private employers, and other organizations in the shared delivery of employment services. This research is part of a larger study of the design and implementation of reforms for employment services. The Danish “Putting People to Work” initiative has entailed stronger emphasis on job placement over other safety net benefits, increased reliance on third-parties for employment training and assistance services, and more entrepreneurial approach of employment service agencies. Conceptual Issues
Collaborative arrangements entail modification of traditional provision of services through governmental hierarchies. Some like Salamon (2002) suggest that these new forms of governance reflect fundamental transformations in the basic forms of governmental services, labeling this “the new governance.” Others like Hill and Lynn (2005) suggest that the changes are less dramatic with governmental organizations still at the core of service delivery, labeling this “polycentric governance.” Consideration of the different forms of collaboration, the reasons for and limits to cooperation among organizations, and the perceived impacts of collaboration are central to our research. Coordination and Collaboration
The literature addressing collaborative arrangements is largely about different ways of coordinating service delivery and is reflective of a longer-standing debate in the organization literature (also see Milward and Provan 2000a). Collaborative arrangements assume multi-organizational delivery of services that is aptly described as a network of organizations. Collaborative arrangements differ in terms of the structure of the network and the degree of formality that binds the organizations within the network. A variety of mechanisms can be used to coordinate activities. The public management networking literature has mainly focused on exchanges that involve formal contracts for service provision by non-governmental organizations. These vary in terms of funding arrangements and contract terms that by definition establish principal-agent relationships (Milward and Provan 2000b). Less attention has been paid to other mechanisms that can be usefully thought of as comprising a continuum that varies from lesser to higher degrees of inter-organizational involvement and that can take on a variety of forms (see Hill and Lynn 2003). The specific exchanges depend on the type of service provision. For example, in studying collaboration activities of American cities, Agranoff and McGuire (2003: 68-85) catalog interactions that include information seeking, adjustments to rules or policies, policymaking assistance, resource exchange, and project- specific actions. In studying delivery of mental health services, Milward and Provan (1998) examine referrals received, referrals sent, case coordination activities, joint The character of collaboration is also important to consider. The presumption in much of the literature is that collaboration is purposeful and that the relevant organizations are willing to cooperate in achieving those ends. But like any partnership, the relationships can be conflict ridden, competitive, cooperative, or neutral. Agranoff and McGuire (2003: 4) suggest collaboration should not be confused with cooperation in that partners are not necessarily helpful to each other. Milward and Provan (2000b) suggest that a challenge for network management is overcoming social dilemmas in which one or more partners’ short-term interests undermine the broader policy objectives. As such, it is useful to remember that each partner in a collaborative undertaking has something at stake and brings in a host of preconceived notions to the partnership. The stakes may be as ethereal as reputation, but often entail more substantive considerations as resources (people and funds), turf, autonomy, or control (Bardach 1998). Bases for Collaboration
There is no shortage of frameworks that have been employed in studying bases for collaboration. A common theme among these is that collaborative partnerships are purposive. Organizations collaborate in order to advance goals of the organization, of the leaders of the organization, or of principals that mandate collaboration. Yet as Weiss (1987) notes, cooperation is not a natural imperative of organizations (also see Hill and Lynn 2003). With this basic caution, we consider several bases for collaboration with particular attention to factors that might explain different patterns in collaboration. These bases are not mutually exclusive. As such, they reinforce each other in ways that cannot be easily disentangled. Our research interest is identifying the relative contribution of Organizational Benefits
The traditional perspective is that organizations will not enter into collaborative arrangements unless they perceive greater benefits than costs from their involvement. This perspective has been discussed with reference to rational choice perspectives (Hill and Lynn 2003), transactions cost frameworks (Feiock et al. 2005, Kruger and McGuire 2005, Lubell et al. 2002), and resource dependency (Lubell 2004, Lundin 2005, O'Toole and Montjoy 1984, O’Toole 2003). We consider hypotheses that relate to the overall benefits of collaboration and more specific problem-solving benefits. H1a: Resource Benefits – Organizations will seek collaborative relationships with others when there are perceived tangible net benefits from the relationship. While the specifics of the rational choice, transaction cost, and resource dependence theories differ, the central point is organizations will not collaborate unless they perceive tangible gains. Resource dependence theories emphasize organizational interdependencies and the benefits or resources that are extracted from collaboration. Transaction-cost perspectives emphasize the costs involved in entering into and maintaining collaborative relationships. An important part of this hypothesis is that the decisions about collaboration are based on perceptions of net benefits, rather than actual benefits. Tangible benefits include new personnel, funding, information, or political support. Relevant costs include the negotiation and administrative costs (management time, shared personnel, out of pocket costs) of entering into and maintaining collaborative H1b: Problem-Solving Benefits – Organizations are more likely to seek collaborative relationships when the service delivery task presents stronger challenges. This hypothesis reflects a functionalist organizational problem-solving perspective. One strand of this theorizing emphasizes the search for solutions to problems that stem from difficulties in the task environment (Weiss 1987). Organizations seek to reduce discordance in their task environment that stems from strong or new external challenges. This may consist of relatively large client loads for which organizations seek to offload those loads to other organizations, a problematic—or increasingly problematic—mix of clients for which organization seek other organizations’ expertise, or other external circumstances (e.g. rise in unemployment rates) that make it difficult to perform business Social Capital
A competing, but not mutually exclusive, perspective to organizational interdependence is that collaboration is based on mutual non-pecuniary relationships among organizations. These foster “social capital” that is a primary basis for overcoming collective action problems in recruitment and retention of members of voluntary networks (Bardach 1998, Lubell et al. 2002, Lubell 2004, Lundin 2005). At issue is what fosters and sustains the mutual relationships. Among the strongest bases for social H2: Trust – Organizations will seek collaborative relationships with others that they trust to follow through on their commitments. This hypothesis addresses at a key psychological basis for cooperation, which is the foundation of much theorizing about collective action institutions (see Leach and Sabatier 2005, Lubell 2004). The basic argument is that organizations, just as individuals, are more willing to cooperate with those they trust to follow through on their commitments. That sense is, in turn, based at least in part on experience with a given organization for which trust is built or destroyed over time (Bardach 1998). Capabilities
How organizations view the benefits of collaboration also depends on their situation. Those with more resources and abilities are presumably less likely to perceive benefits from collaboration than those with fewer resources and greater demands. However, as H3: Capacity – Organizations that have access to greater resources are less likely to cooperate with other organizations. The general presumption is that organizations that have stronger resources do not need to seek out additional help through collaboration. In contrast, smaller organizations with fewer staff and less specialization are more likely to band together (a form of collaboration) or to seek out additional resources through collaboration. In this respect, capacity can be considered both as a matter of organizational expertise and as a matter of Some evidence exists, however, that is contrary to the general presumption in showing that organizations with greater resources are more likely to contract with third parties for services. In studies of municipal outsourcing in Denmark and school district outsourcing in Washington state, Thomas Pallesen (2004, 2006) hypothesized that outsourcing is less of a threat to employees when a local government is wealthy than when it is short of resources. Pallesen, and in separate research O’Toole and Meier (2004a) in a study of school district contracting in Texas, found that more wealthy local governments were more likely to outsource service to third party providers. These researchers argue that contracting is used as a buffer, which can easily be reduced in financially bad times. These findings suggests that the difference between contracting for services with third parties (i.e., as a fiscal buffer) and collaboration over such things as joint service provision needs to be taken into account when considering the influence of Outcomes of Collaboration
It is often taken as a truism that collaboration fosters better outcomes. Despite the plethora of literature about collaboration, relatively few studies of collaborative networks address outcomes. This no doubt reflects the difficulty of measuring them. We address outcomes by considering the extent to which employment managers perceive better outcomes for clients. We assess the extent to which different collaborative arrangements H4 – Perceived Outcomes: Managers of organizations that have healthy collaborative relationships with other organizations perceive better outcomes. In reviewing 65 articles that address influences on public service performance, Boyne (2003) finds “weak” evidence that contracting out improves services and “slightly more” evidence that networked service delivery provides improvements. In one of the earliest studies of network effectiveness, Provan and Milward (1995) found that service improvements for community mental health centers depended on the specifics of network management and stability more than upon the existence of the network. Consistent with this, O’Toole and Meier (2004b) found in studying the influence of networking and managerial behaviors for school performance that performance improvements were associated with higher degrees of managerial networking along with stronger managerial quality and stabilizing features of the network such as personnel stability. These studies suggest that the existence of a collaborative relationship in itself does not necessarily lead to better outcomes. Relationships that are not strong or that are conflict ridden will presumably have less beneficial outcomes than healthier collaborative relationships. As such, we hypothesize that the “health” of the relationship is an important consideration in affecting outcomes. Following Bardach (1998), we conceptualize the health of collaboration as a function of the extent to which mangers are actively involved in the collaboration and the extent to which the relationship is viewed as being supportive. Active involvement of managers serves as the glue for the collaboration. By definition, arrangements that collaborators view as supportive are healthier than ones that they view as conflictual. The reinforcing nature of these forces in producing healthy collaboration suggests they are appropriately modeled with interaction The Setting: Danish Employment Services
The context of this study is the collaboration of Danish municipalities in the implementation of employment policy. The 269 Danish municipalities are multi-purpose local government entities that have responsibility for implementing many different national policies, including administering social assistance and employment measures for unemployed persons that are not eligible for unemployment insurance. The main municipal tasks are checking eligibility for and paying social assistance, giving advice on job search and career and vocational guidance, checking availability for work, and placing unemployed clients into employment or employment promoting measures. Part of the municipal costs for employment services are paid by the national government. Municipal employment services work to varied degrees with governmental and non- governmental partners in the delivery of employment services. A key governmental entity is the national Public Employment Service (PES) that focuses on unemployed clients who are eligible for unemployment insurance benefits. The PES provides employment services for these clients as well as for non-insured recipients of social assistance. Historically, there has not been strong collaboration between municipalities and the PES, and many municipalities have been very critical towards the PES (Larsen et al. 2001). A second set of governmental actors is neighbor municipalities. Some, particularly small, municipalities have formed inter-municipal collaborative employment administrations. Other municipalities collaborate with each other in more specific issue- areas, in buying services from one another, or for exchanging information. Three different types of non-governmental partners are relevant. One is 31 private Unemployment Funds and their affiliated trade unions. These funds administer unemployment insurance, check unemployed member’s availability for work, and offer career guidance. A second group is employer associations that together with unions give advice to municipalities and the PES on local employment policy and perform bridging between these local authorities and firms. A final group is comprised of private for profit, non-profit, and public third-party providers of education, training, case management, and job placement services typically under contract with municipalities the Until 1990 municipalities had to rely on labor market related services from the national PES that had a monopoly of contacting private firms for jobs or activation. From 1990 municipalities had permission to contact private firms for jobs or employment-training offers. Most municipalities have contacts with private firms for the provision of jobs and subsidized employment-training offers, and some have contact with local unions, unemployment funds, and employers associations (Damgaard 2006, Changes in national employment policy under the “Putting More People into Work” reform enacted in 2002 (Damgaard 2003) place greater emphasis on collaboration in employment service delivery. Among other considerations, the reform puts more emphasis on getting unemployed person into jobs more quickly and on monitoring that clients are available for work and regularly looking for jobs. The reform urges municipalities and the Public Employment Service to cooperate more closely. Separate local government reforms to take effect in 2007 reduce the number of municipalities to 98 by amalgamations and mandate municipalities to cooperate with local and regional PES offices by partly merging services in order to achieve a national policy objective of a more unified employment policy implementation system. To a varying extent municipalities have responded to this challenge prior to 2007 by increasing their cooperation with the PES. Also as part of the “Putting More People into Work” reform, both the PES and municipalities have been asked to contract out more services to third- Data and Measures
Much of the prior research on collaboration focuses on the network as a whole and the degree to which organizations cooperate. We break this down in considering collaborative relationships between one type of organization—municipal employment services—and a range of prospective partners. Our analyses follow from the preceding conceptualization of collaboration and bases for collaborative relationships. We analyze data concerning municipal-level implementation of employment reforms for which our data are primarily based on a national survey of middle managers for municipal employment services. These are the most relevant respondents for providing information about collaborative relationships since they are the ones who are responsible for those relationships. The survey responses have been supplemented by secondary data based on register data on population size, resources, and task-difficulty.1 For most municipalities the relevant respondent is a middle manager with responsibility for employment measures for recipients of social assistance who are available for work. In those small municipalities that have no middle manager, the respondent is the chief executive officer for social affairs and employment services. Relevant respondents were identified by telephone calls to every municipality. We sent two reminders by email and Our analyses are based on 204 Internet-based survey responses collected from mid December 2005 through May 2006. The response rate is 75 percent in relation to a total of 269 municipalities. The collected survey data are representative of all Danish municipalities in terms of population size and difficulty of the employment task. Responses for middle managers from municipalities with less than 10,000 inhabitants are marginally underrepresented by 3.8 percentage points in comparison to census distribution, while those from municipalities with between 20,000 and 45,000 inhabitants are slightly overrepresented by 3 percentage points. Among the 204 municipalities that we consider, 31 are part of eight inter-municipal employment centers. These centers provide job services for the participating municipalities, which tend to be smaller municipalities. These centers and the remaining municipalities either provide services on their own or enter into collaborative Two sets of potential concerns arise from the use of these survey data. One is that reliance on middle managers of municipal employment services as informants about collaborative relationships and perceived outcomes of employment services leads to biases in our characterization of collaboration. Some respondents may consistently provide a rosy picture while others are less enthusiastic. However, we do not find evidence of any systematic biases as might be expected due to age or length of time serving as a middle manager of employment services. We fail to find meaningful correlations between these variables and our measures of extent of cooperation with different organizations, degree of trust in different organizations, and perceived outcomes.2 Nor do we find evidence of systematic biases when comparing mean scores for the same collaboration variables between respondents who are males and females.3 Questions were phrased in ways to minimize any response method effects in responding to the questionnaires, aided by the fact that the Internet-based survey administration prevented respondents from viewing multiple questions on the same screen. A second potential concern is our reliance on perceived outcomes rather than actual outcomes of employment services. Middle managers may have incentives to report more positive outcomes than is actually the case. But, our concern is the relative variation in outcomes and not the absolute levels. Moreover, as discussed below, we ask about perception of different outcomes in employment services and not about outcomes of collaborative relationships per se. This phrasing, along with far separation in the question ordering, disconnects any cognitive link between responses about the nature of the collaboration with the perceived outcomes. Actual outcome data would, of course, be preferred. But, these are not available until well after the collection of our survey data. The limited independent survey data about client outcomes that are available within a relevant timeframe do not provide sufficient information for municipal-level analyses. Measures
We conceptualize collaboration as a combination of extent of cooperation and effort that goes into collaboration with each of the relevant potential partners.4 Municipalities that cooperate regularly in sharing information should rate lower for collaboration than those that cooperate with more intensive activities like sharing personnel. We get at this by computing an overall collaboration score that is the product of the score for extent of cooperation and the score for the effort put into collaboration. This has a scale of 0 to 75. The extent of cooperation is measured by respondent rating of frequency of interaction on a scale of 1 (none) to 5 (regular) for each organization. Collaboration effort is measured by assigning a score of 0 if there is no collaboration and increasing scores from 1 to 5 for each of the different types of collaboration: information sharing, client referrals, receive clients, joint programs, and sharing of personnel. This ordering reflects increasing degrees of effort that is involved in collaboration for which information sharing and client referrals involve relatively little effort while undertaking joint programs and sharing of personnel involves substantially more effort. Assignment of scores on a 1 to 5 scale is a simple way of recognizing these differences. Because any given municipality may engage in one or more of the different types of collaboration, the potential score for Our theorizing suggests several factors that help account for variation in collaboration. One consideration is the degree of organizational benefits that an agency receives from collaboration. This entails perceived resource benefits, which we measure as the municipal respondent rating on a scale of 1 (not important) to 5 (very important) of the “importance of the organization for fulfillment of our goals” with reference to each potential collaborator. Organizational benefits also include more specific problem- solving benefits from collaboration. We get at this with an overall measure of task difficulty under the logic that municipalities that face more difficult employment services tasks will have greater problem-solving benefits from collaboration. The measure is an index of the expected mean duration of temporary cash benefits for all adult citizens in each municipality in 2004 based on characteristics of the population and local labor market conditions (e.g. unemployment rate). Higher scores indicate more problematic task environments.5 Natural log values are employed to address skewed data. The degree to which organizations trust potential collaborators is another relevant consideration. This has been measured in a variety of ways in the literature (see Lundin 2005 in particular). Rather than employing a generalized measure of trust, we consider perceptions of the degree to which respondents trust potential collaborators to meet their obligations. We measure this as the municipal respondent rating on a scale of 1 (low) to 5 (very high) of the degree to which “you trust these other authorities and organizations to follow through on their commitments in carrying out employment services.” Separate ratings were provided for each potential collaborator. We consider two aspects of the capacity of municipalities to address employment problems. One is municipal size of population following the logic that larger municipalities will have more options for addressing employment problems. A second measure gets at the resource capacity of the municipality measured as the municipal budgeted basis of taxable income and land value per capita for 2005 with a correction for central government grants and inter-municipal transfers. A related consideration is whether a municipality is a member of a municipal employment center or not. As we discuss below, some smaller municipalities have banded together to form combined employment centers as a way of sharing their resources. We create a dummy variable to indicate whether a given municipality is part of a municipal employment center or not. In order to examine consequences of collaboration for perceived outcomes, we employ an index of perceived outcome based on the responses of middle managers about the extent to which the municipality has succeeded in getting clients to search for jobs, to be available for work, and to enter ordinary employment on a scale of 1 (not at all) to 10 We consider two factors that we hypothesize above as affecting the influence of collaboration on perceived employment outcomes. One is the degree of manager involvement with each collaborative entity. We measure this each manager’s rating of the frequency with which they personally met with representatives of each potential collaborator within the past year on a scale of 1 (not at all) to 6 (once a week).7 The second consideration is the character of the collaborative relationship. We measure this as the employment services manager’s rating for each collaborator of the relationship as being conflict ridden (score 1), competitive (score 2), neutral (score 3), supportive (score Findings
We present our findings in first considering patterns in collaboration between municipal employment services and other organizations. We next consider factors that account for variation in collaboration. We then consider the influence of different types of collaboration and other factors on perceived employment outcomes. Collaboration Patterns
Table 1 summarizes the extent of cooperation and types of collaboration for the 204 municipalities in our study. For each prospective collaborator, we report the mean extent of cooperation, the percentage of municipalities that engages in different forms of collaboration, the resultant mean effort scores, and the combined collaboration score. The ordering of municipal involvement with different collaborators is the same regardless of which summary score is considered. However, there is clearly greater variation in collaborative efforts than in the reported extent of cooperation. For this reason, we argue the overall collaboration score is a better measure than the extent of cooperation score. Table 1. Cooperation and Collaboration
Municipal Cooperation and Collaboration with: a Collaboration Score f
17.24 14.39 8.39
a Cell entries are mean scores or percentages, as indicated, for municipal collaboration with each of the actors indicated in the column headings. Scores are based on 204 municipal respondents. b Services provided by for profit, non-profit, and public organizations such as consulting firms and training institutes. c Mean score for extent of cooperation on a scale of 1 (none) to 5 (regular). d Percentage of municipalities that report each form of cooperation; multiple responses were allowed. e Mean value for summary score for types of involvement with each organization on a scale of 0 to 15, where no collaboration is scored 0, information sharing scored 1, client referrals scored 2, receive clients scored 3, joint programs scored 4, and sharing of personnel scored 5. f Mean value for summary collaboration score that is product of extent of cooperation and collaboration effort scores on a potential scale of 0 to 75. Municipal collaboration is greatest with neighboring municipalities that includes the inter-municipal center collaboration. As indicated by the type of collaboration, these arrangements run the gamut of different forms of collaboration with a relatively strong emphasis on joint programs. The second greatest municipal collaboration is with the national Public Employment Service. These arrangements emphasize client referrals and receipts with some joint programs and sharing of personnel. Thirty one of the responding municipalities have formed joint labor market centers with the PES. Six of those centers include more than one municipality. The center collaboration with the PES includes 12 government sponsored pilot schemes with joint labor market centers with municipalities and the local PES that are forerunners for the mandated future joint job centers. We assume that some of the collaboration between municipalities and the PES has been stimulated recently by the planned merger of the two organizations. Municipal collaboration with other organizations is less extensive. Cooperation with unions and their affiliated unemployment funds and with third party providers (such as consultants and training institutes) is on average fairly frequent, but the types of collaboration and resultant collaboration effort are more limited. Municipal cooperation with employer associations is the least frequent with the emphasis on information sharing. As mentioned above, the Danish government have urged municipalities to use more third party providers and indicated that municipalities will be required to do so in the planned joint job centers with the Public Employment Service. Bases for Collaboration
Our theorizing about collaborative arrangements leads us to consider the role of organizational benefits, social capital, and different aspects of capacity in explaining variation in the extent to which municipalities collaborate with different actors. As noted in our theorizing, we consider these to be contributing bases for collaboration that are not mutually exclusive. Of research interest is the relative contribution of each consideration. The cross-sectional nature of the data do not permit analyses about the dynamics of collaboration in reinforcing trust and subsequent collaboration or of changes in collaborative relationships overtime as the result of different outcomes of the Table 2 presents regression models for these factors in explaining variation in the extent of collaboration with each collaborator. The dependent variable is the relevant collaboration score that combines extent of cooperation and collaboration effort. Given the construction of this measure, it is reasonable to treat it as an interval variable. Higher scores indicate greater degrees of collaboration. The models have been estimated using Ordinary Least Squares regression with appropriate transformations of relevant variables to meet assumptions of linear relationships. Appropriate visual inspections and statistical tests were conducted to verify that OLS regression assumptions were met for these models. The cell entries are the standardized coefficients. Keeping in mind differences in distributions of the independent variables, these suggest the relative magnitude of influence of each. The models explain a reasonable amount of variation in the extent of collaboration with greater explanatory power for those organizations with which municipalities tend to collaborate more. One caveat about these models is the differences in sample sizes that arise from the fact that we only gauged the degree of trust in other organizations when there was some form of collaboration. If no collaboration existed with a potential partner, the trust variable was coded as missing. Excluding non-collaborative relationships potentially biases key relationships especially for collaboration with employer associations and other actors for which the extent of collaboration with municipalities is more limited. We assessed the degree of this potential bias by computing regression models that use mean- value substitution for missing values. Except for the trust variable, the effects and significant levels of other factors are virtually the same as reported in Table 2. The influence of trust is reduced and becomes statistically non-significant for the models of collaboration with employer associations and with other actors. Given the arbitrariness of mean-value substitution, we report results as computed with appropriate caveats.8 Table 2. Explaining Variation in Collaboration
Regression Models for Collaboration Addressing a
Explanatory Factors
Employment
Neighbor
Unemployment
Employer
Municipalities
Funds/Unions
Associations
Organizational Benefits
Social Capital
Capacity
Model Statistics
Notes: * p < .10, ** p < .05, *** p < .01 one-tailed t-test, except R2 values for F-test of model fit. a Dependent variable is the relevant collaboration score (sq root values used to meet linearity assumptions). Cell entries are standardized values with absolute t-values in parentheses. The first set of entries consider the role of organizational benefits for which we hypothesized that increased resource benefits (H1a) and increased problem-solving benefits (H1b) would contribute to greater collaboration. The hypothesis is supported for resource benefits as indicated by the magnitude of the coefficients for resource benefits, which we measure as the dependence on the relevant organizations. One consideration that we could not address is that as organizations collaborate they become more The problem-solving benefit hypothesis is supported for collaboration with the Public Employment Service (although at a lower statistical significance) and for collaboration with third-party providers. We fail to detect an influence of problem solving benefits on the degree of collaboration with neighbor municipalities, unemployment funds/unions, or employer associations. These findings suggest that municipalities with difficult task environments collaborate more with the PES and third-party providers such as training institutes, but not with other potential collaborators. This makes sense because localities and clienteles with more social problems and difficult labor markets are likely to generate more cases in which both the municipality and the PES are involved (e.g. long term unemployed insured workers loosing eligibility for unemployment benefits to be replaced by public assistance, more health problems leading to municipal sickness allowances for insured unemployed workers.) By the same token, it might be tempting for municipalities with a problematic task environment to ask for help from third-party providers that might have specialized in that kind of clientele. Following one of the major findings of the collaboration literature, we also hypothesized that organizations are more likely to collaborate with other organizations that they trust (H2). The social capital entry in Table 2 gets at this in examining the influence of trust, measured as perception that other organizations fulfill their commitments, on extent of collaboration. The hypothesis is supported for municipal collaboration with each of the potential collaborators. However, this relationship is likely overstated for two methodological reasons. One, as discussed above, is that we have no measure of trust when collaborative relationships do not exist. Trust is presumably greater when collaborations exist than when they do not, leading to a stronger effect than would otherwise be the case. 9 The second related consideration is that it is reasonable to assume that increased collaboration leads to greater trust (see Isett and Provan 2005), suggesting that some of the strength of the observed relationship is due to the reciprocal relationship. Given these caveats, it is surprising that the trust relationship with neighbor municipalities with whom collaboration is more extensive is relatively weak when compared with the Public Employment Service in particular. We hypothesized that increased capacity of municipal employment services would lessen incentives to collaborate (H3). As shown in the bottom of Table 2, we address three aspects of capacity: municipal size, resources (revenue base), and whether a municipality is part of a inter-municipal employment center or not. Other than municipal collaboration with employer associations, we fail to detect an influence of size on collaboration. As expected, municipalities with greater resources collaborate less with other municipalities. But, they also tend to collaborate more with employer associations. This makes sense as larger and wealthier municipalities may have stronger administrative capacity for cooperation with employers associations, and employers in larger municipalities are likely to be better organized. Taken together, these results about municipal size and resources suggest that their influences are not as strong as presumed. These findings are consistent with those of Weiss (1987) in studying school-district collaboration. She found that resource capacity was much more frequently mentioned as a rhetorical argumentation by school-districts than evidenced in actual collaboration. The findings for whether a municipality is part of a municipal center make sense in that they reflect the nature of this form of collaboration and of the role of the centers. By definition, municipalities that are part of such centers have greater degrees of collaboration with neighbor municipalities as indicated by the coefficient for the neighbor municipality collaboration. The greater cooperation that municipalities, which are members of inter-municipal centers, have with the Public Employment Service and third parties might due to the greater administrative capacity of centers for establishing that kind of collaboration. The PES and third party organizations might also perceive inter-municipal centers as more attractive partners for collaboration than smaller units of single municipalities that have smaller production scales. Perceived Outcomes
We noted in the introduction that it is often taken as a truism that collaboration fosters better outcomes. We assess this by considering municipal employment managers’ perceptions of employment outcomes and how those differ for varied levels of municipal collaboration. We theorized above (H4) that benefits of collaboration are not automatic since they require “healthy” relationships comprised of active involvement of managers and supportive interactions. A key caveat to this discussion, noted in our discussion of data and measures, is that our outcome measures are perceived outcomes by the managers of municipal employment services rather than actual outcomes. Our measure of perceived outcomes is an index based on ratings of three items: success in getting clients to search for jobs, availability for work, and entering ordinary employment. Some 85 percent of the respondents to each of three items (on a scale of 1 to 10) indicated a score Table 3 reports regression models that explain perceived outcomes as a function of collaboration with each type of organizations and different contextual considerations. For each model, the dependent variable is the index of perceived outcomes. This is a well-behaved continuous measure for which higher scores indicate better perceived outcomes. The models have been estimated using Ordinary Least Squares regression with appropriate transformations of relevant variables to meet assumptions of linear relationships. Appropriate visual inspections and statistical tests were conducted to verify that OLS regression assumptions were met for these models. The cell entries are the standardized coefficients. The column headings indicate the relevant municipal collaborator for the model that is entered as an independent variable either alone or as part of an interaction term as shown under the collaboration influence row heading. The interaction terms are explained below. Put differently, the models differ with respect to the municipal collaborating organization that is considered in explaining variation in perceived outcomes but they are the same with respect to other contextual factors entered Table 3. Explaining Variation in Perceived Outcomes
Regression Models for Perceived Outcomes
involving each collaborator a
Explanatory Factors
Employment
Neighbor
Unemployment
Employer
Municipalities
Funds/Unions
Associations
Collaboration
Influences
Without interaction b
Context f
Model Statistics
Notes: * p < .10, ** p < .05, *** p < .01 one-tailed t-test, except R2 values for F-test of model fit. a Dependent variable is an index of perceived extent to which the municipality has succeeded in getting clients to search for jobs, to be available for work, and to enter ordinary employment. Cell entries are standardized values with absolute t-values in parentheses. The relevant collaborator for each model is shown as the column heading for the model. b The relevant collaboration variable is the column heading (sq root values). c Separate modeling that includes the product of the relevant collaboration score (sq root) times the degree of direct manager involvement in the collaboration. d Separate modeling that includes the product of the relevant collaboration score (sq root) times the degree of manager direct involvement in the collaboration times an index of the character of the relationship e Change in adjusted R2 between the model without interaction terms and the model with the 3-way interaction. f Context explanatory coefficients are reported for the models involving the collaboration interaction term. g Extent to which the manager respondent indicated that caseworkers should focus on the goal of getting clients into jobs more quickly rather than improving their employability in the long run. The number of observations is reduced in these models because only 72 percent of the respondents provided responses for the items comprising the index of perceived outcomes. Non-respondents for the perceived outcome measures are from municipalities that have lower collaboration scores with other partners.11 This suggests that the impacts of collaboration are likely to be overstated by these results, which reinforces the point we make below about limited impacts. Evidence for this is provided by observing that the impact of collaboration on perceived outcomes is reduced when substituting mean values for missing values of the index of perceived outcomes in the regression models. We report the models without mean-value substitution given the arbitrariness of that These models suggest that the various forms of municipal collaboration account for relatively little variation in perceived outcomes. A separate regression model containing only the collaboration variables explains only 2 percent of the variation in perceived outcomes.12 As shown in the first row under collaboration influences (without interactions), there is no detectable influence of collaboration on perceived outcomes when considering collaboration with the Public Employment Service and relatively limited influence when considering collaboration with unemployment funds/unions and employer associations (as gauged by the magnitude of the standardized coefficients). These findings are contrary to the general presumption that collaboration leads to We theorized, however, that the health of the collaborative arrangement needs to be taken into account (H4). This is reflected by the coefficients for the interaction terms that are reported as the second and third entries under the heading, collaboration influences. Each of these involves an interaction term that was entered into a separate model explaining variation in perceived outcomes while controlling for the designated contextual factors. The first interaction term is the interaction of the municipal collaboration score with a given actor and the frequency with which the municipal manager has direct contact with the collaborating entity. The second interaction term also includes the measure of the character of the collaborative relationship. These factors are scored so that higher scores indicate “healthier” relationships. The entry for change in R2 shows the change in adjusted-R2 from the model with the basic collaboration score to the model involving the three-way interaction. The findings for the interaction terms clearly show that taking the frequency of manager involvement and character of the collaborative arrangement into account makes a difference in explaining variation in perceived outcomes. These explain an additional 2 to 5 percent of the variation in perceived outcomes, depending on which collaborating entity is considered. Consideration of the three-way interaction also leads to a detectable influence for each collaborating organization while that influence was not as apparent when considering only the collaboration score. Nonetheless, the relative magnitude of influence and of the explanatory power of these variables is less for the Public Employment Service and for unemployment funds/unions than for other collaborating The lesser influence of collaboration with the national Public Employment Service on perceived outcomes is surprising to advocates of the Danish employment policy of merging municipalities and local PES offices into local job centers. One interpretation is that inter-organizational collaborations between organizations that are ‘pooled’—those that are relatively self-contained and independent––are less likely to improve outcomes. In contrast, collaboration or more exchanges in inter-organizational relationships that are either sequential (where one organization is unilaterally dependent on resources or inputs from another organization) or reciprocal (with mutual dependence) may lead to better outcomes because resources and benefits can be exchanged (O’Toole and Montjoy 1984, O’Toole 2003). Municipalities and the PES offices have gradually moved towards forming two relatively independent, parallel systems that share some similarity to pooled inter-organizational relations under which relatively self-contained organizations each address their own clientele. Another interpretation is that much of the municipal collaboration with the PES is relatively new as an adaptation to the future demands of merging into job centers. New collaboration with organizations which have not previously trusted each other very much is likely to imply substantial transaction costs that may not be compensated by better outcomes in the short run. The findings for contextual factors are consistent across models as the same scores entered in each model. Not surprisingly, the perceived outcomes are poorer for municipalities that have more challenging client mixes (problem extent). Consistent with the findings in the literature about the importance of managerial factors in service outcomes (e.g. Boyne 2003 and O’Toole and Meier 2004a), respondents from municipalities that report emphasizing getting clients into jobs more quickly as a managerial priority in turn perceived that they were achieving better outcomes.13 We fail to find influences of municipal size and resources on perceived outcomes. The lack of influence of size on employment outcomes might be surprising to advocates of the Danish local government structural reform policy of merging municipalities in order to increase effectiveness. However, this finding is consistent with other analyses that failed to document any such consistent results for services in general and employment services in particular (Arendt 2004, Bengtsson 2004, Groes and Olsen Conclusions
This research addresses different patterns of collaboration for Danish employment services. We address collaborative relationships between one type of organization— municipal employment services—and a range of partners that include the national Public Employment Service, unions and their affiliated unemployment funds, employer associations, other municipalities, and third-party providers that include private for profit, non-profit, and public providers of education, training, case management, and job placement services. Three key sets of findings emerge from this research. One is different patterns in collaboration. Municipalities vary considerably in the extent and forms with which they collaborate with other organizations for employment services. This suggests that collaboration should not be thought of as a generic activity with other organizations as the nature of the partners and their prospective roles needs to be considered. In addition, measures of collaboration need to account for more than just the extent or frequency of cooperation. The effort that goes into different types of collaborative arrangements is also important to consider. Our findings show that considering frequency of collaboration alone would give a different understanding of the collaborative arrangements than taking into account the types of collaboration. A second set of findings concerns the factors that explain variation in the extent of collaboration. As found in prior research, we find organizational benefits (dependence and problem-solving benefits), social capital, and various aspects of capacity help shape municipal collaboration with other actors for employment services. However, the role of these considerations differs somewhat for the organizations with which municipalities collaborate. Our findings suggest that each collaborative arrangement is induced (or benefits from) perceptions that there are resource benefits from collaborating. This makes sense since organizations are more likely to collaborate if they perceive the collaboration as extending their expertise, personnel, or other capabilities. Yet, problem- solving benefits only appear to be relevant for collaboration with the Public Employment Service—although weakly—and to a greater extent with third-party providers such as private consultants, non-profit organizations, and public training institutes. Collaborating with third-party providers typically entail contractual relationships that involve a different collaborative dynamic than for the other organizations. This suggests that municipalities are very specific in seeking out collaborators that they think will provide specific A third set of findings concerns how the perceived employment outcomes are affected by municipal collaboration with other organizations. Municipal employment managers generally report that employment outcomes for their clients have been relatively good in the prior year with respect to job search, availability for work, and employment. However, our modeling of these perceived outcomes suggests that the various forms of collaboration account for relatively little variation; 2 percent without taking health of the relationship into account. These findings suggest that the benefits of collaboration per se are over-stated, at least in the general literature on the subject. We show that “healthier” collaborative relationships, which directly involve managers and foster positive relationships, have stronger perceived outcomes. Indeed, an additional 2 to 5 percent of the variation in perceived outcomes is explained when taking health of the collaborative relationship into account. This observation is consistent with the findings of other research (Provan and Milward 1995, O’Toole and Meier 2004b) that managerial factors are important in determining the outcomes of collaborative arrangements. Also important are the extent to which the organization has embraced a goal of getting clients into jobs and the presence of a supportive task environment for This research contributes to the empirical understanding of collaborative arrangements for service delivery. Although the findings are proscribed by attention to one setting, the results have good face validity and are generally consistent with our theorizing. This gives us confidence of our choices about conceptualization of collaboration and our measures. Nonetheless, we recognize a number of caveats to our findings that arise from our reliance on data from a single source (municipal middle managers), limitations in our analyses due to missing data, and reliance on perceived measures of the outcomes of employment services. We have discussed the implications of these limitations for specific findings. One key consequence is that the effects of collaboration on perceived outcomes are if anything overstated in these data. This further reinforces our sense that collaborative benefits are presumed to be greater than they may Our examination of collaborative relationships in the delivery of employment services highlights a number of issues that are relevant for future theorizing and empirical examination of collaborative service delivery. One is greater attention to different aspects of collaboration. Our findings show that the specifics of collaborative arrangements—in terms of who is collaborating with whom, the type of collaboration, the frequency of managerial involvement, and the character of the relationship—need to be taken into account in order to understand the potential for and limits to collaboration in service delivery. A second issue is sorting out connections among dependence, trust, and collaboration with particular attention to the interactions over time. It may be that the collaboration fosters both greater degrees of trust and dependence, rather than the reverse. A third issue is addressing the relationship between outcomes and subsequent collaboration. It also may be that perceived outcomes of collaboration either enhance (positive outcomes) or detract (negative outcomes) from subsequent collaboration. Clearly, these relationships cannot be sorted out with cross-sectional data. A final direction is more refined analyses of the impacts of collaboration for service delivery outcomes. Simply put, not all collaboration is the same and collaboration itself is not a Notes

1 Data about municipal population size and the extent of municipal resources have been provided
2 The relevant measures are discussed in the text that follows. None of the correlations was statistically significant at a p-value of .1 or less for age using a two-tailed test. The strongest correlations involving experience were -.12 for extent of cooperation with neighbor municipalities (p = .09) and -.16 for trust in employer associations (p = .07). 3 Only two independent-sample t-tests were statistically significant at conventional levels of the 11 that we conducted. Male respondents on average report slightly greater degrees of cooperation with unions and with employer associations than do female respondents; amounting to an average difference of .3 on a 5-point scale (p-values < .05). This may be because of stronger gender- 4 Ideally, we would have separate scores for the extent of each form of collaboration with each potential actor. However, we only have a measure of the extent of overall cooperation with each 5 The measure was obtained from the Danish Institute of Local Government Studies based on rich Danish register data from Statistics Denmark. The details of the calculation yet made for the future, larger municipalities can be found in Clausen et al. (2006). 6 The index is a principal component score for the items that make up this dimension of outcomes. As such, the index is a weighted average of the scores on each item. The Cronbach reliability coefficient for this is .90. 7 The response options for each organization are: not at all, once a year, once every half year, once every quarter, once a month, and once a week. 8 The alternative of analyzing the smallest subset of data (128 observations) violates the random selection of observations and does not address the missing value issue. 9 This is evidenced by the fact that the effect of trust when using mean-value substitution for missing data is no longer statistically significant for the models of collaboration with employer 10 These are outcomes that are consistent with the national government emphasis in employment policy in “getting people to work.” These are clearly not the only potential outcomes, or necessarily the desired municipal outcomes, for employment services. Some employment programs put more emphasis on improving clients’ longer-term prospects for obtaining meaningful work. We emphasize the more immediate outcomes here because of their importance 11 The p values are less than .01 for independent sample t-tests comparing means of collaboration scores for each partner between respondents and non respondents to the perceived outcome 12 The collaboration scores are entered separately in Table 3 rather than in combination. The latter introduces multi-collinearity problems given that municipalities often collaborate with more 13 While these may seem like self-serving responses, the two questions were far apart in the questionnaire. As such, it seems unlikely that respondents equated the two responses. References
Agranoff, Robert and Michael McGuire. 2003. Collaborative Public Management, New Strategies for Local Government. Washington, DC: Georgetown University Press. Andersen, J. and Torfing. 2004. Netværksstyring af kommunernes arbejdsmarkedsrettede indsats. Arendt, Jacob Nielsen. 2004. ”Kommunestørrelse og indsatsen for kontanthjælpsmodtagere,” pp. 217-30 in Nils Groes and Leif Olsen (eds.) Brudstykker eller sammenhæng?: – en antologi om forudsætningerne for en offentlig strukturreform. Copenhagen: Fokus. Bardach, Eugene. 1998. Getting Agencies to Work Together: The Practice and Theory of
Managerial Craftsmanship. Washington DC: Brookings Institution Press. Bengtsson, Steen (ed.) 2004. Kommunestørrelsens betydning: Resultater af analyser for Strukturkommissionen. Copenhagen: Danish National Institute of Social Research 04:01. Boyne, George A. 2003. “Sources of Public Service Improvement: A Critical Review and Research Agenda.” Journal of Public Administration Research and Theory 13(3): 367-394. Damgaard, Bodil. 2003. Social- og arbejdsmarkedssystemerne. Copenhagen: Danish National Damgaard, Bodil. 2006. “Does Policy Network Lead to Network Governance?” Public Feiock Richard C. and Hyung-Jun Park, with Annette Steinacker and Jaehoon Kim. 2005. Institutional Collective Action and Economic Development Joint Ventures. Prepared for the
National Public Management Research Association Conference, October 2005, Los Angeles. Forbes, Melissa and Laurence L. Lynn, Jr. 2005. “How Does Public Management Affect Governmental Performance? Findings from International Research,” Journal of Public Administration Research and Theory 15 (4): 559-584. Goldsmith, Stephen and William D. Eggers. 2004. Governing by Network, The New Shape of the Public Sector. Washington, DC: Brookings. Groes, Nils and Leif Olsen eds. 2004. Brudstykker eller sammenhæng?: – en antologi om forudsætningerne for en offentlig strukturreform. Copenhagen: Fokus. Hill, Carolyn and Laurence L. Lynn, Jr. 2005. “Is Hierarchical Governance in Decline? Evidence from Empirical Research,” Journal of Public Administration Research and Theory Hill, Carolyn and Laurence L. Lynn, Jr. 2003. “Producing Human Services, Why do Agencies Collaborate?” Public Management Review 5 (1): 63-81.
Isett, Kimberley Roussin and Keith G. Provan. 2005. “The Evolution of Dyadic Interorganizational Relationships in a Network of Publicly Funded Nonprofit Agencies.” Journal of Public Administration Research and Theory 15 (1): 149-165. Kreuger, Skip and Michael McGuire. 2005. “A Transaction Costs Explanation of Interlocal
Government Collaboration” Prepared for the National Public Management Research
Association Conference, October 2005, Los Angeles. Larsen, Flemming Nikolaj. Abildgaard, Thomas Bredgaard, and Lene Dalsgaard. 2001. Kommunal aktivering. Aalborg: Aalborg Universitetsforlag. Leach, William D. and Paul A. Sabatier. 2005. “To Trust an Adversary: Integrating Rational and Psychological Models of Collaborative Policymaking,” American Political Science Review Lubell, Mark. 2004. “Collaborative Watershed Management: A View from the Grassroots,” Policy Studies Journal 32 (3): 341-361. Lubell, Mark, Mark Schneider, John T. Scholz, and Mihrye Mete. 2002. “Watershed Partnerships and the Emergence of Collective Action Institutions,” American Journal of Political Science Lundin, M. (2007) “Explaining Cooperation: How Resource Interdependence, Goal Congruence, and Trust Affect Joint Actions in Policy Implementation,” Journal of Public Administration Research and Theory 17 (4): 651 – 672. Milward, H. Brinton and Keith G. Provan. 1998. “Measuring Network Structure,” Public Administration 76 (Summer): 386-407. Milward, H. Brinton and Keith G. Provan. 2000a. “How Networks are Governed.” In Carolyn J. Heinrich and Larence E. Lynn, Jr. (eds.) Governance and Performance, New Perspectives. Washington DC: Georgetown University Press, pp. 238-262. Milward, H. Brinton and Keith G. Provan. 2000b. “Governing the Hollow State,” Journal of Public Administration Research and Theory 10(2): 359-379. Osborne, David and Peter Plastrik. 1998. Banishing Bureaucracy, The Five Strategies for Reinventing Government. New York: Penguin Group, Plum Books. O’Toole, Laurence J. Jr. 2003. “Interorganizational Relations in Implementation, pp. 234-244. In B. Guy Peters and John Pierre (eds.) Handbook of Public Administration. Thousand Oaks, O’Toole, Laurence J. Jr. and Kenneth J. Meier. 2004a. “Parkinson’s Law and New Public Management? Contracting Determinants and Service-Quality Consequences in Public Education.” Public Administration Review, 64 (3): 342-352. O’Toole, Laurence J. Jr. and Kenneth J. Meier. 2004b. “Public Management in Intergovernmental Networks: Matching Structural Networks and Managerial Networking,” Journal of Public Administration Research and Theory 14 (4): 469-494. O'Toole, Laurence J., Jr. and Robert S. Montjoy. 1984. “Interorganizational Policy Implementation: A Theoretical Perspective,” Public Administration Review, 44 (6): 491-503. Pallesen, Thomas. 2004. “A Political Perspective on Contracting Out: The Politics of Good Times. Experiences from Danish Local Governments,” Governance 17 (4): 573-587. Pallesen, Thomas. 2006. “The Politics of Contracting Out in School Districts: The Case of Washington State,” State and Local Government Review 38 (1): 34-40. Provan, Keith G., and H. Brinton Milward. 1991. “Institutional-Level Norms and Organizational Involvement in a Service Implementation Network,” Journal of Public Administration Research and Theory 1(4): 391–417. Provan, Keith G. and H. Brinton Milward. 1995. “A Preliminary Theory of Interorganizational Network Effectiveness: A Comparative Study of Four Community Mental Health Systems.” Administrative Science Quarterly 40 (1): 1-33. Salamon, Lester M. 2002. “The New Governance and the Tools of Public Action: An Introduction,” in L. Salamon (ed.) The Tools of Government, A Guide to the New Governance. New York: Oxford, pp. 1-47. Weiss, Janet A. 1987. “Pathways to Cooperation among Public Agencies,” Journal of Policy Analysis and Management 7 (1): 40-65.

Source: http://www.campbellcenter.dk/Files/Filer/SFI/Pdf/Working_papers/WP_22_2007_b.pdf