Geoforum 50 (2013) 76–87

Contents lists available at ScienceDirect

Geoforum

journal homepage: www.elsevier .com/locate /geoforum

Municipal managers’ engagement in multi-level governancearrangements: An empirical analysis grounded in relational economicgeography

0016-7185/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.geoforum.2013.08.003

⇑ Corresponding author at: University of the Basque Country (UPV/EHU), Instituteof Applied Business Economics, Avenida Lehendakari Agirre 83, 48015 Bilbao, Spain.

E-mail addresses: josemaria.barrutia@ehu.es, jose.barrutia@sant.ox.ac.uk (J.M.Barrutia), carmen.etxebarria@ehu.es, carmen.echebarria@sant.ox.ac.uk (C. Echebar-ria), patrick.hartmann@ehu.es (P. Hartmann), vanessa.apaolaza@ehu.es (V. Apaola-za-Ibáñez).

Jose M. Barrutia a,b,⇑, Carmen Echebarria a,b, Patrick Hartmann a, Vanessa Apaolaza-Ibáñez a

a University of the Basque Country (UPV/EHU), Institute of Applied Business Economics, Avenida Lehendakari Agirre 83, 48015 Bilbao, Spainb University of Oxford, European Studies Centre (St. Antony’s College), 62 Woodstock Road, Oxford OX2 6JF, UK

a r t i c l e i n f o a b s t r a c t

Article history:Received 6 December 2012Received in revised form 29 July 2013Available online 9 September 2013

Keywords:Environmental multi-level governanceRelational economic geographyPlanning toolsEngagementSupport benefitsInteraction benefits

Diverse empirical evidence regarding Local Agenda 21 (LA21) seems to indicate that, in territories wheremulti-level governance arrangements (networks) have emerged, LA21 dissemination tends to be higher.However, it is not entirely clear how and why these networks work. We take a step towards covering thisgap by adopting a relational economic geography perspective and focusing on the benefits that emergefrom the interactions and relationships between actors within networks. We focus on two crucial net-work actors: the initiator or promoter of the network (a provincial government agency) and the purpo-sive participants (the municipalities). Our research is intended to explain why the promoter is successfulin attracting participants and engaging them in intra-network interactions. Our findings suggest that spe-cific network-driven benefits have varying outcomes which could be used to guide implementationapproaches based upon the desired effect. We use data collected from 163 municipal managers whoare members of an LA21 network in Girona, Catalonia.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Municipalities have a central role in promoting sustainability,given their closeness to the causes and solutions of many of theproblems associated with this major goal. As the level of gover-nance that is closest to people, local councils play a vital role ineducating, mobilizing and responding to the public concerning is-sues that promote sustainability (Evans et al., 2006). Meanwhile,strategic planning may contribute important benefits for munici-palities and be a vehicle for inclusive public management in a dem-ocratic society (Bryson and Alston, 2011). In consequence, diverseinternational forums have emphasized the need to promote long-term participatory strategic planning processes that address localsustainability.

In particular, the countries that participated in the Rio Summit(1992) subscribed, at least nominally, to the United Nations pro-posal to promote the devising and implementation of local sustain-ability strategies, known as Local Agenda 21 (LA21). LA21 isunderstood as a municipality-led, community-wide participatory

effort to establish a comprehensive local strategic plan for tacklingenvironmental, social, economic and cultural issues that lead toquality-of-life improvement (Barrutia et al., 2007). However, theresponse of local authorities is far from generalized, and becamea major concern surrounding the 2012 Rio+20 Summit (Echebarriaet al., 2009). In consequence, research efforts addressed at indicat-ing possible paths towards a more across-the-board diffusion of lo-cally-based SD strategies are needed. This research seeks torespond to this need by analyzing the experience of a specific prov-ince in Catalonia (Spain), Girona, which has developed a network-ing experience.

Previous research has studied LA21 processes in Europe andconcluded that when LA21 is considered to be the sole responsibil-ity of municipalities, higher levels of LA21 dissemination are unli-kely to be achieved (Echebarria et al., 2009; Evans et al., 2006). Alack of sustainability- and LA21-related resources and capacitieshas been considered as an important brake on the spread of LA21(Evans et al., 2006; Echebarria et al., 2009). In spite of these diffi-culties, however, diverse empirical evidence regarding LA21 seemsto indicate that, in territories where multi-level governancearrangements have emerged, LA21 dissemination tends to be high-er. These multi-level governance arrangements have usually beenreferred to as networks in LA21 literature. We adopt this termfor practical reasons. Networks seem to constitute a launch padfor the explosion of initiatives of this kind (Barrutia et al., 2007;Eckerberg and Dahlgren, 2007).

J.M. Barrutia et al. / Geoforum 50 (2013) 76–87 77

However, it is not entirely clear how and why these networkswork. We take a step towards covering this gap by adopting a rela-tional economic geography perspective (Bathelt and Glückler,2003; Bathelt, 2009; Boggs and Rantisi, 2003; Yeung, 2005) andfocusing on the actors, their strategies and intentions, and the ben-efits that emerge from the interactions and relationships betweenactors within networks. We focus on two crucial network actors:the promoter of the network (in our empirical case a provincialgovernment agency) and the purposive participants (the munici-palities). Our research is intended to explain why the promoter issuccessful in attracting participants and engaging them in intra-network interactions.

As LA21 is a voluntary endeavor, and interactions involve costs(i.e., in terms of time spent, effort and energy), we can infer thatmunicipal managers might perceive additional benefits when net-working processes are implemented. However, the benefits emerg-ing from LA21 networks have not been clearly and systematicallyidentified. Building on multi-level environmental geography (Davisand Reed, 2013) and various related research streams (in particu-lar, LA21, network management, collaborative management, andparticipation in networks), we identify two broad types of benefits:support benefits and interaction benefits. We study the joint effectof support benefits and interaction benefits on various municipalmanagers’ perceptions of engagement.

Engagement means that municipal managers’ perceptions ofboth the LA21 and the network are positive and even enthusiastic,and are therefore integrated in the local managers’ mind-set and inmunicipal life (Bakker et al., 2008). Support benefits refer to arange of advantages that municipal managers perceive from theinvolvement of higher levels of government within the network.Higher levels of government are able to contribute various re-sources needed by municipalities in order to properly implementLA21, which include a range of complements to the focal LA21 toolsuch as financial support, training, methodologies, consultants andbi-directional communication channels with municipalities tojointly decide the support needed (Burch and Robinson, 2007).Interaction benefits refer to a realm of advantages arising from richinteraction between network members in municipality to munici-pality relations, and also at a municipality-to-promoter level.These advantages include learning, social enhancement, and thepurposive benefits primarily pursued (risk sharing and having auseful medium term plan) (Adger, 2003; Davis and Reed, 2013;Dholakia et al., 2004; Eden, 2009; Störmer, 2008).

The remainder of the paper is structured as follows. The nextsection refers to the specific context of this research. The third sec-tion deals with the conceptual background that serves as a basis forthis research. The fourth section develops the model and hypothe-ses that were quantitatively tested. The fifth and sixth sections ex-plain in detail the major concepts used and measured in thisresearch. The seventh section describes the data collection process.The eighth section deals with the model specification. The ninthsection refers to the results of the empirical test. The final sectionpresents discussion and conclusions and offers avenues for furtherresearch.

2. Research context: Girona network for the environment

The 1978 Spanish Constitution gave municipalities new respon-sibilities for the management of their environment. These powersinclude: territorial planning, water and lighting supply, waste col-lection and treatment, sewage and wastewater treatment, publicpassenger transport, environmental protection, and protection ofpublic health. These enhanced competences gave rise to certaindiscrepancies, dysfunctions, and problems with other publicauthorities, and provincial councils and Autonomous Communities

in particular, due in the main to the lack of a clear definition ofeach of their powers.

Municipal, provincial and regional authorities became depen-dent, therefore, on cooperation and joint resource mobilization.Higher levels of government are aware that many important pow-ers of relevance to sustainability are held by the municipalities.And they recognize that local councils are particularly adept atmaintaining proximity to citizens and establishing effective com-munication with them. They are also aware that no LA21 strategywould work without contributions from municipalities, whichwould ultimately have to work on designing and establishing ac-tions to improve sustainability. On the other hand, the municipal-ities are integrated within a provincial council, on which animportant part of their financing depends. In addition, significantpowers for achieving sustainability targets, such as inter-municipaltransport, are in the hands of provincial or regional councils (Bar-rutia et al., 2007).

As a consequence, LA21 has been approached in Spain throughregional and provincial networking. Regional- and provincial-levelnetworking has positioned Spain as one of the leading countries interms of LA21 dissemination in Europe (Echebarria et al., 2009).While dissemination is not necessarily equivalent to stronginvolvement and effectiveness, in Spain LA21 has brought localsustainability planning and new knowledge into play as part ofthe range of tools that may be used to assist sustainability goals.

One of the pioneering Spanish networks is the ‘Girona networkfor the environment’. Girona is a province in north-eastern Spain,in the northern part of the Autonomous Community of Catalonia.It is bordered by the provinces of Barcelona and Lleida, and byFrance and the Mediterranean Sea. This province has 221 munici-palities and 732 thousand people live in its territory. In 1999, theprovince of Girona created the Council of Local Initiatives for theEnvironment (in the Catalan language, Consell d’Inicitatives Localspel Medi Ambient, CILMA). Its objectives are to promote sustain-able development within and among local authorities, to facilitatetheir work in elaborating LA21 and to advise them on any aspectrelated to sustainability. More specifically, the main CILMA aimsare to: (1) constitute a cooperation and exchange tool for the sus-tainability of the municipalities of Girona, and for the implementa-tion of LA21; (2) promote the integration of environmental aspectsin the territorial, social and economic policies of municipalities; (3)disseminate updated information of good sustainability practices;(4) provide technical decision makers and local authorities withappropriate models, methods and techniques; (5) promote and en-hance joint actions with other networks working in the field of sus-tainability; and (6) mobilize efforts for the sake of a common goal,by promoting the inclusion of all sectors and social agents of themunicipalities in the LA21 process.

The fact that there was only a relatively scarce experience andtradition of sustainability presented an important difficulty forthe spread of LA21 processes in the province, when the networkwas set up in 1999. Yet what has been achieved is a broad dissem-ination of LA21 processes. The network has 171 members repre-senting 77.4% of the municipalities in the region. The provincialcouncil had to act as a promoter of the network.

Motivation to participate in the network is fuelled by differentbenefits in the form of funding, learning, methodologies and recog-nition, among others. The promoter and the municipalities arejointly working in various areas. For instance, they collaborate toprevent global warming by reducing municipal energy consump-tion, increasing use of renewable energy resources, making a moreappropriate use of water, implementing green purchasing policies,and promoting sustainable mobility.

The running of the network is shared by all the stakeholders.Consensus is sought by joint planning involving discussions on arange of issues and the collective ironing out of obstacles. Two

78 J.M. Barrutia et al. / Geoforum 50 (2013) 76–87

steering bodies (a General Assembly and an Executive Committee)were created within the network. The first, consisting of politicalofficials, is made up of the 171 mayors of the member municipal-ities and three representatives of the provincial government. Itsmost important duties include approval of the annual budget andthe rules for the internal running of the network. The second bodyis formed by 18 political officials: a president, three representa-tives of the provincial council and 14 representatives of the munic-ipalities. Municipalities are chosen by considering that allgeographical areas and municipality sizes had to be representedin the Executive Committee. The Executive Committee representsthe network and contracts for goods and services that the networkneeds (e.g. consulting firms).

The day-to-day activities of the network are conducted by theTechnical Secretariat which consists of a technical secretary, atechnician and a clerk. Tasks of Technical Secretariat include facil-itating and coordinating the work spaces, acting as a channel ofinternal and external communication, stimulating the exchangeof experiences, developing guidelines, criteria and technical guides,creating mechanisms for citizen participation and developing col-laboration agreements.

For technical environment-related purposes a Technical Com-mittee and six Working Committees were created, specializing inthe areas of water, coastline, energy efficiency and renewable en-ergy, infrastructure, environment, mobility, environmental qualityand waste, and environmental education. The Technical Commit-tee, formed by technical staff from the provincial governmentand the municipalities, is responsible for implementing theplanned activities. Open forums were also promoted to encouragethe participation of civil society.

3. Conceptual background

This research is inspired by and tries to contribute to relationaleconomic geography and multi-level environmental governance.As quantitative research in these areas is scarce we also draw onother research streams to develop the research model and opera-tionalize it. This section is addressed to synthesize relational eco-nomic geography and multi-level governance literatures from theperspective of this research. Other literature streams are intro-duced as they are needed in the next sections.

3.1. Relational economic geography

Traditional economic geography approaches have tended to fo-cus on regions and other spatial representations as objects of theirinquiry. Such research has neglected the role of actors who activelyshape their environment through processes of involving other ac-tors (Bathelt, 2009). As a consequence of these and other problems,relational perspectives have been developed (Bathelt and Glückler,2003; Boggs and Rantisi, 2003; Yeung, 2005). These views arebased on micro-level approaches, focusing on actors involved ineconomic and social processes which result in uneven outcomes,such as levels of agglomeration, economic specialization or devel-opment. In this relational perspective, ‘‘actors and the dynamicprocesses of change and development engendered by their rela-tions are central units of analysis’’ (Boggs and Rantisi, 2003, p.109).

Bathelt and Glückler (2003) developed a relational frameworkfor analysis which systematically focuses on economic actors andtheir action and interaction. In their view any analysis in relationaleconomic geography should be based on an understanding of theintentions and strategies of economic actors and the patterns ofhow they behave. They also stated that a relational frameworkshould emphasize the contextuality of human action. They viewed

economic action as embedded in a specific socio-institutional con-text, which creates opportunities for economic action and interac-tion that would otherwise not exist, as a process situated in timeand place.

Contingency and agency are core components of relational eco-nomic geography (Bathelt and Glückler, 2003; Yeung, 2005), whichfits Giddens (1984) conception of ‘structuration’ in which the dis-cursive consciousness and reflexivity of human agency (i.e., thecapacity of individuals to act independently and to make free deci-sions) can arguably shape the structures that both enable and con-strain their activities.

For the purposes of this research, we stress the role of endoge-nous factors in accounting for change. Whereas a great deal ofexisting literature on networks focuses on exogenous variables,we focus on the beliefs of the relevant actors (in our case the net-work promoter) about the world they live in and the dilemmasthey face. We adopt an agency perspective and focus on goal-ledmanagement that leads to the promotion of the LA21 network bya higher level of government (i.e. an agent, which makes free deci-sions). The new institutional context favored by the network is ad-dressed to motivate ongoing relations between actors and enablethem to be reproduced. However, the municipalities, as reflexiveactors, have their own goals and strategies which might not be inline with those of the leading actors and, thus, perform differentlyby deciding not to engage in the network and LA21 processes(Bathelt, 2009).

3.2. Multi-level governance

Liebler and Ferri (2004) described networks as: (1) organic (i.e.created as a natural result of the interaction between members) or(2) engineered (i.e. fostered by a promoter for a specific purpose).We study an engineered or promoted network. The term ‘‘engi-neered’’ should not be understood as incorporating a hierarchicalconnotation. The adoption of the LA21 cannot be hierarchically im-posed. The promoter has greater access to human, financial andlegislative resources and this advantage lends it special weightwithin the network. However, as explained in Section 2, the pro-moter and the remaining actors maintain mutual dependence rela-tionships. The promoter is viewed as a social marketer (instead ofas a hierarch) that pursues a social goal and achieves it by buildingan appropriate environment in which positive attitudes towardsthe new tool are generated. Although power is not uniformly dis-tributed and the provincial government may have the ultimateauthority to make a decision, the goal is to achieve some degreeof consensus among stakeholders (i.e., network management isconsensus-oriented). Ansell and Gash (2008) use the term consen-sus-oriented as collaborative forums may not succeed in reachingconsensus. ‘‘However, the premise of meeting together in a delib-erative, multilateral, and formal forum is to strive toward consen-sus or, at least, to strive to discover areas of agreement’’ (p. 547).

Therefore, the provincial government agency promotes a multi-scale and multi-actor form of governance that scholars have vari-ously identified as metagovernance and multi-level governance.As noted by Davis and Reed (2013), metagovernance research ismostly concerned with experiences characterized by higher levelsof government which devolve responsibility without offering newpowers. It focuses on practices and procedures that secure govern-mental influence, command and control within governance re-gimes (Whitehead, 2003). Therefore, metagovernance has beendefined as the ‘government of governance’ (Bell and Park, 2006,p. 63). In the metagovernance of networks, the government re-serves the ability to set the agenda, determine which stakeholdersare involved, allocate resources (including information) to the net-work, and decide on how much power or authority is to be sharedwith partners in the network (Bell and Park, 2006). Multi-level

J.M. Barrutia et al. / Geoforum 50 (2013) 76–87 79

governance, however, focuses on experiences involving the effec-tive decentralization of power. The term refers to negotiated,non-hierarchical relationships between institutions at differentlevels (transnational, national, regional and local) based on inter-dependency (Davis and Reed, 2013).

While we believe that the experience studied fits the multi-le-vel governance concept, as decision-making in relation to the strat-egies and activities undertaken by the network is shared betweenthe provincial government and the municipalities, this researchdoes not focus on power relations between network partners (forsuch a perspective, see Faulconbridge and Hall, 2009). And, forpractical reasons, we will use the general term network for refer-ring to the experience studied.

4. Model and hypotheses

Our research problem has to do with the adoption of a newpractice/innovation by municipalities. LA21 is viewed by munici-pal managers as an innovative and complex practice (Evans et al.,2006; Eckerberg and Dahlgren, 2007). Overall, the framework wepropose is consistent with previous research regarding innovationdiffusion, which has identified three common clusters of factorsthat affect innovation adoption: (1) the characteristics of the inno-vation itself, (2) the internal characteristics of the organizationmaking the adoption decision or internal factors, and (3) the envi-ronment in which the adoption occurs, or external factors (Jun andWeare, 2010). While we focus on the effect of external factors fa-vored by the network promoter, we also control for the influenceof the internal characteristics of the municipalities, and conductan in-depth review of LA21 literature to understand how the char-acteristics of the innovation itself affect engagement (Echebarriaet al., 2009; Eckerberg and Dahlgren, 2007; Evans et al., 2006).

An important consideration here is that most previous innova-tion adoption research has approached adoption in terms of a yes/no question. However, public organizations may innovate in searchof legitimacy and not fully engage in innovation. Because weak dif-fusion is insufficient to answer the complexity associated with sus-tainability, we need to measure engagement. Engagement permitsus to expect that the municipalities will take full advantage of thenew tool, use it properly and have a relevant influence on munici-pal practices and approaches in a core sense.

A characteristic of this research is that LA21 adoption takesplace in a networking setting. The network is viewed by the pro-moter as a way of enhancing the perception of benefits by munic-ipalities in relation to the new tool. Therefore, we needed to reviewa broad body of literature addressed to explain the reasons/bene-fits that lead people to participate in networks, such as learningand social enhancement (Dholakia et al., 2004; Störmer, 2008).These benefits can be propitiated by the promoters (Bryson et al.,2006; Gage and Mandell, 1990). For instance, Bryson et al. (2006)emphasize the prominence of wise design and use of forums, are-nas and courts to promote successful collaboration.

On the basis of the above literature, we believe the promotercould develop two major functions: (1) ensuring the provision ofsupport (i.e., it should provide economic and technical supportand guidance, obtain additional support from other higher levelsof government, and establish bi-directional communication chan-nels with municipalities to jointly define the needed support)(Evans et al., 2006; Burch and Robinson, 2007; Eckerberg and Dahl-gren, 2007); (2) encouraging and facilitating a rich interactionenvironment in which important benefits emerge for membersthrough the achievement of the purposive benefits primarily pur-sued, as well as other important benefits such as learning and so-cial enhancement (Adger, 2003; Dholakia et al., 2004). While thenetwork label is broadly used by researchers and practitioners,many experiences labeled as networks fail to provide appropriate

support and stimulate rich interaction environments, which oughtto affect network engagement.

The model to be tested is depicted in Fig. 1. It explains the ante-cedents of five different outcomes that we group within the con-cept of engagement. Two of them, identification and trust, arelinked to the network itself and the concept of social capital, andthe remaining three outcomes, value, satisfaction and behavioralintentions, are more linked to the specific tool that is being pro-moted (i.e., LA21). The main explanatory variables are support ben-efits, which have been stressed by LA21 literature, and interactionbenefits, which have mostly been emphasized by network manage-ment and participation in network research streams.

The major assumptions that underlie the model are based onthe rationality of municipal managers:

Hypothesis 1. The higher the perceived support benefits, thegreater the engagement-related perceptions of municipalmanagers.

Hypothesis 2. The higher the perceived interaction benefits, thegreater the engagement-related perceptions of municipalmanagers.

The positive relationship between support benefits, interactionbenefits and engagement, assumed in hypotheses 1 and 2, is ex-plained in depth in Section 5.

The control variables include interaction density and variableslinked to the municipality profile that should have an effect onengagement (Echebarria et al., 2009; Eckerberg and Dahlgren,2007). The expected relationship between the control variablesand engagement is explained in depth in Section 6.

5. Conceptualization of engagement, support benefits andinteraction benefits

5.1. Engagement

We argue that managers should pursue the engagement of mu-nicipal managers in relation to the good practice that is being pro-moted (and to the network itself). Our concept of engagement isbuilt on prior literature (Bakker et al., 2008). Engagement meansthat the LA21 and the network are enthusiastically integrated inthe local managers’ mind-set and in municipal life. A conceptualframework has not been proposed to capture and systematizethe concept of engagement in networks addressed to disseminatea new practice. However, management literature has proposed dif-ferent outcomes that are intended to measure the assessments,attitudes and intentions of users in relation to products, services,communities and practices (Chiu et al., 2006; Dholakia et al.,2004; Dwyer et al., 1987; Lam et al., 2004). Building on this litera-ture we have identified five outcomes related to the concept ofengagement: identification, trust, value, satisfaction and behav-ioral intentions. While the two former variables are linked to theconcept of social capital, the latter three are associated with themore specific LA21 tool.

5.1.1. Attitudes towards the network and network members: Socialcapital

Nahapiet and Ghoshal (1998) define social capital as ‘‘the sumof the actual and potential resources embedded within, availablethrough, and derived from the network of relationships possessedby an individual or social unit’’ (p. 243). Several research worksempirically show that social capital fosters knowledge transfer,innovation outcomes, and growth (Chiu et al., 2006). Liebler andFerri (2004) studied a wide range of sustainability-related net-working experiences and stressed the importance of social capital

Fig. 1. Model.

80 J.M. Barrutia et al. / Geoforum 50 (2013) 76–87

in the effective functioning of networks. They argued that evennetworks that are initiated by external forces (i.e., engineered net-works) have the potential to generate high levels of social capital,when a great deal of care is taken to foster and develop social cap-ital. Network managers should therefore pursue high levels of socialcapital, since this should positively affect network performance.

Social capital exists when members have a strong identificationwith the collective and trust others within it (Chiu et al., 2006). Theconcept of identification refers to the feeling of attachment to thenetwork by its members. Social identity theorists posit that identi-fication with social groups is derived, first and foremost, from theirfunctionality (Hogg and Abrams, 1988) (i.e., individuals identifywith groups to the extent that groups fulfill important needs ofthe individuals concerned). In turn, identification reinforces func-tionality. Given that valuable knowledge is embedded in individu-als and people usually tend to hoard knowledge, individuals wouldnot contribute their knowledge unless another individual were rec-ognized as a group-mate and the contribution were conducive tothe first person’s welfare (Chiu et al., 2006; Nahapiet and Ghoshal,1998). Trust is viewed in management literature as a set of specificbeliefs dealing primarily with the integrity, benevolence, and abil-ity of another party (Mayer et al., 1995).

5.1.2. Attitudes and intentions towards the LA21-toolUnder this category we include outcome variables referring to

municipal managers’ attitudes and intentions towards the specificLA21 tool. An attitude is a favorable or unfavorable evaluation ofsomething. Attitudes are expected to change as a function of experi-ence. In particular, two different and important attitude-relatedconcepts are considered: value and satisfaction. Value can be con-ceptualized as a weighted comparison between what is ‘obtained’and what is ‘given’ (Lam et al., 2004). It is possible to understand sat-isfaction as a positive affective state resulting from evaluation of allthe aspects associated with a particular activity or relation (Lamet al., 2004). Behavioral intentions refer to purposes in relation tofuture behaviors, such as repeating or intensifying an activity(Dwyer et al., 1987). Value, satisfaction and behavioral intentionsmay be respectively viewed as the cognitive–affective, affective–cognitive and behavioral components of attitude towards LA21.

5.2. Support benefits

Initially, LA21 is viewed as a complex tool and generates astrong sense of insecurity and uncertainty in local authorities

(Eckerberg and Dahlgren, 2007; Evans et al., 2006). Most munici-palities are small and may not be able to call on the human andknowledge resources required. They are also afraid of participationfrom civil society (Coenen, 2009). They wonder what would hap-pen, for instance, if they were not able to meet the commitmentsundertaken. As a consequence, the perception of benefits has tobe reinforced in the municipalities, and becomes a crucial factorin policy success.

On the basis of previous research regarding LA21 processes(Echebarria et al., 2009) we focus on four elements of the manage-ment approach of network promoters that have a relevant effect onnetwork success: comprehensiveness, promoter–member relation-ship, complements to the focal LA21 tool, and means-end integra-tion. Perception of support benefits should be associated withfavorable perceptions in relation to the outcome variables.

Comprehensiveness is defined as the extent to which relevant ac-tors with resources and capacities for achieving sustainability areworking in coordination to foster LA21 (Echebarria et al., 2009).As an optimum, each partner should provide, in accordance withits abilities, one or more of the ingredients vital to the successfulfunctioning of the network. In our case, most relevant partners in-clude provincial, regional and state governments, and internationalbodies (such as the International Council on Local EnvironmentalInitiatives, ICLEI).

Promoter–member relationship: We propose that municipalmanagers’ perceptions of the network will depend on the existenceof fluid relationships, on their appreciation of the prestige of thepromoter/s, and the level of interaction between the promoter/sand the local manager, as shown by LA21 experiences in Europe(Eckerberg and Dahlgren, 2007), and other studies on networking(Gage and Mandell, 1990).

Complements to the focal LA21 tool take in financial support,training and human resources provision and so forth. Collaborativegovernance literature emphasizes the importance of complements(support) to induce collaboration (Ansell and Gash, 2008). Thus,O’Leary and Bingham (2009) found that one of the most importantfactors for participation in collaborative experiences is resourcesavailability (especially of an informational kind). This conclusionis consistent with studies of LA21 implementation in Europe,which point towards the need for support from the higher levelsof government (Kern et al., 2007).

Integration is understood as the extent to which the means(lines of support) are properly designed to cover LA21 ends

J.M. Barrutia et al. / Geoforum 50 (2013) 76–87 81

(Echebarria et al., 2009). Some complements can sometimes goagainst the objective of LA21 dissemination processes, even thoughtheir producers maintain this is not their aim. This happens be-cause the network promoters do not take municipal opinion suffi-ciently into account when they define value complements, orbecause they try to fulfill several objectives (Eckerberg and Dahl-gren, 2007). In the context of European LA21 experience somecases have been reported in which support from central and sub-State governments coexists, but this cohabitation occurs withinan atmosphere of competition instead of collaboration, leading tooverlapping and opposite effects to those intended (Eckerbergand Dahlgren, 2007).

5.3. Interaction benefits

When properly managed by the network promoter (Brysonet al., 2006) interaction should provide several benefits for munici-pal managers. We built on previous literature covering participa-tion in networks and teams and identified three major categoriesof networking benefits, which should emerge from appropriatelyconducted LA21 networks: (1) purposive benefits, (2) learning,and (3) social enhancement. Perception of interaction benefitsshould be associated with favorable perceptions in relation to theoutcome variables.

Within the category of purposive benefits, we include those thatare defined prior to network creation and facilitate achievement ofspecific LA21-related end-state goals (Dholakia et al., 2004). Onemajor benefit pursued by the LA21 movement is to provide munic-ipalities with a sustainability-led plan developed from a detailedmunicipal diagnostic which should guide the everyday actions ofthe municipalities. The benefits of strategic planning for publicorganizations have been emphasized by prior literature. Brysonand Alston (2011) suggest that strategic planning allows municipalmanagers to make today’s decisions in the light of their future con-sequences, to think strategically, clarify future direction, and de-velop a coherent and defensible basis for decision-making. Inaddition, municipalities may improve their efficacy and effective-ness. A second purposive benefit of networking is risk sharing(Jarillo, 1988). LA21-related methodologies and routines adaptedto the specificities of the regional territory are jointly developedand tested. Finally, municipalities should achieve higher levels ofeffective citizen participation, since it is inherent to the LA21 con-cept, and this could lead to better decisions (Coenen, 2009).

Learning benefits refer to LA21-related learning, that is, a betterunderstanding and knowledge of the design of LA21, its underlyingprocesses, and its implementation. The LA21 network holds valu-able collective knowledge of LA21 and its implementation that isgenerated and shared through continued member interactions(Dholakia et al., 2004; Störmer, 2008; Taylor and de Loë, 2012).Learning benefits have been emphasized as major benefits derivedfrom inter-organizational networks (Jarillo, 1988). Learning is par-ticularly important in the context of LA21. Charter 28 of Agenda 21,devoted to LA21, gives little guidance on how local communitiesshould proceed with a LA21 process. Although ICLEI developed ageneral LA21 methodology, each community has to find its ownappropriate way of matching its specific cultural, economic, social,geographic, and demographic characteristics (Coenen, 2009). Acomplex mix of knowledge is needed to implement LA21, whichmust include methodology for designing strategies, awareness ofthe content of environmental, social and economic departmentsand their interrelations, and of the functioning of public institu-tions and their interrelations, as well as the creation and runningof forums that encourage participation from civil society (Echebar-ria et al., 2009).

Social enhancement benefits stem from gains in reputation orstatus and the achievement of a sense of self-efficacy (Dholakia

et al., 2004). Networks serve as a venue for participants to exhibittheir LA21-related knowledge and problem-solving skills. By con-tributing to LA21 support, municipalities and their managers canenhance their expertise-related status and reputation among peermunicipalities as well as in the eyes of the network promoter(Dholakia et al., 2004). Through their contributions, municipalmanagers influence the behavior of their peers towards LA21 aswell as the improvement plans of LA21 promoters.

6. Conceptualizing municipal characteristics and interactiondensity (control variables)

Control variables include interaction density, and six character-istics that have a relevant effect on municipal attitude towardsLA21: risk aversion, sustainability tradition in the municipality,presence of key individuals in the municipality, attitude towardscitizen participation, and municipal size. Other economic, fiscaland political variables were considered for inclusion as controlvariables. Finally, they were not included for various reasons, suchas being relatively homogeneous in our research context.

6.1. Interaction density

Interaction density refers to the quantity and length of contactsbetween municipalities, which are supposed to work together inmany areas. Overall, previous empirical studies provide evidencethat tends to support the positive effect of interaction density onnetwork performance (Meier and O’Toole, 2001).refers to thequantity and length of contacts between municipalities, whichare supposed to work together in many areas. Overall, previousempirical studies provide evidence that tends to support the posi-tive effect of interaction density on network performance (Meierand O’Toole, 2001).

6.2. Risk aversion

The adoption of LA21 is a risky endeavor. A strong awareness ofcosts and risks, such as the risk of promising actions to citizens thatmight not be deliverable, or the existence of a winner–loser sce-nario at local level, has been detected in countries where munici-palities have very different profiles (Kern et al., 2007; Eckerbergand Dahlgren, 2007). Risk aversion has usually been considered abarrier to innovation adoption and, therefore, should be negativelyassociated with engagement (Jun and Weare, 2010).

6.3. Sustainability tradition

Sustainability tradition constitutes a precursor for LA21 pro-cesses. Eckerberg and Dahlgren (2007), for instance, with regardto Sweden, refer to a wide range of experiences developed in the1960s and 1970s that might appear crucial for explaining theadoption of LA21 in this country. Collaborative governance litera-ture also stresses the relevance of previous knowledge as a capac-ity needed for joint action (Ansell and Gash, 2008). Therefore,sustainability tradition should be positively associated withengagement.

6.4. Key individuals

Several authors have highlighted the presence of ‘‘key individu-als’’ who can act as key LA21 agents in municipalities. The Euro-pean research project DISCUS (Evans et al., 2006), in particular,shows that numerous cases can be found where mayors or otheragents endowed with sufficient charisma and commitment haveacted as drivers for the promotion of LA21s, and have even adopted

82 J.M. Barrutia et al. / Geoforum 50 (2013) 76–87

unpopular decisions, on frequent occasions, in order to prioritizelong-term SD targets. Therefore, the presence of key individualsshould be positively associated with engagement.

6.5. Attitude towards citizen participation

One differentiating component of LA21 is citizen participation.However, effective citizen participation requires a favorable atti-tude by municipalities. As Coenen (2009), among others, states,attitude towards citizen participation may be one of the main fac-tors that drive or put a brake on its adoption. Municipalities areclosest to the citizen but have not always incorporated citizen par-ticipation within their management style (Pattie and Johnston,2013). Therefore, a favorable attitude towards citizen participationshould be positively associated with engagement.

6.6. Municipality size

The availability of resources and also of certain capacities isgenerally linked to the size of each municipality. This is, therefore,a factor repeatedly highlighted by LA21 literature (Kern et al.,2007). Case-study supported evidence shows that one importantstep towards achieving a broader diffusion of LA21 is to providesmall municipalities with sustainability experts (Kern et al.,2007). Overall size should be positively associated withengagement.

7. Data collection and measures

The approach to data collection consisted of two stages. First,we reached to an agreement with CILMA. The incentive for theGirona network managers was the promise of providing them withthe sample statistics obtained from the survey. We guaranteed theconfidentiality of the individual responses. In return, we obtained alist of the individuals responsible for LA21 in the 171 municipali-ties that were members of the network. This list constituted oursampling frame. Network managers encouraged municipal author-ities to participate in the study.

Secondly, surveys were administered by telephone. Prior to thecomplete development of the data collection stage the surveyquestionnaire was subjected to a pilot test using a random sampleof 11 network members. The pilot test subjects provided descrip-tive comments on the survey (e.g., on the ambiguity of itemdescriptions) that were used to further refine the item wordings.Finally, the full process was conducted and 163 (out of 171) usablequestionnaires were obtained. Our sample can be considered ashighly representative of the region.

The measures for the study constructs were mostly adaptedfrom existing scales (to fit the study context). This procedure pro-vides some confidence in relation to the reliability and validity ofthe scales. Likert-type scales with scores between 0 (completelydisagree) and 10 (completely agree) were used. Due to space lim-itations, the measures used and the sources for all the study con-structs, as well as the descriptive statistics, are summarized inTables 1 and 2. Altogether 48 items were used to capture the per-ceptions of municipal managers. For instance, as shown in Table 1,trust was measured by adapting three items from Chiu et al. (2006)referring to the extent to which people in the network keep theirpromises, behave consistently and are trustworthy. Likewise,behavioral intentions were measured by using two items fromDwyer et al. (1987) referring to the intentions of municipal manag-ers to keep working with LA21 and intensify activities in the future.

As we adapted items from existing scales, most first-orderfactors could be defined as expected. Eleven factors used in this re-search showed high-levels of convergent validity, with Cronbach’s

alpha ranging from .771 to .942. As an exception, the Cronbach’s al-pha for the control variable risk aversion is only marginally accept-able (.583). However, we decided to maintain it due to its relevanceand singularity.

The second-order factors of support benefits and interactionbenefits also showed high-levels of convergent validity (with Cron-bach’s alpha of .838 and .934 respectively; see Table 3). In addition,we generated second-order constructs to measure the two engage-ment variables considered: attitudes towards the network (socialcapital) and attitudes towards the LA21-tool. These constructsshowed appropriate levels of convergent validity (with Cronbach’salpha of .733 and .908 respectively; see Table 3).

8. Model specification

We used ordinary least squares (OLS) to test the model. Wemodeled variables reflecting support benefits (SupBen) and inter-action benefits (IntBen) as determinants of the outcome variables(OutVar). Due to the lack of insights from previous literature, wedid not develop hypotheses in relation to possible quadratic andinteraction effects of the variables for support benefits and interac-tion benefits on engagement-related outcomes. We did, however,control them in an exploratory way. In addition we controlled forpossible interaction effects between these variables and the vari-able for interaction density. An interaction between the three vari-ables was also considered. Quadratic (QuadTerm) and interaction(IntTerm) terms were therefore included in the model specification.

Our model specification was as follows:

OutVari ¼ a1 þ b1ðSupBeniÞ þ b2ðIntBeniÞ þ b3ðConVariÞþ b4ðQuadTermiÞ þ b5ðIntTermiÞei ð1Þ

whereOutcome variables (OutVar) include attitude towards the net-work/social capital (SocCap), and attitude towards the LA21 tool(AtLA21). Social capital (SocCap) was calculated as a summatedscale that included the variables: identification (Iden) and trust(Trust). Attitude towards the LA21 tool (AtLA21) was calculatedas a summated scale that included the variables: value (Val), sat-isfaction (Sat), and behavioral intentions (BehInt).Support benefits (SupBen) were calculated as a summated scalethat included the variables: complements (Compl), means-endintegration (MEInt), promoter–member relationships (PMRel)and comprehensiveness (Compreh). Likewise, interaction bene-fits (IntBen) were calculated as a summated scale that includedthe variables: purposive benefits (PurpBen), learning (Lear), andsocial enhancement (SocEnh).Control variables include: interaction density (IntDen), riskaversion (Risk), key individuals (KeyInd), sustainability tradi-tion (SusTrad), attitude towards citizen participation (CitPart),and municipality size (MunSize).

Three quadratic terms were tested to explore the curvilinearityof the variables for support benefits (SupBen), interaction benefits(IntBen) and interaction density (IntDen). Likewise, to explore pos-sible interactions between the latter variables, four interactionterms were considered.

9. Finding

Stata 11 statistical software was used to analyze the model. Apreliminary step was to analyze the descriptive statistics. The re-sults are reported in Tables 1–3. Overall first-order factors relatedto interaction benefits offered higher scores (average score = 6.69)than those related to support benefits (5.37). In particular, the

Table 1Analysis of unidimensionality, convergent validity and reliability: constructs measuring outcome variables and support benefits.

Factor/item Source (adapted from...) Mean Std. Dev. Alphaa

Identification Chiu et al. (2006) 6.26 2.33 .917Close to network people 5.86 2.61 .904Positive feeling towards the network 6.67 2.40 .877Sharing many points of view 6.25 2.53 .861

Trust Chiu et al. (2006) 5.46 2.28 .94Keep their promises 5.34 2.37Members behave consistently 5.59 2.32Members are trustworthy 6.27 2.17 Excl.

Value Lam et al. (2004) 6.32 2.44 .918More benefits than costs 6.09 2.52LA21 worth the trouble 6.55 2.55

Satisfaction Lam et al. (2004) 5.85 2.41 .946Very satisfied 5.95 2.48Meet expectations 5.75 2.46

Behavioral intentions Dwyer et al. (1987) 6.97 2.59 .928Keep working with LA21 6.97 2.59We will intensify 6.24 2.77

Complements Echebarria et al. (2009) 4.96 2.50 .942A lot of support at present 4.51 2.57 .935Easily accessible support 4.98 2.67 .896Quality support 5.37 2.69 .915

Integration (means-ends) Echebarria et al. (2009) 5.57 2.17 .908Well-targeted means 5.79 2.41 .863General means-ends adequacy 5.82 2.37 .838Joint means-ends consideration 5.51 2.47 .902Adequate support lines 5.16 2.62 Excl.

Promoters–members relationship Echebarria et al. (2009) 5.64 2.32 .927Fluid relationships 5.64 2.72 .911Prestigious promoter 6.17 2.37 .902Reference promoter 5.90 2.35 .914We collaborate with promoters 5.80 2.76 .901We contribute ideas 4.69 3.00 .922

Comprehensiveness (actors) Echebarria et al. (2009) 5.06 2.13 .878Promoters’ work is coordinated 5.30 2.44 .847No-one missing 5.33 2.62 .882All relevant higher levels of government are supporting 4.72 2.41 .825Promoters collaborate 4.87 2.50 .817

Scale reliability without the item (regular); Std. Dev. = Standard Deviation; n = 163.a Cronbach’s alpha = Scale reliability coefficient (bold).

J.M. Barrutia et al. / Geoforum 50 (2013) 76–87 83

purposive benefits primarily pursued achieved the highest scores(7.22). Interestingly, the item ‘we get citizen participation’achieved the lowest score between all the items addressed to mea-sure interaction benefits (5.75), which is consistent with previousresearch that indicates that LA21 processes are, in practical terms,weak in the area of effective citizen participation. However, atti-tude towards citizen participation achieved a relatively high score(7.28). These results seem to suggest that, according to municipalmanagers, deficiencies in terms of citizen participation are moreexplained by the difficulties in achieving effective citizen participa-tion than by the unwillingness of municipalities.

Outcome variables offered intermediate-high scores. Amongthem, behavioral intentions received the highest score among theoutcome variables (6.97), while trust scored the lowest (6.10).One surprising result was the low score that municipal managersattached to interaction density (3.75). As interaction benefits of-fered relatively high scores, we interpreted this finding in the sensethat interaction quality might be relatively higher than interactiondensity, which was consistent with our qualitative evidence. Thehigh standard deviation of this variable (2.88) should also be no-ticed, suggesting that interaction density is more dependent onmunicipalities and municipal manager characteristics than on pro-moter activities.

Likewise, the correlations between factors were checked (seeTable 4). Interestingly, the outcome variables are positively and

significantly correlated with all explanatory variables. As anexception, the risk variable only shows a significant, but relativelylow, correlation with the outcome variable trust.

Prior to testing the model we checked for normality. TheShapiro–Francia test showed that several variables seemed to benon-normally distributed. Consequently, we used robust standarderrors to test the model.

Although we have a highly representative sample (virtually acensus), we needed to control for the potential effect of a relativelysmall sample in our results. We measured the probability of find-ing a significant effect if it does exist in the population (i.e. power).A-priori statistical power was .882 and post hoc (observed) statis-tical power was 1, which is indicative that sample size could not bea concern (effect size = .15).

Table 5 summarizes the OLS estimation results. As shown in Ta-ble 5, the variables for support benefits (H1; b = .236; p < .01),interaction benefits (H2; b = .597; p < .01), and attitude towardscitizen participation (b = .146; p < .05) have a positive effect onperceived social capital (R-sq. = 72.73%).

The variables for support benefits (H1; b = .449; p < .01), inter-action benefits (H2; b = .365; p < .01), perceived risk (b = �.119;p < .05), attitude towards citizen participation (b = �.082; p < .10),and municipality size (b = �.210; p < .01) have a significant effecton the attitudes of municipal managers towards the LA21 tool(R-sq = 75.26%). Interestingly, as shown in Fig. 2, the effect of sup-

Table 2Analysis of unidimensionality, convergent validity and reliability: constructs measuring interaction benefits and municipality profile.

Factor/item Source (adapted from...) Mean Std. Dev. Alphaa

Purposive benefits Dholakia et al. (2004) 7.22 2.04 .841A medium term plan is very useful 7.37 2.21We get citizen participation 5.75 2.37 ExcludedBy working together we reduce risks 7.08 2.19

Learning Chiu et al., 2006 6.13 2.23 .941We get important information 6.11 2.30 .926We learn 6.29 2.35 .877We resolve problems 5.98 2.42 .940

Reputation Dholakia et al. (2004) 6.89 2.09 .963Reinforces our reputation 6.85 2.29Reinforces our credibility 6.61 2.30

Interaction density Echebarria et al., 2009 3.75 2.88 .940Frequent contacts 4.17 3.10 ExcludedPhone conversations, emails, etc. 3.84 3.14 .930Meetings 3.46 2.93 .889Long conversations 3.95 3.06 .917

Risk aversion Echebarria et al. (2009) 5.89 2.19 .583We prefer ‘wait and see’ 5.10 2.87Precaution regarding new tools 6.68 2.31

Key individuals Evans et al. (2006) 5.79 2.58 .916Influential individuals 5.98 2.65Reference models 5.60 2.72

Tradition Echebarria et al. (2009) 6.22 2.36 .838We are an SD benchmark 6.19 2.61Long SD tradition 6.24 2.49

Citizen participation Echebarria et al. (2009) 7.28 2.04 .771Advantages > Disadvantages 6.73 2.44Worth the trouble 7.84 2.06

Scale reliability without the item (regular); Std. Dev. = Standard Deviation; n = 163.a Cronbach’s alpha = Scale reliability coefficient (bold).

Table 3Analysis of unidimensionality, convergent validity and reliability: second order constructs measuring support benefits and interaction benefits.

Factor/item Source (adapted from...) Mean Std. Dev. Alphaa

Second-order predictorsSupport quality See Tables 1 and 2 5.37 1.88 .838Complements 4.96 2.50 0.816Means-end integration 5.71 2.22 0.815Promoters–members relationship 5.64 2.32 0.773Comprehensiveness 5.06 2.13 0.816

Interaction quality See Tables 1 and 2 6.69 1.89 .834Purposive benefits 7.22 2.04 0.79Learning 6.13 2.23 0.77Reputation 6.73 2.25 0.74

Second-order engagement-related perceptionsSocial capital See Tables 1 and 2 5.82 2.05 .733Identification 6.26 2.33Trust 5.46 2.28

LA21-related perceptions See Tables 1 and 2 6.29 2.23 .908Value 6.32 2.44 86.60Satisfaction 5.85 2.41 86.24Behavioral intentions 6.97 2.59 87.64

Scale reliability without the item (regular); Std. Dev. = Standard Deviation; n = 163.a Cronbach’s alpha = Scale reliability coefficient (bold).

84 J.M. Barrutia et al. / Geoforum 50 (2013) 76–87

port benefits depends on the level of interaction benefits and viceversa. Additional levels of both variables have a stronger effectwhen the level of the other variable is lower, as shown by the neg-ative sign of the interaction term (b = �.336; p < .01).

Interaction density does not have a significant direct effect onthe attitudes of municipal managers towards the LA21 tool, whichmay be considered as surprising. However, it (positively) moder-ates the effect of interaction benefits on behavioral intentions(b = .273; p < .01). As prior research does not explicitly considerthe effect of interaction benefits, we interpret this result in the

sense that the effect of interaction density on outcomes mightnot be as important when interaction benefits are considered.

Finally, we tested for a possible mediator effect of social capitalon attitudes towards the LA21 tool (i.e., a possible relationship be-tween the two outcome variables considered), as this kind of rela-tionship has been found in previous studies. We followed theprocedure outlined by Baron and Kenny (1986), and the conditionsfor mediation were not fulfilled. We concluded, therefore, that bothoutcome variables are better treated as non-directly related. Socialcapital might possibly act as a mediated variable if the dependent

Table 4First-order factors correlations.

(Iden) (Trust) (Val) (Sat) (BehInt) (Compl) (PMRel) (VertInt) (Compr) (PurpB) (Lear) (SocEnh) (IntDen) (Risk) (KeyInd) (Trad)

(Iden) 1(Trust) .5449* 1(Val) .6251* .5094* 1(Sat) .5815* .4627* .7799* 1(BehInt) .5909* .4866* .7581* .7637* 1(Compl) .2965* .3656* .4111* .6066* .4512* 1(PMRel) .5188* .6077* .4929* .6331* .5284* .5526* 1(VertInt) .4848* .5231* .5528* .6710* .5333* .7126* .6343* 1(Compr) .5172* .5715* .6435* .7048* .6176* .4801* .5653* .5901* 1(PurpBen) .6721* .4572* .4793* .4836* .5540* .2457* .4166* .3680* .3980* 1(Lear) .6954* .6827* .5663* .5928* .6293* .4919* .6104* .5566* .5277* .5996* 1(SocEnh) .7417* .5838* .6130* .5905* .6554* .3220* .5601* .5821* .5340* .6998* .6925* 1(IntDen) .4355* .3629* .4074* .4883* .3829* .5179* .4707* .5376* .3990* .3196* .5081* .3325* 1(Risk) �.0201 .1798* .0039 .0791 .0123 .1037 .2459* .1436 .2717* .0035 .071 .1098 .1407 1(KeyInd) .4847* .4656* .3707* .4647* .4686* .3624* .4762* .5547* .3725* .4578* .5899* .5309* .4500* �.0753 1(Trad) .2616* .3186* .2718* .3729* .3225* .3632* .3284* .4736* .4217* .2066* .2882* .3346* .3745* .1531 .3532* 1(CitPar) .4065* .4161* .3418* .2846* .3213* .0441 .2857* .1786* .2610* .4113* .2989* .4027* .0874 .0451 .1977* .2942*

Identification = (Iden); trust = (Trust); value = (Val); satisfaction = (Sat); behavioral intentions = (BehInt); complements (Compl); promoter–member relationship = (PMRel);comprehensiveness = (Compr); purposive benefits = (PurpBen); learning = (Lear); social enhancement (SocEnh); interaction density = (IntDen); risk = (Risk); key individu-als = (KeyInd); tradition = (Trad); attitude towards citizen participation = (CitPart).* Significant at 5% level.

Table 5Determinants of engagement-related perceptions.

Social capital (identification/trust) LA21 tool-related perceptions (value, satisfaction and behavioral intentions)

Coef. P > t Coef. P > t

Main effectsSupport benefits (SupBen) .236*** .003 .449*** .000Interaction benefits (IntBen) .597*** .000 .365*** .000

Control variables: Interaction density and municipalities profileInteraction density (IntDen) �.027 .676Risk �.119** .042Key IndividualsTraditionCitizen participation .146** .016 .082* .087Municipality size �.210*** .000Constant 4.80*** .000 6.62*** .000

Interaction, moderator and curvilinear effects(SupBen) * (IntDen)(SupBen) * (IntBen) �.336*** .002(IntDen) * (IntBen) .273*** .006(SupBen) * (IntBen) * (IntDen)(SupBen)-Squared(IntDen)-Squared(IntBen)-Squared

R-Squared .7273 .000 .7526 .000

Robust standard errors. Beta coefficients.* Significant at the 10%level.** Significant at the 5% level.*** Significant at the 1% level.

J.M. Barrutia et al. / Geoforum 50 (2013) 76–87 85

variable to be explained were that for the final network outcomes(i.e., environmental achievements, number of innovations, and soon). Unfortunately these data are not available and are not consid-ered in this research.

In short, our data do not reject hypotheses 1 (positive effect ofsupport benefits on engagement) and 2 (positive effect of interac-tion benefits on engagement). Interestingly, interaction benefitshave a stronger effect on social capital, while support benefits havemore impact on attitudes towards the LA21 tool.

10. Discussion and conclusions

Research is needed on the governance framework to be imple-mented, in order for local governments to successfully contribute

to global sustainability by implementing sustainable planningtools. We respond to this need by analyzing the experience of aspecific province in Europe, Girona, which has developed a suc-cessful experience, at least if we use LA21 dissemination in the re-gion as an indicator of success. This experience is based onprovincial networking, mainly involving provincial and municipalcouncils. Previous LA21-related literature contributes empiricalevidence that networking leads to high levels of LA21 dissemina-tion in different European countries and regions. However, system-atic studies explaining why LA21 networks are successful have notbeen undertaken. As a consequence, it is unclear how to manageeffective LA21 networks. This research takes a step towards fillingthis gap. We build upon previous literature and identify twosecond-order factors: (1) support benefits; and (2) interaction

Fig. 2. Attitude towards LA21 as a function of interaction benefits at low (�1 SD) and high (+1 SD) level of support benefits. SD = Standard Deviation.

86 J.M. Barrutia et al. / Geoforum 50 (2013) 76–87

benefits, and analyze the effect of these factors on two outcomevariables related to the concept of engagement.

From an academic perspective this research is inspired by andtries to contribute to relational economic geography by designinga conceptual model, and operationalizing and testing it. We makeexplicit the benefits that emerge from the interactions and rela-tionships between actors within networks. Therefore, unlike mostprevious research supported in relational economic geography, ourresearch is not only relational in a thematic sense (Yeung, 2005).

Previous quantitative research has shown the importance ofsupport benefits for explaining favorable attitudes towards LA21.This research shows that interaction benefits are also importantfor explaining the attitudes of municipal managers towards thenetwork and network members, and towards the LA21 tool.

Another conclusion of this research is that support benefits andinteraction benefits are expected to have different effects on differ-ent outcomes. While interaction benefits have a greater effect onsocial capital, support benefits have a stronger effect on attitudesand intentions towards the LA21 tool. Therefore, network manag-ers who focus on the dissemination of a specific tool could priori-tize actions aimed at increasing support benefits. However,network managers who focus on creating social capital couldundertake actions aimed at augmenting networking benefits.

In determining the focus of their actions, whether that be socialcapital or the specific LA21 tool, network managers might considerthat attitudes towards the LA21 tool might be more calculative andless durable than social capital. Social capital might bring deeperand more enduring levels of involvement that could be useful forboosting other good practices in the provincial territory in thefuture.

When explaining attitude towards the LA21 tool, an interestinginteraction effect between support benefits and interaction bene-fits was found. The effect of interaction benefits on attitudes to-wards the LA21 tool is dependent on the level of support benefitsand vice versa. As shown by a negative coefficient of the interac-tion term, the effect of interaction benefits on attitudes towardsthe LA21 tool is greater when support benefits are low than whenthey are high (and vice versa). This finding seems to suggest thatsupport benefits and interaction benefits may be viewed as substi-tutes to some extent. Therefore, network managers might stresssupport benefits or interaction benefits according to their particu-lar resources and context.

We distinguish between interaction benefits and interactiondensity. This differentiation is important, because the participa-tion of municipal managers in the network might not be fully vol-

untary, but be affected by the rewards offered by the provincialcouncil and other higher levels of government. Municipal manag-ers, therefore, might have only a symbolic presence in networkactivities (i.e., participation without involvement). Our findingsshow that interaction density acts as a moderator to explain atti-tudes towards the LA21 tool. In particular, interaction densityreinforces the effect of additional levels of interaction benefitson satisfaction. Rationality is straightforward. When interactionembodies quality (and, as a consequence, the benefits of theinteraction are high), the interaction density will have a greatereffect on satisfaction.

The effect of municipal characteristics on the studied outcomesis relatively limited. Attitude towards citizen participation posi-tively affects both engagement-related outcome variables, and riskaversion and municipality size negatively affect attitudes towardsthe LA21 tool. A plausible interpretation for the relatively limitedeffect of municipal characteristics on engagement-related out-comes could be that we are studying a small geographical contextin which municipalities are relatively homogeneous. Differentfindings might be obtained in broader geographical contexts.

The negative effect of municipality size is particularly interest-ing, as prior literature has considered that LA21 adoption may beeasier for larger municipalities, which are equipped with more re-sources (Kern et al., 2007). Nevertheless, something that might nothave received sufficient attention in previous literature is thatLA21 might also be more difficult to implement in larger munici-palities. Appropriate involvement from civil society might be moredifficult to achieve in large municipalities, and strategies and ac-tions more complex to design and implement. Furthermore, LA21could prove to be more specific, useful and handy in a small munic-ipality. These considerations might negatively affect the percep-tions of the LA21 tool experienced by managers in largemunicipalities.

This research is the first systematic analysis of the joint effect ofsupport benefits and interaction benefits on networks addressed tobuild engagement in relation to good managerial practice. There-fore, more research is needed.

The findings of our study are conditioned by its context; theyare context-dependent. As an epistemological perspective for rela-tional economic geography, Bathelt and Glückler (2003) proposedcritical realism, and we adopt this view. Critical realism accountsfor context-specificity in human action and proposes that humanaction cannot be explained by universal laws. Future investigationsare required in broader geographical contexts in order to assess theimportance of the context on the results achieved.

J.M. Barrutia et al. / Geoforum 50 (2013) 76–87 87

The research has a number of limitations. A potential concernfor studies using a survey as the unique data-source is commonmethod variance. This study is cross sectional, covering only onepoint in time. Longitudinal studies might provide additional in-sights. For instance, we see identification and trust as a result ofrich interaction between network members. However, identifica-tion and trust may also be considered as a prerequisite for involv-ing municipal managers in interactions (Adger, 2003; Davis andReed, 2013; ). Longitudinal studies might offer interesting insightsin relation to the evolution of identification and trust perceptionsin time.

Our research focuses on intraregional government-to-govern-ment interaction. Future research could consider other scales(Gibbs and Jonas, 2000; Wilson, 2012). Regional and provincialLA21 networks should be connected with other state and interna-tional networks, as in the case of the Girona network, which main-tains ties with other Spanish regional networks and internationalorganizations, such as the ICLEI.

Acknowledgements

The authors would like to thank the Editor and the referees fortheir helpful comments and suggestions. Financial support fromthe Spanish Government (Grant number ECO 2010-20792), theBasque Government (Grant Number GIC12/57 – IT 601-13) andthe University of the Basque Country (Grant Numbers UFI11/51and GIU11/17) is gratefully acknowledged.

References

Adger, W.N., 2003. Social capital, collective action, and adaptation to climatechange. Economic Geography 79 (4), 387–404.

Ansell, C., Gash, A., 2008. Collaborative governance in theory and practice. Journal ofPublic Administration Research and Theory 18, 543–571.

Bakker, A.B., Schaufeli, W.B., Leiter, M.P., Taris, T.W., 2008. Work engagement: anemerging concept in occupational health psychology. Work & Stress: AnInternational Journal of Work, Health & Organisations 22 (3), 187–200.

Baron, R.M., Kenny, D.A., 1986. The moderator–mediator variable distinction insocial psychological research: conceptual, strategic, and statisticalconsiderations. Journal of Personality and Social Psychology 51, 1173–1182.

Barrutia, J.M., Aguado, I., Echebarria, C., 2007. Networking for Local Agenda 21implementation: Learning from experiences with Udaltalde and Udalsarea inthe Basque autonomous community. Geoforum 38 (1), 33–48.

Bathelt, H., 2009. Knowledge generation, economic action and relational economicgeography. In: Das Wissen der Städte. VS Verlag für Sozialwissenschaften, pp.47–58.

Bathelt, H., Glückler, J., 2003. Toward a relational economic geography. Journal ofEconomic Geography 3, 117–144.

Bell, S., Park, A., 2006. The problematic metagovernance of networks: water reformin New South Wales. Journal of Public Policy 26 (1), 63–83.

Boggs, J.S., Rantisi, N.M., 2003. The ‘relational turn’ in economic geography. Journalof Economic Geography 3 (2), 109–116.

Bryson, J.M., Alston, F.K. (Eds.), 2011. Creating Your Strategic Plan: A Workbook forPublic and Nonprofit Organizations. John Wiley & Sons, San Francisco, CA.

Bryson, J.M., Crosby, B.C., Stone, M.M., 2006. The design and implementation ofcross-sector collaborations: propositions form the literature. PublicAdministration Review 66, 44–55.

Burch, S., Robinson, J., 2007. A framework for explaining the links between capacityand action in response to global climate change. Climate Policy 7 (4), 304–316.

Chiu, C.M., Hsu, M.H., Wang, E.T.G., 2006. Understanding knowledge sharing invirtual communities: an integration of social capital and social cognitivetheories. Decision Support System 42, 1872–1888.

Coenen, F., 2009. Local Agenda 21: ‘meaningful and effective’ participation? In:Coenen, F. (Ed.), Public Participation and Better Environmental Decisions.Springer, Netherlands, pp. 165–182.

Davis, E. J., & Reed, M. G., (2013). Multi-level governance of British Columbia’smountain pine beetle crisis: The roles of memory and identity. Geoforum, 47,32-41.

Dholakia, U.M., Bagozzi, R.P., Pearo, L.K., 2004. A social influence model of consumerparticipation in network- and small-group-based virtual communities.International Journal of Research in Marketing 21, 241–263.

Dwyer, F.R., Schurr, P.H., Oh, S., 1987. Developing buyer–seller relationships. Journalof Marketing 51, 11–27.

Echebarria, C., Barrutia, J.M., Aguado, I., 2009. The ISC framework: modelling driversfor the degree of Local Agenda 21 implantation in Western Europe.Environment and Planning A 41, 980–995.

Eckerberg, K.A., Dahlgren, A., 2007. Project or process? Fifteen years experiencewith Local Agenda 21 in Sweden. Ekonomiaz 64, 130–149.

Eden, S., 2009. The work of environmental governance networks: traceability,credibility and certification by the Forest Stewardship Council. Geoforum 40,383–394.

Evans, B., Joas, M., Sundback, S., Theobald, K., 2006. Governing local sustainability.Journal of Environmental Planning and Management 49, 849–867.

Faulconbridge, J.R., Hall, S., 2009. Organisational geographies of power: introductionto special issue. Geoforum 40 (5), 785–789.

Gage, R.W., Mandell, M.P., 1990. Strategies for Managing Intergovernmental Policiesand Networks. Praeger, New York.

Gibbs, D., Jonas, A.E.G., 2000. Governance and regulation in local environmentalpolicy: the utility of a regime approach. Geoforum 31, 299–313.

Giddens, A., 1984. The Constitution of Society: Outline of the Theory ofStructuration. Polity Press, Cambridge.

Hogg, M.A., Abrams, D., 1988. Social Identifications: A Social Psychology ofIntergroup Relations and Group Processes. Routledge, London.

Jarillo, J.C., 1988. On strategic networks. Strategic Management Journal 9, 31–41.Jun, K.N., Weare, C., 2010. Institutional motivations in the adoption of innovations:

the case of e-government. Journal of Public Administration Research and Theory21 (3), 495–519.

Kern, K., Koll, C., Schophaus, M., 2007. The diffusion of Local Agenda 21 in Germany:comparing the German federal states. Environmental Politics 16, 604–624.

Lam, S.Y., Shankar, V., Erramilli, M.K., Murthy, B., 2004. Customer value, satisfaction,loyalty, and switching costs: an illustration from business-to-business servicecontext. Journal of the Academy of Marketing Science 32, 293–311.

Liebler, C., Ferri, M., 2004. NGO Networks: Building Capacity in a Changing World’Study Supported by Bureau for Democracy, Conflict and HumanitarianAssistance Office of Private and Voluntary Cooperation. Agency forInternational Development, Washington, United States.

Mayer, R.C., Davis, J.H., Schoorman, F.D., 1995. An integrative model oforganizational trust. Academy of Management Review 20, 709–734.

Meier, K.J., O’Toole, L.J., 2001. Managerial strategies and behavior in networks: amodel with evidence from U.S. public education. Journal of PublicAdministration and Theory 11, 271–293.

Nahapiet, J., Ghoshal, S., 1998. Social capital, intellectual capital, and theorganizational advantage. Academy of Management Review 23, 242–266.

O’Leary, R., Bingham, L., 2009. The Collaborative Public Manager. New Ideas for theTwenty-first Century. Georgetown University Press, Washington, DC.

Pattie, C., Johnston, R., 2013. Personal mobilisation, civic norms and politicalparticipation. Geoforum 45, 178–189.

Störmer, E., 2008. Greening as strategic development in industrial change – whycompanies participate in eco-networks. Geoforum 39, 32–47.

Taylor, B., de Loë, R.C., 2012. Conceptualizations of local knowledge in collaborativeenvironmental governance. Geoforum 43 (6), 1207–1217.

Whitehead, M., 2003. ‘In the shadow of hierarchy’: meta-governance, policy reformand urban regeneration in the West Midlands. Area 35 (1), 6–14.

Wilson, G.A., 2012. Community resilience, globalization, and transitional pathwaysof decision-making. Geoforum 43 (6), 1218–1231.

Yeung, H.W-c., 2005. Rethinking relational economic geography. Transactions of theInstitute of British Geographers 30, 37–51.