The Lifecycle of a Voluntary Policy Innovation : The Case of Local Agenda 21

Local Agenda 21 (LA21) emerged 23 years ago as a voluntary policy innovation for local governments aiming at sustainability and has now completed its lifecycle. We aim at a second look at LA21 from the standpoint of the institutional and innovation diffusion theories and with Portugal as case study. Results show a three moment lifecycle for LA21, each with distinct diffusion patterns. The Dawn, stymied by lack of regional and national leadership, was likely fuelled by a learning mechanism. It lasted 10 years and involved a mere 1% of the potential adopters. The Zenith took place when other countries had already come full circle. During this phase 27% of the local governments became active and both coercion and competition stand out as relevant engines. Twilight, most probably powered through coercion, competition and imitation mechanisms, took LA21 to a steady state with an additional 19% of local governments enrolling. Since then LA21 has shown departures in several different directions, including oblivion. We speculate, based on preliminary data that, although most LA21 are no longer active, a durable setting was created that promotes further innovation and public participation.


Introduction
Local Agenda 21 (LA21) is considered to be a voluntary policy innovation for local governments (LG) with a primary goal of activating the local level to create sustainable communities (Joas & Gronholm, 2004;Seyfang & Smith, 2007).
However, contrary to expectations and in stark contrast to most European LA21 'pioneer' and even 'follower' nations where LA21 was nearing a stationary phase (Gomila, 2000a;Joas & Gronholm, 2004;Lafferty, 2001), a late and unexpected blossoming occurred in Portugal.How this came to happen and what lessons it holds for future local policy innovation diffusion waves elsewhere has not been subject to analysis.
Although focused on a single country, this is the right time-two decades after Agenda 21 was first written-to reflect with the benefit of hindsight on the institutionalization and full lifecycle of this innovation, with a particular focus on determinants and patterns of diffusion (Bingham, 1977) and its role as a source of further novelties (Geissel, 2009).To analyze change we draw upon the institutional theory and the diffusion of innovation theory as main theoretical frameworks.

Local Agenda 21 as a Local Policy Innovation
In 1992 over 178 governments met for the United Nations Conference on Environment and Development, (UNCED) widely known as Rio Summit and adopted Agenda 21 (A21), a transnational program including both sustainable development and governance objectives (Joas & Gronholm, 2004).It includes a clear call for local authorities to adopt a Local Agenda 21, a policy innovation with the ultimate goal of improving local sustainable development prospects (Joas & Gronholm, 2004) through citizen involvement (Elgert & Krueger, 2012).The classic-but not standardized-sequence of the implementation of LA21 follows:  Step 1: Political commitment to LA21 (affecting human and financial resources, engaging stakeholders);  Step 2: Baseline review (elaboration of the sustainability report);  Step 3: Local action plan (definition of sustainability targets and actions to accomplish them);  Step 4: Implementation (taking forward the planned actions);  Step 5: Evaluation and reporting (regular assessment of the results) (Garzillo & Kuhn, 2007).

Diffusion of Innovation Theory
Rogers (1983,2004) defines diffusion as the process by which an innovation (an idea perceived as new) is communicated through certain channels among the members of a population of potential adopters over time and asserts that the adoption of an innovation follows a normal, bell-shaped curve when plotted over time on a frequency basis and a S-shaped curve when cumulative number of adopters is considered.The S-shaped distribution rises slowly at first and then accelerates to a maximum until half of the individuals in the system have adopted.It then increases at a gradually slower rate as the few remaining individuals finally adopt.Each fraction of this curve represents a different adopter category based on time of adoption, which can be classified as follows (Rogers, 1983(Rogers, , 2002)) The relative speed with which an innovation is adopted depends of the perceived attributes of the innovation as well as the way information about the innovation is communicated, the type of innovation decision (collective or authoritative), the characteristics of the social system (degree of interconnectedness and values) and the role of change agents (Rogers, 1983).
A variety of diffusion mechanisms for innovation diffusion have been identified.However, according to Shipan & Volden (2008) and Gilardi (2012), a general consensus has formed around these four core paths:  coercion-where outside pressure compels the adoption of a policy;  competition-where rivalry between organizations prompts one to try to stay ahead of the other;  learning-where knowledge is acquired mainly through peer networking;  imitation-where one organization's innovations are copied by another seeking to look better.
Innovation carries many advantages, and likely drawbacks as well.For the purpose of the current narrative innovation matters since it provides a strategy of legitimation for organizations that sense the evolution in overarching institutions against which their own performance is measured (Damanpour & Schneider, 2006;DiMaggio & Powell, 1983).

Institutional Theory
According to institutional theory (Scott, 1995), institutions (government, family, religion, etc.) are resilient normative systems that organize social, political and economic life and stand on three pillars: regulatory, cognitive, and normative.Change in any of these pillars exerts pressure on related organizations.It has been argued that the narrative of sustainable development was the engine behind one such institutional change.
Ever since the Brundtland Report was published in 1987 the concept of sustainable development has entrenched itself firmly as a research object, a political concept and a set of skills, tools and practices.It has in fact become institutionalized-new values, perceptions, social norms and routines have been generated-and organizations at all levels (central governments, local governments, corporations) adopted them in order to maximize legitimacy and maintain social backing (Jennings & Zandbergen, 1995;Llamas-Sanchez, Garcia-Morales, & Martin-Tapia, 2013;Sneddon & Norgaard, 2006).
Over the decades the values, norms and civic involvement around sustainability never ceased to grow (Llamas-Sanchez, Munoz-Fernandez, & Maraver-Tarifa, 2011), to the point where local decision makers' discourses started to incorporate appropriate wording (Schmidt, Nave, & Guerra, 2005).

Local Agenda 21: Adoption and Diffusion Behavior
With the A21 agreement local authorities were recognized as an arena through which sustainability should be pursued (Kusakabe, 2013;Selman, 2010), and within the first decade a significant effort was made to move LA21 into practice.In 1994 the bottom-up European Sustainable Cities and Towns Campaign, consisting mainly of LG, was launched and played a key role in producing the Aalborg Charter (guidelines for sustainable European cities).ICLEI (Local Governments for Sustainability) constituted another engine of LA diffusion worldwide, creating a vast knowledge base and investing in numerous dissemination initiatives (Elgert & Krueger, 2012).
Five years after the Rio Summit, LA21 activities were underway in more than 1800 LG in 64 countries worldwide (ICLEI-International Council of Local Environmental Initiatives, 1997).At this time, over 80% of the reported actions were taking place in 20 countries and a staggering 90% of initiatives happened in high-income nations.In 2002, LA21 had diffused to 6400 LG in 113 countries worldwide, a more than three-fold increase in less than five years.Over 80% of these LG were located in Europe (Rok & Kuhn, 2012) one of the world regions where LA21 best took root (Evans & Theobald, 2003;Smardon, 2008).However, the response from European local authorities to LA21 in Europe is far from homogeneous (Barrutia & Echebarria, 2011;Joas & Gronholm, 2004).Also, exactly how LA21 is implemented varies to a significant degree between countries and different political and administrative systems (Joas & Gronholm, 2004).
In some European countries the LA21 implementation started soon after Rio.These 'pioneer' nations include Sweden, the United Kingdom and the Netherlands (Gomila, 2000a;Lafferty, 2001), whose national governments or LG associations significantly pushed for the diffusion of LA21 processes (Gomila, 2000a;Joas & Gronholm, 2004;Jorby, 2002;Lafferty, 2001).All local authorities in Sweden had initiated LA21 process by early 1996 (Jorby, 2002) and by 1997 roughly 73% of UK local authorities were involved in a LA21 process (Selman, 2000).These countries were then followed by their Scandinavian counterparts, such as Denmark, Finland and Norway, who had showed very little activity prior to 1996 and are considered 'followers' (Gomila, 2000a;Lafferty, 2001).
There is still a notable lack of research on the diffusion of LA21, that being the case with other policy innovations at the local level as well (Shipan & Volden, 2008;Wolman & Page, 2002).It has however, albeit to a limited extent, been studied in Germany (Kern et al., 2007) and in Spain (Llamas-Sanchez et al., 2013;Llamas-Sanchez et al., 2011).
Two decades later, however, times have changed.Most notably, the results of Rio+20 expressed in the outcome document titled 'The Future We Want' (United Nations Conference on Sustainable Development, 2012) contain no clear reference to LA21.It has also been reported by several authors that 'LA21' as a label has been losing momentum, at least in the UK (Rowe, 2000;Selman, 2000), Sweden (Eckerberg & Dalgren, 2007cited in Echebarria et al., 2009;Keskitalo & Liljenfeldt, 2012) and Germany (Geissel, 2009;Kern et al., 2007).
Research has correspondingly dwindled.ICLEI reported on the frequency of the term 'LA21' as it appeared in English language literature between 1987 and 2008 (Rok & Kuhn, 2012) and concluded that, after the enormous rise in popularity in the 1990s, there was a sharp decrease in its usage in the wake of Rio+20.
That LA21 is not longer diffusing can be ascertained using ICLEI's data relative to adoption of LA21 in European countries in 1997 and 2002, as well as information about LA21 adoption in Germany for 2006 (Kern et al., 2007), and estimating Europe's data point for 2008 by adjusting for Germany's increase (considering that most other countries had no significant additions at this time (Gomila, 2000a;Joas & Gronholm, 2004;Lafferty, 2001)).With these numbers a graph for both Europe and Germany can be charted that fits-tentatively-the S-shaped curve of cumulative innovation diffusion (Rogers, 1983) where the final stationary phase represents the saturation point (Figure 1).What happens after this point varies according to the innovations' own sustainability (Rogers, 1983).LA21 has been discontinued due to replacement in, for instance, the UK, where Sustainable Community Strategies took over (Fidelis & Pires, 2009;Lucas, Ross, & Fuller, 2001;Rowe, 2000;Selman, 2000) and Sweden, where Climate Change Adaptation Plans were adopted (Nilsson, Swartling, & Eckerberg, 2012).Discontinuance may also occur through disenchantment, as was the case in Bergen, Norway (Harvold, 2003), among others.
Where innovations do take hold as such, two evolutionarily divergent paths can be taken.Routinization, or mainstreaming, happened in Ferrara and Fano, in Italy, and in Hanover, Germany (Evans et al., 2006), where LA21 became integrated into the local government's routine and somehow shed its separate self.
In the case of Botkyrka, Sweden, a redefinition, or restructuring, took place.Here the adoption led to the development of internal sustainability aims and work plans and a system of sustainability certification that aims at accrediting administrative units at the municipality (Keskitalo & Liljenfeldt, 2012).It is called the 'Botkyrka model' as it joins together LA21 and standard management and ISO-like quality control systems.
Restructuring can alter the innovation itself, leading to new trends.Both Atkinson (2009) and O'Riordan (2001), among others, have argued that LA21 has given rise to the Transition Movement, a widening network of citizen initiatives seeking to address life after peak oil while avoiding the worst of climate change (Haxeltine & Seyfang, 2009).It has also been suggested that LA21 has guided the way into the Covenant of Mayors and current public participation initiatives, such as participatory budgets (Kemp & Rotmans, 2009;Seyfang & Smith, 2007;Truffer & Coenen, 2012).

Methodology
Previous research on LA21 made it clear that self-assessment processes should be avoided (Rok & Kuhn, 2012).
One way to overcome such constraints is to use the indirect observation method, after the qualitative methods in social sciences (Bernard, 2000;Hewson, Yule, Laurent, & Vogel, 2003).It consists of studying the physical traces a specific behavior leaves behind (Bernard, 2000).One of the new and interesting research settings to conduct indirect observation is the Internet (Bordia, 1996;Hewson et al., 2003), an efficient environment to retrieve suitable and relevant data (Hewson et al., 2003).The Internet is also a low-cost and efficient information sharing technology, unavoidable in dealing with people and organizations (Joseph & Taplin, 2012), prompting public participation (J.Scott, 2006) and improving levels of citizen trust (Grimmelikhuijsen, 2012).Hence, here, the extent of LA21 adoption by Portuguese LG was assessed through the Internet, as the best mirror available for the citizen interaction entailed.
A first database of LA21 processes in Portugal was generated in 2006.For the purpose of this research it was fully revised twice (in 2011 and again in 2012).Data about each LA21 was initially sought by examining the LG' webpages looking for 'LA21' and 'local sustainability'.When no hits were found the Google search engine was used with 'LG name'+'Local Agenda 21'.
We analyzed and crosschecked all search outcomes.Results included local and national media outlets, sustainability baseline reports, action plan reports, official financial reports, local government webpages, consultant teams' webpages, seminar and conference proceedings and other academic publications.These sources provided information about LA21 processes, promoters, financing sources, year of adoption, step of implementation and population exposed.It should be highlighted that, for the purpose of this analysis, a LA21 process is any such self-designated initiative as well as any that, albeit not using the LA21 label, is comparable to it as long as its scope includes the three sustainable development pillars (social, environmental, economic).Excluded from this list were all environmental planning processes, even if they use a LA21 approach or label.Only processes led by LG or prompted and supported by them were taken into account.Some LA21, even though municipal in nature, were implemented only in specific areas (e.g.district).In such cases we considered only the number of inhabitants within the target area.
For Transition Initiatives, Covenant of Mayors, Eco-Management and Audit Scheme (EMAS) and ISO 14 001 Environmental management systems membership was retrieved from the respective official webpages.The number of participatory budgets being implemented was accessed in www.op-portugal.org,a webpage that covers such experiences in the country.

Results and discussion
The degree of LA21 adoption by Portuguese LG (also known as municipalities) ranges from 0.3% in 1997, with only one of the 308 municipalities with LA21, to 47% in 2011(Figure 2).Two other LA21 processes happened during this period but both took place outside municipalities and as such were not considered.In 2002, ten years after Rio, the adoption of LA21 in Portugal still had not reached 1% of LG.That year was, however, a turning point: several LA21 processes were started and a highpoint was attained in 2006.
A minor peak is perceived in 2009 and since then LA21 adoption slowed to a halt.The horizontal S-shaped curve over time described by Rogers (1983) is evident from the data, indicating the innovation saturation point was reached.
Chronologically the diffusion map of LA21 in Portuguese municipalities takes the shape presented in Figure 3.It becomes clear that several neighboring municipalities are adopting LA21 within the same year, revealing that diffusion is taking place, prompted by regional municipality networks.Our data reveal an additional clear trend: average municipality population size adopting LA21 decreases with time (Figure 4).This is consistent with diffusion pattern research on LA21 in other European countries such as Germany and Italy, which reports that pioneers tend to be middle-sized or large cities (Kern et al., 2007;Sancassiani, 2005).
Figure 4. Average number of inhabitants living in municipalities that adopted LA21 processes per year.The number of residents living in municipalities with the highest population (max) and the lowest population effective (min) that adopted LA21 is also represented.The horizontal line in the plot area is the average number of inhabitants per municipality in Portugal (34 272).
The Portuguese late adoption trend parallels other southern European nations such as Italy, Greece and Spain, which also show a significant number of processes only after 2000 (ARE-Federal Office for Spatial Development, 2005;Gomila, 2000b;Sancassiani, 2005).The main features of LA21 of each of these sections are described below.

3.1
The Dawn (1992Dawn ( -2002) ) Literature supports the idea that recent democracies-such as Portugal's-display lower levels of implementation of participatory processes towards sustainability (Schmidt et al., 2006).This translates into politician resistance to regular citizen interaction (Schmidt et al., 2006), with citizens unaware of public participation's true potential (Guerra & Schmidt, 2009).The Portuguese also display low levels of social and political trust (Newton, 2001).This could have hindered LA21 adoption, as Owen and Videras (2008) suggest.Voluntary cooperation may be less likely to occur and policy innovations, such as LA21, which require coordination among multiple stakeholders, do not thrive.
Additionally, Portugal lacks an environmental culture (Schmidt et al., 2006;Silva & Gonçalves, 2005).The first National Environmental Plan was published only in 1995.A parallel set of circumstances have been used to interpret Spain's poor LA21 early adoption record (Gomila, 2000a).Contrary to other European countries (e.g.Sweden), after Rio the Portuguese central government did not enforce, recommend or facilitate LA21 adoption (Aguado et al., 2007;Barrutia & Echebarria, 2011;Carter et al., 2000;Echebarria et al., 2009;Fidelis & Pires, 2009;Guerra & Schmidt, 2009;Schmidt et al., 2005).Likewise, the National Association of Portuguese Municipalities did not step up to the role of LA21 transfer agency (Fidelis & Pires, 2009), as opposed to what happened, for example, in England, with the Local Government Management Board (Gomila, 2000a).
Despite the limitations, three LA21 did start before 2003.The municipalities that individually overcame the 'risk perception' associated with LA21 (Barrutia & Echebarria, 2011) were Cascais (1997), Oeiras (2001) and Odivelas ( 2002), all of them located within the capital city region, showcasing strong international ties and over 100.000 inhabitants.We suggest learning was the most relevant diffusion mechanism for these innovators, the reason being that both Cascais and Oeiras have long-term LA21 processes which are still active at the moment of this writing.When learning is the main mechanism behind adoption, longer-term effects and better results can be expected (Joas & Gronholm, 2001;Shipan & Volden, 2008).
The weak adoption of LA21 during this period certainly hampered its rate of diffusion, since it is accepted that a wide early adoption of policy innovations and network size are two crucial factors (Barrutia & Echebarria, 2011).In Portugal critical mass was not reached during the Dawn period.
Networks and organizations such as the European Sustainable Cities and Towns Campaign, that create direct links between the local and the international levels, have been reported to be of particular importance in countries with no national support for local sustainability processes (Rok & Kuhn, 2012), because they allow LG to find an alternative framework for cooperation and exchange (Rok & Kuhn, 2012;Smardon, 2008), inspire action (Ramiery & Fiorentini, 1998cited in Gomila, 2000a) and, therefore, affect the degree of LA21 dissemination (Echebarria et al., 2009;Joas & Gronholm, 2001).In Portugal, however, none of the 3 LA21 adopting municipalities subscribed the Aalborg pledges during this period.
At the end, in 2002, only 5% of the Portuguese population was under a LA21 process.To adopt innovation, organizations should exchange information with the surrounding environment in order to improve knowledge, prepare members and facilitate assimilation into routines (Damanpour & Schneider, 2006) but, if organizational leaders are not themselves ready, they cannot prompt for adoption.The lack of awareness of LA21 by local politicians has been specifically noted as a serious obstacle to adoption in Portugal (Carter et al., 2000), while Schmidt, Nave and Guerra (2005) observed that a mere 10% of local decision makers fully understood the concept of a LA21.
This, together with a lack of financial resources (Schmidt et al., 2005(Schmidt et al., , 2006) ) and the fact that Portuguese local governance style is still quite old-fashioned (39% of mayors declare that decision making should be reserved for elected politicians and 31% believe that citizens are not ready to play a more active role in local governance (Schmidt et al., 2006)), dampened most activity and can successfully explain the low interest in LA21 adoption during the Dawn period.Zenith (2003Zenith ( -2006) ) Beginning 2003 Portuguese LA21 started bourgeoning: the annual average growth rate of LA21 adoption is 168% for the 2002-2006 period.At the end 85 LA21 processes were being implemented, involving 28% of municipalities and 31% of the population.

The
These figures suggest LG started seeing increased value and the availability of financial resources must have played a major role.Resources from European Union (EU) funded programs, such as the European Regional Development Fund (ERDF) and the European Social Fund-regionally administered-helped set up some LA21 initiatives.Our data reveal that 48% of all LA21 started in 2003-2006 were co-financed by EU programs, the ERDF being the funding source for 74% of those that did receive outside money.
The role of 'agenda transfer' institutions at the regional level (Kern et al., 2007) was finally taken up by regional municipality networks, which launched 63% of all LA21 processes during this time.Their main task was the application for and management of EU grants (Guerra & Schmidt, 2009) but they also provided frameworks, information, technical support and a network enabling environment.
To avoid being left behind, LG may have felt they had no choice but to join the LA21 bandwagon-a sure sign of competition (Barrutia & Echebarria, 2013).Consistently, Fidélis and Pires (2009) assert the 'agenda transfer' effect is the main engine behind LA21 during this period.
Universities and private companies also had a role in LA21 diffusion by providing expertise and training, together with field work.Contracts usually ran for 24 months and encompassed the first three phases of LA21.Of all LA21 born within this period, 95% were associated with an external team.
A perceived coercive mechanism may have been also at play during this period: some Portuguese municipalities expected that a working LA21would improve their chances of grant proposal approval under the then forthcoming National Strategic Reference Framework (2007-2013) that framed EU funding for Portugal (Assembleia Municipal de Pinhel, 2007).

The Twilight (2007-2011)
The final period is characterized by a deceleration in LA21 adoption.The average growth rate of LA21 adoption is now 11% yearly and the steady state is attained before half of the potentially adopting municipalities get involved.During this Twilight 59 new LA21 were created and co-financing reached its apex (59% of these new LA21 received external funds).For the first time EU funding is explicitly available for municipalities wishing to implement LA21.Half of all new LA21 are adopted within a regional framing and almost all (83%) involve external teams.The peak observed in 2009 (with 23 new LA21) is ultimately a consequence this financing, since 21 of those benefited from it.
As regards diffusion mechanisms we suggest that coercion, competition and imitation mechanisms must all have been at play.The first two as a result of the EU funding opportunities and the role of the 'agenda transfer' agencies, as explained previously, the latter because 'latecomers'-usually small municipalities as observed from our data (63% of all new LA21 encompass less than 20.000 inhabitants)-may be looking up to their bigger neighbors and trying to keep up (Shipan & Volden, 2008).This was illustrated by an official at a council meeting: 'our municipality should not defer any longer the adoption […] many municipalities have already done it by now' (Assembleia Municipal de Penacova, 2010).
By the end of the Twilight period the diffusion of LA21 seems to have run its course.That the end would come so fast may be a consequence of the government announcing structural changes targeting municipalities, such as mergers, mandate changes, budget cuts, reduced autonomy and forced internal reorganization.This major move was announced at the 2011general election.Instability ensued and the local political center of gravity shifted.Decreased autonomy, for instance, is considered a factor that hinders LA21adoption (Joas et al., 2007;Jorby, 2000Jorby, , 2002)).The economic crisis that hit Portugal at the time compounded the issue and could also help explain the loss of momentum in LA21 adoption.

Beyond the Twilight
What happened to all these LA21?Did they deliver real outcomes?Our research suggests only 32 of all 144 (22%) municipal LA21 are currently active, i.e., have real up-to-date Internet existence.In Great Britain, on the other hand, even though over 90% of LG saw LA21 as worthy, only about one third had firmly and durably embraced it (Tuxworth, 1996cited in Rowe, 2000).
The Internet shows that not all 82 LA21 launched during the Zenith period fare well after their 5 to 8 year existence.
One would expect that after this length of time all would have already started 'implementation' (Step 4) and 'evaluation and revision' (Step 5).However, in 2011, only 2% of those 82 processes reached Step 5 and only 11% got to Step 4. Data reveals low levels of action, with 52% jammed at the 'plan of action' stage (Step 3).

Beyond
Step 3 the inability to implement the plan takes its toll, partly as a consequence of the retreat of the external teams.One mayor states their LA21 'is not dead but moribund because the external team is no longer available' (Assembleia Municipal do Bombarral, 2010).Another mayor clarifies that 'LA21 is largely abandoned because there is lack of money although some activities are still being implemented' (Camara Municipal de Manteigas, 2012).It is widely known this shortage of funds was partially unleashed and extensively aggravated by the 2008 bank meltdown and the 2010 Euro Zone crisis whose consequences, at least in Portugal, still ripple through everyday life many years later.Being as it is a voluntary policy, LA21 must have come high on the list of much needed spending cuts.
Interestingly, the percentage of all Portuguese LA21 that manage to reach Step 4 and 5 (10%) does not differ between LA21 promoted by regional municipality networks and those initiated by individual municipalities.The 'agenda transfer' effect of regional associations of municipalities is clearly prompting diffusion but not the long term sustainability of the innovation, contrary to what other authors have reported (Barrutia et al., 2007;Joas & Gronholm, 2001).
Some LA21, on the other hand, are still clearly active, having undergone a re-invention process over time (Rogers, 1983(Rogers, , 2004) ) and developed deep roots in municipalities such as Cascais and Oeiras, for instance, who were two of the early adopters.
Could it be that the incentive to further LA21 was drained through some related political measure?One of the most important new international agreements during the period under consideration was the Convention on Access to Information, Public Participation in Decision-Making and Access to Justice in Environmental Matters, better known as the Aarhus Convention.It enshrines, among others, the government's duty to ensure transparency and public participation in environmental issues and, as such, does overlap somewhat with LA21.This instrument, however, was adopted in 1998 and entered into force in 2001, and it was after that time that most LA21 were started and implemented in Portugal, meaning a direct cause and effect relationship is not apparent.
We looked into EMAS, the European Eco-Management and Audit Scheme, but only a single Portuguese LG was certified through this system (Ponta Delgada), meaning LA21 did not spark this kind of interest.The same can be said of ISO 14 001, where a mere two LG had some of their departments certified (Esposende and Maia).It's likely these much more structured administrative approaches crossed too far from the participatory and holistic nature of most LA21 initiatives.
But other paths may have been spurred.Our data reveal that, of all Portuguese municipalities previously enrolled in a LA21 process, 25% later entered the Covenant of Mayors whereas only 14% of the non-LA21 adopters subscribed.As for municipalities with a participatory budget, 15% came with a LA21 background while only 8% were non-LA21 adopters.Also, two of the three official Transition Towns had LA21 programs.These numbers cautiously point to LA21 as an enabling settingfor some models of further policy innovation and grassroots initiatives.

Conclusions
Out of tune with most other European countries, LA21 started diffusing in Portugal five years after it was first approved.What took place afterwards survived governmental neglect and lack of a common vision, reaching a respectable 47% diffusion level.When no central leadership materialized some LG stepped in which, alone or within regional networks-and somewhat prodded through financing and expertise availability-found the necessary commitment to drive the process.That such results could be attained without organized governmental intervention is a tribute to the LA21 concept's intrinsic worth and places differing views (among others Barrutia & Echebarria, 2011;Coenen, 2001) into perspective.The importance of leadership cannot be underestimated, however, and our results show that existence of an active regional coalition does seem to increase the chances of any individual LG of that region engaging in the process.
Overall, diffusion mechanisms identified throughout the LA21 lifecycle seemed sufficient to explain the measured evolution.Each period was marked by a specific subset of those mechanisms, which helped imprint the resulting pattern.
Innovations come and go, and what lives on after newer tides have swept people's attention defines the true sustainability of each trend.Portugal did not reach a major integration and so the national LA21 sustainability level can be classified as weak, even though a facilitating effect for further innovation could be detected.
Maximizing sustainability ought to be high on any government's or agency's agenda when it comes to promoting political innovations such as LA21, and our case study signals some of the variables involved.
Funding is a case in point.In each successive innovation round, when external funds do become available, past performance could be evaluated as criterion for differential sharing.Overall sustainability, as measured by actual innovation integration, should shape such an analysis.
As for external teams, they would do well to refocus the mandate from mere executors to capacity builders.The ability to internalize appropriate skills within the target institution that remain long after the temporary teams have left could well become the litmus test of their true success.
Finally, diversity should be upheld in the overall innovation cycle.Some stakeholders will always feel more motivated than others when faced with each particular option, and it is the motivated ones that have the best chance of moving further and internalizing deeper.By adding carrots and sticks increased numbers may follow along, but the innovation's sustainability level may not improve.If leeway is allowed-within basic constraints-internal motivation may prove the best button to push where innovation diffusion is concerned.
As mentioned by Leach et al. (2012), innovation itself must go through innovation by opening itself to the contributions of local grassroots innovation actors.By providing a safe operating space to such stakeholders an alternative engine might be started that keeps each innovation's sustainability running at maximum speed.

Figure 2 .
Figure 2. Early adoption of LA21 in Portugal and cumulative number of processes

Figure 3 .
Figure 3. Diffusion of LA21 processes in Portugal (Continental Portugal, Madeira Archipelago and Azores Archipelago)