Participatory-deliberative processes and public policy agendas: lessons for policy and practice

Abstract Participatory and deliberative processes have proliferated over recent decades in public administration. These seek to increase the effectiveness and democratic quality of policy making by involving citizens in policy. However, these have mainly operated at local levels of governance, and democratic theorists and practitioners have developed an ambition to scale these up in order to democratize higher tiers of government. This paper draws policy lessons from research on a “multi-level” process that held a similar ambition. The Sustainable Communities Act sought to integrate the results of various locally organized citizen deliberations within the policy development processes of central UK government. In doing so, it aimed to democratize central government problem definition and agenda-setting processes. The paper distinguishes between achievements and failures explained by process design, and more fundamental obstacles to do with broader contextual factors. As such, it identifies lessons for the amelioration of design features, while recognizing constraints that are often beyond the agency of local practitioners. The findings offer practical insights for policy workers and democratic reformers seeking to institutionalize participatory and deliberative innovations.


Introduction
This paper draws lessons from a study of legislation that aimed to institutionalize such ambitions (Bua 2014(Bua , 2017. The Sustainable Communities Act (SCA) sought to integrate the results of various locally organized citizen deliberations within central government policy processes. As well as connecting governance tiers and instigating greater interaction among local communities, local governments and politicians, and policy makers in Westminster and Whitehall (where UK central government offices are based), the SCA sought to contribute to the agenda-setting processes of central government. Drawing on analysis of the first 5 years of the process (2007)(2008)(2009)(2010)(2011)(2012), this paper identifies learning for practitioners, in particular policy workers and entrepreneurs involved in institutional reform.
The paper is organized in five parts. First, we develop our analytical framework based on the distinction between process and contextual factors. Second, we describe our case study and explain its importance and originality vis-a-vis our objectives in this paper. Third, we outline findings that are explained by the process design and (proximate) institutional conditions, moving on in the third section to findings related to contextual factors, including higher-order institutional conditions. Finally, a fifth section discusses the relevance of these findings for practitioners, as well as upon the broader objective to democratize central policy making processes through PDPs. A methodological Appendix is included at the end.

Analyzing PDPs
Elstub and Escobar (forthcoming) define democratic innovations as processes or institutions developed to reimagine and deepen the role of citizens in governance processes by increasing opportunities for participation, deliberation and influence (see also Smith 2009). PDPs represent a large subset of initiatives within the field of democratic innovation. PDPs combine both direct and discursive forms of participation and thus encompass a wide range of processes from participatory budgeting, to mini-publics and some forms of online crowdsourcing.
Over recent decades, there has been a mushrooming of the use of PDPs in a wide variety of contexts (Smith 2009;Fung and Warren 2011;Escobar and Elstub, 2017). While ad hoc experimentation still characterizes these developments, we are now also witnessing a wave of institutionalization of PDPs. This is taking the field from a focus on isolated, one-off PDPs to a broader agenda that seeks to design and embed these processes as more stable features of the institutional system. Recent examples include legislation such as the Tuscany Law of Participation in Italy (2007), the Community Empowerment Act in Scotland (2015) and the Sustainable Communities Act (2007) in England and Wales which is the focus of this article.
With institutionalization comes increased and more continuous contact with broader political and social structures as well as higher order institutions within which PDPs are embedded (Parkinson 2012). As such, scholars looking to investigate the conditions shaping the development of democratic innovations have distinguished between factors related to process design, and interaction with context (Font et al. 2014: 6-8). In this paper, we will make use of, and slightly elaborate upon this distinction, to develop an analytical framework to understand our case, the SCA.
Our analysis differentiates between factors related to process design on the one hand, and political and socio-economic context, on the other hand. Design factors relate to the way in which specific participatory-deliberative processes are designed. Designers make a series of choices including who takes part and how participation is organized. This aspect of our study reflects institutionalist perspectives on democratic reform which, as Fung (2003) explains, argue "that the key to enhancing participation is to be found in better institutional designs: in rules and decision making processes that encourage actors to participate".
Contextual factors have to do with the context in which PDP's are embeddedthe institutions and structures of representative democracy and capitalist political economy. In the case of the UK, which we consider here, contextual factors include the majoritarian representative democratic, and heavily centralized, political system, underpinned by a liberal market economy (Hall and Soskice 2001;Copus et al. 2017;Faucher-King and Le Gales 2010). These factors produce constraints that vary in their degree of resolvability. On one end of a continuum we have contextual features such as representative democratic institutions and the relatively fluid opening and closing of opportunities associated with, for example, different electoral outcomes. These can be influenced by the agency of policy actors and practitioners operating at higher levels than those directly involved in PDPs. Thus, Fung's (2003) insight also applies at this level. On the other end, constraints are more structural in nature, requiring significant social transformation to be resolved, such as those placed upon political and democratic possibilities by capitalism (Dryzek 1996;Streeck 2013).
The analytical distinction between process design and contextual factors is useful for the study of PDPs because it offers a heuristic to examine the interplay between two domains that, albeit overlapping, have distinct practical implications for policy workers. This allows us to map barriers and enablers to the institutionalization of PDPs that connect different levels of policy work in complex governancea much needed line of enquiry in this field.

The SCA
The SCA is UK legislation resulting from campaigns by civil society organizations in favor of "democratic decentralization"that is, to devolve powers and competences in ways that increase political participation. The legislation, therefore, set up a process whereby local authorities were provided with an opportunity to directly influence central government policy processes on the condition that they (a) "try to reach agreement" with (b) a panel of citizens which were broadly representative of local communities. Importantly, at national level, the legislation also implemented a duty upon central government (represented by the Secretary of State for Communities and Local Government) to "try to reach agreement" on the implementation of proposals with a panel of representatives from the Local Government Association (LGA)an organization that represents Local Government in England and Wales.
A brief chronology of the process: (a) from 2008 to 2009, 100 local authorities in England and Wales developed proposals through various forms of citizen engagement; (b) a total of 301 proposals were submitted to the Local Government Association, which had legal authority to vet requests on judicial and technical viability grounds, a process which ended in December 2009; and (c) the LGA and central government tried to reach agreement on the proposals accepted by the LGAa process which was interrupted by a general election, and change in government in 2010, and ended with the publication of a decision rationale in December 2010 (DCLG 2010). This is the period covered by our research.
The SCA essentially sought to enable local communities to define problems, propose solutions, and communicate these to central government. It represents an important case of innovation in the design of PDPs for three main reasons, (a) it focused on agenda setting, (b) it operated across governance levels, and (c) it attempted to institutionalize a link between process outcomes and binding policy decisions. Each of these features is explained in Box 1.

It's an agenda setting process:
Policy-making is traditionally understood as a "problem-solving" enterprisewhereby policy issues or problems are well-defined external inputs to, and stimulants of, the policy process (Hoppe 2010). PDPs have mainly conformed to this model, where the "choice of subject" (Fung 2003: 343) is to be made by process designers who, effectively, control the agenda. But this glosses over the power involved in the ability to define problems and influence agendas (Bua 2012). This is a facet of power which political science has been pre-occupied with at least since Bachrach and Baratz's (1962) seminal essay. By providing a space for citizens to identify and define problems, and channel these into central government, the SCA attempts to democratize agenda setting, something that scholars of democracy and democratic innovation have often called for (Parkinson 2006).

It's a multi-level process:
Participatory and deliberative innovations have mostly operated at local levels of governance, and it has been argued that "to be more fully legitimate, deliberative democracy needs to be much bigger than local consultations on local issues. It needs to be integrated into central government as well" (Parkinson 2006: 177). This presents well-known problems of scalabilityas the jurisdiction covered increases in size it becomes harder to ensure equal participation for all, and forms of representation become necessary (Dahl 1989). This is evident in perhaps the most paradigmatic of PDPsthe participatory budget, where direct participation in neighborhood forums is complemented by forms of representation at higher tiers when the process interacts more directly with policy making at City level (Smith 2009). By abstracting from results of local participation and connecting these to central government, the SCA represents a more ambitious attempt to link scalesan endeavor which has been deemed necessary to improve governance (Elstub 2014).
3. It has a statutory link to decision-making: PDPs tend to be exercises in public consultation. Despite their relative sophistication there is rarely a defined, or binding, link to authoritative decision, especially at higher tiers of government. Their main role is to increase the "input legitimacy" of policy processes. However, the non-binding nature of participatory exercises has long led to citizen skepticism about participation in these (Lowndes et al. 2001). In the design of the SCA, the statutory duty to "try to reach agreement" represents an acceptance that participation should go beyond consultation. In Smith's (2009) terms, this aims to increase "popular control" of the process and ensure a degree of responsiveness that is usually lacking in consultative exerciseswhere there is a pervasive, and legitimate, concern that decisions have already been made and, therefore, the role of citizens is tokenistic.
These three innovative aspects and ambitions of the process make it an important one to analyze in order to contribute to understanding to what extent, and how, PDPs can contribute to democratizing governance.

Findings: process design
At the level of process design there are three main findings: a. Permissive regulation and oversight over local citizens panels and subsequent proposal selection processes undermined the procedural legitimacy of the SCA.
b. Proposals were poorly integrated within central government policy making, and were sidelined in central government. This limited the impact of the SCA on policy agendas and facilitated its subsumption within the processes it intended to reform. c. Despite these limitations, our qualitative evidence suggest that the duty "to try to reach agreement" led to a deeper consideration of certain proposals than would have been the case in its absence. This is especially so regarding proposals that made contributions around the margins of broader policy direction in areas such as decentralization and public services.
Regarding the first finding, it is notable that local citizens panels lay at the core of the democratic purpose of the SCAin the spirit of the legislation it was through participation in local panels that proposals should have been developed, and the requirement to reach agreement in these implies deliberative dynamics of discussion and decision in forming policy proposals. Deliberative quality depends on the level of inclusion, diversity, scrutiny, respect, and reciprocity accomplished in communication amongst citizens, and with experts and authorities (Roberts and Escobar 2015: 89-91). High deliberative standards seek to ensure that decisions are made on the basis of the best evidence and reasons available, as determined through engaged dialogue and scrutiny across perspectives (Escobar 2011). However, more often than not, the reality of implementation of the SCA fell very short of such deliberative standards.
Comparative analysis of our data resulted in the identification of four categories that differ in the forms of engagement across three stages of the deliberative processraising ideas, developing proposals, and deciding which proposals to make. Accordingly, we see four functional uses of SCA in practice 1 : a. Council lobbying tool -characterized by the absence of, or minimal, citizen participation or deliberation in raising ideas, developing, and deciding upon proposals. For example, Kent County Council used the process to promote a long-standing campaign for investment in infrastructure, and submitted the proposal through the process having surveyed its official sounding board (a sample of 1200 citizens that is statistically representative of the local population) on the appropriateness of the SCA as a channel. b. Expert-elite problem solvingcharacterized by the development of ideas and proposals to be decided upon by local politicians, council officers and council-linked stakeholder networks (mainly the "Local Strategic Partnerships" of the time). For example, West Dorset County Council raised ideas by approaching council-linked groups as well as through a local press release. A group of officers then filtered these for viability and a shortlist was put to a subgroup of the Local Strategic Partnership, who selected the proposals to put through the process. c. Sophisticated consultationideas were raised through non-deliberative forms of engagement (e.g. a survey) and developed by council officers and politicians. Citizen deliberation was used to decide which proposals to make, but not to develop the content of these. For example, Bristol City Council raised ideas through a variety of sources, including public meetings, an SCA website, and by approaching council-linked groups. These were filtered by the Council's legal team, who put a shortlist of proposals to an independently constituted and facilitated panel of 30 citizens, selected to be broadly representative of the local population. These met on two separate days to prioritize ideas. The Council's Cabinet convened to discuss the panel's recommendations and made the final decision on which to put through. d. Analytic-deliberative processthis use of SCA entailed close collaboration among local officials, experts, and citizen panels in raising ideas, developing proposals and deciding which proposals to make. For example, Sheffield City Council raised ideas through a relatively large, widely advertised and SCA-specific public event.
A subgroup of participants were recruited to discuss and develop the ideas raised in a series of meetings, with deliberations chaired by the Council leader and supported by policy officers. The decision on which proposals to submit was made as a group.
Fifteen out of the 22 cases analyzed in this research fell within the first two categories. This meant that participation was mostly limited to local elites, which represents a gross diversion from the original spirit of the Act. Many proposals were developed without, or with minimal, public participation and deliberation. Many of the reasons for this relate to contextual factors and are to be explained in the next section. However, regarding process design, this problem was compounded by the by the way in which the LGA panel vetted the proposals it received. As documented in greater detail elsewhere (Bua 2017), the criteria applied was soft on "consultation requirements" meaning that many inappropriately developed proposals made it through the processa clear sign of lack of deliberative scrutiny. This led to a degree of antipathy towards the process by policy makers who objected that proposals had been developed through questionable forms of participation and thus lacked legitimacy. Many also suspected that the SCA had been instrumentalized by local elites to "lobby" for changes in government policy. They argued that the process had failed in sifting these illegitimate proposals out, which is precisely what deliberative engagement should excel at.
Central government policy makers felt legitimized in their resistance to the process. Low oversight over the participatory regulations of the process is problematic because the strength of the claim that PDPs can make to influence policy is based on the democratic legitimacy generated by public deliberation (Bohman 1996). This legitimacy acts as an important political resource for actors that seek to advocate impact, a class of actors that Fung (2005) pithily calls "deliberative activists". This was deficient in the SCAdepriving it of both a source of legitimacy and "communicative power" (Dryzek 1996;Habermas 1996). As we will draw out in the discussion, close attention to the participatory-discursive processes through which citizens engage is a critical factor in PDPs.
The second design aspect to consider was the way in which the results of participation were processed and fed into central government policy makingwhat Dryzek (2010) refers to as "transmission mechanisms". The beginning of this was the LGA selection process. While, as explained above, there was low priority given to "consultation requirements", a "viability" criterion, which essentially judged the political feasibility of proposals, was prioritized. Interviewees complained that this resulted in the elimination of "politically difficult" proposals (interviews 40,42,Local Works Campaigners). This led to another major finding of the studythat the transmission mechanisms operated to shape the outcomes of the SCA in ways that cohered with existing political agendas, and especially government policy. As such, the space for the SCA to open the way for new policy agendas was considerably constrained. This finding is confirmed by the content analysis of our data. As detailed in Bua (2017), when processing requests, central government actors also tended to favor proposals that cohered with policies already in development. Outright rejections were often made on the grounds that policies clearly went against the grain of government prioritiessuch as a proposal for the state to offer banking facilities through the (then publicly owned) post office network, which was rejected on the grounds of the need to reduce the public debt and deficit. However, in many other cases, policies which were presented as "accepted" in the response document (DCLG 2010) were, upon closer inspection, in fact already in development or about to be implemented. For example, a raft of measures proposing to allow local communities greater involvement in planning were to be implemented under the Localism Act (2011)a central piece of legislation at the time.
There are important contextual factors to be taken up in the following section that explain coherence with existing agendas. However, process and proximate institutional design faults partially explain this. First, there seems to have been a lack of incentives for policy makers to give more salience to proposals. This is partially explained by a low involvement of "institutional referees" (Parkinson 2012: 157-578) or actors, organizations and institutions that are external to the process, but ensure scrutiny and accountabilitysuch as the media or judicial institutions. Civil society advocates of the SCA attempted to carry out this role, but suffered from lack of capacity. The transmission of proposals became a technical and de-politicized process, carried out by elites with little external scrutiny. Second, the legislation presumed that central government would speak with one voice, when it was in fact represented by a relatively weak department that had no authority to make decisions outside its purview. Interaction with the siloed structure of central government has been found to be a wider problem in PDPs (Cooper and Smith 2012). This is also true of the SCAas it became largely isolated within DCLG and proposal integration was subsumed by inter-departmental politics.
Despite these limitations, there is one notable achievement that is down to proximate institutional design. Qualitative data suggest that the duty for central government and the LGA panel to "try to reach agreement" on which proposals to implement led to a deeper consideration of proposals than would have been the case in its absence. There was space to catalyze new policy development in cases of proposals that did not challenge the fundamentals of government policy at the time, but suggested a new policy within these parameters or communicated a problem which government was unaware of. These constituted 5.8% of positive responses in total (Bua 2017). One group within these made demands that were small in scale and required no legislative change, and were directly implementedsuch as the commitment to extend funding for a local social enterprise that cohered with the ambitions of a then flagship policy called the "Big Society". In the case of larger scale proposals, the response was to initiate an advisory process such as piloting implementation (as in the case of a proposal to grant local authorities powers to run post office networks) and initiating a formal consultation (as in the case of a proposal on the prohibition of restrictive covenants on community amenities).
In these instances, the SCA enabled placing issues and policy proposals on the public agenda. This is an important achievement which the research suggests is directly linked to the clause establishing a duty to "try to reach agreement" within the legislation. Where proposals were deemed too large in scale, this led central government policy makers to look for alternative solutions, rather than provide a formulaic response. One respondent put this in terms of picking "apart the answer to the proposal … encourage the government to ask … how do we work with the council and local people to address this issue?" (interview 37, LGA Panel Support Officer), another recalled the government's openness to offer support and guidance and explore alternatives (interview 44, Selector Panel Member) and another recalled that this was "pretty unique to the SCA" (interview 19, DCLG Civil Servant).
This deeper consideration given to proposals is an important achievement, indicating a form of democratic responsiveness which is realized not when inputs trigger outputs, but when they catalyze genuine deliberation amongst authorized policy makers (Chambers 2012). In practice, this means that the SCA had capacity to foster deliberative quality at systemic level by creating an incentive for considered engagement in agenda-setting across levels of governance.

Findings: contextual factors
The context within which the SCA process was implemented had considerable influence on the performance of the process. There are two headline findings: a. First, the capturing of many proposal development processes by local state elites (discussed above) is an example of the SCA becoming entangled in the heavily centralized dynamics of British politics and democracy. b. Second the SCA was to a large degree subsumed by developments in the policy making and electoral cycle.
First, we described above how in 15 out of our 22 cases demonstrated minimal levels of participation, clearly contravening the spirit of the process. There are important contextual reasons for this which are primarily related to the heavily centralized nature of politics in the UK, and the peripheral position of local government. One of the results of this is that local authorities and politics become a scenario for playing out of national politics (Copus et al. 2017). This influenced the instrumentalization of the process as a "council lobbying tool", something that was strongly suspected by national politicians and policy makers we interviewed. Moreover, a great part of the process designers interviewed were skeptical that central government would engage with proposals in anything beyond a tokenistic way, generating a clear incentive to develop speculative proposals and not invest too much resource in engagement. Indeed, during the heavily centralist (Faucher-King and Le Gales 2010) New Labour years (1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010), a great degree of institutional experimentation around local participatory governance was encouraged, and imposed, by central government. Many council officers across thus reported seeing in the SCA a good opportunity to propose to government ideas which had arisen through these, but which the council had no power to act upon. In this sense, the SCA could be said to have acted as a conduit of information between a powerful and distant central government, and local authorities that had been engaging in much public participation but were too disempowered to act on many of the issues arising from these. The important analytical point is that the SCA was shaped by the central-local relations created by the centralism of its host state and polity.
Second, the general election in 2010 and subsequent change in government significantly delayed the SCA process and interrupted its continuity. Electoral deadlock in a 2010 general election was resolved through a coalition between the Conservative Party and the country's third party, the Liberal Democratswhich would form part of government for the first time since its founding in 1988. A long period of uncertainty followed significantly delaying the SCA. Whitehall civil servants had prepared for this by anticipating possible outcomes, and making links between electoral manifestoes and SCA policy proposals. These related to a wide variety of policy areas, across multiple departments. A key role in the process was carried out by a team of civil servants in DCLG who were tasked with coordinating the government response. Their approach was to search for developments in the policy process that offered opportunities for promoting SCA proposals.
Coalition government presented an opportunity for the SCA because both coalition parties had supported the legislation, the spirit of which cohered with major policy initiatives at the time such as the Conservative's "Big Society" and the Liberal Democrat's "community politics", which focused on decentralizing institutional reform. However, it also represented a constraint, because policy was relatively unspecified particularly in those areas unrelated to fiscal consolidation and economic growth. For example, the development of public sector austerity policies led to the rejection of proposals that called for an extension of state intervention, such as one to create a public banking system through the British Post Office, proposed by four local authorities.
The SCA thus encountered significant contextual constraints. Principal amongst these was a policy development "bottle-neck", leading to a significant delay in the process. Proposals calling for new legislation were faced with the problem of the limited legislative processing capacity of representative institutions. Respondents highlighted the importance of policy and law "in the pipeline" (interview 19, DCLG Officer) or "legislative vehicles" (interview, LGA Officer) which could "piggy-back" (interview 61, Local Process Manager) proposals through legislation. The SCA had thus become to a great degree caught up in the legislative and policy cycles of representative institutions, illustrating how the continued dominance of representative democracy can displace PDPs like the SCA.
Finally, tensions between the SCA and the higher-order representative institutions constrained the extent to which the SCA influenced policy development. Such barriers had to do with higher order political institutions, and mirrored many other analyses of PDPs which conclude that there is a fine line between seeking to compliment representative institutions and being subsumed by these (Barnes et al. 2007;Baiocchi and Ganuza 2017), hence remarks that policy proposals had "vaporized" (interview 55, Process Participant, Sheffield) or disappeared into a "black hole" (interview 61, Process Manager, South Somerset). The problems the SCA faced in feeding various policy proposals into the variegated organizational make up of central government are a direct result of these contextual influences to do with higher order representative institutions. In fact, scholars of democratic innovation have highlighted that PDPs are sites of competition between actors in representative institutions (Goldfrank 2007;Goldfrank and Schneider 2007), the influence of these upon PDPs should play an important part of institutional analysis of PDPs.
Paying heed to this consideration, it is important to take into account the impact that practioners and policy agents can have at different institutional levelsfrom process managers making immediate design decisions, to government officials and politicians dealing with the inputs generated by PDPs.
Discussion and conclusions: the prospects for democratizing agenda setting through PDP's Before explaining our conclusions, we want to emphasize that this research was not designed to be generalizable in a conventional sense (see Methodological Appendix). The point of a focused case study is (1) to offer contextualized findings that speak to policy workers (scholars and practitioners) operating in policy domains resonant with the case under study; and (2) to inform further research, both in-depth and comparative. In doing so, this paper has sought to add to a limited, but growing pool of case studies that examines the institutionalization of PDPs (e.g. Lewanski 2013;Ravazzi 2016;Avritzer 2017). Our focus on the SCA offers the advantages and limitations of a single-n case (see Appendix), and contributes to the groundwork necessary to enable a programme of comparative research. A potentially promising line of enquiry may be to compare the SCA to other legislative attempts at democratic reform towards citizen-led governance (e.g. Tuscany Law of Participation; Scottish Community Empowerment Act).
Having said this, what lessons can be drawn from the findings above that can be of use to practitioners?
First, our analysis points to some important limitations, which primarily revolve around the tendency for process outputs to be subsumed by representative institutions. This echoes previous studies which have shown the clash of political cultures at the interface between existing bureaucratic and representative institutions and democratic innovations that foreground participation and deliberation (Escobar 2017). It is noteworthy here that design factors can help attenuate these problems. Research on the paradigmatic case of Participatory Budgeting (PB) in Porto Alegre (Baiocchi and Ganuza 2017) puts much of the success of the process down to broader institutional reforms which made PB the only mechanism for influencing the City Budget. This limited competition from other inputs and made it harder for policy makers to deflect demands.
Our second point concerns procedural legitimacy. Permissive regulations and low levels of oversight of the proposal generating process led to important deficits here which significantly undermined the legitimacy and strength of the claim that the SCA could make to influence government agendas. Designers and managers of PDPs should be especially attentive to the quality of public participation and deliberation processesan important source of "communicative power" (Dryzek 1996;Habermas 1996), which can underpin the work of "deliberative activists" (Fung 2005) and thus generate scrutiny and maximize impact. In particular, the quality of participation hinges on high standards along three dimensions of any PDP process, namely high level of diversity and inclusion, high deliberative standards of communication, and clear connection to relevant decision making arenas (Fung 2006: 66).
Third, and (building on the previous point) the SCA's "transmission mechanisms" (Dryzek 2010) into policy making were rather poor and the process became entangled in political and bureaucratic cycles. Intelligent design can mitigate these issues. As Smith (2009: 49-53) explains about PB, much of its success can be put down to the its accomplishment in mobilizing significant constituencies around proposals, generating a valuable source of external scrutiny, or "countervailing power" (Fung and Wright 2003), that kept co-option tendencies in check. Alongside appropriate "institutional referees" (Parkinson 2012), these were largely missing from the SCAthe proposal integration process became a non-political and technical endeavor carried out between bureaucratic elites with little external scrutiny.
Contextual factors at institutional level are important here. The effect of the May 2010 election was to significantly block the process, and made it depend heavily on the agency of civil servants. Once again, this points towards the need to investigate the interplay between institutional actors (i.e. process designers, policy workers, and politicians) across different levels of governance, which remains understudied in the field of democratic innovation. It is indeed inevitable that constraints will be imposed by higher order institutions and agendas. However, one key feature of intelligent and efficient design is that of compatibility with encompassing systems (Goodin 1996). In this respect, we point to one process design featurethe duty to try to reach agreementwhich was modestly successful in generating policy influence. Importantly, this delivered the intended effect of increasing democratic responsiveness. Crucial to this was the openness of government officials to look for alternative solutions, and engage in more depth with proposals than would have been the case with a consultative process.
The case of the SCA illustrates that institutional democratic innovations offer a way of opening up the black box of decision making where the public policy agenda is set. However, the case also illustrates the challenge faced by institutional entrepreneurs and civic innovators. There are a range of design and contextual factors that must be taken into account when carving up space for new forms of democratic governance. Our main conclusion is that the SCA was shaped by the political dynamics of the processes that it sought to reform, revealing certain tensions between the SCA and higher order institutions. This confirms the fine line found in the broader literature on democratic innovation between seeking to compliment representative institutions and being absorbed by them. Crucially, democratic innovations must strive to accommodate the growing aspirations of engaged citizenship alongside the slowchanging culture culture of public administration in representative democracies. Cases like the SCA illuminate the challenge of institutionalizing participatory and deliberative democracy in the reformist project of upgrading our representative institutions. Notes 1. See Appendix item 1 for a description of the approach to typology development and comparison. 2. For a full account of the comparative rationale and research design see Bua (2014: 72-82;104-111). 3. Six cases were reclassified and the final names for categories developed once a more precise picture of engagement had been developed through interview data, but the four types roughly correspond as such -"council lobbying tool" (low openness, low collaboration), "sophisticated consultation" (high openness, low collaboration), "expertelite problem solving" (low openness, high collaboration), and "analytic-deliberative process" (high openness, high collaboration).
As noted in the introduction to this paper, the institutionalization of PDPs is a recent phenomenon that is only now beginning to draw attention in the field of democratic innovation. As such, the field needs the development of a pool of in-depth case studies as a precursor to comparative approaches. We are mindful here of Kuhn's warning that a discipline without "a large number of thoroughly executed case studies is a discipline without systematic production of exemplars, and that a discipline without exemplars is an ineffective one" (in Flyvbjerg 2004: 432). A key strength of studying exemplars and building case studies is the potential for relatability and transferability: the future understanding and discussion of cases elsewhere in the light of knowledge drawn here. Our case study does not, therefore, seek generalizability/external validity but rather relatability/transferabilitywhich are the appropriate criteria to assess the quality of research in interpretive approaches to policy analysis (Yanow and Schwartz-Shea 2012;Fischer et al. 2015).
The research is mainly based on a mixture of quantitative data, drawn from a "qualitative content analysis" (Mayring 2000) of the government response document (DCLG 2010) and qualitative data drawn from a total of 75 interviews with actors involved in different stages and parts of the process. These ranged from professional advocates of the SCA, citizen panel participants, and local and national politicians and civil servants.
The content analysis followed established procedures (Mayring 2000) and consistency checks were well within the threshold for reliability, considered by Vieira and Garrett (2005) to be 0.75. Qualitative data were gathered through a combination of snowball and theoretical sampling, allowing for an iterative relationship between analysis and sampling (Ragin 2004), and carried out using a code and retrieve approach to thematic analysis (Ritchie and Lewis 2003). Quantitative and qualitative data play different roles in the analysis. The content analysis preceded qualitative interviewing. It was used to develop a rough measure of the policy impact of the SCA, and elicit themes and questions to explore through interviews. The main analytical work depends on qualitative data, which explored the logics and motivations of different actors and sought to elicit the main mechanisms at play throughout the SCA process.
The qualitative analysis was carried out at two levels. First, interviews with politicians and policy actors taking part in the policy integration process at the level of national state policy making were carried out, using an iterative sampling and analysis strategy based on snowballing and theoretical comparison (Charmaz 2006). Research at the second level also made use of this sampling strategy, plus a comparative component, which it is worth explaining. A sample of 22 cases was selected from a population of 63 cases for which data was available. A comparative analysis of process documents which described engagement processes was carried out to categorize local authorities into a typology of engagement. 2 This was done by separating the engagement process into four stages (idea raising; developing and refining proposals; prioritizing proposals; deciding proposals) and measuring the degree to which the process was "open" (high/low) to influence by citizens, and the level of collaboration (high/low) between actors operating across each stage. This produced a 4 type taxonomy which, following an approach which Ragin (2004) describes as "casing" was refined iteratively in conversation new data (mainly from interviews) to result in the development of the four types used here. 3