Who is Engaged and Why? Testing an Instrumental Perspective on Stakeholder Engagement

From an instrumental perspective, public administrators are expected to analyze their environment. They are also expected to take proactive steps to manage environmental impacts. These actions are intended to ensure that public administrators are able to achieve their goals. Part of analyzing their environment involves the engagement of stakeholders. Stakeholder engagement is believed to provide public administrators with insights into how they can better evaluate innovative policy options that allow them to overcome limited administrative capacity. Stakeholder engagement is also believed to allow public administrators with opportunities to mobilize support in light of political opposition. The purpose of this article is to test these claims within the policy area of municipal contracting out. In the article, we find strong support for these ideas. Thus, our findings should inform public administration theories about the role and importance of stakeholder engagement.

The present study seeks to examine and test this instrumental perspective.Specifically, we explore the extent to which public administrators engage stakeholders purposefully; and, we provide insight into who public administrators engage and why.To do so, we make use of novel data from the 2007 International City/County Management Association (ICMA) Alternative Service Delivery (ASD) survey.Although the primary purpose of this survey was to explore the nature of contracting out in local governments, a portion of the survey focused on stakeholder engagement.Therefore, we use this data along with data from the U.S. Census Bureau to explore the extent to which stakeholders within this particular policy area are engaged.We also use this data to explore how conditions within the decision environment, along with opposition and administrative capacity, can be used to explain patterns of engagement within municipalities.
We focus on the contracting out decision in municipalities for theoretical and methodological reasons.From a theoretical perspective, the contracting out decision represents a ubiquitous and fundamental question for local governments (e.g., Boyne 1998;Brown & Potoski, 2003;Warner & Hebdon 2001;Zhang, Gibson, & Schafer, 2018).That is: How do you offer services?The decision to contract out, then, provides a context that is both highly salient to local governments and fairly universal as a substantive policy question that local governments must address.
Indeed, the topic of contracting out is often described in the strategy literature as the adoption of innovation given that it is viewed as a new approach to service delivery by municipal governments (Damanpour & Schneider, 2008;Mandell & Steelman, 2003;Walker, 2006).Innovative policies tend to produce strong reactions within communities (Borins, 2002;Damanpour & Schneider 2008;Hess & Adams, 2007;Walker, Avellaneda, & Berry, 2011).As a result, not only do innovative policies test government's instrumentality (Borins, 2001;Damanpour & Schneider, 2008;Vigoda-Gadot, Shoham, Schwabsky, & Ruvio, 2008) they also provide a particularly acute setting for studying instrumental approaches to public management.
Although previous research (e.g., Schafer & Zhang, 2016) has suggested a relationship between stakeholder engagement and local government contracting outcomes, a more nuanced probe of this relationship remains unexplored.Thus, contracting out provides a unique policy environment in which to explore the hypotheses later put forth in this study.
From a methodological perspective, ICMA has produced seven datasets on municipal contracting out since 1982 (ICMA, 2012).Although we focus on only the 2007 data for this study, the questions asked on the survey have been repeated over time and were developed by the leading practitioner-oriented organization focused on researching the practices of public administration (ICMA, 2012).We, therefore, have confidence in the reliability and validity of the questions included on the survey.
Within the policy context of contracting out, we attempt to make a two-fold contribution to the literature on stakeholder engagement in the public sector.First, we examine whether public administrators actively shape the policy process using an instrumental approach (as opposed to react to it) (Feldman & Khademian, 2002;Svara, 1998;Wilson, 1887).This examination will allow us to determine the extent to which public administrators actively and purposively engage stakeholders to achieve a particular goal.Second, we explore who is engaged as a stakeholder and why.This will allow us to uncover the many purposes that stakeholders serve in the policy process and highlight new ways of thinking about stakeholders' role in governance.This exploration also allows us to provide an answer to the following call to researchers, "To understand the new governance, we cannot simply examine tools; we must understand the role of human kind-the citizens, stakeholders, and public administrators who are the tool makers and tool users" (Bingham, Nabatchi, & O'leary, 2005, p. 548).
Our findings provide strong support that environmental conditions lead to more external stakeholder engagement.Specifically, we find that increased political opposition within the external environment increases the extent to which stakeholders are engaged.Additionally, we find that a lack of administrative capacity regarding the policy decision under consideration increases stakeholder engagement.We also find various patterns of engagement, with particular clusters of stakeholders conditioned by their external environment.Overall, our findings show that public administrators (at least those in this study) recognize that certain stakeholders can have value in specific circumstances in the policy process.We conclude by arguing that stakeholder engagement is a purposeful action undertaken by these administrators to manage their policy environment.This finding supports an instrumental perspective to the engagement of stakeholders.

Purposeful Stakeholder Engagement
While definitions of stakeholders can vary, particularly in their inclusivity (Bryson, 2010), stakeholders can broadly be defined as "…any group or individual who can affect or is affected by the achievement of an organization's objectives" (Freeman, 1984, p. 46).Nabatachi (2012) further adds that stakeholders have "active and legitimate interests…by virtue of their professional role or involvement in formal organization" (p.21).In this way, the definition of stakeholders is inclusive of citizens but focuses more on the organized, professional, and active interests of stakeholders within and outside the community (Nabatachi 2012).
An instrumental perspective of stakeholder engagement assumes that in a policy process stakeholders will be engaged strategically in order to serve some clear purpose in the governance process (Freeman, 1984;Gergen, 1969;Moynihan, 2003;Preston, 1990;Soma & Vatn, 2014).This perspective does not view stakeholders as being engaged in order to advance lofty democratic ideals about the role of citizens in a democracy.Moreover, this perspective does not view government as a mere pawn in a political system that is only responsive to those that are most able to work within the confines of the system (Boviard, 2005;Stivers, 1992;Vigoda, 2002).Rather, stakeholders from an instrumental perspective are engaged in order to achieve goals and yield benefits to the agency (Skelcher, Rynck, Klijn, & Votes, 2008).Participation among stakeholders is not just something to be done according to this perspective, it is one of many inputs in a complex system that can offer clear benefits and provide assurance that goals are being achieved (Soma & Vatn, 2014).Although the instrumental perspective of stakeholder engagement has not often been studied (Reed, 2008;Reed et al., 2009) this perspective may help to explain patterns of engagement that can reconcile competing views of administrative action in democratic societies (Feldman & Khademian, 2002;Vigoda, 2002).
Beginning with the NPM movement, government actions toward stakeholders have been described as not only purposeful, strategic, innovative, and entrepreneurial, but also endowed with discretion and capacity to seek out stakeholders as partners to bring meaningful and demanded change (Moore, 1995;Osborne & Gabler, 1992).This view is in contrast to some earlier explanations of bureaucratic behavior that once considered government to be purely responsive to stakeholders and easily captured by stakeholder interests (Stivers, 1992).Contemporary views now see public administrators as collaborative and strategic since they often rely on stakeholder resources and information to achieve their goals (Bryson, 2004).
From an instrumental perspective, then, one might expect public administrators to weigh the costs and benefits of stakeholder engagement.Indeed, public administrators may more purposefully rely on stakeholder involvement as a tool to achieve objectives and accrue meaningful benefits to their agency (Bryson, 2004;Bryson, 2010;Irvin & Stansbury, 2004;Moynihan, 2003).In doing so, these public administrators may be more thoughtful about whom to engage and under what circumstances.This may result in at least two conditions in government's decision environment that are able to be managed more effectively.These conditions are strong political opposition and insufficient administrative capacity.

Political Opposition
The inherently political system in which government operates means that every policy area is likely characterized as having a range of oppositional, and oftentimes conflictual, opinions.The characterization of the environment as being oppositional may be especially true within the decision context that this research examines-that is, the contracting out of government services (Brown & Potoski, 2003;Peters & Pierre, 1998) and the adoption of innovations (Borins, 2000(Borins, , 2002)).While opposition may be ubiquitous in these environments, the instrumental perspective assumes that clear actions will be taken in order to lessen the impact of opposition.
Stakeholder engagement can aid in lessening the impact of an oppositional environment by building support (Behn, 1978;Bryson, 2004;Moore, 1995;Moynihan, 2003;Simon, Smithburg, & Thompson, 1950).This support can increase the acceptability, legitimacy, and sustainability of any policy action eventually chosen (Beierle, 2002;Bryson et al., 2011, Nutt, 2002).Cultivating support and building winning coalitions are key to the administrative task of making sure that whatever course of action is chosen does not mar the implementation process as a result of costly delays or active efforts to delegitimize a policy (Lawrence & Degan, 2001).
Identifying, mobilizing, and building coalitions have all been highlighted as key tasks of public administration (McGuire, 2002;Moore, 1995;Purdy, 2012).For instance, Skelcher and colleagues (2008) have suggested that the development of governance networks serves to shape and develop coalitions that can ensure realization of political objectives.Moreover, Irvin and Stansbury (2004) have argued that engaging stakeholders can help build alliances.When these alliances consist of influential community members, they argue, these individuals can spread their enthusiasm for a policy and diffuse opposition.Weible (2007), in an analysis of stakeholders for marine protected areas, demonstrated how stakeholders provide political support by lobbying, mobilizing resources, pushing for change, and spending time convincing the public to adopt new policy initiatives.Krueathep, Suwanmala, and Bureekul (2012) also found that when stakeholder interests were in conflict more collaboration with external network actors occurred.Moreover, Wang (2001) found that the extent of political competition among different groups in a city had a significant and positive impact on who was involved in administrative decision-making.Thus, engagement of stakeholders can be used as an instrument to manage oppositional political environments.Indeed, by engaging stakeholders outside of the core institutional actors we expect that the administrative face of government will purposefully work to create coalitions that can be used to coalesce public opinion in support of administrative decisions.
Hypothesis1: As oppositional intensity increases, engagement with stakeholders will also increase.

Administrative Capacity
The second environmental condition that may give rise to increased stakeholder engagement is a lack of administrative capacity.Feiock and West (1993) define the administrative capacity of government as having the requisite knowledge, technical skills, and expertise of an organization to evaluate feasibility among a set of policy alternatives.Similarly, Hilderbrand and Grindle (1994) noted that administrative capacity consists of having requisite information and the analytical ability to process that information.Although administrative capacity is critical to achieving the goals of an agency, the capacity to make informed and intelligent decisions varies by agency and policy context (Bowman & Kearney, 2011;Head, 2008;Honadle, 1981).For instance, Howlett (2009) demonstrated that implementing new initiatives, as compared to simply advancing old initiatives, demands greater administrative capacity.Moreover, Head (2008) found that collaborative networks were more likely to emerge in policy contexts where simple technical solutions for an issue were neither relevant nor feasible.
Notably, previous research has shown that the contracting out decision can be information and analytic capacity intensive.Brown and Potoski (2003) demonstrated that the decision to contract out has numerous costs associated with careful and rational analysis.They found that, to a large extent, many governments are not equipped to evaluate the full costs of contracting out.Thus, the high information and processing costs associated with the evaluation of the contracting out decision can make this decision environment one in which administrative capacity is demanded of organizations but is not actually present (Smirnova & Leland, 2014).
Engaging with stakeholders when faced with a difficult decision like contracting out can provide governments with information to overcome risks and costs associated with a particular decision process.Accessing this information, however, is not always easy given that it tends to be decentralized (Thomas & Poister, 2009).Therefore, to maximize the acquisition of information, managers must often reach out to several stakeholders.Reed (2008) has argued that by increasing inputs, such as the number of people providing information on a policy system, more robust and higher quality information can be obtained; and, although greater information is not synonymous with good information, Bierele (2002) found that information received from stakeholders is typically of high quality.
In the context of contracting out, there are multiple elements of a service (or service area) that stakeholders can provide information about that may not be under the direct purview of government.For example, stakeholders can provide more information about service quality or elements of the supply chain that are needed in order to provide a quality service.Thus, stakeholders can provide government with information to overcome obstacles in asymmetric contracting out environments.
Hypothesis2: As insufficiency in administrative capacity increases, engagement with stakeholders will also increase.
The general process of stakeholder analysis begins with identifying the focus, the issue, or the intervention for which stakeholder analysis is being done (Bryson, 2010;Prell, Hubacek, & Reed, 2009;Reed et al., 2009).In this way, the identification of stakeholders is usually limited to the decision context or "system boundaries" under consideration (Reed et al., 2009(Reed et al., , p. 1947)).In this study, we were limited in the identification of stakeholders to those that ICMA had already identified as being engaged in the contracting out decision.This limitation, however, does not inhibit our ability to theorize about the role that individual stakeholders play in a given context (Soma & Vatn, 2014;Weible, 2007).Indeed, Jennings and Hall (2011) identified a range of stakeholders in state level policy administration and theorized about the different types of information that these stakeholders provided government agencies.
Building on the beliefs that stakeholders are engaged to manage strong political opposition and insufficient administrative capacity, we expect that some stakeholders will be engaged for coalition building purposes.On the other hand, we expect that some stakeholders will be engaged for their information value.Thus, we hypothesize that: Hypothesis3: As opposition increases in an environment, stakeholders that are able to build coalitions and provide political support will more likely be engaged.
Hypothesis4: As insufficiency of administrative capacity within a government agency increases, stakeholders that are able to provide information will more likely be engaged.

Data and Methods
Our data on stakeholder engagement comes from the 2007 ICMA Alternative Service Delivery (ASD) survey.Although the purpose of the survey was to examine contracting out in city and county governments, the survey also included several questions about the environmental and internal conditions of agencies as well as the processes that these agencies use to facilitate contracting out (including external stakeholder engagement).Although this data has been widely used by contracting out scholars (e.g., Brown & Potoski, 2003;Fernandez & Ryu, 2008;Hefetz & Warner, 2012;Levin & Tadelis, 2010;Warner & Hebdon, 2001), there has been limited use of this data to explore stakeholder engagement in the contracting out decision (for an exception see, Schafer & Zhang, 2016).
The survey that we obtain our data from was distributed to administrators of approximately 6,000 municipalities and counties in the United States (US).The response rate was 26.6% (n=1,599).For the purposes of this study, we limit our analysis to only municipalities (n=874) so that we can examine the data within the context of demographic and financial data from the US Census Bureau.

Dependent Variables
Our key dependent variable is the type of stakeholder engaged in each municipality.To construct this variable, we created an additive index using the survey question: Who outside your local government organization was involved in evaluating the feasibility of private service delivery?
Responses included: 1) potential service deliverers, 2) professionals/consultants with expertise in the particular service area, 3) service recipients/consumers, 4) managers/ Chief Administrative Officers (CAOs) of other local governments who have experience using private service delivery, 5) citizen advisory committees, and 6) state agencies, leagues, or associations.An option for "other" was also provided.If a municipality indicated that they engaged three of these stakeholders, the external stakeholder engagement index for that municipality was calculated as three.
Since this index may not be inclusive of all stakeholders, the content validity of the index can certainly be questioned.However, as previously noted, most studies of stakeholders are limited by the decision context in which they are examined.Consequently, most studies of stakeholders by design likely do not include the full range of stakeholders that they could (Reed et al., 2009).Still, when we examined prior studies that have listed external stakeholders within a given policy area (e.g., Schalk, 2011;Thomas & Poister, 2008;Walker, Avellaneda and Berry, 2011;Weible, 2007), we found that all major categories of stakeholders in these prior studies were also included on the ASD list of external stakeholders.For instance, Thomas and Poister (2008) organized the Georgia Department of Transportation's stakeholders into customers, advocacy and interest groups, suppliers and business partners, policymaking and oversight bodies, similar agencies with different jurisdictions, and partner agencies.All of these categories of stakeholders are included in the list of external stakeholders provided on the ASD.
In addition, we undertook careful analysis of the "other" stakeholder responses in order to ensure that all applicable stakeholders had been included in our analysis.Only 17 responses were written in and several of these were not external stakeholders.However, two municipalities did note that their local Chamber of Commerce was a stakeholder category; and, one municipality noted that nonprofits were a stakeholder category.We did not create additional categories based on these responses, but this did help us to confirm that most external stakeholders had already been included on the ASD.
Hypotheses 1 and 2 were examined using the index of external stakeholders described above.In order to explore hypotheses 3 and 4, factor scores based on external stakeholders were generated.This allowed us to identify patterns of stakeholder engagement within each municipality.We conducted a factor analysis (Torres-Reyna, 2010) of the dummy variables that were used to construct the initial index.Table 1 reports the findings of this analysis.
Two factors with eigenvalues equal to or higher than 1 were retained based on the Kaiser criterion (Torres-Reyna, 2019).The factor-loading matrix (using varimax rotation) is presented in Table 2.The loadings are listed in rank order.The matrix shows that there are two clusters (or categories) of stakeholders that municipalities engage.Factor1 accounts for 31% of the observed variance and consists of "Managers/ CAOs of other local governments," "Potential service delivers," and "State agencies, leagues, or associations."Factor2 accounts for 18% of the observed variance.This factor consists of "Citizen advisory committees," "Service recipients/Consumers," and "Professionals/Consultants." In a separate analysis undertaken to explore hypotheses 3 and 4, we saved the factor scores as new variables for each municipality.We then used these scores as our dependent variables in models 2 and 3 (in Table 6).

Independent Variables
To measure our two key independent variables of interest, political opposition and administrative capacity, we created two additional additive indices using the ASD survey.Our index for political opposition was created based on the survey question asking: "Has your local government encountered any obstacle in adopting private service delivery?"The response options that we used in the creation of this index were: 1) opposition from citizens, 2) opposition from elected officials, 3) opposition from local government line employees, 4) opposition from department heads, and 5) restrictive labor contracts/agreements.We added only the "yes" responses to each question to construct the index.Thus, these scores range from 0 to 5. A "yes" to the last response (i.e., restrictive labor contracts/agreements) could indicate a political environment that is more opposed to contracting (Fernandez, Smith, & Wenger, 2007;Hefetz & Warner, 2004).For this reason, we included it in our political opposition index.
We statistically examined our political opposition index using factor analysis.The index is unidimensional, based on the Kaiser criterion that an eigenvalue be greater than one in order to be retained.We, therefore, have confidence that this index accurately captures a singular concept.In addition, using the 1992 and 1997 versions of the ASD survey, Hefetz and Warner (2004) used similar questions to construct an "opposition index."They did not, however, examine this index in relation to external stakeholder engagement as we do in the present study.
The index for administrative capacity is also based on the question: "Has your local government encountered any obstacle in adopting private service delivery?" However, the response options that we included in the construction of this index were: 1) lack of staff with sufficient expertise in contract management, 2) lack of empirical evidence on the effectiveness of private alternatives, 3) lack of precedent or institutional rigidities, 4) problems with contract specifications, and/or 5) lack of adequate contract monitoring systems.This index is, again, additive of all "yes" responses.Thus, as the value for the index increases this indicates that municipalities have less administrative capacity.In contrast, a lower value on the index indicates greater administrative capacity.
The first two responses included in our administrative capacity index clearly relate to our earlier definition of administrative capacity-particularly since these items pertain to the expertise and evidence available to staff in order to evaluate alternatives.The third response is a measure of capacity, as it captures the extent to which the agency has the ability to learn and adapt.As earlier noted, the contracting out decision is particularly information intensive due to contract specification and monitoring (Brown & Potoski, 2003).The fourth and fifth responses pertain to the administrative capacity to deal with specific attributes of contract specification and monitoring.
We statistically examined the validity of our administrative capacity index using factor analysis.
In doing so, we found that this index was also unidimensional.We, therefore, have confidence that the index accurately captures a singular concept, which is administrative capacity.
Admittedly, there are a number of limitations with using these indices.First, the construct validity of both indices can be questioned as they have not been externally validated to accurately capture the concepts that we have attributed to them.However, as noted previously, ICMA has consistently reproduced the ASD survey over the years.Each iteration of the survey has included the questions that we focus on in this study.This indicates, to some extent, that the constructs have face validity.We are also able to demonstrate the construct validity of the indices from the results of our factor analysis.
Another potential issue in this analysis relates to the additive construction of our measures.
Concerning the stakeholder engagement variable, the survey asked only if stakeholders were engaged.Respondents were not asked how often they were engaged.Thus, we are unable to measure the intensity of stakeholder engagement.We are also unable to establish whether certain stakeholders are engaged more often than others or whether different stakeholders are engaged for different reasons.
Finally, it should be noted that we made efforts to examine each variable individually (and as indices).We also considered weighting variables.However, there were no strong statistical or theoretical reasons to do so.

Control Variables
There are a number of other variables that may also explain a municipality's engagement of external stakeholders.Therefore, we included a range of control variables consistent with literature on stakeholder engagement, the decision environment of contracting out, and the broader concept of public engagement.
The first set of control variables includes specific factors related to contracting out that might influence stakeholder engagement.The instrumental perspective offered here suggests that stakeholder engagement may vary based on the degree to which the policy under consideration is innovative (Borins, 2002;Howlett, 2009).Although there is evidence suggesting that the contracting out decision is innovative (see review of literature, above), some municipalities may already have a percentage of their services contracted out.Thus, we included a control variable to measure the percentage of services already outsourced.Considering that each service has unique aspects, we expect that the level of contracting out already taking place will lead to an increase in stakeholder engagement.This may be especially true if administrators have already learned about the value of stakeholder engagement.
The second set of control variables included are municipal characteristics.These include political ideology, form of government, highest level of education, household median income, population size, and the number of full-time employees within the municipality.
A community's overall political ideology is an important consideration in the adoption of any policy.To assess political ideology, we included a measure of the perception of demand for smaller government.When a community demands smaller government there may be fewer impediments to contracting out.On the ASD survey, respondents were asked about their perceptions of the community's demand for smaller government.A "yes" response was coded as "1."A no response was coded as "0" and was used as the referent category.
Form of government is a commonly used variable in studies of general citizen engagement (e.g., Ebdon, 2002;Wang, 2001;Wang, Hawkins, & Berman, 2014;Yang & Callahan, 2007) as well as in studies of external stakeholder engagement (e.g., Boxelaar, Paine, & Beilin, 2006;Handley & Howell-Moroney, 2010).Therefore, we account for form of government with a variable that indicates whether the city's form of government is: 1) mayor-council, 2) city-manager, or 3) commission.The mayor-council form is used as the referent category in our analyses.
Consistent with the logic of instrumentality, we expect the city-manager form of government, compared to the commission and the mayor-council forms of government, to engage external stakeholders more since professional managers are trained to treat involvement as a tool of governance (Handley & Howell-Moroney, 2010;Yang & Callahan, 2005;).However, previous research on citizen engagement has shown that the commission and the mayor-council forms of government tend to engage citizens more (Yang & Callahan, 2007;Zhang & Yang, 2009).In the context of external stakeholder engagement, we refer to the assumption of instrumental action; and, we expect that the manager-council form of government will be most likely to engage external stakeholders given that their professional and technocratic knowledge base may make them more proactive and aware of the benefits of doing so.
To account for city size, we included a measure of each city's population in the year the survey was distributed as well as the number of full-time employees in each city.Yang and Callahan (2005) have suggested that smaller cities may be the more likely to engage citizens since "…residents in smaller jurisdictions tend to have more interactions with elected officials, are more knowledgeable about issues, possess a greater sense of community, and therefore are more likely to participate" (p.196).However, they did not find support for this hypothesis.Instead they found that larger municipalities had higher levels of citizen engagement.Therefore, we expect larger communities will be more likely to engage external stakeholders.We also expect that the number of full-time employees will limit the extent of external stakeholder engagement, as there will be more internal resources to manage the contracting out process.
We measured education level of the population as the percentage of persons within each municipality with a high school degree or above.We expect education level will have a positive effect on external stakeholder engagement.Indeed, several scholars have noted that one of the limitations of working with the public is the public's lack of education and training about the inner workings of government.This type of training, some have suggested, would help to ensure the usefulness of public input (Callahan & Yang, 2005;Nabatachi & Amsler, 2014).It should come as no surprise, then, that Yang and Callahan (2007) found that the level of education in an area had a positive effect on public involvement in local government, as perceived by public managers.Thus, a more educated citizenry is likely to be associated with increased stakeholder engagement by municipalities.
We measured income level of the population using household median income.Wealthier citizens are believed to have more time to dedicate to participation in governance activities.As such, these citizens may be more willing to offer their time to participate in governance activities once engaged.Carr and Halvorsen (2001) found that across mechanisms for public engagement, citizens that participated in governance activities had higher than average income levels.In the context of stakeholder engagement, then, community income level may also increase the degree to which stakeholders are engaged in the policymaking process.
We also included several controls for the geographic location of municipal respondents.Regions were categorized based on ICMA's region codes and included: "1-Northeast" (which consists of Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island, Vermont, Delaware, District  California, Hawaii, Nevada, Oregon, and Washington).
Finally, we included a measure of the metropolitization of each municipality.A central city in a metropolitan statistical area (MSA) is coded as "1," a suburban municipality is coded as "2," and municipalities not located in an MSA are coded as "3."This is an important control variable since the metro status of a city is often a critical consideration in contracting out decisions (Mohr, Deller, & Halstea, 2010) and innovative policy implementation (Homsy & Warner, 2015).
Table 3 includes variable descriptions, sources, and the hypothesized direction of all independent variables.Table 4 includes summary statistics of the dependent and independent variables.

Models
The dependent variable in Table 5 is the additive index of external stakeholder engagement ranging from 0 to This variable is a count of the types of stakeholders that are engaged in each municipality when "evaluating the feasibility" of contracting out.The appropriate modeling approach for count data is Poisson regression (Dunteman & Ho, 2006).However, there are a large number of zeros in our data and the variance (1.01) marginally exceeds the mean.For this reason, we considered the use of either a zero-inflated Poisson regression model or a negative binomial regression model.Both of these models are used for count data with high dispersion.However, after running post estimation commands in Stata for both models, we found that the Poisson model (despite the value of the variance) was the best modeling approach.As such, we proceeded with a Poisson model in Table 5.Consistent with Cameron and Trivedi (2009), though, we used robust standard errors.
For models 2 and 3 in Table 6, we used the factor scores generated from our factor analysis analyzing the component questions of the external stakeholder engagement index.For the purposes of this analysis, these indices represent continuous variables.Thus, we use Ordinary Least Squares (OLS) regression to estimate these models.The independent variables in models 2 and 3 are the same as those used in Table 5. Robust standard errors are also obtained for these models in order to account for heteroscedastic residuals.

Results
Table 5 includes the Poisson regression results, incidence rate ratios, and the effect of a one-unit change of each independent variable on the dependent variable of interest-which is a count of the types of external stakeholders engaged in the contracting out decision.Consistent with hypotheses 1 and 2, our indices for political opposition and administrative capacity are statistically significant and positively correlated with external stakeholder involvement.Specifically, a one-unit increase in the political opposition index (indicating increasing opposition to contracting out) increases the number of external stakeholders engaged by 26.18%.A one-unit increase in the administrative capacity index (indicating decreasing administrative capacity) increases the number of external stakeholders engaged by 20.24%.Figures 1 and 2 graphically display the relationship between these variables and the increase in the predicted count of external stakeholders engaged.
Several of our control variables are also significant and in the expected direction.The percentage of municipal services outsourced and citizens' demand for smaller government both increase the extent of external stakeholder engagement.It is worth noting that with a one-unit increase in the percentage of services outsourced, local government officials are 97% more likely to engage   external stakeholders.By contrast, when citizens demand smaller government local government officials are 51% more likely to engage external stakeholders.
In addition, the results from our Poisson regression model show that the household median income variable is statistically significant.The magnitude of this effect, however, is trivial.Interestingly, form of government failed to show any statistical significance in the model.In fact, the council-manager form of government did not (statistically) behave any differently than the mayor-council form of government in relation to the number of stakeholders engaged.This finding is inconsistent with findings from previous research showing that the council-manager form of government is more likely to engage citizens.
From a geographical perspective, we did not find a distinct pattern across municipalities in their stakeholder engagement practices.The only exception to this was that local governments located in the North Central region were more likely to engage stakeholders than those located in the Northeast region.However, we did find that governments located in center cities engaged more stakeholders than governments located in either suburban areas or governments located in nonmetro municipalities.This was expected as central city governments tend to be more diversified and often have greater need to engage stakeholders in decision-making.
The results of our Poisson regression model are consistent with an instrumental theory of stakeholder engagement.That is, as political opposition and a lack of administrative capacity increase, the types of stakeholders that governments engage also increase.To unpack this finding, the next step in our analysis was to estimate two OLS regression models (see model 2 and model 3 in Table 6).The dependent variables in both models are the factor scores of external stakeholders engaged.
Model 2 presents the results of each independent variable's effect on factor1.As previously noted, factor1 consists of the following stakeholders: 1) potential service deliverers, 2) managers and CAOs of other local governments, and 3) state agencies, leagues, or associations.The results of the regression analysis show that political opposition increases the engagement of these external stakeholders by 14%.A lack of administrative capacity increases the engagement of these external stakeholders by 18%.Findings for the control variables for model 2 are consistent with the findings for the control variables shown in the Poisson regression model in Table 5.
Model 3 in Table 6 presents the results of each independent variable's effect on factor2.As previously noted, factor2 consists of the following stakeholders: 1) professionals/consultants, 2) service recipients, and 3) citizen advisory committees.The results of this analysis show that political opposition increases the engagement of these external stakeholders by 14%.Moreover, a lack of administrative capacity is an insignificant predictor of engagement with this cluster of external stakeholders.
The results of models 2 and 3 indicate that some external stakeholders are indeed engaged to overcome a lack of administrative capacity while others are engaged to overcome political opposition.The results further suggest that the public administrators in this study actively manage their environments by engaging external stakeholders.

Discussion
Models 2 and 3 show that a lack of administrative capacity is associated with factor1; however, lack of administrative capacity is not associated with factor2.This finding provides some evidence that public administrators actively shape the policy process as opposed to merely react to it.
The factor1 cluster of stakeholders are likely to provide public managers with useful information concerning the contracting out policy decision that will help them to overcome a lack of administrative capacity.For example, potential service deliverers may be able to offer well developed bids for municipalities to help them identify hidden costs (Brown & Potoski, 2003;Hefetz & Warner, 2004).Moreover, managers of other local and state governments, leagues, and/or associations may serve as repositories for best practices.These external stakeholders can all provide public managers with insights into substantive issues (Jennings & Hall, 2012).
In contrast, the stakeholders included in our factor2 cluster were engaged as a result of strong political opposition.Service recipients, in particular, are critical to political suasion given that they have a vested interest in the outcome (Rowley & Modoveanu, 2003).Thus, service recipients can be expected to mobilize around an issue.Citizen advisory committees can also act as coalition builders because they serve as a liaison between government and the public (Schaller, 1964).These committees provide leadership and voice within a community.
Interestingly, we found that consultants and professionals were also in this category (although they had the lowest factor loading).While some may not consider consultants to be political actors who are engaged as a result of strong political opposition, consultants have often been thought to "depoliticize" environments (Beveridge, 2012).These experts (in their own right) are often used to help build support for political ideas.They are also often used to provide political coverage for politicians.Indeed, consultants allow politicians to refer to the consultants' findings as the basis for their policy decisions, rather than a politician's own ideologies.Thus, consultants make it easier for politicians to adopt a politically contentious policy in the name of technical necessity (Saint-Martin, 1998, 2005).Consistent with this argument, Howlett and Migone (2013) in their survey of Canadian government consultants found that consulting reports provided more process information regarding how to adopt a policy-including advice about the acceptability of policy proposals-as compared to more technical or substantive policy content.This could mean that consultants help to build coalitions and political support around opposition.
Taken together, the results of this study suggest that public managers should think critically about the environment in which the contracting out decision occurs.When evaluating the feasibility of contracting out, public managers should strategically identify stakeholders that can ameliorate the specific issue(s) they face (whether political opposition, administrative capacity, or both).
While some may find it obvious that public managers would act strategically, we see this issue as far from settled-particularly in the context of innovative policy decisions.

Conclusions
In this study we found that municipalities with political opposition to contracting out and less administrative capacity were more likely to engage multiple external stakeholders as part of the policy decision process (while controlling for a range of relevant factors).Moreover, we found that clusters of stakeholders were engaged for specific purposes.Some stakeholders were engaged in order to provide political and coalitional support to municipalities, while others were engaged to provide information and analytic capacity to municipalities.
Our findings, thus, contribute to the limited literature specifically on stakeholder engagement (and to some extent the broader literature on public engagement) by public administrators.While stakeholders have long been recognized as powerful actors in local governments their influence has not always been viewed as providing benefits to government.Often, like public engagement, stakeholder engagement has been viewed as either a normative value of government that must be legally accounted for or as a depleting source of frustration that must be attended to.However, our findings cast new light on these views.In particular, we demonstrate that stakeholders have instrumental benefit to the administration of government.Viewing stakeholders and the public as meaningful and helpful contributors to the governing process may be a pivotal and motivating step toward advancing engagement processes in government.
We were also able to overcome the issue of limited theory development on the topic of stakeholder engagement as we categorized stakeholders by the purpose(s) that they served in the contracting out policy decision.Although further research (across a variety of policy areas) is needed to confirm the patterns of engagement that we found in this study, our findings highlight the fact that by focusing on a specific policy area scholars are not necessarily limited in their ability to develop theory about stakeholders.This should be encouraging as broader ideas about the value of stakeholders are needed in order to increase our understanding of the role that stakeholders play in the work of public administration.
Second, the findings from this study are consistent with an instrumental perspective on stakeholder engagement.Some municipalities view stakeholders as sources of strategic importance; and, in order to achieve their goals these municipalities have sought to elicit stakeholder participation in policy decision processes (e.g., contracting out).This suggests that some municipalities may be strategic in their approach to engaging stakeholders as well as their approach to policymaking more generally.These municipalities are likely to recognize the value of stakeholders in decision processes and, as such, they may proactively elicit stakeholder participation.
Finally, our findings add to what we know from contemporary theories of public administration, such as NPM, that suggest public administrators are proactive in their engagement with stakeholders.Indeed, our findings indicate that public administrators actively manage their environments (e.g., by engaging with external stakeholders).They also recognize systems level issues that can influence policy outcomes.
Still, it is important to point out that nearly 50% of the municipalities in our study did not indicate that they attempted to engage any type of external stakeholder at all in evaluating the feasibility of contracting out.Thus, while our theory of instrumental stakeholder engagement seems to explain the behavior of some municipalities it does not explain the behavior of all municipalities.
As such, there are a number of questions that remain; and, future research should further explore the hypotheses in this study.Ideally, this research will use different policy areas and different measures of proactive management.

Figure 1 .
Figure 1.Marginal Effect of Political Opposition Index on External Stakeholders Engaged

Figure 2 .
Figure 2. Marginal Effects of Administrative Capacity Index on External Stakeholders Engaged

Table 4 .
Summary Statistics

Table 5 .
Poisson Regression Results: Count of the Types of External Stakeholders Engaged

Table 6 .
OLS Regression Results: Factors 1 and 2 Based on Stakeholders Engaged