Printer Friendly

Municipal shared service collaboration in the Alberta Capital Region: the case of recreation.

Why do municipalities engage in local governmental cooperation and, more specifically, in shared services arrangements (SSAs) to provide certain municipal services? (1) This question is becoming increasingly relevant. Despite a substantial body of literature on the delivery of local government services, inter-local government cooperation is receiving additional attention. (On the U.S., see Mildred Warner and Robert Hebdon [2001] and Curtis Wood [2006]; on Australia, see Brian Dollery, Warrick Moppett and Lin Crase [2006]; on Europe, see R. Hulst and A. Van Montfort [2007]; and, on Canada, see Robert Bish [1999] and Neil Hepburn, Edward LeSage and Melville McMillan [2004]. Much of the earlier literature on the organization of local service delivery focused on contracting-out, often either specifically on contracting with private providers or with little or no distinction made between contracts with private agents and contracts with other governments. (2) Perhaps appreciating that most external arrangements for local government services are with other governments or public authorities, that they have a long history, and, possibly, that the motivation for inter-governmental cooperation has increased, numerous recent papers on local service delivery have focused specifically on inter-governmental arrangements and cooperation (Pagano 1999; Warner and Hebdon 2001; Wood 2006).

This article is an empirical examination of shared service arrangements for recreational and cultural services among municipalities within the Edmonton metropolitan region. Our purpose is to understand better the role and potential for inter-municipal service cooperation by identifying factors that promote or discourage (i.e., the determinants of) municipal participation in inter-municipal agreements for service delivery. (3) We focus on four important considerations in municipal decision-making: 1) potential cost-/service-delivery economies expected to be closely related to population; 2) municipal fiscal well-being; 3) concerns for local control motivated by the desire to have services corresponding well to local preferences; and 4) available and appealing partners. The data simplifies the analysis in various ways. All the centres are within a single metropolitan area and within one province, so the regional "market" and legal/institutional factors are the same. We investigate a single service. All the arrangements are inter-governmental. All the municipalities were members of the Alberta Capital Region Alliance (ACRA), the common venue for interaction, and ACRA was comprehensive in its coverage of the municipalities in the region.

This study provides insight into some less commonly examined features. First, we not only examine the determinants of partnering as affected by the characteristics of that individual municipality but we also explore the characteristics of the potential partners as determinants of the existing SSAs to gain insight into the selection of partners. Municipalities wishing to collaborate want worthy partners. E.C. Shaeffer and W.C. Bryant (1983) list some of the considerations--the adequacy of potential partners' resources, organizational limitations, skills and intellectual resources, and structures that will facilitate collaborative action. Willingness and enthusiasm for collaboration is, of course, critical. That relies heavily on mutual trust (Einbinder et al. 2000), supported by the "glue" of shared values and norms (Alter and Hage 1993; Thurmaier and Wood 2002) and facilitated by organizational and political culture (Visser 2002) and network linkages (LeRoux and Carr 2007). Shared values are often associated with greater homogeneity. Reto Steiner (2003) found evidence that greater homogeneity promotes inter-local agreements in Switzerland, but Michael Nelson (1990) found some evidence to the contrary in metropolitan areas in the United States. In an analysis of school district consolidations in Ohio, David Brasington (1999, 2003) found there that cost-saving opportunities overwhelmed concern for socio-economic heterogeneity in the consolidation decision. The importance of attractive partners spurs us to look at the characteristics of potential partners as an influence on inter-local agreements.

Second, our analysis highlights small municipalities within a metropolitan environment. The U.S. studies that dominate literature on the determinants of local government cooperation and contracting tend to focus on relatively large municipalities. (4) In contrast to most other studies (Wood 2006, being an exception), our data set is small and is dominated by small municipalities. Our data comes from twenty-two municipalities in the Alberta Capital Region, and only four have a population exceeding 25,000. In particular, the modest size of many of the communities within the Alberta Capital Region can be expected to make potential cost-savings related to economies of scale, scope and sharing more important.

This research is of interest to many for several reasons. First, it assesses the traditional determinants of inter-local collaboration in a Canadian metropolitan region. Second, the attention to small municipalities

makes the work relevant to the large portion of Canadians who live in small municipalities, many within metropolitan areas. Finally, our exploration into the attractiveness of potential partners is unusual.

The article begins with a discussion of our model and data and then presents empirical results relating to the proposed determinants. That is followed by an interpretative discussion and conclusions.

Model and data

We systematically tested several prospective determinants of recreational and cultural agreements among the twenty-two Edmonton-area municipalities. There were twenty-one such agreements identified in our 2002 survey of SSAs in the Edmonton region (see Table 1). They covered such services as ice arenas, pools, playing fields and halls. The vast majority of these SSAs (eighteen of twenty-one) were shared cost agreements, and three covered planning and coordination. All were formal written agreements. Thirty-nine municipalities were party to these agreements. All but one agreement was a two-party arrangement between neighbouring municipalities. The exception was the tripartite agreement among the City of Spruce Grove, the Town of Stony Plain, and Parkland County covering the TransAlta Tri-Leisure Centre, a multiservice recreational facility serving the three communities.

Municipalities in the data analysed include not only those involved in existing SSAs but also the municipalities that might have been party to potential agreements but between which no agreement had emerged. The unit of analysis in other determinant studies is each municipality, and the dependent variable is typically whether it is in an SSA (or contract) or not, the number of such arrangements participated in, or the share of the budget devoted to such arrangements. The regression results provide information as to the factors affecting participation in SSAs (or contracts) or the extent of those agreements. We use parallel data with the dependent variable, whether the municipality participates or not in a recreation SSA. However, like Brasington, we also include in our observations information on the partnering municipality, plus data on potential partnerships in SSAs that were not realized. By this approach, we gain insight not only into why municipalities might be inclined to partner (as other studies) but also into what may determine with whom they partner. All neighbouring municipalities--that is, those sharing a common border--were regarded as potential SSA partners. There were thirty-one bordering situations. In addition, we looked at each possible partnership from the perspective of each player. That is, we included both parties in a potential partnership dyad as separate observations. That gives sixty-two possible partners in two-party agreements. The number of observations is twice the number of possible agreements because we are interested in the appeal of each agreement from the perspective of each partner. The Tri-Leisure facility agreement was included as a special case because the players there are engaged in other agreements. This case was treated as three situations of one potential partner joining with the other two; hence, it adds three more observations. Thus, using the sixty-five existing and potential agreements, we sought to identify factors determining the probability of a particular municipality belonging to a shared service recreational agreement with a neighbouring municipality and, given the alternatives, which neighbour is the more appealing partner.

We used a logit model to analyse the data. The logit model is a regression method used to analyse data when a discrete (and particularly a binary) choice is made--for example, vote or not vote, purchase a product or not, attend college or not- and the objective is to explain the decision. Hence, it is a method well suited for explaining the decision to enter into an SSA. (5) The dependent variable indicates whether a municipality has an SSA for recreational services with a neighbouring municipality. The value is one if an agreement exists and zero if not. As noted above, various explanatory variables have been suggested in the literature. Because our municipalities vary greatly in population, from 380 to 666,105 persons, we are especially interested in economies of "scale." Hence, we give particular attention to the populations in the potential partnering communities.

Many of the SSAs are between a county government and one of the urban centres within its boundaries. (6) Because county residents typically use recreational facilities in a nearby urban municipality, only a portion of the county's residents represents the relevant population for sharing in the use of facilities in the particular neighbouring community. Hence, the relevant population of the county as a potential partner is not always that of the whole county even when the county is formally the partner in an agreement. Therefore, it is necessary for three counties to allocate county residents among (or associate them with) the urban centres in the county. This distribution is simple in the case of Leduc County but, as explained in the appendix at the end of this article, rather more complicated in the case of Parkland County and Sturgeon County.

Various fiscal, demographic and economic variables have been identified as possible determinants of municipalities' participation in recreation SSAs. We systematically explore the alternatives. Those data come 1) from the municipal profiles and, when required, the more detailed municipal financial information system sources available at the Alberta Municipal Affairs web site,; and 2) from the community profiles database of Statistics Canada, at

Estimation results

The literature leads us to four hypotheses concerning what motivates (or restrains) municipalities to engage in SSAs: 1) economies in service delivery; 2) fiscal considerations; 3) local control; and 4) available and appealing partners. We tested these, progressively looking for evidence supporting each hypothesis. Given the size of the municipalities dominating our data, we anticipated that the potential benefits from realizing economies of scale are an important incentive for entering into recreation SSAs. Hence, we began by examining the impact of those economies.

Hypothesis 1: Economies in service delivery

As is typically assumed in this literature, population is taken to reflect the ability to realize economies in service delivery. (7) Municipalities with larger populations are expected to be better able to achieve those economies and smaller municipalities to be somewhat disadvantaged. Hence, the initial regression includes just "own" population and the neighbour's population. The results for this case are the first reported (those under Hypothesis 1) in Table 2. The negative and significantly-different-from-zero coefficient for own population indicates that the larger the own population is the smaller the probability of its being in a recreation SSA. The impact on the probability varies with population size. The coefficient for the neighbouring municipality's population is also negative and significant but about one third as large. Thus, larger populations, certainly of one's own municipality and to a lesser degree that of the neighbour's, diminish the likelihood of belonging to an SSA. Thus, recreation SSAs are more appealing to smaller municipalities. That outcome lends support to Hypothesis 1: that is, that the potential to realize service-delivery economies (whether taken as cost reductions and/or service improvements) is an important determinant of participation in SSAs.

The two population variables explain much of the variance, 0.804 or 80.4 per cent of the normalized success rate. (8) Hence, municipal populations explain much of the SSA behaviour in this data. Because the population variables explain so much of the variation among observations, there is little variance for additional variables to explain; thus, there is limited opportunity (19.6 per cent of the variance remains to be explained) for other variables to contribute significantly to the regression. To address this problem, we ran a series of regressions of three variables in which one other (or additional) variable was included with the population variables. (9) The other (or added) variable is described in the left-hand column of Table 2. Following this procedure, it is possible to examine the other hypotheses.

Hypothesis 2: Fiscal considerations

Hypothesis 2 proposes that municipalities in more difficult fiscal situations are more inclined to participate in SSAs. We report on two variables that we regard as the best available measures of fiscal pressure (in the sense of reflecting demand on municipal fiscal capacity). They are the equalized tax rate and the debt to debt limit ratio. Those results are reported in Table 2 under Hypothesis 2.

The equalized property tax rate was expected to be a sound indicator of fiscal pressure. Because municipal tax rates (and assessments) are not necessarily comparable across municipalities, for levying the provincial school property tax, the province utilizes equalized tax rates that, while the rates may vary among municipalities, result in comparable tax burdens per dollar of equalized assessed value. When we use the equalized tax rate for residential and farm property as a variable measuring local tax burden, the coefficient is positive and significant (at the twenty- or, more accurately, at about the twelve-per-cent level). While statistically weaker than for the population variables, this result suggests that a relatively larger tax rate (and relative tax burden) encourages SSA participation, a result consistent with our expectations.

For an indicator of the burden of debt on a municipality, we used the ratio of debt to the provincially determined debt limit. (10) A higher debt limit per capita reflects the ability of the municipality to assume debt and, more generally, greater fiscal capacity, but, the greater the portion of the debt limit utilized, the greater the burden whatever the debt limit. If actual debt is large relative to the debt limit, there is greater fiscal pressure on a municipality and so it could be more predisposed to SSAs. When the debt to debt limit ratio is the additional variable, the coefficient is positive and significant (at the ten-per-cent level); a result again supporting the hypothesis that fiscal conditions matter.

Hypothesis 3: Local control

Hypothesis 3 predicts that the desire to retain local control is negatively associated with SSA participation. Everything else the same, communities with higher incomes are expected to enjoy greater discretion and so to be more willing to make choices favouring local control. Also, demographic variables are often seen to reflect the demand for services and/or the voters' interest in those services, including their concern for local control. Owner-occupants of residences are often felt to be more politically active and more concerned about taxes and services than renters. A similar supposition is held for the elderly (i.e., those sixty-five or older) except that, for them, taxes may dominate services. Particularly for recreational services, the number of youth (e.g., persons under twenty) may be a determinant of demand. To the extent that higher incomes, owner-occupancy, elderliness, and youth are logically associated with municipal preferences for control, we should expect to see Hypothesis 3 confirmed where authorities are differentiated on these variables. When these variables were introduced, only median family income was found to have a coefficient that differed significantly from zero and only at the twenty-per-cent level of significance, though negative as expected. Hence, we concluded that the prediction of Hypothesis 3 is not supported by our data. That is, local control, as we measure it, does not seem to be an important factor here.

Hypothesis 4: Available and appealing partners

Partnerships require agreement between at least two parties. Therefore, the characteristics of the potential partners can be expected to be possible determinants of whether or not an SSA exists. Neighbour's population size has already been included and discussed. In order to gain further insight into the potential influences of fiscal factors and local control variables, their values in the neighbouring community are also introduced (individually) into the regressions. A difficulty emerges here in that the coefficients of these variables in the neigbouring community may reflect two separate influences. First, they should be picking up the influence of that variable on the neighbouring community's decision to engage or not in SSAs. Second, these variables may also reflect the attractiveness of the neighbouring community as a potential partner. If the two influences operate in opposing directions, the interpretation of the results is complicated.

Adding the neighbour's fiscal and control variables to the regression was not helpful, so the results are not reported. Differences in the values of the variables, as an alternative to their levels, were introduced in an effort to determine if homogeneity (heterogeneity) promoted (discouraged) cooperation in SSAs, but they too were not important. Of course, population sizes (including neighbour's population) continued to dominate as determinants of SSA participation.


Because the sample is small and the analysis necessarily partial, the econometric results are tentative. Beyond the population variable that supports the economies hypothesis, the results are weak. There is essentially no evidence in our data that preferences for local control matter. However, the two indicators of fiscal health do suggest that greater pressure on a municipality's fiscal capacity may positively influence the propensity to engage in SSAs. Beyond population, the neighbour's characteristics appear to have little influence on its appeal to a potential SSA partner. Their impact and interpretation, however, are complicated by the push and pull forces that some characteristics may have and by the influence the variables have on the neighbour's own decision.

In spite of the qualifications relating to the other variables and hypotheses, the results for own (especially) and neighbour's population variables are relatively strong and quite robust, and they are consistent with the hypothesis that, for small communities, a potential for realizing economies in service delivery is important. The coefficient for the own population variable is typically about -0.25 and significant at the (standard) rive-per-cent level. The coefficient for neighbour's population is consistently smaller, in the range of- 0.078 to -0.10, and is usually significant at the rive-per-cent level. These results, plus their contribution towards explaining the variance, indicate that "own" population and neighbour's population are important determinants of the probability of entering into a recreational SSA in this environment. The impacts of the population variables on that probability are demonstrated in Table 3. When the two communities are of equal size, the predicted probability of belonging to a recreation SSA is almost one for communities of 6,000 persons (approximately the median population in our data) or less, but it declines rapidly as population increases--to about 0.87 at 12,000, 0.49 at 18,000 and 0.12 at 24,000. If own population varies but that of the neighbour's is constant at 6,000, the estimated probabilities for the larger communities increase--to about 0.91, 0.71 and 0.37 for communities of 12,000, 18,000 and 24,000, respectively. These results support the contention that small communities have much to gain through cooperation and in sharing recreational facilities and their operations. It is logical to expect that benefits come in terms of cost-savings and/or service improvements from the economies of scale and economies of sharing that SSAs may achieve.

These results were not unexpected. The academic literature and responses to our survey both suggested that economies in service delivery can be important for small local authorities. George Boyne (1998b) is correct in arguing that population can be a poor measure of scale of public output and that many papers exploring public-service contracting overlook that fact. This problem has been recognized (and occasionally addressed) in the literature that examines the "publicness" (or congestion) of local publicly provided outputs. Wallace Oates (1988) highlights the problem through his "zoo effect." That is, it is difficult to distinguish economies of scale or sharing when savings from serving a larger population may be translated into improved services rather than expenditure reductions. For example, with a sufficiently large population, even small per capita savings on other recreational activities may enable provision of a zoo at the same per capita outlay. That is, the recreational services have been expanded and improved with population but at a constant per capita cost. Ideally, for identifying economies of scale, one wants to measure outlays for constant quality services. Holding quality of public services constant or even having an index of their quality is rare. Melville McMillan (1989) takes advantage of the tire insurance ratings of municipal tire departments to test for publicness in fire services--that is, test for per capita cost reductions with population increases when service quality is held constant. He found publicness, or what is often referred to as economies of scale (but which may involve scope or sharing), in small Ontario municipalities (e.g., those with populations under 10,000). Even without the benefit of indices of service quality, M. McMillan, W. Wilson and L. Arthur (1981) found evidence of publicness (i.e., economies with increasing population size) in service delivery in small Ontario urban centres but (like others) little or no evidence of such publicness for larger urban areas. Larger centres may, however, still realize certain economies with population, but it is more difficult to detect (unless the quality of service is measured) because the gains achieved are often used to improve service quality.

The prevailing importance of population as a determinant of intermunicipal cooperation observed here is consistent with the results of others who have examined small municipalities. David Brasington (1999) found that cost-savings resulting from the economies associated with larger populations, "scale" economies, were the driving determinant of the decision of Ohio municipalities to establish joint school districts. Similarly, he also found that once population economies were accounted for, socio-demographic factors (notably inter-jurisdictional income and racial differences) were not found to be important determinants. (11) From his analysis of Swiss municipalities (mean population of 2,501), Reto Steiner (2003) observed that municipalities' participation in SSAs was inversely correlated with population size and that SSAs served as an alternative to merger for many.

Responses to our survey were also consistent with the existence of economies with population size for small communities. Our survey asked municipalities for their main motivation for participating in SSAs. The most frequent response was that it provided a new opportunity to provide a service that did not previously exist (45.5 per cent of respondents). Second was that the SSA was seen as a way of reducing cost while maintaining service levels (22.7 per cent). A modest 13.6 per cent reported that the SSA was seen as a way of improving service without increasing cost. Finally, 18.2 per cent of respondents reported that use of an SSA was necessary to avoid losing the service altogether. Clearly, the opportunity to maintain, improve or expand services dominated pure cost-savings as the primary motivation for our municipalities engaging in SSAs. However, those opportunities were only possible because of the economies offered through collaboration in an SSA. (12) Those potential economies are greater and more important for small municipalities.

The advantages for smaller communities entering into SSAs can be illustrated. Many of the recreational agreements involve ice arenas. Even a relatively basic facility requires a significant population to ensure effective use. To illustrate, the City of Edmonton plans an ice arena for about every 22,000 persons. Currently, these are quality facilities and typically include two ice rinks. Similarly, St. Albert, with a population of 53,080, has four indoor ice surfaces and Ft. Saskatchewan, with 13,120 persons, has two. In contrast, eight urban centres of the twenty-two municipalities in the metropolitan area have populations less than 2,500. Clearly, there are potential advantages for them to cooperate with the surrounding rural areas, the people in the county, to provide and operate recreational facilities.

Shared services arrangements are not limited to the small communities. A number of the large centres are also involved in agreements. However, there is a fairly distinct critical population level separating those that are engaged in SSAs and those that are not. The cities of Edmonton (666,105), St. Albert (53,080), and Ft. Saskatchewan (13,120) plus the (specialized) County of Strathcona (71,985) are not involved in any recreational SSAs. Meanwhile, the cities of Leduc (15,030) and Spruce Grove (15,980) and the Town of Stony Plain (9,585) are the largest of the centres with SSAs. Ft. Saskatchewan stands out as being more comparable in size to the latter group but without an SSA in this area. Its low equalized tax rate (reflecting high non-residential assessments), moderate debt burden and relatively high income probably--all are consistent with our empirical results--contribute to its independence. The larger centres have the populations to make effective use of typical facilities, thereby keeping the costs to taxpayers and users reasonable. Indeed, they often need to duplicate facilities to provide adequate service to their residents--that is, horizontally integrate services among similar geographically dispersed units and users. Furthermore, where populations are sufficiently large, those municipalities can also realize economies of scope in providing recreational facilities offering a wider range of activities and services. (13) For example, by collaborating, Stony Plain, Spruce Grove, and Parkland County were able to provide a much superior facility in their area. Even so, St. Albert and Sturgeon County have debated cooperating to provide a similar facility to their communities, and the City of Leduc and Leduc County twice considered but rejected such an option.

St. Albert and Ft. Saskatchewan do not report formal recreation cost-sharing agreements but sharing is ongoing nonetheless. From our surveys, we found that these two municipalities make no attempt to restrict access to their recreational facilities by residents of the surrounding counties or to charge differential fees. When asked about this, one municipal recreation director stated that the additional rural users did not constitute a sufficiently large proportion of their total users to worry about. Various reasons may underlie this response. First, it could imply that the potential partners were not expected to contribute in a major way towards costs even with an SSA. Second, it is quite possible that through the normal charges (that all users pay) the rural users already make what is felt to be an adequate contribution. Third, rural residents may already be seen to be contributing indirectly as important customers of the urban centre's (taxpaying) businesses. Fourth, engaging in SSAs or imposing special charges on non-residents involves costs that must be balanced against the benefits.

The results of our study suggest that population economies are important in determining participation by small municipalities in shared recreational service agreements. This result is consistent with the literature focusing on public-service provision in small communities. In fact, when using a small data set encompassing many small municipalities, population size appears as the dominant determinant. Other factors do not appear important (or, in the case of fiscal considerations, not nearly as important), but their contribution is likely obscured by the small database.


Collaboration requires extra effort, so municipalities, like other organizations, are not inclined towards shared service arrangements or other forms of inter-municipal cooperation as a first option. Real gains must be expected if municipalities are to be so motivated. This article investigated factors contributing to municipal collaboration--specifically, the determinants of inter-municipal SSAs for recreational services. The literature suggested four categories of influential factors--potential cost-/service-delivery economies, fiscal considerations, maintenance of local control, and the appeal of potential partners. Previous empirical studies usually excluded or underrepresented small municipalities. That is unfortunate in that there are many small municipalities and for them service sharing might be expected to be attractive. Small municipalities comprise the bulk our observations. Hence, to our knowledge, our analysis provides a rare instance of research of this type incorporating small municipalities. (14) In addition--although with limited success--we extended the analysis beyond earlier work to consider the characteristics of potential partners in an effort to gain insight into the selection of partners (i.e., who partners with whom).

Population is found to be the dominant determinant of municipal engagement in SSAs. Given the "lumpiness" of the recreational facilities typical of those covered by the agreements analysed (e.g., ice arenas), additional users generate economies of scale and sharing and, thereby, reduce unit costs. Thus, small municipalities with a single facility can be expected to be more inclined towards SSAs, while large municipalities, where facilities are often being replicated and the potential economies are small, are less inclined towards SSAs. The importance of population in our results contrasts with its weaker performance in other studies because, we believe, of the greater relative potential for cost-savings for low population jurisdictions. In turn, the larger the population of a potential partner, the lower the likelihood of the two partnering in an SSA. However, this effect is much smaller, only about one third of the effect of own population, and statistically somewhat weaker. Thus, for recreation agreements in the Edmonton area, service-delivery economies linked to both own and neighbour's population appear to be important determinants of SSA participation.

Constrained by the limited number of observations, it was not possible to obtain conclusive results for variables representing the other factors. Still, measures of fiscal pressure (equalized property tax rates and debt to debt limit ratios) suggest that fiscal pressure encourages participation in SSAs. However, there is no evidence from our results that characteristics widely accepted as reflecting a preference for maintaining local control matter. Also, aside from population, no characteristics that might make potential partners more or less appealing (such as inter-jurisdictional heterogeneity in characteristics) became evident. While small municipalities may be more appealing to each other as SSA partners, it is most likely that it is their small populations that make them individually more willing to participate in SSAs.

While the results were not as conclusive as we would have liked, it appears clear that pursuit of economies in service delivery leads to collaboration among municipalities and that those opportunities are particularly important for municipalities with small populations. These results, the broad range of municipalities investigated, and the examination of partners invite further research.



Regression model

The logit model described (see Pindyck and Rubinfeld 1981, for details) is suitable but not ideal. A deficiency arises because agreements are observed only when both parties agree. Reasons underlying lack of agreements where none exist--whether one or the other or both parties were not interested--are not observed. The statistical model designed specifically to address this type of partial observability problem, the Poirier bivariate probit model, is a two equation model solved simultaneously (e.g., see Brasington 1999 and 2003). Efforts to apply that model here were largely unsuccessful as the models often failed to converge (or converge to meaningful parameters) or, when converged, the parameters were usually not statistically significant at conventional levels. These problems resulted from the small number of observations (thirty-four) for each equation. Setting up the estimation, as done here, with observations on both potential partners to be estimated by a single equation, affords data on all the players and provides about twice as many observations. The estimates are consistent but are inefficient because they neglect the inter-relationships between the error terms of the separate equations. That interaction, however, does not seem to be a problem here for parameter estimation. Experiments with random selections of sixty-five observations or selections of subsets of the sixty-five observations, done to reduce possible inter-relations of the residuals, yield results similar to those presented except for the expected reduced statistical significance of the parameter values when the number of observations is smaller.

County service areas

Leduc county has defined recreational areas surrounding the urban communities in the county. These recreational areas cooperate through a joint board with the neighbouring urban centre for delivery of recreational services to their combined resident populations. The populations of those rural recreational areas are used as the relevant county population for the analysis. Parkland County and Sturgeon County do not have such recreational areas. Parkland, however, keeps a record (for cost-sharing purposes) of its residents' use of recreational services in the urban centres in the county. That information was the basis for allocating Parkland County residents to agreements with the county's urban centres. Sturgeon County has cost-sharing arrangements with its urban centres but does not track county residents' usage. For Sturgeon County, the number of county residents paired with each urban centre for the analysis was determined by size of the centre with allowance for notable variation in the nearby density of the county's population. Appling the same method to Parkland County resulted in an allocation closely aligned to actual usage so this assignment is believed to approximate reasonably well the Sturgeon County population serviced by facilities in the neighbouring urban centres. In contrast to the other counties, Strathcona County, a "specialized" municipality, is (and is treated as) a single entity. That county provides all recreational services to both rural and urban residents within the county's boundaries. The 2001 populations are those used for the analysis.


Alter, C., and J. Hage. 1993. Organizations Working Together. Newbury Park, Calif.: Sage Publications.

Bish, Robert L. 2000. "Evolutionary alternatives for metropolitan areas: The capital region of British Columbia." Canadian Journal of Regional Science 23 (1) Spring: 73-87.

--. 1999. "Local government service production in the capital region." Report. Victoria: University of Victoria, School of Public Administration, Local Government Institute. Available at

Boyne, George A. 1998a. "Bureaucratic theory meets reality: public choice and service contracting in U.S. local government." Public Administration Review 58 (6) November/December: 474-84.

--. 1998b. "The determinants of variations in local service contracting: Garbage in, garbage out?" Urban Affairs Review 34(1) September: 150-63.

Brasington, David M. 1999. "Joint provision of public goods: the consolidation of school districts." Journal of Public Economics 73 (3) September: 373-93.

--. 2003. "Snobbery, racism, or mutual distaste: What promotes and hinders cooperation in local public good provision?" Review of Economics and Statistics 85 (4) November: 874-83.

Dollery, Brian, Warrick Moppett, and Lin Crase. 2006. "Spontaneous structural reform in Australian local government: The case of the Gulgandra co-operative model in New South Wales." Australian Geographer 37 (3) November: 395-409.

Einbinder, Susan D., Peter J. Robertson, Armando Garcia, Gojko Vuckovic, and Rino J. Patti. 2000. "Interorganizational collaboration in social service organizations: A study of prerequisites to success." Journal of Children and Poverty 6 (2) September: 119-40.

Ferris, James M. 1986. "The decision to contract out." Urban Affairs Quarterly 22 (2) December: 289-311.

Hepburn, Neil, Edward LeSage, and Melville McMillan. 2004. Shared Service Arrangements: Determinants of Success. Information Bulletin No. 78. Edmonton: Western Centre for Economic Research, University of Alberta.

Hulst, R., and A. van Montfort (eds.). 2007. Inter-Municipal Cooperation in Europe. New York: Springer Netherlands.

Joassart-Marcelli, Pascale, and Juliet Musso 2005. "Municipal service provision choices within a metropolitan Area." Urban Affairs Review 40 (4) March: 492-519.

LeRoux, Kelly, and Jered B. Carr. 2007. "Explaining local government cooperation on public works: Evidence from Michigan." Report. Detroit: Political Science Working Group on Interlocal Services Cooperation, Wayne State University.

McDavid, James C., and Eric G. Clemens. 1995. "Contracting out local government services: The B.C. experience." Canadian Public Administration 38 (2) Summer: 177-93.

McMillan, Melville L. 1989. "On measuring congestion of local public goods." Journal of Urban Economics 26 (2) September: 131-37.

McMillan, Melville L., W.R. Wilson, and L.M. Arthur. 1981. "The publicness of local public goods: Evidence from Ontario municipalities." Canadian Journal of Economics 14 (4) November: 596-608.

Morgan, David R., and Michael W. Hirlinger. 1991. "Intergovernmental service contracts: A multivariate explanation." Urban Affairs Quarterly 27 (1) September: 128-44.

Nelson, Michael A. 1990. "Decentralization of the subnational public sector: An empirical analysis of the determinants of local government structure in metropolitan areas in the U.S." Southern Economic Journal 57 (2) October: 443-57.

--. 1997. "Municipal government approaches to service delivery: An analysis from a transactions cost perspective." Economic Inquiry 35 (1) January: 82-96.

Oates, Wallace E. 1988. "On the measurement of congestion in the provision of local public goods." Journal of Urban Economics 24 (1) July: 85-94.

Pagano, Michael. 1999. "Metropolitan limits: Intermetropolitan disparities and governance in U.S. laboratories of democracy." In Governance and Opportunity in Metropolitan American, edited by the National Research Council (U.S.), Committee on Improving the Future of U.S. Cities through Improved Metropolitan Area Governance. Washington, D.C.: National Academy Press: 252-95. Prepared by Alan Altshuler et al.

Pindyck, Robert S., and Daniel L. Rubinfeld. 1981. Econometric Models and Economic Forecasts. New York: McGraw-Hill.

Shaeffer, E.C., and W.C. Bryant. 1983. Structures and Processes for Effective Collaboration among Local Schools, Colleges and Universities: A Collaborative Project of Kannapolis Schools. Charlotte: Livingstone College, University of North Carolina.

Sonenblum, Sidney, John J. Kirlin, and John C. Ries. 1977. How Cities Provide Services: An Evaluation of Alternative Delivery Structures. Cambridge, Mass.: Ballinger Publishing Company.

Statistics Canada 2002. A National Overview: Population and Dwelling Counts No. 93-360-XPB. Ottawa: Statistics Canada.

Steiner, Reto. 2003. "The causes, spread and effects of intermunicipal cooperation and municipal mergers in Switzerland." Public Management Review 5 (4) December: 551-71.

Thurmaier, Kurt, and Curtis Wood. 2002. "Interlocal agreements as overlapping social networks: Picket-fence regionalism in Metropolitan Kansas City." Public Administration Review 62 (5) September/October: 585-98.

Visser, James A. 2002. "Understanding local government cooperation in urban regions: Towards a cultural model of interlocal relations." American Review of Public Administration 32 (1) March: 40-65.

Warner, Mildred, and Robert Hebdon. 2001. "Local government restructuring: privatization and its alternatives." Journal of Policy Analysis and Management 20 (2) Spring: 315-36.

Wood, Curtis. 2006. "Scope and patterns of metropolitan governance in urban America: Probing the complexities in the Kansas City Region." American Review of Public Administration 36 (3) 1 September: 337-53.


(1) For our purposes a "shared service" is defined as a service, good or policy that is produced and/or delivered through a principal-agent arrangement or a partnership by two or more local governments. Some (e.g., Steiner 2003: 553) define local government cooperation more broadly to encompass local governments cooperating to contract with private suppliers.

(2) See James Ferris (1986), Sidney Sonenblum, John Kirlin and John Ries (1977) and George Boyne (1998a) for an overview. In contrast, David Morgan and Michael Hirlinger (1991) examine inter-governmental contracts.

(3) A significant portion of the expanding literature has sought to identify the determinants of local government cooperation using statistical analysis (e.g., Joassart-Marcelli and Musso 2005; LeRoux and Carr 2007; Warner and Hebdon 2001; and Wood 2006). There is a core of established determinants that is typically represented in one form or another. George Boyne (1998b) defined four major categories of determinants--fiscal stress; economies of scale and market structure; public preferences; and the power of public employees. Additional factors are emerging--political and bureaucratic networks and interaction (e.g., Thurmaier and Wood 2002; Visser 2002; LeRoux and Carr 2007) and transactions costs (Nelson 1997).

(4) To illustrate, David Morgan and Michael Hirlinger (1991) and James Ferris (1986) analyse only municipalities with populations exceeding 25,000, and larger municipalities (10,000 to 100,000) are greatly overrepresented in Michael Nelson (1997) and may be overrepresented in Mildred Warner and Robert Hebdon (2001). Sidney Sonenblum, John Kirlin and John Ries (1977) exclude those with less than 5,000 people. More representative sets of observations are studied by Pascale Joassart-Marcelli and Juliet Musso (2005), Kelly LeRoux and Jered Carr (2007), and Curtis Wood (2006), but population is not a variable in Wood's regression and the implications of population in the other results are not consistent.

(5) As explained in the appendix on methodology, the logit method is not ideal but it is a functional approach when, due to the limited data, the more complex preferred method does not perform adequately.

(6) County governments in Alberta are general-purpose rural municipal (district) governments, not regional authorities as in some provinces and states. That is, the counties and the urban municipalities within their boundaries are independent local governments with separate non-overlapping jurisdictions and constituents. As noted below, Strathcona County is a "specialized" municipality characterized by not having any separate urban municipalities despite the significant urban areas there.

(7) It is widely recognized that population is, or at least may be, a poor proxy for measuring output and economies of scale (e.g., Boyne 1998b). This problem has been addressed, in part, by the literature focusing on measuring publicness or congestion in local public services (e.g., Oates 1988; McMillan 1989). At this point, we demonstrate that population matters and address the implications for economies of scale further in the article, under "Discussion."

(8) Three measures of explanatory power are reported. McFadden's [R.sup.2] (adjusted for degrees of freedom) indicates that 77.1 per cent of the variation in this data is explained by own and neighbour's populations. Prediction rates are another measure. The model predicts 92.3 per cent of the observations correctly. However, the naive prediction rate from the data (the actual portion in SSAs) is 60 per cent so the normalized success rate of the model is a better indicator of its explanatory power.

(9) One would like to run a model including the variables representing all the hypotheses (so as to simultaneously control for the entire set of expected relevant factors) but the small number of observations and the limited variation here prevented that.

(10) The debt limit is 1.5 times the sum of municipal total revenues less capital transfers from the provincial and federal governments less the principle outstanding of loans made by the municipality (under Section 265 of the Municipal Government Act, R.S.A. 2000, c. M-26).

(11) Subsequent analysis (Brasington 2003) found that income and racial differences had some identifiable effect but through rather complex interactions.

(12) This response contrasts with the greater importance given to cost reduction in the responses obtained by James McDavid and Eric Clemens (1995) from local governments in B.C. This difference does not necessarily imply inconsistency. Shared services arrangements are important avenues for achieving cost reductions in both instances, but, for our respondents, applying those savings to service improvements seems more important.

(13) It is suspected that, as noted by Reto Steiner (2003), most of the cost-savings are converted to or realized as service improvements. The fact that municipalities appear to realize economies at relatively modest populations is likely the reason that recreation is observed as a local service having a low probability of being provided through shared or contract arrangements (e.g., Joassart-Marcelli and Musso 2005; Nelson 1997; Steiner 2003; Wood 2006).

(14) Note, however, that Robert Bish (1999, 2000), from studies of the region surrounding Victoria, British Columbia, observes that various inter-municipal cooperative arrangements and other forms of contracting allow many municipalities to achieve an efficient scale of service despite their small sizes. This result is facilitated by the presence of the regional district (a general purpose government with a flexible mandate that overlies the twelve area municipalities).

Edward C. LeSage Jr. is professor emeritus, Faculty of Extension, University of Alberta. Melville L. McMillan is professor, Department of Economics, University of Alberta. Neil Hepburn is lecturer, Department of Social Sciences, Augustana Faculty, University of Alberta. The authors thank the support of Alberta Capital Region Alliance (ACRA) and Alberta Municipal Affairs for making this article possible. They thank Ken Woitt, executive director of ACRA, for his cooperation. They also kindly acknowledge Junaid Jahangir and Min Li for their research assistance and express their gratitude to the editor and the two anonymous referees for their valuable guidance.
Table 1. Alberta Capital Region Alliance Municipalities and their
Populations, 2001

Beaumont 7,010
Bon Accord 1,530
Bruderheim 1,202
Calmar 1,900
Devon 4,965
Edmonton 666,105
Ft. Sask. 13,120
Gibbons 2,655
Leduc 15,030
Leduc Co. 12,530
Legal 1,060
Morinville 6,545
New Sarepta 380
Parkland 27,250
Redwater 2,170
Spruce Grove 15,980
St. Albert 53,080
Stony Plain 9,585
Strathcona 71,985
Sturgeon 18,065
Thorsby 800
Wabamun 605
Warburg 560

Note: Information is from Alberta Capital Region Alliance and
Statistics Canada. Populations are those reported from the 2001
census. Populations used in the analysis differ slightly, as they
came from the 2001 official population list from Alberta Municipal
Affairs. Bruderheim was not included in the regression analysis
because its service agreement with Strathcona County was not specific
to recreational services. While ACRA still maintains a web site
(, its members became part of the Alberta
government's Capital Region Board (see
when ACRA officially closed on 30 April 2008.

Table 2. Results for "Own" Variables (a)


Other variable Own Neighbour
Hypothesis 1
 -0.2436 (-2.352) (c) -0.0780 (-2.059)
Hypothesis 2
Equalized tax rate (d) -0.1826 (-2.073) -0.1060 (-1.644)
Debt to debt limit ratio -0.2608 (-2.731) -0.1002 (-1.455)

Hypothesis 3
Median family income -0.2490 (-2.270) -0.0915 (-1.971)
(in thousands)
Percentage owner- occupied -0.2469 (-2.048) -0.0777 (-2.031)
Percentage 65 or older -0.2507 (-2.082) -0.0774 (-2.039)
Percentage under 20 -0.2460 (-2.323) -0.0784 (-2.028)

 Adjusted (b)
 Own other McFadden
Other variable variable [R.sup.2]
Hypothesis 1
 -- 0.771
Hypothesis 2
Equalized tax rate (d) 811.47 (1.624) 0.818
Debt to debt limit ratio 18.784 (1.744) 0.847

Hypothesis 3
Median family income -0.2104 (-1.346) 0.789
(in thousands)
Percentage owner- occupied 0.0069 (0.054) 0.767
Percentage 65 or older -0.0377 (-0.126) 0.767
Percentage under 20 0.1419 (0.290) 0.768

 Prediction Normalized
 success success
Other variable rate rate
Hypothesis 1
 0.923 0.804
Hypothesis 2
Equalized tax rate (d) 0.954 0.856
Debt to debt limit ratio 0.969 0.888

Hypothesis 3
Median family income 0.938 0.822
(in thousands)
Percentage owner- occupied 0.923 0.804
Percentage 65 or older 0.923 0.805
Percentage under 20 0.923 0.805


(a) Results are for a series of regressions including own and
neighboring municipality's population (in thousands) and usually one
other variable. Results here are those for when the other variable is
for the own municipality. The values of the constant in the
regressions are not reported.

(b) McFadden [R.sup.2] is adjusted for degrees of freedom.

(c) t-values are in brackets. Critical values are 2.600, 2.000,
1.671 and 1.296 at the 1 per cent, 5 per cent, 10 per cent and
20 per cent levels of significance.

(d) Equalized residential and farm property tax rate in mills.
Property tax rates are equalized to make them comparable across
municipalities for provincial school property tax purposes.

Table 3. Probability of SSA Participation as Influenced by Own and
Neighbour's Population (a)

 Neighbour's population Neighbour's population
Ownpopulation equals own population equals 6,000

3,000 0.992 0.990
6 000 (b) 0.979 0.979
12,000 0.870 0.914
18,000 0.493 0.712
24,000 0.123 0.365


(a) Probabilities are based on the logit regression
5.7588-0.2436 x (own population) - 0.0780 x (neighbour's population)
where population is in thousands.

(b) The median population of the communities in the data is 6,226.
COPYRIGHT 2008 Institute of Public Administration of Canada
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2008 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:LeSage, Edward C., Jr.; McMillan, Melville L.; Hepburn, Neil
Publication:Canadian Public Administration
Geographic Code:1CANA
Date:Sep 1, 2008
Previous Article:Racial inequality in employment in Canada: Empirical analysis and emerging trends.
Next Article:When markets fail to deliver: an examination of the privatization and de-privatization of water and wastewater services delivery in Hamilton, Canada.

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters |